Mercurial > emacs
comparison src/regex.c @ 18262:e5e99de79a88
Fix up whitespace.
author | Richard M. Stallman <rms@gnu.org> |
---|---|
date | Sun, 15 Jun 1997 19:05:59 +0000 |
parents | a642c99198ec |
children | 5e9d099a4751 |
comparison
equal
deleted
inserted
replaced
18261:2f0b00246056 | 18262:e5e99de79a88 |
---|---|
9 the Free Software Foundation; either version 2, or (at your option) | 9 the Free Software Foundation; either version 2, or (at your option) |
10 any later version. | 10 any later version. |
11 | 11 |
12 This program is distributed in the hope that it will be useful, | 12 This program is distributed in the hope that it will be useful, |
13 but WITHOUT ANY WARRANTY; without even the implied warranty of | 13 but WITHOUT ANY WARRANTY; without even the implied warranty of |
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | 14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
15 GNU General Public License for more details. | 15 GNU General Public License for more details. |
16 | 16 |
17 You should have received a copy of the GNU General Public License | 17 You should have received a copy of the GNU General Public License |
18 along with this program; if not, write to the Free Software | 18 along with this program; if not, write to the Free Software |
19 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, | 19 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, |
20 USA. */ | 20 USA. */ |
21 | 21 |
22 /* AIX requires this to be the first thing in the file. */ | 22 /* AIX requires this to be the first thing in the file. */ |
23 #if defined (_AIX) && !defined (REGEX_MALLOC) | 23 #if defined (_AIX) && !defined (REGEX_MALLOC) |
24 #pragma alloca | 24 #pragma alloca |
25 #endif | 25 #endif |
26 | 26 |
27 #undef _GNU_SOURCE | 27 #undef _GNU_SOURCE |
28 #define _GNU_SOURCE | 28 #define _GNU_SOURCE |
29 | 29 |
30 /* Converts the pointer to the char to BEG-based offset from the start. */ | 30 /* Converts the pointer to the char to BEG-based offset from the start. */ |
31 #define PTR_TO_OFFSET(d) \ | 31 #define PTR_TO_OFFSET(d) \ |
32 POS_AS_IN_BUFFER (MATCHING_IN_FIRST_STRING \ | 32 POS_AS_IN_BUFFER (MATCHING_IN_FIRST_STRING \ |
33 ? (d) - string1 : (d) - (string2 - size1)) | 33 ? (d) - string1 : (d) - (string2 - size1)) |
34 #define POS_AS_IN_BUFFER(p) ((p) + 1) | 34 #define POS_AS_IN_BUFFER(p) ((p) + 1) |
35 | 35 |
36 #ifdef HAVE_CONFIG_H | 36 #ifdef HAVE_CONFIG_H |
37 #include <config.h> | 37 #include <config.h> |
38 #endif | 38 #endif |
39 | 39 |
40 /* We need this for `regex.h', and perhaps for the Emacs include files. */ | 40 /* We need this for `regex.h', and perhaps for the Emacs include files. */ |
41 #include <sys/types.h> | 41 #include <sys/types.h> |
42 | 42 |
43 /* This is for other GNU distributions with internationalized messages. */ | 43 /* This is for other GNU distributions with internationalized messages. */ |
44 #if HAVE_LIBINTL_H || defined (_LIBC) | 44 #if HAVE_LIBINTL_H || defined (_LIBC) |
45 # include <libintl.h> | 45 # include <libintl.h> |
46 #else | 46 #else |
47 # define gettext(msgid) (msgid) | 47 # define gettext(msgid) (msgid) |
48 #endif | 48 #endif |
83 char *malloc (); | 83 char *malloc (); |
84 char *realloc (); | 84 char *realloc (); |
85 #endif | 85 #endif |
86 | 86 |
87 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow. | 87 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow. |
88 If nothing else has been done, use the method below. */ | 88 If nothing else has been done, use the method below. */ |
89 #ifdef INHIBIT_STRING_HEADER | 89 #ifdef INHIBIT_STRING_HEADER |
90 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY)) | 90 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY)) |
91 #if !defined (bzero) && !defined (bcopy) | 91 #if !defined (bzero) && !defined (bcopy) |
92 #undef INHIBIT_STRING_HEADER | 92 #undef INHIBIT_STRING_HEADER |
93 #endif | 93 #endif |
192 /* Jim Meyering writes: | 192 /* Jim Meyering writes: |
193 | 193 |
194 "... Some ctype macros are valid only for character codes that | 194 "... Some ctype macros are valid only for character codes that |
195 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when | 195 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when |
196 using /bin/cc or gcc but without giving an ansi option). So, all | 196 using /bin/cc or gcc but without giving an ansi option). So, all |
197 ctype uses should be through macros like ISPRINT... If | 197 ctype uses should be through macros like ISPRINT... If |
198 STDC_HEADERS is defined, then autoconf has verified that the ctype | 198 STDC_HEADERS is defined, then autoconf has verified that the ctype |
199 macros don't need to be guarded with references to isascii. ... | 199 macros don't need to be guarded with references to isascii. ... |
200 Defining isascii to 1 should let any compiler worth its salt | 200 Defining isascii to 1 should let any compiler worth its salt |
201 eliminate the && through constant folding." */ | 201 eliminate the && through constant folding." */ |
202 | 202 |
203 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII)) | 203 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII)) |
204 #define ISASCII(c) 1 | 204 #define ISASCII(c) 1 |
205 #else | 205 #else |
206 #define ISASCII(c) isascii(c) | 206 #define ISASCII(c) isascii(c) |
233 #endif | 233 #endif |
234 | 234 |
235 /* We remove any previous definition of `SIGN_EXTEND_CHAR', | 235 /* We remove any previous definition of `SIGN_EXTEND_CHAR', |
236 since ours (we hope) works properly with all combinations of | 236 since ours (we hope) works properly with all combinations of |
237 machines, compilers, `char' and `unsigned char' argument types. | 237 machines, compilers, `char' and `unsigned char' argument types. |
238 (Per Bothner suggested the basic approach.) */ | 238 (Per Bothner suggested the basic approach.) */ |
239 #undef SIGN_EXTEND_CHAR | 239 #undef SIGN_EXTEND_CHAR |
240 #if __STDC__ | 240 #if __STDC__ |
241 #define SIGN_EXTEND_CHAR(c) ((signed char) (c)) | 241 #define SIGN_EXTEND_CHAR(c) ((signed char) (c)) |
242 #else /* not __STDC__ */ | 242 #else /* not __STDC__ */ |
243 /* As in Harbison and Steele. */ | 243 /* As in Harbison and Steele. */ |
271 #else /* not __GNUC__ */ | 271 #else /* not __GNUC__ */ |
272 #if HAVE_ALLOCA_H | 272 #if HAVE_ALLOCA_H |
273 #include <alloca.h> | 273 #include <alloca.h> |
274 #else /* not __GNUC__ or HAVE_ALLOCA_H */ | 274 #else /* not __GNUC__ or HAVE_ALLOCA_H */ |
275 #if 0 /* It is a bad idea to declare alloca. We always cast the result. */ | 275 #if 0 /* It is a bad idea to declare alloca. We always cast the result. */ |
276 #ifndef _AIX /* Already did AIX, up at the top. */ | 276 #ifndef _AIX /* Already did AIX, up at the top. */ |
277 char *alloca (); | 277 char *alloca (); |
278 #endif /* not _AIX */ | 278 #endif /* not _AIX */ |
279 #endif | 279 #endif |
280 #endif /* not HAVE_ALLOCA_H */ | 280 #endif /* not HAVE_ALLOCA_H */ |
281 #endif /* not __GNUC__ */ | 281 #endif /* not __GNUC__ */ |
318 | 318 |
319 #define REGEX_ALLOCATE_STACK alloca | 319 #define REGEX_ALLOCATE_STACK alloca |
320 | 320 |
321 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \ | 321 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \ |
322 REGEX_REALLOCATE (source, osize, nsize) | 322 REGEX_REALLOCATE (source, osize, nsize) |
323 /* No need to explicitly free anything. */ | 323 /* No need to explicitly free anything. */ |
324 #define REGEX_FREE_STACK(arg) | 324 #define REGEX_FREE_STACK(arg) |
325 | 325 |
326 #endif /* not REGEX_MALLOC */ | 326 #endif /* not REGEX_MALLOC */ |
327 #endif /* not using relocating allocator */ | 327 #endif /* not using relocating allocator */ |
328 | 328 |
329 | 329 |
330 /* True if `size1' is non-NULL and PTR is pointing anywhere inside | 330 /* True if `size1' is non-NULL and PTR is pointing anywhere inside |
331 `string1' or just past its end. This works if PTR is NULL, which is | 331 `string1' or just past its end. This works if PTR is NULL, which is |
332 a good thing. */ | 332 a good thing. */ |
333 #define FIRST_STRING_P(ptr) \ | 333 #define FIRST_STRING_P(ptr) \ |
334 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) | 334 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) |
335 | 335 |
336 /* (Re)Allocate N items of type T using malloc, or fail. */ | 336 /* (Re)Allocate N items of type T using malloc, or fail. */ |
337 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t))) | 337 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t))) |
338 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t))) | 338 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t))) |
339 #define RETALLOC_IF(addr, n, t) \ | 339 #define RETALLOC_IF(addr, n, t) \ |
340 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t) | 340 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t) |
341 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) | 341 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) |
342 | 342 |
343 #define BYTEWIDTH 8 /* In bits. */ | 343 #define BYTEWIDTH 8 /* In bits. */ |
344 | 344 |
345 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0)) | 345 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0)) |
346 | 346 |
347 #undef MAX | 347 #undef MAX |
348 #undef MIN | 348 #undef MIN |
354 #define true 1 | 354 #define true 1 |
355 | 355 |
356 static int re_match_2_internal (); | 356 static int re_match_2_internal (); |
357 | 357 |
358 /* These are the command codes that appear in compiled regular | 358 /* These are the command codes that appear in compiled regular |
359 expressions. Some opcodes are followed by argument bytes. A | 359 expressions. Some opcodes are followed by argument bytes. A |
360 command code can specify any interpretation whatsoever for its | 360 command code can specify any interpretation whatsoever for its |
361 arguments. Zero bytes may appear in the compiled regular expression. */ | 361 arguments. Zero bytes may appear in the compiled regular expression. */ |
362 | 362 |
363 typedef enum | 363 typedef enum |
364 { | 364 { |
365 no_op = 0, | 365 no_op = 0, |
366 | 366 |
367 /* Succeed right away--no more backtracking. */ | 367 /* Succeed right away--no more backtracking. */ |
368 succeed, | 368 succeed, |
369 | 369 |
370 /* Followed by one byte giving n, then by n literal bytes. */ | 370 /* Followed by one byte giving n, then by n literal bytes. */ |
371 exactn, | 371 exactn, |
372 | 372 |
373 /* Matches any (more or less) character. */ | 373 /* Matches any (more or less) character. */ |
374 anychar, | 374 anychar, |
375 | 375 |
376 /* Matches any one char belonging to specified set. First | 376 /* Matches any one char belonging to specified set. First |
377 following byte is number of bitmap bytes. Then come bytes | 377 following byte is number of bitmap bytes. Then come bytes |
378 for a bitmap saying which chars are in. Bits in each byte | 378 for a bitmap saying which chars are in. Bits in each byte |
379 are ordered low-bit-first. A character is in the set if its | 379 are ordered low-bit-first. A character is in the set if its |
380 bit is 1. A character too large to have a bit in the map is | 380 bit is 1. A character too large to have a bit in the map is |
381 automatically not in the set. */ | 381 automatically not in the set. */ |
382 charset, | 382 charset, |
383 | 383 |
384 /* Same parameters as charset, but match any character that is | 384 /* Same parameters as charset, but match any character that is |
385 not one of those specified. */ | 385 not one of those specified. */ |
386 charset_not, | 386 charset_not, |
387 | 387 |
388 /* Start remembering the text that is matched, for storing in a | 388 /* Start remembering the text that is matched, for storing in a |
389 register. Followed by one byte with the register number, in | 389 register. Followed by one byte with the register number, in |
390 the range 0 to one less than the pattern buffer's re_nsub | 390 the range 0 to one less than the pattern buffer's re_nsub |
391 field. Then followed by one byte with the number of groups | 391 field. Then followed by one byte with the number of groups |
392 inner to this one. (This last has to be part of the | 392 inner to this one. (This last has to be part of the |
393 start_memory only because we need it in the on_failure_jump | 393 start_memory only because we need it in the on_failure_jump |
394 of re_match_2.) */ | 394 of re_match_2.) */ |
395 start_memory, | 395 start_memory, |
396 | 396 |
397 /* Stop remembering the text that is matched and store it in a | 397 /* Stop remembering the text that is matched and store it in a |
398 memory register. Followed by one byte with the register | 398 memory register. Followed by one byte with the register |
399 number, in the range 0 to one less than `re_nsub' in the | 399 number, in the range 0 to one less than `re_nsub' in the |
400 pattern buffer, and one byte with the number of inner groups, | 400 pattern buffer, and one byte with the number of inner groups, |
401 just like `start_memory'. (We need the number of inner | 401 just like `start_memory'. (We need the number of inner |
402 groups here because we don't have any easy way of finding the | 402 groups here because we don't have any easy way of finding the |
403 corresponding start_memory when we're at a stop_memory.) */ | 403 corresponding start_memory when we're at a stop_memory.) */ |
404 stop_memory, | 404 stop_memory, |
405 | 405 |
406 /* Match a duplicate of something remembered. Followed by one | 406 /* Match a duplicate of something remembered. Followed by one |
407 byte containing the register number. */ | 407 byte containing the register number. */ |
408 duplicate, | 408 duplicate, |
409 | 409 |
410 /* Fail unless at beginning of line. */ | 410 /* Fail unless at beginning of line. */ |
411 begline, | 411 begline, |
412 | 412 |
413 /* Fail unless at end of line. */ | 413 /* Fail unless at end of line. */ |
414 endline, | 414 endline, |
415 | 415 |
416 /* Succeeds if at beginning of buffer (if emacs) or at beginning | 416 /* Succeeds if at beginning of buffer (if emacs) or at beginning |
417 of string to be matched (if not). */ | 417 of string to be matched (if not). */ |
418 begbuf, | 418 begbuf, |
419 | 419 |
420 /* Analogously, for end of buffer/string. */ | 420 /* Analogously, for end of buffer/string. */ |
421 endbuf, | 421 endbuf, |
422 | 422 |
423 /* Followed by two byte relative address to which to jump. */ | 423 /* Followed by two byte relative address to which to jump. */ |
424 jump, | 424 jump, |
425 | 425 |
426 /* Same as jump, but marks the end of an alternative. */ | 426 /* Same as jump, but marks the end of an alternative. */ |
427 jump_past_alt, | 427 jump_past_alt, |
428 | 428 |
429 /* Followed by two-byte relative address of place to resume at | 429 /* Followed by two-byte relative address of place to resume at |
430 in case of failure. */ | 430 in case of failure. */ |
431 on_failure_jump, | 431 on_failure_jump, |
432 | 432 |
433 /* Like on_failure_jump, but pushes a placeholder instead of the | 433 /* Like on_failure_jump, but pushes a placeholder instead of the |
434 current string position when executed. */ | 434 current string position when executed. */ |
435 on_failure_keep_string_jump, | 435 on_failure_keep_string_jump, |
436 | 436 |
437 /* Throw away latest failure point and then jump to following | 437 /* Throw away latest failure point and then jump to following |
438 two-byte relative address. */ | 438 two-byte relative address. */ |
439 pop_failure_jump, | 439 pop_failure_jump, |
440 | 440 |
441 /* Change to pop_failure_jump if know won't have to backtrack to | 441 /* Change to pop_failure_jump if know won't have to backtrack to |
442 match; otherwise change to jump. This is used to jump | 442 match; otherwise change to jump. This is used to jump |
443 back to the beginning of a repeat. If what follows this jump | 443 back to the beginning of a repeat. If what follows this jump |
444 clearly won't match what the repeat does, such that we can be | 444 clearly won't match what the repeat does, such that we can be |
445 sure that there is no use backtracking out of repetitions | 445 sure that there is no use backtracking out of repetitions |
446 already matched, then we change it to a pop_failure_jump. | 446 already matched, then we change it to a pop_failure_jump. |
447 Followed by two-byte address. */ | 447 Followed by two-byte address. */ |
448 maybe_pop_jump, | 448 maybe_pop_jump, |
449 | 449 |
450 /* Jump to following two-byte address, and push a dummy failure | 450 /* Jump to following two-byte address, and push a dummy failure |
451 point. This failure point will be thrown away if an attempt | 451 point. This failure point will be thrown away if an attempt |
452 is made to use it for a failure. A `+' construct makes this | 452 is made to use it for a failure. A `+' construct makes this |
453 before the first repeat. Also used as an intermediary kind | 453 before the first repeat. Also used as an intermediary kind |
454 of jump when compiling an alternative. */ | 454 of jump when compiling an alternative. */ |
455 dummy_failure_jump, | 455 dummy_failure_jump, |
456 | 456 |
457 /* Push a dummy failure point and continue. Used at the end of | 457 /* Push a dummy failure point and continue. Used at the end of |
458 alternatives. */ | 458 alternatives. */ |
459 push_dummy_failure, | 459 push_dummy_failure, |
460 | 460 |
461 /* Followed by two-byte relative address and two-byte number n. | 461 /* Followed by two-byte relative address and two-byte number n. |
462 After matching N times, jump to the address upon failure. */ | 462 After matching N times, jump to the address upon failure. */ |
463 succeed_n, | 463 succeed_n, |
464 | 464 |
465 /* Followed by two-byte relative address, and two-byte number n. | 465 /* Followed by two-byte relative address, and two-byte number n. |
466 Jump to the address N times, then fail. */ | 466 Jump to the address N times, then fail. */ |
467 jump_n, | 467 jump_n, |
468 | 468 |
469 /* Set the following two-byte relative address to the | 469 /* Set the following two-byte relative address to the |
470 subsequent two-byte number. The address *includes* the two | 470 subsequent two-byte number. The address *includes* the two |
471 bytes of number. */ | 471 bytes of number. */ |
472 set_number_at, | 472 set_number_at, |
473 | 473 |
474 wordchar, /* Matches any word-constituent character. */ | 474 wordchar, /* Matches any word-constituent character. */ |
475 notwordchar, /* Matches any char that is not a word-constituent. */ | 475 notwordchar, /* Matches any char that is not a word-constituent. */ |
476 | 476 |
477 wordbeg, /* Succeeds if at word beginning. */ | 477 wordbeg, /* Succeeds if at word beginning. */ |
478 wordend, /* Succeeds if at word end. */ | 478 wordend, /* Succeeds if at word end. */ |
479 | 479 |
480 wordbound, /* Succeeds if at a word boundary. */ | 480 wordbound, /* Succeeds if at a word boundary. */ |
481 notwordbound /* Succeeds if not at a word boundary. */ | 481 notwordbound /* Succeeds if not at a word boundary. */ |
482 | 482 |
483 #ifdef emacs | 483 #ifdef emacs |
484 ,before_dot, /* Succeeds if before point. */ | 484 ,before_dot, /* Succeeds if before point. */ |
485 at_dot, /* Succeeds if at point. */ | 485 at_dot, /* Succeeds if at point. */ |
486 after_dot, /* Succeeds if after point. */ | 486 after_dot, /* Succeeds if after point. */ |
487 | 487 |
488 /* Matches any character whose syntax is specified. Followed by | 488 /* Matches any character whose syntax is specified. Followed by |
489 a byte which contains a syntax code, e.g., Sword. */ | 489 a byte which contains a syntax code, e.g., Sword. */ |
490 syntaxspec, | 490 syntaxspec, |
491 | 491 |
492 /* Matches any character whose syntax is not that specified. */ | 492 /* Matches any character whose syntax is not that specified. */ |
493 notsyntaxspec, | 493 notsyntaxspec, |
494 | 494 |
495 /* Matches any character whose category-set contains the specified | 495 /* Matches any character whose category-set contains the specified |
496 category. The operator is followed by a byte which contains a | 496 category. The operator is followed by a byte which contains a |
497 category code (mnemonic ASCII character). */ | 497 category code (mnemonic ASCII character). */ |
498 categoryspec, | 498 categoryspec, |
499 | 499 |
500 /* Matches any character whose category-set does not contain the | 500 /* Matches any character whose category-set does not contain the |
501 specified category. The operator is followed by a byte which | 501 specified category. The operator is followed by a byte which |
502 contains the category code (mnemonic ASCII character). */ | 502 contains the category code (mnemonic ASCII character). */ |
542 int temp = SIGN_EXTEND_CHAR (*(source + 1)); | 542 int temp = SIGN_EXTEND_CHAR (*(source + 1)); |
543 *dest = *source & 0377; | 543 *dest = *source & 0377; |
544 *dest += temp << 8; | 544 *dest += temp << 8; |
545 } | 545 } |
546 | 546 |
547 #ifndef EXTRACT_MACROS /* To debug the macros. */ | 547 #ifndef EXTRACT_MACROS /* To debug the macros. */ |
548 #undef EXTRACT_NUMBER | 548 #undef EXTRACT_NUMBER |
549 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) | 549 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) |
550 #endif /* not EXTRACT_MACROS */ | 550 #endif /* not EXTRACT_MACROS */ |
551 | 551 |
552 #endif /* DEBUG */ | 552 #endif /* DEBUG */ |
555 SOURCE must be an lvalue. */ | 555 SOURCE must be an lvalue. */ |
556 | 556 |
557 #define EXTRACT_NUMBER_AND_INCR(destination, source) \ | 557 #define EXTRACT_NUMBER_AND_INCR(destination, source) \ |
558 do { \ | 558 do { \ |
559 EXTRACT_NUMBER (destination, source); \ | 559 EXTRACT_NUMBER (destination, source); \ |
560 (source) += 2; \ | 560 (source) += 2; \ |
561 } while (0) | 561 } while (0) |
562 | 562 |
563 #ifdef DEBUG | 563 #ifdef DEBUG |
564 static void | 564 static void |
565 extract_number_and_incr (destination, source) | 565 extract_number_and_incr (destination, source) |
578 | 578 |
579 #endif /* DEBUG */ | 579 #endif /* DEBUG */ |
580 | 580 |
581 /* Store a multibyte character in three contiguous bytes starting | 581 /* Store a multibyte character in three contiguous bytes starting |
582 DESTINATION, and increment DESTINATION to the byte after where the | 582 DESTINATION, and increment DESTINATION to the byte after where the |
583 character is stored. Therefore, DESTINATION must be an lvalue. */ | 583 character is stored. Therefore, DESTINATION must be an lvalue. */ |
584 | 584 |
585 #define STORE_CHARACTER_AND_INCR(destination, character) \ | 585 #define STORE_CHARACTER_AND_INCR(destination, character) \ |
586 do { \ | 586 do { \ |
587 (destination)[0] = (character) & 0377; \ | 587 (destination)[0] = (character) & 0377; \ |
588 (destination)[1] = ((character) >> 8) & 0377; \ | 588 (destination)[1] = ((character) >> 8) & 0377; \ |
589 (destination)[2] = (character) >> 16; \ | 589 (destination)[2] = (character) >> 16; \ |
590 (destination) += 3; \ | 590 (destination) += 3; \ |
591 } while (0) | 591 } while (0) |
592 | 592 |
593 /* Put into DESTINATION a character stored in three contiguous bytes | 593 /* Put into DESTINATION a character stored in three contiguous bytes |
594 starting at SOURCE. */ | 594 starting at SOURCE. */ |
595 | 595 |
596 #define EXTRACT_CHARACTER(destination, source) \ | 596 #define EXTRACT_CHARACTER(destination, source) \ |
597 do { \ | 597 do { \ |
598 (destination) = ((source)[0] \ | 598 (destination) = ((source)[0] \ |
599 | ((source)[1] << 8) \ | 599 | ((source)[1] << 8) \ |
606 /* Size of bitmap of charset P in bytes. P is a start of charset, | 606 /* Size of bitmap of charset P in bytes. P is a start of charset, |
607 i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */ | 607 i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */ |
608 #define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F) | 608 #define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F) |
609 | 609 |
610 /* Nonzero if charset P has range table. */ | 610 /* Nonzero if charset P has range table. */ |
611 #define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80) | 611 #define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80) |
612 | 612 |
613 /* Return the address of range table of charset P. But not the start | 613 /* Return the address of range table of charset P. But not the start |
614 of table itself, but the before where the number of ranges is | 614 of table itself, but the before where the number of ranges is |
615 stored. `2 +' means to skip re_opcode_t and size of bitmap. */ | 615 stored. `2 +' means to skip re_opcode_t and size of bitmap. */ |
616 #define CHARSET_RANGE_TABLE(p) (&(p)[2 + CHARSET_BITMAP_SIZE (p)]) | 616 #define CHARSET_RANGE_TABLE(p) (&(p)[2 + CHARSET_BITMAP_SIZE (p)]) |
617 | 617 |
618 /* Test if C is listed in the bitmap of charset P. */ | 618 /* Test if C is listed in the bitmap of charset P. */ |
619 #define CHARSET_LOOKUP_BITMAP(p, c) \ | 619 #define CHARSET_LOOKUP_BITMAP(p, c) \ |
620 ((c) < CHARSET_BITMAP_SIZE (p) * BYTEWIDTH \ | 620 ((c) < CHARSET_BITMAP_SIZE (p) * BYTEWIDTH \ |
621 && (p)[2 + (c) / BYTEWIDTH] & (1 << ((c) % BYTEWIDTH))) | 621 && (p)[2 + (c) / BYTEWIDTH] & (1 << ((c) % BYTEWIDTH))) |
622 | 622 |
623 /* Return the address of end of RANGE_TABLE. COUNT is number of | 623 /* Return the address of end of RANGE_TABLE. COUNT is number of |
624 ranges (which is a pair of (start, end)) in the RANGE_TABLE. `* 2' | 624 ranges (which is a pair of (start, end)) in the RANGE_TABLE. `* 2' |
625 is start of range and end of range. `* 3' is size of each start | 625 is start of range and end of range. `* 3' is size of each start |
626 and end. */ | 626 and end. */ |
627 #define CHARSET_RANGE_TABLE_END(range_table, count) \ | 627 #define CHARSET_RANGE_TABLE_END(range_table, count) \ |
628 ((range_table) + (count) * 2 * 3) | 628 ((range_table) + (count) * 2 * 3) |
629 | 629 |
630 /* Test if C is in RANGE_TABLE. A flag NOT is negated if C is in. | 630 /* Test if C is in RANGE_TABLE. A flag NOT is negated if C is in. |
631 COUNT is number of ranges in RANGE_TABLE. */ | 631 COUNT is number of ranges in RANGE_TABLE. */ |
632 #define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \ | 632 #define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \ |
633 do \ | 633 do \ |
634 { \ | 634 { \ |
635 int range_start, range_end; \ | 635 int range_start, range_end; \ |
667 | 667 |
668 /* If DEBUG is defined, Regex prints many voluminous messages about what | 668 /* If DEBUG is defined, Regex prints many voluminous messages about what |
669 it is doing (if the variable `debug' is nonzero). If linked with the | 669 it is doing (if the variable `debug' is nonzero). If linked with the |
670 main program in `iregex.c', you can enter patterns and strings | 670 main program in `iregex.c', you can enter patterns and strings |
671 interactively. And if linked with the main program in `main.c' and | 671 interactively. And if linked with the main program in `main.c' and |
672 the other test files, you can run the already-written tests. */ | 672 the other test files, you can run the already-written tests. */ |
673 | 673 |
674 #ifdef DEBUG | 674 #ifdef DEBUG |
675 | 675 |
676 /* We use standard I/O for debugging. */ | 676 /* We use standard I/O for debugging. */ |
677 #include <stdio.h> | 677 #include <stdio.h> |
684 #define DEBUG_STATEMENT(e) e | 684 #define DEBUG_STATEMENT(e) e |
685 #define DEBUG_PRINT1(x) if (debug) printf (x) | 685 #define DEBUG_PRINT1(x) if (debug) printf (x) |
686 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2) | 686 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2) |
687 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3) | 687 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3) |
688 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4) | 688 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4) |
689 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ | 689 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ |
690 if (debug) print_partial_compiled_pattern (s, e) | 690 if (debug) print_partial_compiled_pattern (s, e) |
691 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ | 691 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ |
692 if (debug) print_double_string (w, s1, sz1, s2, sz2) | 692 if (debug) print_double_string (w, s1, sz1, s2, sz2) |
693 | 693 |
694 | 694 |
704 while (i < (1 << BYTEWIDTH)) | 704 while (i < (1 << BYTEWIDTH)) |
705 { | 705 { |
706 if (fastmap[i++]) | 706 if (fastmap[i++]) |
707 { | 707 { |
708 was_a_range = 0; | 708 was_a_range = 0; |
709 putchar (i - 1); | 709 putchar (i - 1); |
710 while (i < (1 << BYTEWIDTH) && fastmap[i]) | 710 while (i < (1 << BYTEWIDTH) && fastmap[i]) |
711 { | 711 { |
712 was_a_range = 1; | 712 was_a_range = 1; |
713 i++; | 713 i++; |
714 } | 714 } |
715 if (was_a_range) | 715 if (was_a_range) |
716 { | 716 { |
717 printf ("-"); | 717 printf ("-"); |
718 putchar (i - 1); | 718 putchar (i - 1); |
719 } | 719 } |
720 } | 720 } |
721 } | 721 } |
722 putchar ('\n'); | 722 putchar ('\n'); |
723 } | 723 } |
724 | 724 |
725 | 725 |
746 { | 746 { |
747 printf ("%d:\t", p - start); | 747 printf ("%d:\t", p - start); |
748 | 748 |
749 switch ((re_opcode_t) *p++) | 749 switch ((re_opcode_t) *p++) |
750 { | 750 { |
751 case no_op: | 751 case no_op: |
752 printf ("/no_op"); | 752 printf ("/no_op"); |
753 break; | 753 break; |
754 | 754 |
755 case exactn: | 755 case exactn: |
756 mcnt = *p++; | 756 mcnt = *p++; |
757 printf ("/exactn/%d", mcnt); | 757 printf ("/exactn/%d", mcnt); |
758 do | 758 do |
759 { | 759 { |
760 putchar ('/'); | 760 putchar ('/'); |
761 putchar (*p++); | 761 putchar (*p++); |
762 } | 762 } |
763 while (--mcnt); | 763 while (--mcnt); |
764 break; | 764 break; |
765 | 765 |
766 case start_memory: | 766 case start_memory: |
767 mcnt = *p++; | 767 mcnt = *p++; |
768 printf ("/start_memory/%d/%d", mcnt, *p++); | 768 printf ("/start_memory/%d/%d", mcnt, *p++); |
769 break; | 769 break; |
770 | 770 |
771 case stop_memory: | 771 case stop_memory: |
772 mcnt = *p++; | 772 mcnt = *p++; |
773 printf ("/stop_memory/%d/%d", mcnt, *p++); | 773 printf ("/stop_memory/%d/%d", mcnt, *p++); |
774 break; | 774 break; |
775 | 775 |
776 case duplicate: | 776 case duplicate: |
777 printf ("/duplicate/%d", *p++); | 777 printf ("/duplicate/%d", *p++); |
778 break; | 778 break; |
779 | 779 |
780 case anychar: | 780 case anychar: |
781 printf ("/anychar"); | 781 printf ("/anychar"); |
782 break; | 782 break; |
783 | 783 |
784 case charset: | 784 case charset: |
785 case charset_not: | 785 case charset_not: |
786 { | 786 { |
787 register int c, last = -100; | 787 register int c, last = -100; |
788 register int in_range = 0; | 788 register int in_range = 0; |
789 | 789 |
790 printf ("/charset [%s", | 790 printf ("/charset [%s", |
791 (re_opcode_t) *(p - 1) == charset_not ? "^" : ""); | 791 (re_opcode_t) *(p - 1) == charset_not ? "^" : ""); |
792 | 792 |
793 assert (p + *p < pend); | 793 assert (p + *p < pend); |
794 | 794 |
795 for (c = 0; c < 256; c++) | 795 for (c = 0; c < 256; c++) |
796 if (c / 8 < *p | 796 if (c / 8 < *p |
797 && (p[1 + (c/8)] & (1 << (c % 8)))) | 797 && (p[1 + (c/8)] & (1 << (c % 8)))) |
798 { | 798 { |
799 /* Are we starting a range? */ | 799 /* Are we starting a range? */ |
800 if (last + 1 == c && ! in_range) | 800 if (last + 1 == c && ! in_range) |
802 putchar ('-'); | 802 putchar ('-'); |
803 in_range = 1; | 803 in_range = 1; |
804 } | 804 } |
805 /* Have we broken a range? */ | 805 /* Have we broken a range? */ |
806 else if (last + 1 != c && in_range) | 806 else if (last + 1 != c && in_range) |
807 { | 807 { |
808 putchar (last); | 808 putchar (last); |
809 in_range = 0; | 809 in_range = 0; |
810 } | 810 } |
811 | 811 |
812 if (! in_range) | 812 if (! in_range) |
813 putchar (c); | 813 putchar (c); |
814 | 814 |
815 last = c; | 815 last = c; |
816 } | 816 } |
817 | 817 |
818 if (in_range) | 818 if (in_range) |
819 putchar (last); | 819 putchar (last); |
820 | 820 |
821 putchar (']'); | 821 putchar (']'); |
824 } | 824 } |
825 break; | 825 break; |
826 | 826 |
827 case begline: | 827 case begline: |
828 printf ("/begline"); | 828 printf ("/begline"); |
829 break; | 829 break; |
830 | 830 |
831 case endline: | 831 case endline: |
832 printf ("/endline"); | 832 printf ("/endline"); |
833 break; | 833 break; |
834 | 834 |
835 case on_failure_jump: | 835 case on_failure_jump: |
836 extract_number_and_incr (&mcnt, &p); | 836 extract_number_and_incr (&mcnt, &p); |
837 printf ("/on_failure_jump to %d", p + mcnt - start); | 837 printf ("/on_failure_jump to %d", p + mcnt - start); |
838 break; | 838 break; |
839 | 839 |
840 case on_failure_keep_string_jump: | 840 case on_failure_keep_string_jump: |
841 extract_number_and_incr (&mcnt, &p); | 841 extract_number_and_incr (&mcnt, &p); |
842 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start); | 842 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start); |
843 break; | 843 break; |
844 | 844 |
845 case dummy_failure_jump: | 845 case dummy_failure_jump: |
846 extract_number_and_incr (&mcnt, &p); | 846 extract_number_and_incr (&mcnt, &p); |
847 printf ("/dummy_failure_jump to %d", p + mcnt - start); | 847 printf ("/dummy_failure_jump to %d", p + mcnt - start); |
848 break; | 848 break; |
849 | 849 |
850 case push_dummy_failure: | 850 case push_dummy_failure: |
851 printf ("/push_dummy_failure"); | 851 printf ("/push_dummy_failure"); |
852 break; | 852 break; |
853 | 853 |
854 case maybe_pop_jump: | 854 case maybe_pop_jump: |
855 extract_number_and_incr (&mcnt, &p); | |
856 printf ("/maybe_pop_jump to %d", p + mcnt - start); | |
857 break; | |
858 | |
859 case pop_failure_jump: | |
860 extract_number_and_incr (&mcnt, &p); | 855 extract_number_and_incr (&mcnt, &p); |
861 printf ("/pop_failure_jump to %d", p + mcnt - start); | 856 printf ("/maybe_pop_jump to %d", p + mcnt - start); |
862 break; | 857 break; |
863 | 858 |
864 case jump_past_alt: | 859 case pop_failure_jump: |
865 extract_number_and_incr (&mcnt, &p); | 860 extract_number_and_incr (&mcnt, &p); |
866 printf ("/jump_past_alt to %d", p + mcnt - start); | 861 printf ("/pop_failure_jump to %d", p + mcnt - start); |
867 break; | 862 break; |
868 | 863 |
869 case jump: | 864 case jump_past_alt: |
870 extract_number_and_incr (&mcnt, &p); | 865 extract_number_and_incr (&mcnt, &p); |
871 printf ("/jump to %d", p + mcnt - start); | 866 printf ("/jump_past_alt to %d", p + mcnt - start); |
872 break; | 867 break; |
873 | 868 |
874 case succeed_n: | 869 case jump: |
875 extract_number_and_incr (&mcnt, &p); | 870 extract_number_and_incr (&mcnt, &p); |
876 extract_number_and_incr (&mcnt2, &p); | 871 printf ("/jump to %d", p + mcnt - start); |
872 break; | |
873 | |
874 case succeed_n: | |
875 extract_number_and_incr (&mcnt, &p); | |
876 extract_number_and_incr (&mcnt2, &p); | |
877 printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2); | 877 printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2); |
878 break; | 878 break; |
879 | 879 |
880 case jump_n: | 880 case jump_n: |
881 extract_number_and_incr (&mcnt, &p); | 881 extract_number_and_incr (&mcnt, &p); |
882 extract_number_and_incr (&mcnt2, &p); | 882 extract_number_and_incr (&mcnt2, &p); |
883 printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2); | 883 printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2); |
884 break; | 884 break; |
885 | 885 |
886 case set_number_at: | 886 case set_number_at: |
887 extract_number_and_incr (&mcnt, &p); | 887 extract_number_and_incr (&mcnt, &p); |
888 extract_number_and_incr (&mcnt2, &p); | 888 extract_number_and_incr (&mcnt2, &p); |
889 printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2); | 889 printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2); |
890 break; | 890 break; |
891 | 891 |
892 case wordbound: | 892 case wordbound: |
893 printf ("/wordbound"); | 893 printf ("/wordbound"); |
894 break; | 894 break; |
895 | 895 |
896 case notwordbound: | 896 case notwordbound: |
897 printf ("/notwordbound"); | 897 printf ("/notwordbound"); |
898 break; | 898 break; |
899 | 899 |
900 case wordbeg: | 900 case wordbeg: |
901 printf ("/wordbeg"); | 901 printf ("/wordbeg"); |
902 break; | 902 break; |
903 | 903 |
905 printf ("/wordend"); | 905 printf ("/wordend"); |
906 | 906 |
907 #ifdef emacs | 907 #ifdef emacs |
908 case before_dot: | 908 case before_dot: |
909 printf ("/before_dot"); | 909 printf ("/before_dot"); |
910 break; | 910 break; |
911 | 911 |
912 case at_dot: | 912 case at_dot: |
913 printf ("/at_dot"); | 913 printf ("/at_dot"); |
914 break; | 914 break; |
915 | 915 |
916 case after_dot: | 916 case after_dot: |
917 printf ("/after_dot"); | 917 printf ("/after_dot"); |
918 break; | 918 break; |
919 | 919 |
920 case syntaxspec: | 920 case syntaxspec: |
921 printf ("/syntaxspec"); | 921 printf ("/syntaxspec"); |
922 mcnt = *p++; | 922 mcnt = *p++; |
923 printf ("/%d", mcnt); | 923 printf ("/%d", mcnt); |
924 break; | 924 break; |
925 | 925 |
926 case notsyntaxspec: | 926 case notsyntaxspec: |
927 printf ("/notsyntaxspec"); | 927 printf ("/notsyntaxspec"); |
928 mcnt = *p++; | 928 mcnt = *p++; |
929 printf ("/%d", mcnt); | 929 printf ("/%d", mcnt); |
930 break; | 930 break; |
931 #endif /* emacs */ | 931 #endif /* emacs */ |
932 | 932 |
933 case wordchar: | 933 case wordchar: |
934 printf ("/wordchar"); | 934 printf ("/wordchar"); |
935 break; | 935 break; |
936 | 936 |
937 case notwordchar: | 937 case notwordchar: |
938 printf ("/notwordchar"); | 938 printf ("/notwordchar"); |
939 break; | 939 break; |
940 | 940 |
941 case begbuf: | 941 case begbuf: |
942 printf ("/begbuf"); | 942 printf ("/begbuf"); |
943 break; | 943 break; |
944 | 944 |
945 case endbuf: | 945 case endbuf: |
946 printf ("/endbuf"); | 946 printf ("/endbuf"); |
947 break; | 947 break; |
948 | 948 |
949 default: | 949 default: |
950 printf ("?%d", *(p-1)); | 950 printf ("?%d", *(p-1)); |
951 } | 951 } |
952 | 952 |
953 putchar ('\n'); | 953 putchar ('\n'); |
954 } | 954 } |
955 | 955 |
997 if (where == NULL) | 997 if (where == NULL) |
998 printf ("(null)"); | 998 printf ("(null)"); |
999 else | 999 else |
1000 { | 1000 { |
1001 if (FIRST_STRING_P (where)) | 1001 if (FIRST_STRING_P (where)) |
1002 { | 1002 { |
1003 for (this_char = where - string1; this_char < size1; this_char++) | 1003 for (this_char = where - string1; this_char < size1; this_char++) |
1004 putchar (string1[this_char]); | 1004 putchar (string1[this_char]); |
1005 | 1005 |
1006 where = string2; | 1006 where = string2; |
1007 } | 1007 } |
1008 | 1008 |
1009 for (this_char = where - string2; this_char < size2; this_char++) | 1009 for (this_char = where - string2; this_char < size2; this_char++) |
1010 putchar (string2[this_char]); | 1010 putchar (string2[this_char]); |
1011 } | 1011 } |
1012 } | 1012 } |
1013 | 1013 |
1014 #else /* not DEBUG */ | 1014 #else /* not DEBUG */ |
1015 | 1015 |
1037 /* Specify the precise syntax of regexps for compilation. This provides | 1037 /* Specify the precise syntax of regexps for compilation. This provides |
1038 for compatibility for various utilities which historically have | 1038 for compatibility for various utilities which historically have |
1039 different, incompatible syntaxes. | 1039 different, incompatible syntaxes. |
1040 | 1040 |
1041 The argument SYNTAX is a bit mask comprised of the various bits | 1041 The argument SYNTAX is a bit mask comprised of the various bits |
1042 defined in regex.h. We return the old syntax. */ | 1042 defined in regex.h. We return the old syntax. */ |
1043 | 1043 |
1044 reg_syntax_t | 1044 reg_syntax_t |
1045 re_set_syntax (syntax) | 1045 re_set_syntax (syntax) |
1046 reg_syntax_t syntax; | 1046 reg_syntax_t syntax; |
1047 { | 1047 { |
1050 re_syntax_options = syntax; | 1050 re_syntax_options = syntax; |
1051 return ret; | 1051 return ret; |
1052 } | 1052 } |
1053 | 1053 |
1054 /* This table gives an error message for each of the error codes listed | 1054 /* This table gives an error message for each of the error codes listed |
1055 in regex.h. Obviously the order here has to be same as there. | 1055 in regex.h. Obviously the order here has to be same as there. |
1056 POSIX doesn't require that we do anything for REG_NOERROR, | 1056 POSIX doesn't require that we do anything for REG_NOERROR, |
1057 but why not be nice? */ | 1057 but why not be nice? */ |
1058 | 1058 |
1059 static const char *re_error_msgid[] = | 1059 static const char *re_error_msgid[] = |
1060 { | 1060 { |
1061 gettext_noop ("Success"), /* REG_NOERROR */ | 1061 gettext_noop ("Success"), /* REG_NOERROR */ |
1062 gettext_noop ("No match"), /* REG_NOMATCH */ | 1062 gettext_noop ("No match"), /* REG_NOMATCH */ |
1075 gettext_noop ("Premature end of regular expression"), /* REG_EEND */ | 1075 gettext_noop ("Premature end of regular expression"), /* REG_EEND */ |
1076 gettext_noop ("Regular expression too big"), /* REG_ESIZE */ | 1076 gettext_noop ("Regular expression too big"), /* REG_ESIZE */ |
1077 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */ | 1077 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */ |
1078 }; | 1078 }; |
1079 | 1079 |
1080 /* Avoiding alloca during matching, to placate r_alloc. */ | 1080 /* Avoiding alloca during matching, to placate r_alloc. */ |
1081 | 1081 |
1082 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the | 1082 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the |
1083 searching and matching functions should not call alloca. On some | 1083 searching and matching functions should not call alloca. On some |
1084 systems, alloca is implemented in terms of malloc, and if we're | 1084 systems, alloca is implemented in terms of malloc, and if we're |
1085 using the relocating allocator routines, then malloc could cause a | 1085 using the relocating allocator routines, then malloc could cause a |
1107 | 1107 |
1108 /* The match routines may not allocate if (1) they would do it with malloc | 1108 /* The match routines may not allocate if (1) they would do it with malloc |
1109 and (2) it's not safe for them to use malloc. | 1109 and (2) it's not safe for them to use malloc. |
1110 Note that if REL_ALLOC is defined, matching would not use malloc for the | 1110 Note that if REL_ALLOC is defined, matching would not use malloc for the |
1111 failure stack, but we would still use it for the register vectors; | 1111 failure stack, but we would still use it for the register vectors; |
1112 so REL_ALLOC should not affect this. */ | 1112 so REL_ALLOC should not affect this. */ |
1113 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs) | 1113 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs) |
1114 #undef MATCH_MAY_ALLOCATE | 1114 #undef MATCH_MAY_ALLOCATE |
1115 #endif | 1115 #endif |
1116 | 1116 |
1117 | 1117 |
1128 #endif | 1128 #endif |
1129 | 1129 |
1130 /* Roughly the maximum number of failure points on the stack. Would be | 1130 /* Roughly the maximum number of failure points on the stack. Would be |
1131 exactly that if always used MAX_FAILURE_ITEMS items each time we failed. | 1131 exactly that if always used MAX_FAILURE_ITEMS items each time we failed. |
1132 This is a variable only so users of regex can assign to it; we never | 1132 This is a variable only so users of regex can assign to it; we never |
1133 change it ourselves. */ | 1133 change it ourselves. */ |
1134 #if defined (MATCH_MAY_ALLOCATE) | 1134 #if defined (MATCH_MAY_ALLOCATE) |
1135 /* 4400 was enough to cause a crash on Alpha OSF/1, | 1135 /* 4400 was enough to cause a crash on Alpha OSF/1, |
1136 whose default stack limit is 2mb. */ | 1136 whose default stack limit is 2mb. */ |
1137 int re_max_failures = 20000; | 1137 int re_max_failures = 20000; |
1138 #else | 1138 #else |
1189 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. | 1189 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. |
1190 | 1190 |
1191 Return 1 if succeeds, and 0 if either ran out of memory | 1191 Return 1 if succeeds, and 0 if either ran out of memory |
1192 allocating space for it or it was already too large. | 1192 allocating space for it or it was already too large. |
1193 | 1193 |
1194 REGEX_REALLOCATE_STACK requires `destination' be declared. */ | 1194 REGEX_REALLOCATE_STACK requires `destination' be declared. */ |
1195 | 1195 |
1196 #define DOUBLE_FAIL_STACK(fail_stack) \ | 1196 #define DOUBLE_FAIL_STACK(fail_stack) \ |
1197 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \ | 1197 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \ |
1198 ? 0 \ | 1198 ? 0 \ |
1199 : ((fail_stack).stack = (fail_stack_elt_t *) \ | 1199 : ((fail_stack).stack = (fail_stack_elt_t *) \ |
1200 REGEX_REALLOCATE_STACK ((fail_stack).stack, \ | 1200 REGEX_REALLOCATE_STACK ((fail_stack).stack, \ |
1201 (fail_stack).size * sizeof (fail_stack_elt_t), \ | 1201 (fail_stack).size * sizeof (fail_stack_elt_t), \ |
1202 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ | 1202 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ |
1203 \ | 1203 \ |
1204 (fail_stack).stack == NULL \ | 1204 (fail_stack).stack == NULL \ |
1205 ? 0 \ | 1205 ? 0 \ |
1206 : ((fail_stack).size <<= 1, \ | 1206 : ((fail_stack).size <<= 1, \ |
1207 1))) | 1207 1))) |
1208 | 1208 |
1209 | 1209 |
1210 /* Push pointer POINTER on FAIL_STACK. | 1210 /* Push pointer POINTER on FAIL_STACK. |
1211 Return 1 if was able to do so and 0 if ran out of memory allocating | 1211 Return 1 if was able to do so and 0 if ran out of memory allocating |
1212 space to do so. */ | 1212 space to do so. */ |
1217 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \ | 1217 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \ |
1218 1)) | 1218 1)) |
1219 | 1219 |
1220 /* Push a pointer value onto the failure stack. | 1220 /* Push a pointer value onto the failure stack. |
1221 Assumes the variable `fail_stack'. Probably should only | 1221 Assumes the variable `fail_stack'. Probably should only |
1222 be called from within `PUSH_FAILURE_POINT'. */ | 1222 be called from within `PUSH_FAILURE_POINT'. */ |
1223 #define PUSH_FAILURE_POINTER(item) \ | 1223 #define PUSH_FAILURE_POINTER(item) \ |
1224 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item) | 1224 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item) |
1225 | 1225 |
1226 /* This pushes an integer-valued item onto the failure stack. | 1226 /* This pushes an integer-valued item onto the failure stack. |
1227 Assumes the variable `fail_stack'. Probably should only | 1227 Assumes the variable `fail_stack'. Probably should only |
1228 be called from within `PUSH_FAILURE_POINT'. */ | 1228 be called from within `PUSH_FAILURE_POINT'. */ |
1229 #define PUSH_FAILURE_INT(item) \ | 1229 #define PUSH_FAILURE_INT(item) \ |
1230 fail_stack.stack[fail_stack.avail++].integer = (item) | 1230 fail_stack.stack[fail_stack.avail++].integer = (item) |
1231 | 1231 |
1232 /* Push a fail_stack_elt_t value onto the failure stack. | 1232 /* Push a fail_stack_elt_t value onto the failure stack. |
1233 Assumes the variable `fail_stack'. Probably should only | 1233 Assumes the variable `fail_stack'. Probably should only |
1234 be called from within `PUSH_FAILURE_POINT'. */ | 1234 be called from within `PUSH_FAILURE_POINT'. */ |
1235 #define PUSH_FAILURE_ELT(item) \ | 1235 #define PUSH_FAILURE_ELT(item) \ |
1236 fail_stack.stack[fail_stack.avail++] = (item) | 1236 fail_stack.stack[fail_stack.avail++] = (item) |
1237 | 1237 |
1238 /* These three POP... operations complement the three PUSH... operations. | 1238 /* These three POP... operations complement the three PUSH... operations. |
1239 All assume that `fail_stack' is nonempty. */ | 1239 All assume that `fail_stack' is nonempty. */ |
1264 do { \ | 1264 do { \ |
1265 char *destination; \ | 1265 char *destination; \ |
1266 /* Must be int, so when we don't save any registers, the arithmetic \ | 1266 /* Must be int, so when we don't save any registers, the arithmetic \ |
1267 of 0 + -1 isn't done as unsigned. */ \ | 1267 of 0 + -1 isn't done as unsigned. */ \ |
1268 int this_reg; \ | 1268 int this_reg; \ |
1269 \ | 1269 \ |
1270 DEBUG_STATEMENT (failure_id++); \ | 1270 DEBUG_STATEMENT (failure_id++); \ |
1271 DEBUG_STATEMENT (nfailure_points_pushed++); \ | 1271 DEBUG_STATEMENT (nfailure_points_pushed++); \ |
1272 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \ | 1272 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \ |
1273 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\ | 1273 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\ |
1274 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\ | 1274 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\ |
1275 \ | 1275 \ |
1276 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \ | 1276 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \ |
1277 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \ | 1277 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \ |
1278 \ | 1278 \ |
1279 /* Ensure we have enough space allocated for what we will push. */ \ | 1279 /* Ensure we have enough space allocated for what we will push. */ \ |
1280 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ | 1280 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ |
1281 { \ | 1281 { \ |
1282 if (!DOUBLE_FAIL_STACK (fail_stack)) \ | 1282 if (!DOUBLE_FAIL_STACK (fail_stack)) \ |
1283 return failure_code; \ | 1283 return failure_code; \ |
1284 \ | 1284 \ |
1285 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \ | 1285 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \ |
1286 (fail_stack).size); \ | 1286 (fail_stack).size); \ |
1287 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\ | 1287 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\ |
1288 } \ | 1288 } \ |
1289 \ | 1289 \ |
1290 /* Push the info, starting with the registers. */ \ | 1290 /* Push the info, starting with the registers. */ \ |
1291 DEBUG_PRINT1 ("\n"); \ | 1291 DEBUG_PRINT1 ("\n"); \ |
1292 \ | 1292 \ |
1324 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \ | 1324 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \ |
1325 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ | 1325 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ |
1326 PUSH_FAILURE_POINTER (pattern_place); \ | 1326 PUSH_FAILURE_POINTER (pattern_place); \ |
1327 \ | 1327 \ |
1328 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \ | 1328 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \ |
1329 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ | 1329 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ |
1330 size2); \ | 1330 size2); \ |
1331 DEBUG_PRINT1 ("'\n"); \ | 1331 DEBUG_PRINT1 ("'\n"); \ |
1332 PUSH_FAILURE_POINTER (string_place); \ | 1332 PUSH_FAILURE_POINTER (string_place); \ |
1333 \ | 1333 \ |
1334 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \ | 1334 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \ |
1371 LOW_REG, HIGH_REG -- the highest and lowest active registers. | 1371 LOW_REG, HIGH_REG -- the highest and lowest active registers. |
1372 REGSTART, REGEND -- arrays of string positions. | 1372 REGSTART, REGEND -- arrays of string positions. |
1373 REG_INFO -- array of information about each subexpression. | 1373 REG_INFO -- array of information about each subexpression. |
1374 | 1374 |
1375 Also assumes the variables `fail_stack' and (if debugging), `bufp', | 1375 Also assumes the variables `fail_stack' and (if debugging), `bufp', |
1376 `pend', `string1', `size1', `string2', and `size2'. */ | 1376 `pend', `string1', `size1', `string2', and `size2'. */ |
1377 | 1377 |
1378 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\ | 1378 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\ |
1379 { \ | 1379 { \ |
1380 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \ | 1380 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \ |
1381 int this_reg; \ | 1381 int this_reg; \ |
1384 assert (!FAIL_STACK_EMPTY ()); \ | 1384 assert (!FAIL_STACK_EMPTY ()); \ |
1385 \ | 1385 \ |
1386 /* Remove failure points and point to how many regs pushed. */ \ | 1386 /* Remove failure points and point to how many regs pushed. */ \ |
1387 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \ | 1387 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \ |
1388 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \ | 1388 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \ |
1389 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \ | 1389 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \ |
1390 \ | 1390 \ |
1391 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ | 1391 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ |
1392 \ | 1392 \ |
1393 DEBUG_POP (&failure_id); \ | 1393 DEBUG_POP (&failure_id); \ |
1394 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \ | 1394 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \ |
1416 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \ | 1416 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \ |
1417 \ | 1417 \ |
1418 if (1) \ | 1418 if (1) \ |
1419 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ | 1419 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ |
1420 { \ | 1420 { \ |
1421 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \ | 1421 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \ |
1422 \ | 1422 \ |
1423 reg_info[this_reg].word = POP_FAILURE_ELT (); \ | 1423 reg_info[this_reg].word = POP_FAILURE_ELT (); \ |
1424 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \ | 1424 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \ |
1425 \ | 1425 \ |
1426 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \ | 1426 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \ |
1427 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ | 1427 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ |
1428 \ | 1428 \ |
1429 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \ | 1429 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \ |
1430 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ | 1430 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ |
1431 } \ | 1431 } \ |
1432 else \ | 1432 else \ |
1433 { \ | 1433 { \ |
1434 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \ | 1434 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \ |
1435 { \ | 1435 { \ |
1461 { | 1461 { |
1462 fail_stack_elt_t word; | 1462 fail_stack_elt_t word; |
1463 struct | 1463 struct |
1464 { | 1464 { |
1465 /* This field is one if this group can match the empty string, | 1465 /* This field is one if this group can match the empty string, |
1466 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ | 1466 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ |
1467 #define MATCH_NULL_UNSET_VALUE 3 | 1467 #define MATCH_NULL_UNSET_VALUE 3 |
1468 unsigned match_null_string_p : 2; | 1468 unsigned match_null_string_p : 2; |
1469 unsigned is_active : 1; | 1469 unsigned is_active : 1; |
1470 unsigned matched_something : 1; | 1470 unsigned matched_something : 1; |
1471 unsigned ever_matched_something : 1; | 1471 unsigned ever_matched_something : 1; |
1522 if (translate) c = (unsigned char) translate[c]; \ | 1522 if (translate) c = (unsigned char) translate[c]; \ |
1523 } while (0) | 1523 } while (0) |
1524 #endif | 1524 #endif |
1525 | 1525 |
1526 /* Fetch the next character in the uncompiled pattern, with no | 1526 /* Fetch the next character in the uncompiled pattern, with no |
1527 translation. */ | 1527 translation. */ |
1528 #define PATFETCH_RAW(c) \ | 1528 #define PATFETCH_RAW(c) \ |
1529 do {if (p == pend) return REG_EEND; \ | 1529 do {if (p == pend) return REG_EEND; \ |
1530 c = (unsigned char) *p++; \ | 1530 c = (unsigned char) *p++; \ |
1531 } while (0) | 1531 } while (0) |
1532 | 1532 |
1533 /* Go backwards one character in the pattern. */ | 1533 /* Go backwards one character in the pattern. */ |
1534 #define PATUNFETCH p-- | 1534 #define PATUNFETCH p-- |
1535 | 1535 |
1547 /* Macros for outputting the compiled pattern into `buffer'. */ | 1547 /* Macros for outputting the compiled pattern into `buffer'. */ |
1548 | 1548 |
1549 /* If the buffer isn't allocated when it comes in, use this. */ | 1549 /* If the buffer isn't allocated when it comes in, use this. */ |
1550 #define INIT_BUF_SIZE 32 | 1550 #define INIT_BUF_SIZE 32 |
1551 | 1551 |
1552 /* Make sure we have at least N more bytes of space in buffer. */ | 1552 /* Make sure we have at least N more bytes of space in buffer. */ |
1553 #define GET_BUFFER_SPACE(n) \ | 1553 #define GET_BUFFER_SPACE(n) \ |
1554 while (b - bufp->buffer + (n) > bufp->allocated) \ | 1554 while (b - bufp->buffer + (n) > bufp->allocated) \ |
1555 EXTEND_BUFFER () | 1555 EXTEND_BUFFER () |
1556 | 1556 |
1557 /* Make sure we have one more byte of buffer space and then add C to it. */ | 1557 /* Make sure we have one more byte of buffer space and then add C to it. */ |
1569 *b++ = (unsigned char) (c1); \ | 1569 *b++ = (unsigned char) (c1); \ |
1570 *b++ = (unsigned char) (c2); \ | 1570 *b++ = (unsigned char) (c2); \ |
1571 } while (0) | 1571 } while (0) |
1572 | 1572 |
1573 | 1573 |
1574 /* As with BUF_PUSH_2, except for three bytes. */ | 1574 /* As with BUF_PUSH_2, except for three bytes. */ |
1575 #define BUF_PUSH_3(c1, c2, c3) \ | 1575 #define BUF_PUSH_3(c1, c2, c3) \ |
1576 do { \ | 1576 do { \ |
1577 GET_BUFFER_SPACE (3); \ | 1577 GET_BUFFER_SPACE (3); \ |
1578 *b++ = (unsigned char) (c1); \ | 1578 *b++ = (unsigned char) (c1); \ |
1579 *b++ = (unsigned char) (c2); \ | 1579 *b++ = (unsigned char) (c2); \ |
1580 *b++ = (unsigned char) (c3); \ | 1580 *b++ = (unsigned char) (c3); \ |
1581 } while (0) | 1581 } while (0) |
1582 | 1582 |
1583 | 1583 |
1584 /* Store a jump with opcode OP at LOC to location TO. We store a | 1584 /* Store a jump with opcode OP at LOC to location TO. We store a |
1585 relative address offset by the three bytes the jump itself occupies. */ | 1585 relative address offset by the three bytes the jump itself occupies. */ |
1586 #define STORE_JUMP(op, loc, to) \ | 1586 #define STORE_JUMP(op, loc, to) \ |
1587 store_op1 (op, loc, (to) - (loc) - 3) | 1587 store_op1 (op, loc, (to) - (loc) - 3) |
1588 | 1588 |
1589 /* Likewise, for a two-argument jump. */ | 1589 /* Likewise, for a two-argument jump. */ |
1590 #define STORE_JUMP2(op, loc, to, arg) \ | 1590 #define STORE_JUMP2(op, loc, to, arg) \ |
1591 store_op2 (op, loc, (to) - (loc) - 3, arg) | 1591 store_op2 (op, loc, (to) - (loc) - 3, arg) |
1592 | 1592 |
1593 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */ | 1593 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */ |
1594 #define INSERT_JUMP(op, loc, to) \ | 1594 #define INSERT_JUMP(op, loc, to) \ |
1595 insert_op1 (op, loc, (to) - (loc) - 3, b) | 1595 insert_op1 (op, loc, (to) - (loc) - 3, b) |
1596 | 1596 |
1597 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */ | 1597 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */ |
1598 #define INSERT_JUMP2(op, loc, to, arg) \ | 1598 #define INSERT_JUMP2(op, loc, to, arg) \ |
1599 insert_op2 (op, loc, (to) - (loc) - 3, arg, b) | 1599 insert_op2 (op, loc, (to) - (loc) - 3, arg, b) |
1600 | 1600 |
1601 | 1601 |
1602 /* This is not an arbitrary limit: the arguments which represent offsets | 1602 /* This is not an arbitrary limit: the arguments which represent offsets |
1603 into the pattern are two bytes long. So if 2^16 bytes turns out to | 1603 into the pattern are two bytes long. So if 2^16 bytes turns out to |
1604 be too small, many things would have to change. */ | 1604 be too small, many things would have to change. */ |
1605 #define MAX_BUF_SIZE (1L << 16) | 1605 #define MAX_BUF_SIZE (1L << 16) |
1606 | 1606 |
1607 | 1607 |
1608 /* Extend the buffer by twice its current size via realloc and | 1608 /* Extend the buffer by twice its current size via realloc and |
1609 reset the pointers that pointed into the old block to point to the | 1609 reset the pointers that pointed into the old block to point to the |
1610 correct places in the new one. If extending the buffer results in it | 1610 correct places in the new one. If extending the buffer results in it |
1611 being larger than MAX_BUF_SIZE, then flag memory exhausted. */ | 1611 being larger than MAX_BUF_SIZE, then flag memory exhausted. */ |
1612 #define EXTEND_BUFFER() \ | 1612 #define EXTEND_BUFFER() \ |
1613 do { \ | 1613 do { \ |
1614 unsigned char *old_buffer = bufp->buffer; \ | 1614 unsigned char *old_buffer = bufp->buffer; \ |
1615 if (bufp->allocated == MAX_BUF_SIZE) \ | 1615 if (bufp->allocated == MAX_BUF_SIZE) \ |
1616 return REG_ESIZE; \ | 1616 return REG_ESIZE; \ |
1617 bufp->allocated <<= 1; \ | 1617 bufp->allocated <<= 1; \ |
1618 if (bufp->allocated > MAX_BUF_SIZE) \ | 1618 if (bufp->allocated > MAX_BUF_SIZE) \ |
1619 bufp->allocated = MAX_BUF_SIZE; \ | 1619 bufp->allocated = MAX_BUF_SIZE; \ |
1620 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\ | 1620 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\ |
1621 if (bufp->buffer == NULL) \ | 1621 if (bufp->buffer == NULL) \ |
1622 return REG_ESPACE; \ | 1622 return REG_ESPACE; \ |
1623 /* If the buffer moved, move all the pointers into it. */ \ | 1623 /* If the buffer moved, move all the pointers into it. */ \ |
1624 if (old_buffer != bufp->buffer) \ | 1624 if (old_buffer != bufp->buffer) \ |
1625 { \ | 1625 { \ |
1626 b = (b - old_buffer) + bufp->buffer; \ | 1626 b = (b - old_buffer) + bufp->buffer; \ |
1627 begalt = (begalt - old_buffer) + bufp->buffer; \ | 1627 begalt = (begalt - old_buffer) + bufp->buffer; \ |
1628 if (fixup_alt_jump) \ | 1628 if (fixup_alt_jump) \ |
1629 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\ | 1629 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\ |
1630 if (laststart) \ | 1630 if (laststart) \ |
1631 laststart = (laststart - old_buffer) + bufp->buffer; \ | 1631 laststart = (laststart - old_buffer) + bufp->buffer; \ |
1632 if (pending_exact) \ | 1632 if (pending_exact) \ |
1633 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ | 1633 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ |
1634 } \ | 1634 } \ |
1635 } while (0) | 1635 } while (0) |
1636 | 1636 |
1637 | 1637 |
1638 /* Since we have one byte reserved for the register number argument to | 1638 /* Since we have one byte reserved for the register number argument to |
1646 | 1646 |
1647 | 1647 |
1648 /* Macros for the compile stack. */ | 1648 /* Macros for the compile stack. */ |
1649 | 1649 |
1650 /* Since offsets can go either forwards or backwards, this type needs to | 1650 /* Since offsets can go either forwards or backwards, this type needs to |
1651 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ | 1651 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ |
1652 typedef int pattern_offset_t; | 1652 typedef int pattern_offset_t; |
1653 | 1653 |
1654 typedef struct | 1654 typedef struct |
1655 { | 1655 { |
1656 pattern_offset_t begalt_offset; | 1656 pattern_offset_t begalt_offset; |
1672 #define INIT_COMPILE_STACK_SIZE 32 | 1672 #define INIT_COMPILE_STACK_SIZE 32 |
1673 | 1673 |
1674 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0) | 1674 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0) |
1675 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) | 1675 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) |
1676 | 1676 |
1677 /* The next available element. */ | 1677 /* The next available element. */ |
1678 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) | 1678 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) |
1679 | 1679 |
1680 | 1680 |
1681 /* Structure to manage work area for range table. */ | 1681 /* Structure to manage work area for range table. */ |
1682 struct range_table_work_area | 1682 struct range_table_work_area |
1683 { | 1683 { |
1684 int *table; /* actual work area. */ | 1684 int *table; /* actual work area. */ |
1685 int allocated; /* allocated size for work area in bytes. */ | 1685 int allocated; /* allocated size for work area in bytes. */ |
1686 int used; /* actually used size in words. */ | 1686 int used; /* actually used size in words. */ |
1687 }; | 1687 }; |
1688 | 1688 |
1689 /* Make sure that WORK_AREA can hold more N multibyte characters. */ | 1689 /* Make sure that WORK_AREA can hold more N multibyte characters. */ |
1690 #define EXTEND_RANGE_TABLE_WORK_AREA(work_area, n) \ | 1690 #define EXTEND_RANGE_TABLE_WORK_AREA(work_area, n) \ |
1691 do { \ | 1691 do { \ |
1709 EXTEND_RANGE_TABLE_WORK_AREA ((work_area), 2); \ | 1709 EXTEND_RANGE_TABLE_WORK_AREA ((work_area), 2); \ |
1710 (work_area).table[(work_area).used++] = (range_start); \ | 1710 (work_area).table[(work_area).used++] = (range_start); \ |
1711 (work_area).table[(work_area).used++] = (range_end); \ | 1711 (work_area).table[(work_area).used++] = (range_end); \ |
1712 } while (0) | 1712 } while (0) |
1713 | 1713 |
1714 /* Free allocated memory for WORK_AREA. */ | 1714 /* Free allocated memory for WORK_AREA. */ |
1715 #define FREE_RANGE_TABLE_WORK_AREA(work_area) \ | 1715 #define FREE_RANGE_TABLE_WORK_AREA(work_area) \ |
1716 do { \ | 1716 do { \ |
1717 if ((work_area).table) \ | 1717 if ((work_area).table) \ |
1718 free ((work_area).table); \ | 1718 free ((work_area).table); \ |
1719 } while (0) | 1719 } while (0) |
1722 #define RANGE_TABLE_WORK_USED(work_area) ((work_area).used) | 1722 #define RANGE_TABLE_WORK_USED(work_area) ((work_area).used) |
1723 #define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i]) | 1723 #define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i]) |
1724 | 1724 |
1725 | 1725 |
1726 /* Set the bit for character C in a list. */ | 1726 /* Set the bit for character C in a list. */ |
1727 #define SET_LIST_BIT(c) \ | 1727 #define SET_LIST_BIT(c) \ |
1728 (b[((unsigned char) (c)) / BYTEWIDTH] \ | 1728 (b[((unsigned char) (c)) / BYTEWIDTH] \ |
1729 |= 1 << (((unsigned char) c) % BYTEWIDTH)) | 1729 |= 1 << (((unsigned char) c) % BYTEWIDTH)) |
1730 | 1730 |
1731 | 1731 |
1732 /* Get the next unsigned number in the uncompiled pattern. */ | 1732 /* Get the next unsigned number in the uncompiled pattern. */ |
1733 #define GET_UNSIGNED_NUMBER(num) \ | 1733 #define GET_UNSIGNED_NUMBER(num) \ |
1734 { if (p != pend) \ | 1734 { if (p != pend) \ |
1735 { \ | 1735 { \ |
1736 PATFETCH (c); \ | 1736 PATFETCH (c); \ |
1737 while (ISDIGIT (c)) \ | 1737 while (ISDIGIT (c)) \ |
1738 { \ | 1738 { \ |
1739 if (num < 0) \ | 1739 if (num < 0) \ |
1740 num = 0; \ | 1740 num = 0; \ |
1741 num = num * 10 + c - '0'; \ | 1741 num = num * 10 + c - '0'; \ |
1742 if (p == pend) \ | 1742 if (p == pend) \ |
1743 break; \ | 1743 break; \ |
1744 PATFETCH (c); \ | 1744 PATFETCH (c); \ |
1745 } \ | 1745 } \ |
1746 } \ | 1746 } \ |
1747 } | 1747 } |
1748 | 1748 |
1749 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ | 1749 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ |
1750 | 1750 |
1751 #define IS_CHAR_CLASS(string) \ | 1751 #define IS_CHAR_CLASS(string) \ |
1767 | 1767 |
1768 static fail_stack_type fail_stack; | 1768 static fail_stack_type fail_stack; |
1769 | 1769 |
1770 /* Size with which the following vectors are currently allocated. | 1770 /* Size with which the following vectors are currently allocated. |
1771 That is so we can make them bigger as needed, | 1771 That is so we can make them bigger as needed, |
1772 but never make them smaller. */ | 1772 but never make them smaller. */ |
1773 static int regs_allocated_size; | 1773 static int regs_allocated_size; |
1774 | 1774 |
1775 static const char ** regstart, ** regend; | 1775 static const char ** regstart, ** regend; |
1776 static const char ** old_regstart, ** old_regend; | 1776 static const char ** old_regstart, ** old_regend; |
1777 static const char **best_regstart, **best_regend; | 1777 static const char **best_regstart, **best_regend; |
1778 static register_info_type *reg_info; | 1778 static register_info_type *reg_info; |
1779 static const char **reg_dummy; | 1779 static const char **reg_dummy; |
1780 static register_info_type *reg_info_dummy; | 1780 static register_info_type *reg_info_dummy; |
1781 | 1781 |
1782 /* Make the register vectors big enough for NUM_REGS registers, | 1782 /* Make the register vectors big enough for NUM_REGS registers, |
1783 but don't make them smaller. */ | 1783 but don't make them smaller. */ |
1784 | 1784 |
1785 static | 1785 static |
1786 regex_grow_registers (num_regs) | 1786 regex_grow_registers (num_regs) |
1787 int num_regs; | 1787 int num_regs; |
1788 { | 1788 { |
1875 /* Place in the uncompiled pattern (i.e., the {) to | 1875 /* Place in the uncompiled pattern (i.e., the {) to |
1876 which to go back if the interval is invalid. */ | 1876 which to go back if the interval is invalid. */ |
1877 const char *beg_interval; | 1877 const char *beg_interval; |
1878 | 1878 |
1879 /* Address of the place where a forward jump should go to the end of | 1879 /* Address of the place where a forward jump should go to the end of |
1880 the containing expression. Each alternative of an `or' -- except the | 1880 the containing expression. Each alternative of an `or' -- except the |
1881 last -- ends with a forward jump of this sort. */ | 1881 last -- ends with a forward jump of this sort. */ |
1882 unsigned char *fixup_alt_jump = 0; | 1882 unsigned char *fixup_alt_jump = 0; |
1883 | 1883 |
1884 /* Counts open-groups as they are encountered. Remembered for the | 1884 /* Counts open-groups as they are encountered. Remembered for the |
1885 matching close-group on the compile stack, so the same register | 1885 matching close-group on the compile stack, so the same register |
1894 if (debug) | 1894 if (debug) |
1895 { | 1895 { |
1896 unsigned debug_count; | 1896 unsigned debug_count; |
1897 | 1897 |
1898 for (debug_count = 0; debug_count < size; debug_count++) | 1898 for (debug_count = 0; debug_count < size; debug_count++) |
1899 putchar (pattern[debug_count]); | 1899 putchar (pattern[debug_count]); |
1900 putchar ('\n'); | 1900 putchar ('\n'); |
1901 } | 1901 } |
1902 #endif /* DEBUG */ | 1902 #endif /* DEBUG */ |
1903 | 1903 |
1904 /* Initialize the compile stack. */ | 1904 /* Initialize the compile stack. */ |
1940 | 1940 |
1941 if (bufp->allocated == 0) | 1941 if (bufp->allocated == 0) |
1942 { | 1942 { |
1943 if (bufp->buffer) | 1943 if (bufp->buffer) |
1944 { /* If zero allocated, but buffer is non-null, try to realloc | 1944 { /* If zero allocated, but buffer is non-null, try to realloc |
1945 enough space. This loses if buffer's address is bogus, but | 1945 enough space. This loses if buffer's address is bogus, but |
1946 that is the user's responsibility. */ | 1946 that is the user's responsibility. */ |
1947 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); | 1947 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); |
1948 } | 1948 } |
1949 else | 1949 else |
1950 { /* Caller did not allocate a buffer. Do it for them. */ | 1950 { /* Caller did not allocate a buffer. Do it for them. */ |
1951 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); | 1951 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); |
1952 } | 1952 } |
1953 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE); | 1953 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE); |
1954 | 1954 |
1955 bufp->allocated = INIT_BUF_SIZE; | 1955 bufp->allocated = INIT_BUF_SIZE; |
1956 } | 1956 } |
1957 | 1957 |
1961 while (p != pend) | 1961 while (p != pend) |
1962 { | 1962 { |
1963 PATFETCH (c); | 1963 PATFETCH (c); |
1964 | 1964 |
1965 switch (c) | 1965 switch (c) |
1966 { | 1966 { |
1967 case '^': | 1967 case '^': |
1968 { | 1968 { |
1969 if ( /* If at start of pattern, it's an operator. */ | 1969 if ( /* If at start of pattern, it's an operator. */ |
1970 p == pattern + 1 | 1970 p == pattern + 1 |
1971 /* If context independent, it's an operator. */ | 1971 /* If context independent, it's an operator. */ |
1972 || syntax & RE_CONTEXT_INDEP_ANCHORS | 1972 || syntax & RE_CONTEXT_INDEP_ANCHORS |
1973 /* Otherwise, depends on what's come before. */ | 1973 /* Otherwise, depends on what's come before. */ |
1974 || at_begline_loc_p (pattern, p, syntax)) | 1974 || at_begline_loc_p (pattern, p, syntax)) |
1975 BUF_PUSH (begline); | 1975 BUF_PUSH (begline); |
1976 else | 1976 else |
1977 goto normal_char; | 1977 goto normal_char; |
1978 } | 1978 } |
1979 break; | 1979 break; |
1980 | 1980 |
1981 | 1981 |
1982 case '$': | 1982 case '$': |
1983 { | 1983 { |
1984 if ( /* If at end of pattern, it's an operator. */ | 1984 if ( /* If at end of pattern, it's an operator. */ |
1985 p == pend | 1985 p == pend |
1986 /* If context independent, it's an operator. */ | 1986 /* If context independent, it's an operator. */ |
1987 || syntax & RE_CONTEXT_INDEP_ANCHORS | 1987 || syntax & RE_CONTEXT_INDEP_ANCHORS |
1988 /* Otherwise, depends on what's next. */ | 1988 /* Otherwise, depends on what's next. */ |
1989 || at_endline_loc_p (p, pend, syntax)) | 1989 || at_endline_loc_p (p, pend, syntax)) |
1990 BUF_PUSH (endline); | 1990 BUF_PUSH (endline); |
1991 else | 1991 else |
1992 goto normal_char; | 1992 goto normal_char; |
1993 } | 1993 } |
1994 break; | 1994 break; |
1995 | 1995 |
1996 | 1996 |
1997 case '+': | 1997 case '+': |
1998 case '?': | 1998 case '?': |
1999 if ((syntax & RE_BK_PLUS_QM) | 1999 if ((syntax & RE_BK_PLUS_QM) |
2000 || (syntax & RE_LIMITED_OPS)) | 2000 || (syntax & RE_LIMITED_OPS)) |
2001 goto normal_char; | 2001 goto normal_char; |
2002 handle_plus: | 2002 handle_plus: |
2003 case '*': | 2003 case '*': |
2004 /* If there is no previous pattern... */ | 2004 /* If there is no previous pattern... */ |
2005 if (!laststart) | 2005 if (!laststart) |
2006 { | 2006 { |
2007 if (syntax & RE_CONTEXT_INVALID_OPS) | 2007 if (syntax & RE_CONTEXT_INVALID_OPS) |
2008 FREE_STACK_RETURN (REG_BADRPT); | 2008 FREE_STACK_RETURN (REG_BADRPT); |
2009 else if (!(syntax & RE_CONTEXT_INDEP_OPS)) | 2009 else if (!(syntax & RE_CONTEXT_INDEP_OPS)) |
2010 goto normal_char; | 2010 goto normal_char; |
2011 } | 2011 } |
2012 | 2012 |
2013 { | 2013 { |
2014 /* Are we optimizing this jump? */ | 2014 /* Are we optimizing this jump? */ |
2015 boolean keep_string_p = false; | 2015 boolean keep_string_p = false; |
2016 | 2016 |
2017 /* 1 means zero (many) matches is allowed. */ | 2017 /* 1 means zero (many) matches is allowed. */ |
2018 char zero_times_ok = 0, many_times_ok = 0; | 2018 char zero_times_ok = 0, many_times_ok = 0; |
2019 | 2019 |
2020 /* If there is a sequence of repetition chars, collapse it | 2020 /* If there is a sequence of repetition chars, collapse it |
2021 down to just one (the right one). We can't combine | 2021 down to just one (the right one). We can't combine |
2022 interval operators with these because of, e.g., `a{2}*', | 2022 interval operators with these because of, e.g., `a{2}*', |
2023 which should only match an even number of `a's. */ | 2023 which should only match an even number of `a's. */ |
2024 | 2024 |
2025 for (;;) | 2025 for (;;) |
2026 { | 2026 { |
2027 zero_times_ok |= c != '+'; | 2027 zero_times_ok |= c != '+'; |
2028 many_times_ok |= c != '?'; | 2028 many_times_ok |= c != '?'; |
2029 | 2029 |
2030 if (p == pend) | 2030 if (p == pend) |
2031 break; | 2031 break; |
2032 | 2032 |
2033 PATFETCH (c); | 2033 PATFETCH (c); |
2034 | 2034 |
2035 if (c == '*' | 2035 if (c == '*' |
2036 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) | 2036 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) |
2037 ; | 2037 ; |
2038 | 2038 |
2039 else if (syntax & RE_BK_PLUS_QM && c == '\\') | 2039 else if (syntax & RE_BK_PLUS_QM && c == '\\') |
2040 { | 2040 { |
2041 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | 2041 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); |
2042 | 2042 |
2043 PATFETCH (c1); | 2043 PATFETCH (c1); |
2044 if (!(c1 == '+' || c1 == '?')) | 2044 if (!(c1 == '+' || c1 == '?')) |
2045 { | 2045 { |
2046 PATUNFETCH; | 2046 PATUNFETCH; |
2047 PATUNFETCH; | 2047 PATUNFETCH; |
2048 break; | 2048 break; |
2049 } | 2049 } |
2050 | 2050 |
2051 c = c1; | 2051 c = c1; |
2052 } | 2052 } |
2053 else | 2053 else |
2054 { | 2054 { |
2055 PATUNFETCH; | 2055 PATUNFETCH; |
2056 break; | 2056 break; |
2057 } | 2057 } |
2058 | 2058 |
2059 /* If we get here, we found another repeat character. */ | 2059 /* If we get here, we found another repeat character. */ |
2060 } | 2060 } |
2061 | 2061 |
2062 /* Star, etc. applied to an empty pattern is equivalent | 2062 /* Star, etc. applied to an empty pattern is equivalent |
2063 to an empty pattern. */ | 2063 to an empty pattern. */ |
2064 if (!laststart) | 2064 if (!laststart) |
2065 break; | 2065 break; |
2066 | 2066 |
2067 /* Now we know whether or not zero matches is allowed | 2067 /* Now we know whether or not zero matches is allowed |
2068 and also whether or not two or more matches is allowed. */ | 2068 and also whether or not two or more matches is allowed. */ |
2069 if (many_times_ok) | 2069 if (many_times_ok) |
2070 { /* More than one repetition is allowed, so put in at the | 2070 { /* More than one repetition is allowed, so put in at the |
2071 end a backward relative jump from `b' to before the next | 2071 end a backward relative jump from `b' to before the next |
2072 jump we're going to put in below (which jumps from | 2072 jump we're going to put in below (which jumps from |
2073 laststart to after this jump). | 2073 laststart to after this jump). |
2074 | 2074 |
2075 But if we are at the `*' in the exact sequence `.*\n', | 2075 But if we are at the `*' in the exact sequence `.*\n', |
2076 insert an unconditional jump backwards to the ., | 2076 insert an unconditional jump backwards to the ., |
2077 instead of the beginning of the loop. This way we only | 2077 instead of the beginning of the loop. This way we only |
2078 push a failure point once, instead of every time | 2078 push a failure point once, instead of every time |
2079 through the loop. */ | 2079 through the loop. */ |
2080 assert (p - 1 > pattern); | 2080 assert (p - 1 > pattern); |
2081 | 2081 |
2082 /* Allocate the space for the jump. */ | 2082 /* Allocate the space for the jump. */ |
2083 GET_BUFFER_SPACE (3); | 2083 GET_BUFFER_SPACE (3); |
2084 | 2084 |
2085 /* We know we are not at the first character of the pattern, | 2085 /* We know we are not at the first character of the pattern, |
2086 because laststart was nonzero. And we've already | 2086 because laststart was nonzero. And we've already |
2087 incremented `p', by the way, to be the character after | 2087 incremented `p', by the way, to be the character after |
2088 the `*'. Do we have to do something analogous here | 2088 the `*'. Do we have to do something analogous here |
2089 for null bytes, because of RE_DOT_NOT_NULL? */ | 2089 for null bytes, because of RE_DOT_NOT_NULL? */ |
2090 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.') | 2090 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.') |
2091 && zero_times_ok | 2091 && zero_times_ok |
2092 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n') | 2092 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n') |
2093 && !(syntax & RE_DOT_NEWLINE)) | 2093 && !(syntax & RE_DOT_NEWLINE)) |
2094 { /* We have .*\n. */ | 2094 { /* We have .*\n. */ |
2095 STORE_JUMP (jump, b, laststart); | 2095 STORE_JUMP (jump, b, laststart); |
2096 keep_string_p = true; | 2096 keep_string_p = true; |
2097 } | 2097 } |
2098 else | 2098 else |
2099 /* Anything else. */ | 2099 /* Anything else. */ |
2100 STORE_JUMP (maybe_pop_jump, b, laststart - 3); | 2100 STORE_JUMP (maybe_pop_jump, b, laststart - 3); |
2101 | 2101 |
2102 /* We've added more stuff to the buffer. */ | 2102 /* We've added more stuff to the buffer. */ |
2103 b += 3; | 2103 b += 3; |
2104 } | 2104 } |
2105 | 2105 |
2106 /* On failure, jump from laststart to b + 3, which will be the | 2106 /* On failure, jump from laststart to b + 3, which will be the |
2107 end of the buffer after this jump is inserted. */ | 2107 end of the buffer after this jump is inserted. */ |
2108 GET_BUFFER_SPACE (3); | 2108 GET_BUFFER_SPACE (3); |
2109 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump | 2109 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump |
2110 : on_failure_jump, | 2110 : on_failure_jump, |
2111 laststart, b + 3); | 2111 laststart, b + 3); |
2112 pending_exact = 0; | 2112 pending_exact = 0; |
2113 b += 3; | 2113 b += 3; |
2114 | 2114 |
2115 if (!zero_times_ok) | 2115 if (!zero_times_ok) |
2116 { | 2116 { |
2117 /* At least one repetition is required, so insert a | 2117 /* At least one repetition is required, so insert a |
2118 `dummy_failure_jump' before the initial | 2118 `dummy_failure_jump' before the initial |
2119 `on_failure_jump' instruction of the loop. This | 2119 `on_failure_jump' instruction of the loop. This |
2120 effects a skip over that instruction the first time | 2120 effects a skip over that instruction the first time |
2121 we hit that loop. */ | 2121 we hit that loop. */ |
2122 GET_BUFFER_SPACE (3); | 2122 GET_BUFFER_SPACE (3); |
2123 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); | 2123 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); |
2124 b += 3; | 2124 b += 3; |
2125 } | 2125 } |
2126 } | 2126 } |
2127 break; | 2127 break; |
2128 | 2128 |
2129 | 2129 |
2130 case '.': | 2130 case '.': |
2131 laststart = b; | 2131 laststart = b; |
2132 BUF_PUSH (anychar); | 2132 BUF_PUSH (anychar); |
2133 break; | 2133 break; |
2134 | 2134 |
2135 | 2135 |
2136 case '[': | 2136 case '[': |
2137 { | 2137 { |
2138 CLEAR_RANGE_TABLE_WORK_USED (range_table_work); | 2138 CLEAR_RANGE_TABLE_WORK_USED (range_table_work); |
2139 | 2139 |
2140 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | 2140 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); |
2141 | 2141 |
2142 /* Ensure that we have enough space to push a charset: the | 2142 /* Ensure that we have enough space to push a charset: the |
2143 opcode, the length count, and the bitset; 34 bytes in all. */ | 2143 opcode, the length count, and the bitset; 34 bytes in all. */ |
2144 GET_BUFFER_SPACE (34); | 2144 GET_BUFFER_SPACE (34); |
2145 | 2145 |
2146 laststart = b; | 2146 laststart = b; |
2147 | 2147 |
2148 /* We test `*p == '^' twice, instead of using an if | 2148 /* We test `*p == '^' twice, instead of using an if |
2149 statement, so we only need one BUF_PUSH. */ | 2149 statement, so we only need one BUF_PUSH. */ |
2150 BUF_PUSH (*p == '^' ? charset_not : charset); | 2150 BUF_PUSH (*p == '^' ? charset_not : charset); |
2151 if (*p == '^') | 2151 if (*p == '^') |
2152 p++; | 2152 p++; |
2153 | 2153 |
2154 /* Remember the first position in the bracket expression. */ | 2154 /* Remember the first position in the bracket expression. */ |
2155 p1 = p; | 2155 p1 = p; |
2156 | 2156 |
2157 /* Push the number of bytes in the bitmap. */ | 2157 /* Push the number of bytes in the bitmap. */ |
2158 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); | 2158 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); |
2159 | 2159 |
2160 /* Clear the whole map. */ | 2160 /* Clear the whole map. */ |
2161 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH); | 2161 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH); |
2162 | 2162 |
2163 /* charset_not matches newline according to a syntax bit. */ | 2163 /* charset_not matches newline according to a syntax bit. */ |
2164 if ((re_opcode_t) b[-2] == charset_not | 2164 if ((re_opcode_t) b[-2] == charset_not |
2165 && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) | 2165 && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) |
2166 SET_LIST_BIT ('\n'); | 2166 SET_LIST_BIT ('\n'); |
2167 | 2167 |
2168 /* Read in characters and ranges, setting map bits. */ | 2168 /* Read in characters and ranges, setting map bits. */ |
2169 for (;;) | 2169 for (;;) |
2170 { | 2170 { |
2171 int len; | 2171 int len; |
2172 boolean escaped_char = false; | 2172 boolean escaped_char = false; |
2173 | 2173 |
2174 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | 2174 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); |
2175 | 2175 |
2176 PATFETCH (c); | 2176 PATFETCH (c); |
2177 | 2177 |
2178 /* \ might escape characters inside [...] and [^...]. */ | 2178 /* \ might escape characters inside [...] and [^...]. */ |
2179 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') | 2179 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') |
2180 { | 2180 { |
2181 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | 2181 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); |
2182 | 2182 |
2183 PATFETCH (c); | 2183 PATFETCH (c); |
2184 escaped_char = true; | 2184 escaped_char = true; |
2185 } | 2185 } |
2186 else | 2186 else |
2187 { | 2187 { |
2188 /* Could be the end of the bracket expression. If it's | 2188 /* Could be the end of the bracket expression. If it's |
2189 not (i.e., when the bracket expression is `[]' so | 2189 not (i.e., when the bracket expression is `[]' so |
2190 far), the ']' character bit gets set way below. */ | 2190 far), the ']' character bit gets set way below. */ |
2191 if (c == ']' && p != p1 + 1) | 2191 if (c == ']' && p != p1 + 1) |
2192 break; | 2192 break; |
2193 } | 2193 } |
2194 | 2194 |
2195 /* If C indicates start of multibyte char, get the | 2195 /* If C indicates start of multibyte char, get the |
2196 actual character code in C, and set the pattern | 2196 actual character code in C, and set the pattern |
2197 pointer P to the next character boundary. */ | 2197 pointer P to the next character boundary. */ |
2198 if (bufp->multibyte && BASE_LEADING_CODE_P (c)) | 2198 if (bufp->multibyte && BASE_LEADING_CODE_P (c)) |
2199 { | 2199 { |
2200 PATUNFETCH; | 2200 PATUNFETCH; |
2201 c = STRING_CHAR_AND_LENGTH (p, pend - p, len); | 2201 c = STRING_CHAR_AND_LENGTH (p, pend - p, len); |
2202 p += len; | 2202 p += len; |
2203 } | 2203 } |
2204 /* What should we do for the character which is | 2204 /* What should we do for the character which is |
2205 greater than 0x7F, but not BASE_LEADING_CODE_P? | 2205 greater than 0x7F, but not BASE_LEADING_CODE_P? |
2206 XXX */ | 2206 XXX */ |
2207 | 2207 |
2208 /* See if we're at the beginning of a possible character | 2208 /* See if we're at the beginning of a possible character |
2209 class. */ | 2209 class. */ |
2210 | 2210 |
2211 else if (!escaped_char && | 2211 else if (!escaped_char && |
2212 syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') | 2212 syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') |
2213 { /* Leave room for the null. */ | 2213 { /* Leave room for the null. */ |
2214 char str[CHAR_CLASS_MAX_LENGTH + 1]; | 2214 char str[CHAR_CLASS_MAX_LENGTH + 1]; |
2215 | 2215 |
2216 PATFETCH (c); | 2216 PATFETCH (c); |
2217 c1 = 0; | 2217 c1 = 0; |
2218 | 2218 |
2219 /* If pattern is `[[:'. */ | 2219 /* If pattern is `[[:'. */ |
2220 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | 2220 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); |
2221 | 2221 |
2222 for (;;) | 2222 for (;;) |
2223 { | 2223 { |
2224 PATFETCH (c); | 2224 PATFETCH (c); |
2225 if (c == ':' || c == ']' || p == pend | 2225 if (c == ':' || c == ']' || p == pend |
2226 || c1 == CHAR_CLASS_MAX_LENGTH) | 2226 || c1 == CHAR_CLASS_MAX_LENGTH) |
2227 break; | 2227 break; |
2228 str[c1++] = c; | 2228 str[c1++] = c; |
2229 } | 2229 } |
2230 str[c1] = '\0'; | 2230 str[c1] = '\0'; |
2231 | 2231 |
2232 /* If isn't a word bracketed by `[:' and `:]': | 2232 /* If isn't a word bracketed by `[:' and `:]': |
2233 undo the ending character, the letters, and | 2233 undo the ending character, the letters, and |
2234 leave the leading `:' and `[' (but set bits for | 2234 leave the leading `:' and `[' (but set bits for |
2235 them). */ | 2235 them). */ |
2236 if (c == ':' && *p == ']') | 2236 if (c == ':' && *p == ']') |
2237 { | 2237 { |
2238 int ch; | 2238 int ch; |
2239 boolean is_alnum = STREQ (str, "alnum"); | 2239 boolean is_alnum = STREQ (str, "alnum"); |
2240 boolean is_alpha = STREQ (str, "alpha"); | 2240 boolean is_alpha = STREQ (str, "alpha"); |
2241 boolean is_blank = STREQ (str, "blank"); | 2241 boolean is_blank = STREQ (str, "blank"); |
2242 boolean is_cntrl = STREQ (str, "cntrl"); | 2242 boolean is_cntrl = STREQ (str, "cntrl"); |
2243 boolean is_digit = STREQ (str, "digit"); | 2243 boolean is_digit = STREQ (str, "digit"); |
2244 boolean is_graph = STREQ (str, "graph"); | 2244 boolean is_graph = STREQ (str, "graph"); |
2245 boolean is_lower = STREQ (str, "lower"); | 2245 boolean is_lower = STREQ (str, "lower"); |
2246 boolean is_print = STREQ (str, "print"); | 2246 boolean is_print = STREQ (str, "print"); |
2247 boolean is_punct = STREQ (str, "punct"); | 2247 boolean is_punct = STREQ (str, "punct"); |
2248 boolean is_space = STREQ (str, "space"); | 2248 boolean is_space = STREQ (str, "space"); |
2249 boolean is_upper = STREQ (str, "upper"); | 2249 boolean is_upper = STREQ (str, "upper"); |
2250 boolean is_xdigit = STREQ (str, "xdigit"); | 2250 boolean is_xdigit = STREQ (str, "xdigit"); |
2251 | 2251 |
2252 if (!IS_CHAR_CLASS (str)) | 2252 if (!IS_CHAR_CLASS (str)) |
2253 FREE_STACK_RETURN (REG_ECTYPE); | 2253 FREE_STACK_RETURN (REG_ECTYPE); |
2254 | 2254 |
2255 /* Throw away the ] at the end of the character | 2255 /* Throw away the ] at the end of the character |
2256 class. */ | 2256 class. */ |
2257 PATFETCH (c); | 2257 PATFETCH (c); |
2258 | 2258 |
2259 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | 2259 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); |
2260 | 2260 |
2261 for (ch = 0; ch < 1 << BYTEWIDTH; ch++) | 2261 for (ch = 0; ch < 1 << BYTEWIDTH; ch++) |
2262 { | 2262 { |
2263 int translated = TRANSLATE (ch); | 2263 int translated = TRANSLATE (ch); |
2264 /* This was split into 3 if's to | 2264 /* This was split into 3 if's to |
2265 avoid an arbitrary limit in some compiler. */ | 2265 avoid an arbitrary limit in some compiler. */ |
2266 if ( (is_alnum && ISALNUM (ch)) | 2266 if ( (is_alnum && ISALNUM (ch)) |
2267 || (is_alpha && ISALPHA (ch)) | 2267 || (is_alpha && ISALPHA (ch)) |
2268 || (is_blank && ISBLANK (ch)) | 2268 || (is_blank && ISBLANK (ch)) |
2269 || (is_cntrl && ISCNTRL (ch))) | 2269 || (is_cntrl && ISCNTRL (ch))) |
2270 SET_LIST_BIT (translated); | 2270 SET_LIST_BIT (translated); |
2271 if ( (is_digit && ISDIGIT (ch)) | 2271 if ( (is_digit && ISDIGIT (ch)) |
2272 || (is_graph && ISGRAPH (ch)) | 2272 || (is_graph && ISGRAPH (ch)) |
2273 || (is_lower && ISLOWER (ch)) | 2273 || (is_lower && ISLOWER (ch)) |
2274 || (is_print && ISPRINT (ch))) | 2274 || (is_print && ISPRINT (ch))) |
2275 SET_LIST_BIT (translated); | 2275 SET_LIST_BIT (translated); |
2276 if ( (is_punct && ISPUNCT (ch)) | 2276 if ( (is_punct && ISPUNCT (ch)) |
2277 || (is_space && ISSPACE (ch)) | 2277 || (is_space && ISSPACE (ch)) |
2278 || (is_upper && ISUPPER (ch)) | 2278 || (is_upper && ISUPPER (ch)) |
2279 || (is_xdigit && ISXDIGIT (ch))) | 2279 || (is_xdigit && ISXDIGIT (ch))) |
2280 SET_LIST_BIT (translated); | 2280 SET_LIST_BIT (translated); |
2281 } | 2281 } |
2282 | 2282 |
2283 /* Repeat the loop. */ | 2283 /* Repeat the loop. */ |
2284 continue; | 2284 continue; |
2285 } | 2285 } |
2286 else | 2286 else |
2287 { | 2287 { |
2288 c1++; | 2288 c1++; |
2289 while (c1--) | 2289 while (c1--) |
2290 PATUNFETCH; | 2290 PATUNFETCH; |
2291 SET_LIST_BIT ('['); | 2291 SET_LIST_BIT ('['); |
2292 | 2292 |
2293 /* Because the `:' may starts the range, we | 2293 /* Because the `:' may starts the range, we |
2294 can't simply set bit and repeat the loop. | 2294 can't simply set bit and repeat the loop. |
2295 Instead, just set it to C and handle below. */ | 2295 Instead, just set it to C and handle below. */ |
2296 c = ':'; | 2296 c = ':'; |
2297 } | 2297 } |
2298 } | 2298 } |
2299 | 2299 |
2300 if (p < pend && p[0] == '-' && p[1] != ']') | 2300 if (p < pend && p[0] == '-' && p[1] != ']') |
2301 { | 2301 { |
2302 | 2302 |
2303 /* Discard the `-'. */ | 2303 /* Discard the `-'. */ |
2313 } | 2313 } |
2314 | 2314 |
2315 if (!SAME_CHARSET_P (c, c1)) | 2315 if (!SAME_CHARSET_P (c, c1)) |
2316 FREE_STACK_RETURN (REG_ERANGE); | 2316 FREE_STACK_RETURN (REG_ERANGE); |
2317 } | 2317 } |
2318 else | 2318 else |
2319 /* Range from C to C. */ | 2319 /* Range from C to C. */ |
2320 c1 = c; | 2320 c1 = c; |
2321 | 2321 |
2322 /* Set the range ... */ | 2322 /* Set the range ... */ |
2323 if (SINGLE_BYTE_CHAR_P (c)) | 2323 if (SINGLE_BYTE_CHAR_P (c)) |
2324 /* ... into bitmap. */ | 2324 /* ... into bitmap. */ |
2325 { | 2325 { |
2326 unsigned this_char; | 2326 unsigned this_char; |
2327 int range_start = c, range_end = c1; | 2327 int range_start = c, range_end = c1; |
2328 | 2328 |
2329 /* If the start is after the end, the range is empty. */ | 2329 /* If the start is after the end, the range is empty. */ |
2330 if (range_start > range_end) | 2330 if (range_start > range_end) |
2336 else | 2336 else |
2337 { | 2337 { |
2338 for (this_char = range_start; this_char <= range_end; | 2338 for (this_char = range_start; this_char <= range_end; |
2339 this_char++) | 2339 this_char++) |
2340 SET_LIST_BIT (TRANSLATE (this_char)); | 2340 SET_LIST_BIT (TRANSLATE (this_char)); |
2341 } | 2341 } |
2342 } | 2342 } |
2343 else | 2343 else |
2344 /* ... into range table. */ | 2344 /* ... into range table. */ |
2345 SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1); | 2345 SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1); |
2346 } | 2346 } |
2347 | 2347 |
2348 /* Discard any (non)matching list bytes that are all 0 at the | 2348 /* Discard any (non)matching list bytes that are all 0 at the |
2349 end of the map. Decrease the map-length byte too. */ | 2349 end of the map. Decrease the map-length byte too. */ |
2350 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) | 2350 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) |
2351 b[-1]--; | 2351 b[-1]--; |
2352 b += b[-1]; | 2352 b += b[-1]; |
2353 | 2353 |
2354 /* Build real range table from work area. */ | 2354 /* Build real range table from work area. */ |
2355 if (RANGE_TABLE_WORK_USED (range_table_work)) | 2355 if (RANGE_TABLE_WORK_USED (range_table_work)) |
2356 { | 2356 { |
2357 int i; | 2357 int i; |
2358 int used = RANGE_TABLE_WORK_USED (range_table_work); | 2358 int used = RANGE_TABLE_WORK_USED (range_table_work); |
2359 | 2359 |
2360 /* Allocate space for COUNT + RANGE_TABLE. Needs two | 2360 /* Allocate space for COUNT + RANGE_TABLE. Needs two |
2361 bytes for COUNT and three bytes for each character. */ | 2361 bytes for COUNT and three bytes for each character. */ |
2362 GET_BUFFER_SPACE (2 + used * 3); | 2362 GET_BUFFER_SPACE (2 + used * 3); |
2363 | 2363 |
2364 /* Indicate the existence of range table. */ | 2364 /* Indicate the existence of range table. */ |
2365 laststart[1] |= 0x80; | 2365 laststart[1] |= 0x80; |
2366 | 2366 |
2367 STORE_NUMBER_AND_INCR (b, used / 2); | 2367 STORE_NUMBER_AND_INCR (b, used / 2); |
2368 for (i = 0; i < used; i++) | 2368 for (i = 0; i < used; i++) |
2369 STORE_CHARACTER_AND_INCR | 2369 STORE_CHARACTER_AND_INCR |
2370 (b, RANGE_TABLE_WORK_ELT (range_table_work, i)); | 2370 (b, RANGE_TABLE_WORK_ELT (range_table_work, i)); |
2371 } | 2371 } |
2372 } | 2372 } |
2373 break; | 2373 break; |
2374 | 2374 |
2375 | 2375 |
2376 case '(': | 2376 case '(': |
2377 if (syntax & RE_NO_BK_PARENS) | 2377 if (syntax & RE_NO_BK_PARENS) |
2378 goto handle_open; | 2378 goto handle_open; |
2379 else | 2379 else |
2380 goto normal_char; | 2380 goto normal_char; |
2381 | 2381 |
2382 | 2382 |
2383 case ')': | 2383 case ')': |
2384 if (syntax & RE_NO_BK_PARENS) | 2384 if (syntax & RE_NO_BK_PARENS) |
2385 goto handle_close; | 2385 goto handle_close; |
2386 else | 2386 else |
2387 goto normal_char; | 2387 goto normal_char; |
2388 | 2388 |
2389 | 2389 |
2390 case '\n': | 2390 case '\n': |
2391 if (syntax & RE_NEWLINE_ALT) | 2391 if (syntax & RE_NEWLINE_ALT) |
2392 goto handle_alt; | 2392 goto handle_alt; |
2393 else | 2393 else |
2394 goto normal_char; | 2394 goto normal_char; |
2395 | 2395 |
2396 | 2396 |
2397 case '|': | 2397 case '|': |
2398 if (syntax & RE_NO_BK_VBAR) | 2398 if (syntax & RE_NO_BK_VBAR) |
2399 goto handle_alt; | 2399 goto handle_alt; |
2400 else | 2400 else |
2401 goto normal_char; | 2401 goto normal_char; |
2402 | 2402 |
2403 | 2403 |
2404 case '{': | 2404 case '{': |
2405 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) | 2405 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) |
2406 goto handle_interval; | 2406 goto handle_interval; |
2407 else | 2407 else |
2408 goto normal_char; | 2408 goto normal_char; |
2409 | 2409 |
2410 | 2410 |
2411 case '\\': | 2411 case '\\': |
2412 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | 2412 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); |
2413 | 2413 |
2414 /* Do not translate the character after the \, so that we can | 2414 /* Do not translate the character after the \, so that we can |
2415 distinguish, e.g., \B from \b, even if we normally would | 2415 distinguish, e.g., \B from \b, even if we normally would |
2416 translate, e.g., B to b. */ | 2416 translate, e.g., B to b. */ |
2417 PATFETCH_RAW (c); | 2417 PATFETCH_RAW (c); |
2418 | 2418 |
2419 switch (c) | 2419 switch (c) |
2420 { | 2420 { |
2421 case '(': | 2421 case '(': |
2422 if (syntax & RE_NO_BK_PARENS) | 2422 if (syntax & RE_NO_BK_PARENS) |
2423 goto normal_backslash; | 2423 goto normal_backslash; |
2424 | 2424 |
2425 handle_open: | 2425 handle_open: |
2426 bufp->re_nsub++; | 2426 bufp->re_nsub++; |
2427 regnum++; | 2427 regnum++; |
2428 | 2428 |
2429 if (COMPILE_STACK_FULL) | 2429 if (COMPILE_STACK_FULL) |
2430 { | 2430 { |
2431 RETALLOC (compile_stack.stack, compile_stack.size << 1, | 2431 RETALLOC (compile_stack.stack, compile_stack.size << 1, |
2432 compile_stack_elt_t); | 2432 compile_stack_elt_t); |
2433 if (compile_stack.stack == NULL) return REG_ESPACE; | 2433 if (compile_stack.stack == NULL) return REG_ESPACE; |
2434 | 2434 |
2435 compile_stack.size <<= 1; | 2435 compile_stack.size <<= 1; |
2436 } | 2436 } |
2437 | 2437 |
2438 /* These are the values to restore when we hit end of this | 2438 /* These are the values to restore when we hit end of this |
2439 group. They are all relative offsets, so that if the | 2439 group. They are all relative offsets, so that if the |
2440 whole pattern moves because of realloc, they will still | 2440 whole pattern moves because of realloc, they will still |
2441 be valid. */ | 2441 be valid. */ |
2442 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; | 2442 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; |
2443 COMPILE_STACK_TOP.fixup_alt_jump | 2443 COMPILE_STACK_TOP.fixup_alt_jump |
2444 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; | 2444 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; |
2445 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; | 2445 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; |
2446 COMPILE_STACK_TOP.regnum = regnum; | 2446 COMPILE_STACK_TOP.regnum = regnum; |
2447 | 2447 |
2448 /* We will eventually replace the 0 with the number of | 2448 /* We will eventually replace the 0 with the number of |
2449 groups inner to this one. But do not push a | 2449 groups inner to this one. But do not push a |
2450 start_memory for groups beyond the last one we can | 2450 start_memory for groups beyond the last one we can |
2451 represent in the compiled pattern. */ | 2451 represent in the compiled pattern. */ |
2452 if (regnum <= MAX_REGNUM) | 2452 if (regnum <= MAX_REGNUM) |
2453 { | 2453 { |
2454 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; | 2454 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; |
2455 BUF_PUSH_3 (start_memory, regnum, 0); | 2455 BUF_PUSH_3 (start_memory, regnum, 0); |
2456 } | 2456 } |
2457 | 2457 |
2458 compile_stack.avail++; | 2458 compile_stack.avail++; |
2459 | 2459 |
2460 fixup_alt_jump = 0; | 2460 fixup_alt_jump = 0; |
2461 laststart = 0; | 2461 laststart = 0; |
2462 begalt = b; | 2462 begalt = b; |
2463 /* If we've reached MAX_REGNUM groups, then this open | 2463 /* If we've reached MAX_REGNUM groups, then this open |
2464 won't actually generate any code, so we'll have to | 2464 won't actually generate any code, so we'll have to |
2465 clear pending_exact explicitly. */ | 2465 clear pending_exact explicitly. */ |
2466 pending_exact = 0; | 2466 pending_exact = 0; |
2467 break; | 2467 break; |
2468 | 2468 |
2469 | 2469 |
2470 case ')': | 2470 case ')': |
2471 if (syntax & RE_NO_BK_PARENS) goto normal_backslash; | 2471 if (syntax & RE_NO_BK_PARENS) goto normal_backslash; |
2472 | 2472 |
2473 if (COMPILE_STACK_EMPTY) | 2473 if (COMPILE_STACK_EMPTY) |
2474 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) | 2474 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) |
2475 goto normal_backslash; | 2475 goto normal_backslash; |
2476 else | 2476 else |
2477 FREE_STACK_RETURN (REG_ERPAREN); | 2477 FREE_STACK_RETURN (REG_ERPAREN); |
2478 | 2478 |
2479 handle_close: | 2479 handle_close: |
2480 if (fixup_alt_jump) | 2480 if (fixup_alt_jump) |
2481 { /* Push a dummy failure point at the end of the | 2481 { /* Push a dummy failure point at the end of the |
2482 alternative for a possible future | 2482 alternative for a possible future |
2483 `pop_failure_jump' to pop. See comments at | 2483 `pop_failure_jump' to pop. See comments at |
2484 `push_dummy_failure' in `re_match_2'. */ | 2484 `push_dummy_failure' in `re_match_2'. */ |
2485 BUF_PUSH (push_dummy_failure); | 2485 BUF_PUSH (push_dummy_failure); |
2486 | 2486 |
2487 /* We allocated space for this jump when we assigned | 2487 /* We allocated space for this jump when we assigned |
2488 to `fixup_alt_jump', in the `handle_alt' case below. */ | 2488 to `fixup_alt_jump', in the `handle_alt' case below. */ |
2489 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); | 2489 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); |
2490 } | 2490 } |
2491 | 2491 |
2492 /* See similar code for backslashed left paren above. */ | 2492 /* See similar code for backslashed left paren above. */ |
2493 if (COMPILE_STACK_EMPTY) | 2493 if (COMPILE_STACK_EMPTY) |
2494 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) | 2494 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) |
2495 goto normal_char; | 2495 goto normal_char; |
2496 else | 2496 else |
2497 FREE_STACK_RETURN (REG_ERPAREN); | 2497 FREE_STACK_RETURN (REG_ERPAREN); |
2498 | 2498 |
2499 /* Since we just checked for an empty stack above, this | 2499 /* Since we just checked for an empty stack above, this |
2500 ``can't happen''. */ | 2500 ``can't happen''. */ |
2501 assert (compile_stack.avail != 0); | 2501 assert (compile_stack.avail != 0); |
2502 { | 2502 { |
2503 /* We don't just want to restore into `regnum', because | 2503 /* We don't just want to restore into `regnum', because |
2504 later groups should continue to be numbered higher, | 2504 later groups should continue to be numbered higher, |
2505 as in `(ab)c(de)' -- the second group is #2. */ | 2505 as in `(ab)c(de)' -- the second group is #2. */ |
2506 regnum_t this_group_regnum; | 2506 regnum_t this_group_regnum; |
2507 | 2507 |
2508 compile_stack.avail--; | 2508 compile_stack.avail--; |
2509 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; | 2509 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; |
2510 fixup_alt_jump | 2510 fixup_alt_jump |
2511 = COMPILE_STACK_TOP.fixup_alt_jump | 2511 = COMPILE_STACK_TOP.fixup_alt_jump |
2512 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 | 2512 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 |
2513 : 0; | 2513 : 0; |
2514 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; | 2514 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; |
2515 this_group_regnum = COMPILE_STACK_TOP.regnum; | 2515 this_group_regnum = COMPILE_STACK_TOP.regnum; |
2516 /* If we've reached MAX_REGNUM groups, then this open | 2516 /* If we've reached MAX_REGNUM groups, then this open |
2517 won't actually generate any code, so we'll have to | 2517 won't actually generate any code, so we'll have to |
2518 clear pending_exact explicitly. */ | 2518 clear pending_exact explicitly. */ |
2519 pending_exact = 0; | 2519 pending_exact = 0; |
2520 | 2520 |
2521 /* We're at the end of the group, so now we know how many | 2521 /* We're at the end of the group, so now we know how many |
2522 groups were inside this one. */ | 2522 groups were inside this one. */ |
2523 if (this_group_regnum <= MAX_REGNUM) | 2523 if (this_group_regnum <= MAX_REGNUM) |
2524 { | 2524 { |
2525 unsigned char *inner_group_loc | 2525 unsigned char *inner_group_loc |
2526 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; | 2526 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; |
2527 | 2527 |
2528 *inner_group_loc = regnum - this_group_regnum; | 2528 *inner_group_loc = regnum - this_group_regnum; |
2529 BUF_PUSH_3 (stop_memory, this_group_regnum, | 2529 BUF_PUSH_3 (stop_memory, this_group_regnum, |
2530 regnum - this_group_regnum); | 2530 regnum - this_group_regnum); |
2531 } | 2531 } |
2532 } | 2532 } |
2533 break; | 2533 break; |
2534 | 2534 |
2535 | 2535 |
2536 case '|': /* `\|'. */ | 2536 case '|': /* `\|'. */ |
2537 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) | 2537 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) |
2538 goto normal_backslash; | 2538 goto normal_backslash; |
2539 handle_alt: | 2539 handle_alt: |
2540 if (syntax & RE_LIMITED_OPS) | 2540 if (syntax & RE_LIMITED_OPS) |
2541 goto normal_char; | 2541 goto normal_char; |
2542 | 2542 |
2543 /* Insert before the previous alternative a jump which | 2543 /* Insert before the previous alternative a jump which |
2544 jumps to this alternative if the former fails. */ | 2544 jumps to this alternative if the former fails. */ |
2545 GET_BUFFER_SPACE (3); | 2545 GET_BUFFER_SPACE (3); |
2546 INSERT_JUMP (on_failure_jump, begalt, b + 6); | 2546 INSERT_JUMP (on_failure_jump, begalt, b + 6); |
2547 pending_exact = 0; | 2547 pending_exact = 0; |
2548 b += 3; | 2548 b += 3; |
2549 | 2549 |
2550 /* The alternative before this one has a jump after it | 2550 /* The alternative before this one has a jump after it |
2551 which gets executed if it gets matched. Adjust that | 2551 which gets executed if it gets matched. Adjust that |
2552 jump so it will jump to this alternative's analogous | 2552 jump so it will jump to this alternative's analogous |
2553 jump (put in below, which in turn will jump to the next | 2553 jump (put in below, which in turn will jump to the next |
2554 (if any) alternative's such jump, etc.). The last such | 2554 (if any) alternative's such jump, etc.). The last such |
2555 jump jumps to the correct final destination. A picture: | 2555 jump jumps to the correct final destination. A picture: |
2556 _____ _____ | 2556 _____ _____ |
2557 | | | | | 2557 | | | | |
2558 | v | v | 2558 | v | v |
2559 a | b | c | 2559 a | b | c |
2560 | 2560 |
2561 If we are at `b', then fixup_alt_jump right now points to a | 2561 If we are at `b', then fixup_alt_jump right now points to a |
2562 three-byte space after `a'. We'll put in the jump, set | 2562 three-byte space after `a'. We'll put in the jump, set |
2563 fixup_alt_jump to right after `b', and leave behind three | 2563 fixup_alt_jump to right after `b', and leave behind three |
2564 bytes which we'll fill in when we get to after `c'. */ | 2564 bytes which we'll fill in when we get to after `c'. */ |
2565 | 2565 |
2566 if (fixup_alt_jump) | 2566 if (fixup_alt_jump) |
2567 STORE_JUMP (jump_past_alt, fixup_alt_jump, b); | 2567 STORE_JUMP (jump_past_alt, fixup_alt_jump, b); |
2568 | 2568 |
2569 /* Mark and leave space for a jump after this alternative, | 2569 /* Mark and leave space for a jump after this alternative, |
2570 to be filled in later either by next alternative or | 2570 to be filled in later either by next alternative or |
2571 when know we're at the end of a series of alternatives. */ | 2571 when know we're at the end of a series of alternatives. */ |
2572 fixup_alt_jump = b; | 2572 fixup_alt_jump = b; |
2573 GET_BUFFER_SPACE (3); | 2573 GET_BUFFER_SPACE (3); |
2574 b += 3; | 2574 b += 3; |
2575 | 2575 |
2576 laststart = 0; | 2576 laststart = 0; |
2577 begalt = b; | 2577 begalt = b; |
2578 break; | 2578 break; |
2579 | 2579 |
2580 | 2580 |
2581 case '{': | 2581 case '{': |
2582 /* If \{ is a literal. */ | 2582 /* If \{ is a literal. */ |
2583 if (!(syntax & RE_INTERVALS) | 2583 if (!(syntax & RE_INTERVALS) |
2584 /* If we're at `\{' and it's not the open-interval | 2584 /* If we're at `\{' and it's not the open-interval |
2585 operator. */ | 2585 operator. */ |
2586 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) | 2586 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) |
2587 || (p - 2 == pattern && p == pend)) | 2587 || (p - 2 == pattern && p == pend)) |
2588 goto normal_backslash; | 2588 goto normal_backslash; |
2589 | 2589 |
2590 handle_interval: | 2590 handle_interval: |
2591 { | 2591 { |
2592 /* If got here, then the syntax allows intervals. */ | 2592 /* If got here, then the syntax allows intervals. */ |
2593 | 2593 |
2594 /* At least (most) this many matches must be made. */ | 2594 /* At least (most) this many matches must be made. */ |
2595 int lower_bound = -1, upper_bound = -1; | 2595 int lower_bound = -1, upper_bound = -1; |
2596 | 2596 |
2597 beg_interval = p - 1; | 2597 beg_interval = p - 1; |
2598 | 2598 |
2599 if (p == pend) | 2599 if (p == pend) |
2600 { | 2600 { |
2601 if (syntax & RE_NO_BK_BRACES) | 2601 if (syntax & RE_NO_BK_BRACES) |
2602 goto unfetch_interval; | 2602 goto unfetch_interval; |
2603 else | 2603 else |
2604 FREE_STACK_RETURN (REG_EBRACE); | 2604 FREE_STACK_RETURN (REG_EBRACE); |
2605 } | 2605 } |
2606 | 2606 |
2607 GET_UNSIGNED_NUMBER (lower_bound); | 2607 GET_UNSIGNED_NUMBER (lower_bound); |
2608 | 2608 |
2609 if (c == ',') | 2609 if (c == ',') |
2610 { | 2610 { |
2611 GET_UNSIGNED_NUMBER (upper_bound); | 2611 GET_UNSIGNED_NUMBER (upper_bound); |
2612 if (upper_bound < 0) upper_bound = RE_DUP_MAX; | 2612 if (upper_bound < 0) upper_bound = RE_DUP_MAX; |
2613 } | 2613 } |
2614 else | 2614 else |
2615 /* Interval such as `{1}' => match exactly once. */ | 2615 /* Interval such as `{1}' => match exactly once. */ |
2616 upper_bound = lower_bound; | 2616 upper_bound = lower_bound; |
2617 | 2617 |
2618 if (lower_bound < 0 || upper_bound > RE_DUP_MAX | 2618 if (lower_bound < 0 || upper_bound > RE_DUP_MAX |
2619 || lower_bound > upper_bound) | 2619 || lower_bound > upper_bound) |
2620 { | 2620 { |
2621 if (syntax & RE_NO_BK_BRACES) | 2621 if (syntax & RE_NO_BK_BRACES) |
2622 goto unfetch_interval; | 2622 goto unfetch_interval; |
2623 else | 2623 else |
2624 FREE_STACK_RETURN (REG_BADBR); | 2624 FREE_STACK_RETURN (REG_BADBR); |
2625 } | 2625 } |
2626 | 2626 |
2627 if (!(syntax & RE_NO_BK_BRACES)) | 2627 if (!(syntax & RE_NO_BK_BRACES)) |
2628 { | 2628 { |
2629 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE); | 2629 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE); |
2630 | 2630 |
2631 PATFETCH (c); | 2631 PATFETCH (c); |
2632 } | 2632 } |
2633 | 2633 |
2634 if (c != '}') | 2634 if (c != '}') |
2635 { | 2635 { |
2636 if (syntax & RE_NO_BK_BRACES) | 2636 if (syntax & RE_NO_BK_BRACES) |
2637 goto unfetch_interval; | 2637 goto unfetch_interval; |
2638 else | 2638 else |
2639 FREE_STACK_RETURN (REG_BADBR); | 2639 FREE_STACK_RETURN (REG_BADBR); |
2640 } | 2640 } |
2641 | 2641 |
2642 /* We just parsed a valid interval. */ | 2642 /* We just parsed a valid interval. */ |
2643 | 2643 |
2644 /* If it's invalid to have no preceding re. */ | 2644 /* If it's invalid to have no preceding re. */ |
2645 if (!laststart) | 2645 if (!laststart) |
2646 { | 2646 { |
2647 if (syntax & RE_CONTEXT_INVALID_OPS) | 2647 if (syntax & RE_CONTEXT_INVALID_OPS) |
2648 FREE_STACK_RETURN (REG_BADRPT); | 2648 FREE_STACK_RETURN (REG_BADRPT); |
2649 else if (syntax & RE_CONTEXT_INDEP_OPS) | 2649 else if (syntax & RE_CONTEXT_INDEP_OPS) |
2650 laststart = b; | 2650 laststart = b; |
2651 else | 2651 else |
2652 goto unfetch_interval; | 2652 goto unfetch_interval; |
2653 } | 2653 } |
2654 | 2654 |
2655 /* If the upper bound is zero, don't want to succeed at | 2655 /* If the upper bound is zero, don't want to succeed at |
2656 all; jump from `laststart' to `b + 3', which will be | 2656 all; jump from `laststart' to `b + 3', which will be |
2657 the end of the buffer after we insert the jump. */ | 2657 the end of the buffer after we insert the jump. */ |
2658 if (upper_bound == 0) | 2658 if (upper_bound == 0) |
2659 { | 2659 { |
2660 GET_BUFFER_SPACE (3); | 2660 GET_BUFFER_SPACE (3); |
2661 INSERT_JUMP (jump, laststart, b + 3); | 2661 INSERT_JUMP (jump, laststart, b + 3); |
2662 b += 3; | 2662 b += 3; |
2663 } | 2663 } |
2664 | 2664 |
2665 /* Otherwise, we have a nontrivial interval. When | 2665 /* Otherwise, we have a nontrivial interval. When |
2666 we're all done, the pattern will look like: | 2666 we're all done, the pattern will look like: |
2667 set_number_at <jump count> <upper bound> | 2667 set_number_at <jump count> <upper bound> |
2668 set_number_at <succeed_n count> <lower bound> | 2668 set_number_at <succeed_n count> <lower bound> |
2669 succeed_n <after jump addr> <succeed_n count> | 2669 succeed_n <after jump addr> <succeed_n count> |
2670 <body of loop> | 2670 <body of loop> |
2671 jump_n <succeed_n addr> <jump count> | 2671 jump_n <succeed_n addr> <jump count> |
2672 (The upper bound and `jump_n' are omitted if | 2672 (The upper bound and `jump_n' are omitted if |
2673 `upper_bound' is 1, though.) */ | 2673 `upper_bound' is 1, though.) */ |
2674 else | 2674 else |
2675 { /* If the upper bound is > 1, we need to insert | 2675 { /* If the upper bound is > 1, we need to insert |
2676 more at the end of the loop. */ | 2676 more at the end of the loop. */ |
2677 unsigned nbytes = 10 + (upper_bound > 1) * 10; | 2677 unsigned nbytes = 10 + (upper_bound > 1) * 10; |
2678 | 2678 |
2679 GET_BUFFER_SPACE (nbytes); | 2679 GET_BUFFER_SPACE (nbytes); |
2680 | 2680 |
2681 /* Initialize lower bound of the `succeed_n', even | 2681 /* Initialize lower bound of the `succeed_n', even |
2682 though it will be set during matching by its | 2682 though it will be set during matching by its |
2683 attendant `set_number_at' (inserted next), | 2683 attendant `set_number_at' (inserted next), |
2684 because `re_compile_fastmap' needs to know. | 2684 because `re_compile_fastmap' needs to know. |
2685 Jump to the `jump_n' we might insert below. */ | 2685 Jump to the `jump_n' we might insert below. */ |
2686 INSERT_JUMP2 (succeed_n, laststart, | 2686 INSERT_JUMP2 (succeed_n, laststart, |
2687 b + 5 + (upper_bound > 1) * 5, | 2687 b + 5 + (upper_bound > 1) * 5, |
2688 lower_bound); | 2688 lower_bound); |
2689 b += 5; | 2689 b += 5; |
2690 | 2690 |
2691 /* Code to initialize the lower bound. Insert | 2691 /* Code to initialize the lower bound. Insert |
2692 before the `succeed_n'. The `5' is the last two | 2692 before the `succeed_n'. The `5' is the last two |
2693 bytes of this `set_number_at', plus 3 bytes of | 2693 bytes of this `set_number_at', plus 3 bytes of |
2694 the following `succeed_n'. */ | 2694 the following `succeed_n'. */ |
2695 insert_op2 (set_number_at, laststart, 5, lower_bound, b); | 2695 insert_op2 (set_number_at, laststart, 5, lower_bound, b); |
2696 b += 5; | 2696 b += 5; |
2697 | 2697 |
2698 if (upper_bound > 1) | 2698 if (upper_bound > 1) |
2699 { /* More than one repetition is allowed, so | 2699 { /* More than one repetition is allowed, so |
2700 append a backward jump to the `succeed_n' | 2700 append a backward jump to the `succeed_n' |
2701 that starts this interval. | 2701 that starts this interval. |
2702 | 2702 |
2703 When we've reached this during matching, | 2703 When we've reached this during matching, |
2704 we'll have matched the interval once, so | 2704 we'll have matched the interval once, so |
2705 jump back only `upper_bound - 1' times. */ | 2705 jump back only `upper_bound - 1' times. */ |
2706 STORE_JUMP2 (jump_n, b, laststart + 5, | 2706 STORE_JUMP2 (jump_n, b, laststart + 5, |
2707 upper_bound - 1); | 2707 upper_bound - 1); |
2708 b += 5; | 2708 b += 5; |
2709 | 2709 |
2710 /* The location we want to set is the second | 2710 /* The location we want to set is the second |
2711 parameter of the `jump_n'; that is `b-2' as | 2711 parameter of the `jump_n'; that is `b-2' as |
2712 an absolute address. `laststart' will be | 2712 an absolute address. `laststart' will be |
2713 the `set_number_at' we're about to insert; | 2713 the `set_number_at' we're about to insert; |
2714 `laststart+3' the number to set, the source | 2714 `laststart+3' the number to set, the source |
2715 for the relative address. But we are | 2715 for the relative address. But we are |
2716 inserting into the middle of the pattern -- | 2716 inserting into the middle of the pattern -- |
2717 so everything is getting moved up by 5. | 2717 so everything is getting moved up by 5. |
2718 Conclusion: (b - 2) - (laststart + 3) + 5, | 2718 Conclusion: (b - 2) - (laststart + 3) + 5, |
2719 i.e., b - laststart. | 2719 i.e., b - laststart. |
2720 | 2720 |
2721 We insert this at the beginning of the loop | 2721 We insert this at the beginning of the loop |
2722 so that if we fail during matching, we'll | 2722 so that if we fail during matching, we'll |
2723 reinitialize the bounds. */ | 2723 reinitialize the bounds. */ |
2724 insert_op2 (set_number_at, laststart, b - laststart, | 2724 insert_op2 (set_number_at, laststart, b - laststart, |
2725 upper_bound - 1, b); | 2725 upper_bound - 1, b); |
2726 b += 5; | 2726 b += 5; |
2727 } | 2727 } |
2728 } | 2728 } |
2729 pending_exact = 0; | 2729 pending_exact = 0; |
2730 beg_interval = NULL; | 2730 beg_interval = NULL; |
2731 } | 2731 } |
2732 break; | 2732 break; |
2733 | 2733 |
2734 unfetch_interval: | 2734 unfetch_interval: |
2735 /* If an invalid interval, match the characters as literals. */ | 2735 /* If an invalid interval, match the characters as literals. */ |
2736 assert (beg_interval); | 2736 assert (beg_interval); |
2737 p = beg_interval; | 2737 p = beg_interval; |
2738 beg_interval = NULL; | 2738 beg_interval = NULL; |
2739 | 2739 |
2740 /* normal_char and normal_backslash need `c'. */ | 2740 /* normal_char and normal_backslash need `c'. */ |
2741 PATFETCH (c); | 2741 PATFETCH (c); |
2742 | 2742 |
2743 if (!(syntax & RE_NO_BK_BRACES)) | 2743 if (!(syntax & RE_NO_BK_BRACES)) |
2744 { | 2744 { |
2745 if (p > pattern && p[-1] == '\\') | 2745 if (p > pattern && p[-1] == '\\') |
2746 goto normal_backslash; | 2746 goto normal_backslash; |
2747 } | 2747 } |
2748 goto normal_char; | 2748 goto normal_char; |
2749 | 2749 |
2750 #ifdef emacs | 2750 #ifdef emacs |
2751 /* There is no way to specify the before_dot and after_dot | 2751 /* There is no way to specify the before_dot and after_dot |
2752 operators. rms says this is ok. --karl */ | 2752 operators. rms says this is ok. --karl */ |
2753 case '=': | 2753 case '=': |
2754 BUF_PUSH (at_dot); | 2754 BUF_PUSH (at_dot); |
2755 break; | 2755 break; |
2756 | 2756 |
2757 case 's': | 2757 case 's': |
2758 laststart = b; | 2758 laststart = b; |
2759 PATFETCH (c); | 2759 PATFETCH (c); |
2760 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); | 2760 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); |
2761 break; | 2761 break; |
2762 | 2762 |
2763 case 'S': | 2763 case 'S': |
2764 laststart = b; | 2764 laststart = b; |
2765 PATFETCH (c); | 2765 PATFETCH (c); |
2766 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); | 2766 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); |
2767 break; | 2767 break; |
2768 | 2768 |
2769 case 'c': | 2769 case 'c': |
2770 laststart = b; | 2770 laststart = b; |
2771 PATFETCH_RAW (c); | 2771 PATFETCH_RAW (c); |
2772 BUF_PUSH_2 (categoryspec, c); | 2772 BUF_PUSH_2 (categoryspec, c); |
2778 BUF_PUSH_2 (notcategoryspec, c); | 2778 BUF_PUSH_2 (notcategoryspec, c); |
2779 break; | 2779 break; |
2780 #endif /* emacs */ | 2780 #endif /* emacs */ |
2781 | 2781 |
2782 | 2782 |
2783 case 'w': | 2783 case 'w': |
2784 laststart = b; | 2784 laststart = b; |
2785 BUF_PUSH (wordchar); | 2785 BUF_PUSH (wordchar); |
2786 break; | 2786 break; |
2787 | 2787 |
2788 | 2788 |
2789 case 'W': | 2789 case 'W': |
2790 laststart = b; | 2790 laststart = b; |
2791 BUF_PUSH (notwordchar); | 2791 BUF_PUSH (notwordchar); |
2792 break; | 2792 break; |
2793 | 2793 |
2794 | 2794 |
2795 case '<': | 2795 case '<': |
2796 BUF_PUSH (wordbeg); | 2796 BUF_PUSH (wordbeg); |
2797 break; | 2797 break; |
2798 | 2798 |
2799 case '>': | 2799 case '>': |
2800 BUF_PUSH (wordend); | 2800 BUF_PUSH (wordend); |
2801 break; | 2801 break; |
2802 | 2802 |
2803 case 'b': | 2803 case 'b': |
2804 BUF_PUSH (wordbound); | 2804 BUF_PUSH (wordbound); |
2805 break; | 2805 break; |
2806 | 2806 |
2807 case 'B': | 2807 case 'B': |
2808 BUF_PUSH (notwordbound); | 2808 BUF_PUSH (notwordbound); |
2809 break; | 2809 break; |
2810 | 2810 |
2811 case '`': | 2811 case '`': |
2812 BUF_PUSH (begbuf); | 2812 BUF_PUSH (begbuf); |
2813 break; | 2813 break; |
2814 | 2814 |
2815 case '\'': | 2815 case '\'': |
2816 BUF_PUSH (endbuf); | 2816 BUF_PUSH (endbuf); |
2817 break; | 2817 break; |
2818 | 2818 |
2819 case '1': case '2': case '3': case '4': case '5': | 2819 case '1': case '2': case '3': case '4': case '5': |
2820 case '6': case '7': case '8': case '9': | 2820 case '6': case '7': case '8': case '9': |
2821 if (syntax & RE_NO_BK_REFS) | 2821 if (syntax & RE_NO_BK_REFS) |
2822 goto normal_char; | 2822 goto normal_char; |
2823 | 2823 |
2824 c1 = c - '0'; | 2824 c1 = c - '0'; |
2825 | 2825 |
2826 if (c1 > regnum) | 2826 if (c1 > regnum) |
2827 FREE_STACK_RETURN (REG_ESUBREG); | 2827 FREE_STACK_RETURN (REG_ESUBREG); |
2828 | 2828 |
2829 /* Can't back reference to a subexpression if inside of it. */ | 2829 /* Can't back reference to a subexpression if inside of it. */ |
2830 if (group_in_compile_stack (compile_stack, c1)) | 2830 if (group_in_compile_stack (compile_stack, c1)) |
2831 goto normal_char; | 2831 goto normal_char; |
2832 | 2832 |
2833 laststart = b; | 2833 laststart = b; |
2834 BUF_PUSH_2 (duplicate, c1); | 2834 BUF_PUSH_2 (duplicate, c1); |
2835 break; | 2835 break; |
2836 | 2836 |
2837 | 2837 |
2838 case '+': | 2838 case '+': |
2839 case '?': | 2839 case '?': |
2840 if (syntax & RE_BK_PLUS_QM) | 2840 if (syntax & RE_BK_PLUS_QM) |
2841 goto handle_plus; | 2841 goto handle_plus; |
2842 else | 2842 else |
2843 goto normal_backslash; | 2843 goto normal_backslash; |
2844 | 2844 |
2845 default: | 2845 default: |
2846 normal_backslash: | 2846 normal_backslash: |
2847 /* You might think it would be useful for \ to mean | 2847 /* You might think it would be useful for \ to mean |
2848 not to translate; but if we don't translate it | 2848 not to translate; but if we don't translate it |
2849 it will never match anything. */ | 2849 it will never match anything. */ |
2850 c = TRANSLATE (c); | 2850 c = TRANSLATE (c); |
2851 goto normal_char; | 2851 goto normal_char; |
2852 } | 2852 } |
2853 break; | 2853 break; |
2854 | 2854 |
2855 | 2855 |
2856 default: | 2856 default: |
2857 /* Expects the character in `c'. */ | 2857 /* Expects the character in `c'. */ |
2858 normal_char: | 2858 normal_char: |
2859 p1 = p - 1; /* P1 points the head of C. */ | 2859 p1 = p - 1; /* P1 points the head of C. */ |
2860 #ifdef emacs | 2860 #ifdef emacs |
2861 if (bufp->multibyte) | 2861 if (bufp->multibyte) |
2862 /* Set P to the next character boundary. */ | 2862 /* Set P to the next character boundary. */ |
2863 p += MULTIBYTE_FORM_LENGTH (p1, pend - p1) - 1; | 2863 p += MULTIBYTE_FORM_LENGTH (p1, pend - p1) - 1; |
2864 #endif | 2864 #endif |
2865 /* If no exactn currently being built. */ | 2865 /* If no exactn currently being built. */ |
2866 if (!pending_exact | 2866 if (!pending_exact |
2867 | 2867 |
2868 /* If last exactn not at current position. */ | 2868 /* If last exactn not at current position. */ |
2869 || pending_exact + *pending_exact + 1 != b | 2869 || pending_exact + *pending_exact + 1 != b |
2870 | 2870 |
2871 /* We have only one byte following the exactn for the count. */ | 2871 /* We have only one byte following the exactn for the count. */ |
2872 || *pending_exact >= (1 << BYTEWIDTH) - (p - p1) | 2872 || *pending_exact >= (1 << BYTEWIDTH) - (p - p1) |
2873 | 2873 |
2874 /* If followed by a repetition operator. */ | 2874 /* If followed by a repetition operator. */ |
2875 || *p == '*' || *p == '^' | 2875 || *p == '*' || *p == '^' |
2876 || ((syntax & RE_BK_PLUS_QM) | 2876 || ((syntax & RE_BK_PLUS_QM) |
2877 ? *p == '\\' && (p[1] == '+' || p[1] == '?') | 2877 ? *p == '\\' && (p[1] == '+' || p[1] == '?') |
2878 : (*p == '+' || *p == '?')) | 2878 : (*p == '+' || *p == '?')) |
2879 || ((syntax & RE_INTERVALS) | 2879 || ((syntax & RE_INTERVALS) |
2880 && ((syntax & RE_NO_BK_BRACES) | 2880 && ((syntax & RE_NO_BK_BRACES) |
2881 ? *p == '{' | 2881 ? *p == '{' |
2882 : (p[0] == '\\' && p[1] == '{')))) | 2882 : (p[0] == '\\' && p[1] == '{')))) |
2883 { | 2883 { |
2884 /* Start building a new exactn. */ | 2884 /* Start building a new exactn. */ |
2885 | 2885 |
2886 laststart = b; | 2886 laststart = b; |
2887 | 2887 |
2888 BUF_PUSH_2 (exactn, 0); | 2888 BUF_PUSH_2 (exactn, 0); |
2889 pending_exact = b - 1; | 2889 pending_exact = b - 1; |
2890 } | 2890 } |
2891 | 2891 |
2892 /* Here, C may translated, therefore C may not equal to *P1. */ | 2892 /* Here, C may translated, therefore C may not equal to *P1. */ |
2893 while (1) | 2893 while (1) |
2894 { | 2894 { |
2895 BUF_PUSH (c); | 2895 BUF_PUSH (c); |
2896 (*pending_exact)++; | 2896 (*pending_exact)++; |
2897 if (++p1 == p) | 2897 if (++p1 == p) |
2898 break; | 2898 break; |
2899 | 2899 |
2900 /* Rest of multibyte form should be copied literally. */ | 2900 /* Rest of multibyte form should be copied literally. */ |
2901 c = *(unsigned char *)p1; | 2901 c = *(unsigned char *)p1; |
2902 } | 2902 } |
2903 break; | 2903 break; |
2904 } /* switch (c) */ | 2904 } /* switch (c) */ |
2905 } /* while p != pend */ | 2905 } /* while p != pend */ |
2906 | 2906 |
2907 | 2907 |
2908 /* Through the pattern now. */ | 2908 /* Through the pattern now. */ |
2909 | 2909 |
2975 return REG_NOERROR; | 2975 return REG_NOERROR; |
2976 } /* regex_compile */ | 2976 } /* regex_compile */ |
2977 | 2977 |
2978 /* Subroutines for `regex_compile'. */ | 2978 /* Subroutines for `regex_compile'. */ |
2979 | 2979 |
2980 /* Store OP at LOC followed by two-byte integer parameter ARG. */ | 2980 /* Store OP at LOC followed by two-byte integer parameter ARG. */ |
2981 | 2981 |
2982 static void | 2982 static void |
2983 store_op1 (op, loc, arg) | 2983 store_op1 (op, loc, arg) |
2984 re_opcode_t op; | 2984 re_opcode_t op; |
2985 unsigned char *loc; | 2985 unsigned char *loc; |
3056 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; | 3056 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; |
3057 | 3057 |
3058 return | 3058 return |
3059 /* After a subexpression? */ | 3059 /* After a subexpression? */ |
3060 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) | 3060 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) |
3061 /* After an alternative? */ | 3061 /* After an alternative? */ |
3062 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); | 3062 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); |
3063 } | 3063 } |
3064 | 3064 |
3065 | 3065 |
3066 /* The dual of at_begline_loc_p. This one is for $. We assume there is | 3066 /* The dual of at_begline_loc_p. This one is for $. We assume there is |
3076 const char *next_next = p + 1 < pend ? p + 1 : 0; | 3076 const char *next_next = p + 1 < pend ? p + 1 : 0; |
3077 | 3077 |
3078 return | 3078 return |
3079 /* Before a subexpression? */ | 3079 /* Before a subexpression? */ |
3080 (syntax & RE_NO_BK_PARENS ? *next == ')' | 3080 (syntax & RE_NO_BK_PARENS ? *next == ')' |
3081 : next_backslash && next_next && *next_next == ')') | 3081 : next_backslash && next_next && *next_next == ')') |
3082 /* Before an alternative? */ | 3082 /* Before an alternative? */ |
3083 || (syntax & RE_NO_BK_VBAR ? *next == '|' | 3083 || (syntax & RE_NO_BK_VBAR ? *next == '|' |
3084 : next_backslash && next_next && *next_next == '|'); | 3084 : next_backslash && next_next && *next_next == '|'); |
3085 } | 3085 } |
3086 | 3086 |
3087 | 3087 |
3088 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and | 3088 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and |
3089 false if it's not. */ | 3089 false if it's not. */ |
3144 | 3144 |
3145 /* Have to increment the pointer into the pattern string, so the | 3145 /* Have to increment the pointer into the pattern string, so the |
3146 caller isn't still at the ending character. */ | 3146 caller isn't still at the ending character. */ |
3147 (*p_ptr)++; | 3147 (*p_ptr)++; |
3148 | 3148 |
3149 /* If the start is after the end, the range is empty. */ | 3149 /* If the start is after the end, the range is empty. */ |
3150 if (range_start > range_end) | 3150 if (range_start > range_end) |
3151 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; | 3151 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; |
3152 | 3152 |
3153 /* Here we see why `this_char' has to be larger than an `unsigned | 3153 /* Here we see why `this_char' has to be larger than an `unsigned |
3154 char' -- the range is inclusive, so if `range_end' == 0xff | 3154 char' -- the range is inclusive, so if `range_end' == 0xff |
3198 /* This holds the pointer to the failure stack, when | 3198 /* This holds the pointer to the failure stack, when |
3199 it is allocated relocatably. */ | 3199 it is allocated relocatably. */ |
3200 fail_stack_elt_t *failure_stack_ptr; | 3200 fail_stack_elt_t *failure_stack_ptr; |
3201 | 3201 |
3202 /* Assume that each path through the pattern can be null until | 3202 /* Assume that each path through the pattern can be null until |
3203 proven otherwise. We set this false at the bottom of switch | 3203 proven otherwise. We set this false at the bottom of switch |
3204 statement, to which we get only if a particular path doesn't | 3204 statement, to which we get only if a particular path doesn't |
3205 match the empty string. */ | 3205 match the empty string. */ |
3206 boolean path_can_be_null = true; | 3206 boolean path_can_be_null = true; |
3207 | 3207 |
3208 /* We aren't doing a `succeed_n' to begin with. */ | 3208 /* We aren't doing a `succeed_n' to begin with. */ |
3209 boolean succeed_n_p = false; | 3209 boolean succeed_n_p = false; |
3210 | 3210 |
3211 /* If all elements for base leading-codes in fastmap is set, this | 3211 /* If all elements for base leading-codes in fastmap is set, this |
3212 flag is set true. */ | 3212 flag is set true. */ |
3213 boolean match_any_multibyte_characters = false; | 3213 boolean match_any_multibyte_characters = false; |
3214 | 3214 |
3215 /* Maximum code of simple (single byte) character. */ | 3215 /* Maximum code of simple (single byte) character. */ |
3216 int simple_char_max; | 3216 int simple_char_max; |
3217 | 3217 |
3218 assert (fastmap != NULL && p != NULL); | 3218 assert (fastmap != NULL && p != NULL); |
3219 | 3219 |
3220 INIT_FAIL_STACK (); | 3220 INIT_FAIL_STACK (); |
3221 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ | 3221 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ |
3222 bufp->fastmap_accurate = 1; /* It will be when we're done. */ | 3222 bufp->fastmap_accurate = 1; /* It will be when we're done. */ |
3223 bufp->can_be_null = 0; | 3223 bufp->can_be_null = 0; |
3224 | 3224 |
3225 while (1) | 3225 while (1) |
3226 { | 3226 { |
3240 } | 3240 } |
3241 else | 3241 else |
3242 break; | 3242 break; |
3243 } | 3243 } |
3244 | 3244 |
3245 /* We should never be about to go beyond the end of the pattern. */ | 3245 /* We should never be about to go beyond the end of the pattern. */ |
3246 assert (p < pend); | 3246 assert (p < pend); |
3247 | 3247 |
3248 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) | 3248 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) |
3249 { | 3249 { |
3250 | 3250 |
3251 /* I guess the idea here is to simply not bother with a fastmap | 3251 /* I guess the idea here is to simply not bother with a fastmap |
3252 if a backreference is used, since it's too hard to figure out | 3252 if a backreference is used, since it's too hard to figure out |
3253 the fastmap for the corresponding group. Setting | 3253 the fastmap for the corresponding group. Setting |
3254 `can_be_null' stops `re_search_2' from using the fastmap, so | 3254 `can_be_null' stops `re_search_2' from using the fastmap, so |
3255 that is all we do. */ | 3255 that is all we do. */ |
3256 case duplicate: | 3256 case duplicate: |
3257 bufp->can_be_null = 1; | 3257 bufp->can_be_null = 1; |
3258 goto done; | 3258 goto done; |
3259 | 3259 |
3260 | 3260 |
3261 /* Following are the cases which match a character. These end | 3261 /* Following are the cases which match a character. These end |
3262 with `break'. */ | 3262 with `break'. */ |
3263 | 3263 |
3264 case exactn: | 3264 case exactn: |
3265 fastmap[p[1]] = 1; | 3265 fastmap[p[1]] = 1; |
3266 break; | 3266 break; |
3267 | 3267 |
3268 | 3268 |
3269 #ifndef emacs | 3269 #ifndef emacs |
3270 case charset: | 3270 case charset: |
3271 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | 3271 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) |
3272 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) | 3272 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) |
3273 fastmap[j] = 1; | 3273 fastmap[j] = 1; |
3274 break; | 3274 break; |
3275 | 3275 |
3276 | 3276 |
3277 case charset_not: | 3277 case charset_not: |
3278 /* Chars beyond end of map must be allowed. */ | 3278 /* Chars beyond end of map must be allowed. */ |
3279 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) | 3279 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) |
3280 fastmap[j] = 1; | 3280 fastmap[j] = 1; |
3281 | 3281 |
3282 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | 3282 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) |
3283 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) | 3283 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) |
3284 fastmap[j] = 1; | 3284 fastmap[j] = 1; |
3285 break; | 3285 break; |
3286 | 3286 |
3287 | 3287 |
3288 case wordchar: | 3288 case wordchar: |
3289 for (j = 0; j < (1 << BYTEWIDTH); j++) | 3289 for (j = 0; j < (1 << BYTEWIDTH); j++) |
3290 if (SYNTAX (j) == Sword) | 3290 if (SYNTAX (j) == Sword) |
3313 | 3313 |
3314 /* Make P points the range table. */ | 3314 /* Make P points the range table. */ |
3315 p += CHARSET_BITMAP_SIZE (&p[-2]); | 3315 p += CHARSET_BITMAP_SIZE (&p[-2]); |
3316 | 3316 |
3317 /* Extract the number of ranges in range table into | 3317 /* Extract the number of ranges in range table into |
3318 COUNT. */ | 3318 COUNT. */ |
3319 EXTRACT_NUMBER_AND_INCR (count, p); | 3319 EXTRACT_NUMBER_AND_INCR (count, p); |
3320 for (; count > 0; count--, p += 2 * 3) /* XXX */ | 3320 for (; count > 0; count--, p += 2 * 3) /* XXX */ |
3321 { | 3321 { |
3322 /* Extract the start of each range. */ | 3322 /* Extract the start of each range. */ |
3323 EXTRACT_CHARACTER (c, p); | 3323 EXTRACT_CHARACTER (c, p); |
3363 if (SYNTAX (j) == Sword) | 3363 if (SYNTAX (j) == Sword) |
3364 fastmap[j] = 1; | 3364 fastmap[j] = 1; |
3365 | 3365 |
3366 if (bufp->multibyte) | 3366 if (bufp->multibyte) |
3367 /* Any character set can possibly contain a character | 3367 /* Any character set can possibly contain a character |
3368 whose syntax is `Sword'. */ | 3368 whose syntax is `Sword'. */ |
3369 goto set_fastmap_for_multibyte_characters; | 3369 goto set_fastmap_for_multibyte_characters; |
3370 break; | 3370 break; |
3371 | 3371 |
3372 | 3372 |
3373 case notwordchar: | 3373 case notwordchar: |
3381 whose syntax is not `Sword'. */ | 3381 whose syntax is not `Sword'. */ |
3382 goto set_fastmap_for_multibyte_characters; | 3382 goto set_fastmap_for_multibyte_characters; |
3383 break; | 3383 break; |
3384 #endif | 3384 #endif |
3385 | 3385 |
3386 case anychar: | 3386 case anychar: |
3387 { | 3387 { |
3388 int fastmap_newline = fastmap['\n']; | 3388 int fastmap_newline = fastmap['\n']; |
3389 | 3389 |
3390 /* `.' matches anything (but if bufp->multibyte is | 3390 /* `.' matches anything (but if bufp->multibyte is |
3391 nonzero, matches `\000' .. `\127' and possible multibyte | 3391 nonzero, matches `\000' .. `\127' and possible multibyte |
3408 /* ... except perhaps newline. */ | 3408 /* ... except perhaps newline. */ |
3409 if (!(bufp->syntax & RE_DOT_NEWLINE)) | 3409 if (!(bufp->syntax & RE_DOT_NEWLINE)) |
3410 fastmap['\n'] = fastmap_newline; | 3410 fastmap['\n'] = fastmap_newline; |
3411 | 3411 |
3412 /* Return if we have already set `can_be_null'; if we have, | 3412 /* Return if we have already set `can_be_null'; if we have, |
3413 then the fastmap is irrelevant. Something's wrong here. */ | 3413 then the fastmap is irrelevant. Something's wrong here. */ |
3414 else if (bufp->can_be_null) | 3414 else if (bufp->can_be_null) |
3415 goto done; | 3415 goto done; |
3416 | 3416 |
3417 /* Otherwise, have to check alternative paths. */ | 3417 /* Otherwise, have to check alternative paths. */ |
3418 break; | 3418 break; |
3422 case wordbound: | 3422 case wordbound: |
3423 case notwordbound: | 3423 case notwordbound: |
3424 case wordbeg: | 3424 case wordbeg: |
3425 case wordend: | 3425 case wordend: |
3426 case notsyntaxspec: | 3426 case notsyntaxspec: |
3427 case syntaxspec: | 3427 case syntaxspec: |
3428 /* This match depends on text properties. These end with | 3428 /* This match depends on text properties. These end with |
3429 aborting optimizations. */ | 3429 aborting optimizations. */ |
3430 bufp->can_be_null = 1; | 3430 bufp->can_be_null = 1; |
3431 goto done; | 3431 goto done; |
3432 #if 0 | 3432 #if 0 |
3433 k = *p++; | 3433 k = *p++; |
3434 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH); | 3434 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH); |
3435 for (j = 0; j < simple_char_max; j++) | 3435 for (j = 0; j < simple_char_max; j++) |
3436 if (SYNTAX (j) == (enum syntaxcode) k) | 3436 if (SYNTAX (j) == (enum syntaxcode) k) |
3478 if (!CHAR_HAS_CATEGORY (j, k)) | 3478 if (!CHAR_HAS_CATEGORY (j, k)) |
3479 fastmap[j] = 1; | 3479 fastmap[j] = 1; |
3480 | 3480 |
3481 if (bufp->multibyte) | 3481 if (bufp->multibyte) |
3482 /* Any character set can possibly contain a character | 3482 /* Any character set can possibly contain a character |
3483 whose category is not K. */ | 3483 whose category is not K. */ |
3484 goto set_fastmap_for_multibyte_characters; | 3484 goto set_fastmap_for_multibyte_characters; |
3485 break; | 3485 break; |
3486 | 3486 |
3487 /* All cases after this match the empty string. These end with | 3487 /* All cases after this match the empty string. These end with |
3488 `continue'. */ | 3488 `continue'. */ |
3489 | 3489 |
3490 | 3490 |
3491 case before_dot: | 3491 case before_dot: |
3492 case at_dot: | 3492 case at_dot: |
3493 case after_dot: | 3493 case after_dot: |
3494 continue; | 3494 continue; |
3495 #endif /* emacs */ | 3495 #endif /* emacs */ |
3496 | 3496 |
3497 | 3497 |
3498 case no_op: | 3498 case no_op: |
3499 case begline: | 3499 case begline: |
3500 case endline: | 3500 case endline: |
3501 case begbuf: | 3501 case begbuf: |
3502 case endbuf: | 3502 case endbuf: |
3503 #ifndef emacs | 3503 #ifndef emacs |
3504 case wordbound: | 3504 case wordbound: |
3505 case notwordbound: | 3505 case notwordbound: |
3506 case wordbeg: | 3506 case wordbeg: |
3507 case wordend: | 3507 case wordend: |
3508 #endif | 3508 #endif |
3509 case push_dummy_failure: | 3509 case push_dummy_failure: |
3510 continue; | 3510 continue; |
3511 | 3511 |
3512 | 3512 |
3513 case jump_n: | 3513 case jump_n: |
3514 case pop_failure_jump: | 3514 case pop_failure_jump: |
3515 case maybe_pop_jump: | 3515 case maybe_pop_jump: |
3516 case jump: | 3516 case jump: |
3517 case jump_past_alt: | 3517 case jump_past_alt: |
3518 case dummy_failure_jump: | 3518 case dummy_failure_jump: |
3519 EXTRACT_NUMBER_AND_INCR (j, p); | 3519 EXTRACT_NUMBER_AND_INCR (j, p); |
3520 p += j; | 3520 p += j; |
3521 if (j > 0) | 3521 if (j > 0) |
3522 continue; | 3522 continue; |
3523 | 3523 |
3524 /* Jump backward implies we just went through the body of a | 3524 /* Jump backward implies we just went through the body of a |
3525 loop and matched nothing. Opcode jumped to should be | 3525 loop and matched nothing. Opcode jumped to should be |
3526 `on_failure_jump' or `succeed_n'. Just treat it like an | 3526 `on_failure_jump' or `succeed_n'. Just treat it like an |
3527 ordinary jump. For a * loop, it has pushed its failure | 3527 ordinary jump. For a * loop, it has pushed its failure |
3528 point already; if so, discard that as redundant. */ | 3528 point already; if so, discard that as redundant. */ |
3529 if ((re_opcode_t) *p != on_failure_jump | 3529 if ((re_opcode_t) *p != on_failure_jump |
3530 && (re_opcode_t) *p != succeed_n) | 3530 && (re_opcode_t) *p != succeed_n) |
3531 continue; | 3531 continue; |
3532 | 3532 |
3533 p++; | 3533 p++; |
3534 EXTRACT_NUMBER_AND_INCR (j, p); | 3534 EXTRACT_NUMBER_AND_INCR (j, p); |
3535 p += j; | 3535 p += j; |
3536 | 3536 |
3537 /* If what's on the stack is where we are now, pop it. */ | 3537 /* If what's on the stack is where we are now, pop it. */ |
3538 if (!FAIL_STACK_EMPTY () | 3538 if (!FAIL_STACK_EMPTY () |
3539 && fail_stack.stack[fail_stack.avail - 1].pointer == p) | 3539 && fail_stack.stack[fail_stack.avail - 1].pointer == p) |
3540 fail_stack.avail--; | 3540 fail_stack.avail--; |
3541 | 3541 |
3542 continue; | 3542 continue; |
3543 | 3543 |
3544 | 3544 |
3545 case on_failure_jump: | 3545 case on_failure_jump: |
3546 case on_failure_keep_string_jump: | 3546 case on_failure_keep_string_jump: |
3547 handle_on_failure_jump: | 3547 handle_on_failure_jump: |
3548 EXTRACT_NUMBER_AND_INCR (j, p); | 3548 EXTRACT_NUMBER_AND_INCR (j, p); |
3549 | 3549 |
3550 /* For some patterns, e.g., `(a?)?', `p+j' here points to the | 3550 /* For some patterns, e.g., `(a?)?', `p+j' here points to the |
3551 end of the pattern. We don't want to push such a point, | 3551 end of the pattern. We don't want to push such a point, |
3552 since when we restore it above, entering the switch will | 3552 since when we restore it above, entering the switch will |
3553 increment `p' past the end of the pattern. We don't need | 3553 increment `p' past the end of the pattern. We don't need |
3554 to push such a point since we obviously won't find any more | 3554 to push such a point since we obviously won't find any more |
3555 fastmap entries beyond `pend'. Such a pattern can match | 3555 fastmap entries beyond `pend'. Such a pattern can match |
3556 the null string, though. */ | 3556 the null string, though. */ |
3557 if (p + j < pend) | 3557 if (p + j < pend) |
3558 { | 3558 { |
3559 if (!PUSH_PATTERN_OP (p + j, fail_stack)) | 3559 if (!PUSH_PATTERN_OP (p + j, fail_stack)) |
3560 { | 3560 { |
3561 RESET_FAIL_STACK (); | 3561 RESET_FAIL_STACK (); |
3562 return -2; | 3562 return -2; |
3563 } | 3563 } |
3564 } | |
3565 else | |
3566 bufp->can_be_null = 1; | |
3567 | |
3568 if (succeed_n_p) | |
3569 { | |
3570 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ | |
3571 succeed_n_p = false; | |
3572 } | 3564 } |
3573 | 3565 else |
3574 continue; | 3566 bufp->can_be_null = 1; |
3567 | |
3568 if (succeed_n_p) | |
3569 { | |
3570 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ | |
3571 succeed_n_p = false; | |
3572 } | |
3573 | |
3574 continue; | |
3575 | 3575 |
3576 | 3576 |
3577 case succeed_n: | 3577 case succeed_n: |
3578 /* Get to the number of times to succeed. */ | 3578 /* Get to the number of times to succeed. */ |
3579 p += 2; | 3579 p += 2; |
3580 | 3580 |
3581 /* Increment p past the n for when k != 0. */ | 3581 /* Increment p past the n for when k != 0. */ |
3582 EXTRACT_NUMBER_AND_INCR (k, p); | 3582 EXTRACT_NUMBER_AND_INCR (k, p); |
3583 if (k == 0) | 3583 if (k == 0) |
3584 { | 3584 { |
3585 p -= 4; | 3585 p -= 4; |
3586 succeed_n_p = true; /* Spaghetti code alert. */ | 3586 succeed_n_p = true; /* Spaghetti code alert. */ |
3587 goto handle_on_failure_jump; | 3587 goto handle_on_failure_jump; |
3588 } | 3588 } |
3589 continue; | 3589 continue; |
3590 | 3590 |
3591 | 3591 |
3592 case set_number_at: | 3592 case set_number_at: |
3593 p += 4; | 3593 p += 4; |
3594 continue; | 3594 continue; |
3595 | 3595 |
3596 | 3596 |
3597 case start_memory: | 3597 case start_memory: |
3598 case stop_memory: | 3598 case stop_memory: |
3599 p += 2; | 3599 p += 2; |
3600 continue; | 3600 continue; |
3601 | 3601 |
3602 | 3602 |
3603 default: | 3603 default: |
3604 abort (); /* We have listed all the cases. */ | 3604 abort (); /* We have listed all the cases. */ |
3605 } /* switch *p++ */ | 3605 } /* switch *p++ */ |
3606 | 3606 |
3607 /* Getting here means we have found the possible starting | 3607 /* Getting here means we have found the possible starting |
3608 characters for one path of the pattern -- and that the empty | 3608 characters for one path of the pattern -- and that the empty |
3609 string does not match. We need not follow this path further. | 3609 string does not match. We need not follow this path further. |
3610 Instead, look at the next alternative (remembered on the | 3610 Instead, look at the next alternative (remembered on the |
3611 stack), or quit if no more. The test at the top of the loop | 3611 stack), or quit if no more. The test at the top of the loop |
3612 does these things. */ | 3612 does these things. */ |
3613 path_can_be_null = false; | 3613 path_can_be_null = false; |
3614 p = pend; | 3614 p = pend; |
3615 } /* while p */ | 3615 } /* while p */ |
3616 | 3616 |
3617 /* Set `can_be_null' for the last path (also the first path, if the | 3617 /* Set `can_be_null' for the last path (also the first path, if the |
3618 pattern is empty). */ | 3618 pattern is empty). */ |
3619 bufp->can_be_null |= path_can_be_null; | 3619 bufp->can_be_null |= path_can_be_null; |
3620 | 3620 |
3621 done: | 3621 done: |
3622 RESET_FAIL_STACK (); | 3622 RESET_FAIL_STACK (); |
3623 return 0; | 3623 return 0; |
3656 regs->num_regs = 0; | 3656 regs->num_regs = 0; |
3657 regs->start = regs->end = (regoff_t *) 0; | 3657 regs->start = regs->end = (regoff_t *) 0; |
3658 } | 3658 } |
3659 } | 3659 } |
3660 | 3660 |
3661 /* Searching routines. */ | 3661 /* Searching routines. */ |
3662 | 3662 |
3663 /* Like re_search_2, below, but only one string is specified, and | 3663 /* Like re_search_2, below, but only one string is specified, and |
3664 doesn't let you say where to stop matching. */ | 3664 doesn't let you say where to stop matching. */ |
3665 | 3665 |
3666 int | 3666 int |
3718 register RE_TRANSLATE_TYPE translate = bufp->translate; | 3718 register RE_TRANSLATE_TYPE translate = bufp->translate; |
3719 int total_size = size1 + size2; | 3719 int total_size = size1 + size2; |
3720 int endpos = startpos + range; | 3720 int endpos = startpos + range; |
3721 int anchored_start = 0; | 3721 int anchored_start = 0; |
3722 | 3722 |
3723 /* Nonzero if we have to concern multibyte character. */ | 3723 /* Nonzero if we have to concern multibyte character. */ |
3724 int multibyte = bufp->multibyte; | 3724 int multibyte = bufp->multibyte; |
3725 | 3725 |
3726 /* Check for out-of-range STARTPOS. */ | 3726 /* Check for out-of-range STARTPOS. */ |
3727 if (startpos < 0 || startpos > total_size) | 3727 if (startpos < 0 || startpos > total_size) |
3728 return -1; | 3728 return -1; |
3765 if (bufp->buffer[0] == begline) | 3765 if (bufp->buffer[0] == begline) |
3766 anchored_start = 1; | 3766 anchored_start = 1; |
3767 | 3767 |
3768 #ifdef emacs | 3768 #ifdef emacs |
3769 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, | 3769 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, |
3770 POS_AS_IN_BUFFER (startpos > 0 | 3770 POS_AS_IN_BUFFER (startpos > 0 |
3771 ? startpos - 1 : startpos), | 3771 ? startpos - 1 : startpos), |
3772 1); | 3772 1); |
3773 #endif | 3773 #endif |
3774 | 3774 |
3775 /* Loop through the string, looking for a place to start matching. */ | 3775 /* Loop through the string, looking for a place to start matching. */ |
3787 == '\n'))) | 3787 == '\n'))) |
3788 goto advance; | 3788 goto advance; |
3789 } | 3789 } |
3790 | 3790 |
3791 /* If a fastmap is supplied, skip quickly over characters that | 3791 /* If a fastmap is supplied, skip quickly over characters that |
3792 cannot be the start of a match. If the pattern can match the | 3792 cannot be the start of a match. If the pattern can match the |
3793 null string, however, we don't need to skip characters; we want | 3793 null string, however, we don't need to skip characters; we want |
3794 the first null string. */ | 3794 the first null string. */ |
3795 if (fastmap && startpos < total_size && !bufp->can_be_null) | 3795 if (fastmap && startpos < total_size && !bufp->can_be_null) |
3796 { | 3796 { |
3797 if (range > 0) /* Searching forwards. */ | 3797 if (range > 0) /* Searching forwards. */ |
3798 { | 3798 { |
3799 register const char *d; | 3799 register const char *d; |
3800 register int lim = 0; | 3800 register int lim = 0; |
3801 int irange = range; | 3801 int irange = range; |
3802 | 3802 |
3803 if (startpos < size1 && startpos + range >= size1) | 3803 if (startpos < size1 && startpos + range >= size1) |
3804 lim = range - (size1 - startpos); | 3804 lim = range - (size1 - startpos); |
3805 | 3805 |
3806 d = POS_ADDR_VSTRING (startpos); | 3806 d = POS_ADDR_VSTRING (startpos); |
3807 | 3807 |
3808 /* Written out as an if-else to avoid testing `translate' | 3808 /* Written out as an if-else to avoid testing `translate' |
3809 inside the loop. */ | 3809 inside the loop. */ |
3810 if (translate) | 3810 if (translate) |
3811 while (range > lim | 3811 while (range > lim |
3812 && !fastmap[(unsigned char) | 3812 && !fastmap[(unsigned char) |
3813 translate[(unsigned char) *d++]]) | 3813 translate[(unsigned char) *d++]]) |
3814 range--; | 3814 range--; |
3815 else | 3815 else |
3816 while (range > lim && !fastmap[(unsigned char) *d++]) | 3816 while (range > lim && !fastmap[(unsigned char) *d++]) |
3817 range--; | 3817 range--; |
3818 | 3818 |
3819 startpos += irange - range; | 3819 startpos += irange - range; |
3820 } | 3820 } |
3821 else /* Searching backwards. */ | 3821 else /* Searching backwards. */ |
3822 { | 3822 { |
3823 register char c = (size1 == 0 || startpos >= size1 | 3823 register char c = (size1 == 0 || startpos >= size1 |
3824 ? string2[startpos - size1] | 3824 ? string2[startpos - size1] |
3825 : string1[startpos]); | 3825 : string1[startpos]); |
3826 | 3826 |
3827 if (!fastmap[(unsigned char) TRANSLATE (c)]) | 3827 if (!fastmap[(unsigned char) TRANSLATE (c)]) |
3828 goto advance; | 3828 goto advance; |
3829 } | 3829 } |
3830 } | 3830 } |
3831 | 3831 |
3832 /* If can't match the null string, and that's all we have left, fail. */ | 3832 /* If can't match the null string, and that's all we have left, fail. */ |
3833 if (range >= 0 && startpos == total_size && fastmap | 3833 if (range >= 0 && startpos == total_size && fastmap |
3834 && !bufp->can_be_null) | 3834 && !bufp->can_be_null) |
3835 return -1; | 3835 return -1; |
3836 | 3836 |
3837 val = re_match_2_internal (bufp, string1, size1, string2, size2, | 3837 val = re_match_2_internal (bufp, string1, size1, string2, size2, |
3838 startpos, regs, stop); | 3838 startpos, regs, stop); |
3839 #ifndef REGEX_MALLOC | 3839 #ifndef REGEX_MALLOC |
3848 if (val == -2) | 3848 if (val == -2) |
3849 return -2; | 3849 return -2; |
3850 | 3850 |
3851 advance: | 3851 advance: |
3852 if (!range) | 3852 if (!range) |
3853 break; | 3853 break; |
3854 else if (range > 0) | 3854 else if (range > 0) |
3855 { | 3855 { |
3856 /* Update STARTPOS to the next character boundary. */ | 3856 /* Update STARTPOS to the next character boundary. */ |
3857 if (multibyte) | 3857 if (multibyte) |
3858 { | 3858 { |
3859 const unsigned char *p = POS_ADDR_VSTRING (startpos); | 3859 const unsigned char *p = POS_ADDR_VSTRING (startpos); |
3860 const unsigned char *pend = STOP_ADDR_VSTRING (startpos); | 3860 const unsigned char *pend = STOP_ADDR_VSTRING (startpos); |
3865 break; | 3865 break; |
3866 startpos += len; | 3866 startpos += len; |
3867 } | 3867 } |
3868 else | 3868 else |
3869 { | 3869 { |
3870 range--; | 3870 range--; |
3871 startpos++; | 3871 startpos++; |
3872 } | 3872 } |
3873 } | 3873 } |
3874 else | 3874 else |
3875 { | 3875 { |
3876 range++; | 3876 range++; |
3877 startpos--; | 3877 startpos--; |
3878 | 3878 |
3879 /* Update STARTPOS to the previous character boundary. */ | 3879 /* Update STARTPOS to the previous character boundary. */ |
3880 if (multibyte) | 3880 if (multibyte) |
3881 { | 3881 { |
3882 const unsigned char *p = POS_ADDR_VSTRING (startpos); | 3882 const unsigned char *p = POS_ADDR_VSTRING (startpos); |
3898 break; | 3898 break; |
3899 | 3899 |
3900 startpos -= len; | 3900 startpos -= len; |
3901 } | 3901 } |
3902 } | 3902 } |
3903 } | 3903 } |
3904 } | 3904 } |
3905 return -1; | 3905 return -1; |
3906 } /* re_search_2 */ | 3906 } /* re_search_2 */ |
3907 | 3907 |
3908 /* Declarations and macros for re_match_2. */ | 3908 /* Declarations and macros for re_match_2. */ |
3909 | 3909 |
3910 static int bcmp_translate (); | 3910 static int bcmp_translate (); |
3911 static boolean alt_match_null_string_p (), | 3911 static boolean alt_match_null_string_p (), |
3912 common_op_match_null_string_p (), | 3912 common_op_match_null_string_p (), |
3913 group_match_null_string_p (); | 3913 group_match_null_string_p (); |
3914 | 3914 |
3915 /* This converts PTR, a pointer into one of the search strings `string1' | 3915 /* This converts PTR, a pointer into one of the search strings `string1' |
3916 and `string2' into an offset from the beginning of that string. */ | 3916 and `string2' into an offset from the beginning of that string. */ |
3917 #define POINTER_TO_OFFSET(ptr) \ | 3917 #define POINTER_TO_OFFSET(ptr) \ |
3918 (FIRST_STRING_P (ptr) \ | 3918 (FIRST_STRING_P (ptr) \ |
3924 #define MATCHING_IN_FIRST_STRING (dend == end_match_1) | 3924 #define MATCHING_IN_FIRST_STRING (dend == end_match_1) |
3925 | 3925 |
3926 /* Call before fetching a character with *d. This switches over to | 3926 /* Call before fetching a character with *d. This switches over to |
3927 string2 if necessary. */ | 3927 string2 if necessary. */ |
3928 #define PREFETCH() \ | 3928 #define PREFETCH() \ |
3929 while (d == dend) \ | 3929 while (d == dend) \ |
3930 { \ | 3930 { \ |
3931 /* End of string2 => fail. */ \ | 3931 /* End of string2 => fail. */ \ |
3932 if (dend == end_match_2) \ | 3932 if (dend == end_match_2) \ |
3933 goto fail; \ | 3933 goto fail; \ |
3934 /* End of string1 => advance to string2. */ \ | 3934 /* End of string1 => advance to string2. */ \ |
3935 d = string2; \ | 3935 d = string2; \ |
3936 dend = end_match_2; \ | 3936 dend = end_match_2; \ |
3937 } | 3937 } |
3938 | 3938 |
3939 | 3939 |
3940 /* Test if at very beginning or at very end of the virtual concatenation | 3940 /* Test if at very beginning or at very end of the virtual concatenation |
3941 of `string1' and `string2'. If only one string, it's `string2'. */ | 3941 of `string1' and `string2'. If only one string, it's `string2'. */ |
3942 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) | 3942 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) |
3943 #define AT_STRINGS_END(d) ((d) == end2) | 3943 #define AT_STRINGS_END(d) ((d) == end2) |
3944 | 3944 |
3945 | 3945 |
3946 /* Test if D points to a character which is word-constituent. We have | 3946 /* Test if D points to a character which is word-constituent. We have |
3947 two special cases to check for: if past the end of string1, look at | 3947 two special cases to check for: if past the end of string1, look at |
3948 the first character in string2; and if before the beginning of | 3948 the first character in string2; and if before the beginning of |
3949 string2, look at the last character in string1. */ | 3949 string2, look at the last character in string1. */ |
3950 #define WORDCHAR_P(d) \ | 3950 #define WORDCHAR_P(d) \ |
3951 (SYNTAX ((d) == end1 ? *string2 \ | 3951 (SYNTAX ((d) == end1 ? *string2 \ |
3952 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \ | 3952 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \ |
3953 == Sword) | 3953 == Sword) |
3954 | 3954 |
3955 /* Disabled due to a compiler bug -- see comment at case wordbound */ | 3955 /* Disabled due to a compiler bug -- see comment at case wordbound */ |
3956 | 3956 |
3957 /* The comment at case wordbound is following one, but we don't use | 3957 /* The comment at case wordbound is following one, but we don't use |
3958 AT_WORD_BOUNDARY anymore to support multibyte form. | 3958 AT_WORD_BOUNDARY anymore to support multibyte form. |
3959 | 3959 |
3960 The DEC Alpha C compiler 3.x generates incorrect code for the | 3960 The DEC Alpha C compiler 3.x generates incorrect code for the |
3961 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of | 3961 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of |
3962 AT_WORD_BOUNDARY, so this code is disabled. Expanding the | 3962 AT_WORD_BOUNDARY, so this code is disabled. Expanding the |
3963 macro and introducing temporary variables works around the bug. */ | 3963 macro and introducing temporary variables works around the bug. */ |
3964 | 3964 |
3965 #if 0 | 3965 #if 0 |
3966 /* Test if the character before D and the one at D differ with respect | 3966 /* Test if the character before D and the one at D differ with respect |
3967 to being word-constituent. */ | 3967 to being word-constituent. */ |
3988 } while (0) | 3988 } while (0) |
3989 #else | 3989 #else |
3990 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */ | 3990 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */ |
3991 #endif /* not MATCH_MAY_ALLOCATE */ | 3991 #endif /* not MATCH_MAY_ALLOCATE */ |
3992 | 3992 |
3993 /* These values must meet several constraints. They must not be valid | 3993 /* These values must meet several constraints. They must not be valid |
3994 register values; since we have a limit of 255 registers (because | 3994 register values; since we have a limit of 255 registers (because |
3995 we use only one byte in the pattern for the register number), we can | 3995 we use only one byte in the pattern for the register number), we can |
3996 use numbers larger than 255. They must differ by 1, because of | 3996 use numbers larger than 255. They must differ by 1, because of |
3997 NUM_FAILURE_ITEMS above. And the value for the lowest register must | 3997 NUM_FAILURE_ITEMS above. And the value for the lowest register must |
3998 be larger than the value for the highest register, so we do not try | 3998 be larger than the value for the highest register, so we do not try |
3999 to actually save any registers when none are active. */ | 3999 to actually save any registers when none are active. */ |
4000 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) | 4000 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) |
4001 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) | 4001 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) |
4002 | 4002 |
4003 /* Matching routines. */ | 4003 /* Matching routines. */ |
4004 | 4004 |
4005 #ifndef emacs /* Emacs never uses this. */ | 4005 #ifndef emacs /* Emacs never uses this. */ |
4006 /* re_match is like re_match_2 except it takes only a single string. */ | 4006 /* re_match is like re_match_2 except it takes only a single string. */ |
4007 | 4007 |
4008 int | 4008 int |
4009 re_match (bufp, string, size, pos, regs) | 4009 re_match (bufp, string, size, pos, regs) |
4010 struct re_pattern_buffer *bufp; | 4010 struct re_pattern_buffer *bufp; |
4019 } | 4019 } |
4020 #endif /* not emacs */ | 4020 #endif /* not emacs */ |
4021 | 4021 |
4022 #ifdef emacs | 4022 #ifdef emacs |
4023 /* In Emacs, this is the string or buffer in which we | 4023 /* In Emacs, this is the string or buffer in which we |
4024 are matching. It is used for looking up syntax properties. */ | 4024 are matching. It is used for looking up syntax properties. */ |
4025 Lisp_Object re_match_object; | 4025 Lisp_Object re_match_object; |
4026 #endif | 4026 #endif |
4027 | 4027 |
4028 /* re_match_2 matches the compiled pattern in BUFP against the | 4028 /* re_match_2 matches the compiled pattern in BUFP against the |
4029 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 | 4029 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 |
4030 and SIZE2, respectively). We start matching at POS, and stop | 4030 and SIZE2, respectively). We start matching at POS, and stop |
4031 matching at STOP. | 4031 matching at STOP. |
4032 | 4032 |
4033 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we | 4033 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we |
4034 store offsets for the substring each group matched in REGS. See the | 4034 store offsets for the substring each group matched in REGS. See the |
4035 documentation for exactly how many groups we fill. | 4035 documentation for exactly how many groups we fill. |
4036 | 4036 |
4037 We return -1 if no match, -2 if an internal error (such as the | 4037 We return -1 if no match, -2 if an internal error (such as the |
4038 failure stack overflowing). Otherwise, we return the length of the | 4038 failure stack overflowing). Otherwise, we return the length of the |
4039 matched substring. */ | 4039 matched substring. */ |
4040 | 4040 |
4041 int | 4041 int |
4042 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) | 4042 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) |
4043 struct re_pattern_buffer *bufp; | 4043 struct re_pattern_buffer *bufp; |
4046 int pos; | 4046 int pos; |
4047 struct re_registers *regs; | 4047 struct re_registers *regs; |
4048 int stop; | 4048 int stop; |
4049 { | 4049 { |
4050 int result; | 4050 int result; |
4051 | 4051 |
4052 #ifdef emacs | 4052 #ifdef emacs |
4053 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, | 4053 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, |
4054 POS_AS_IN_BUFFER (pos > 0 ? pos - 1 : pos), | 4054 POS_AS_IN_BUFFER (pos > 0 ? pos - 1 : pos), |
4055 1); | 4055 1); |
4056 #endif | 4056 #endif |
4057 | 4057 |
4058 result = re_match_2_internal (bufp, string1, size1, string2, size2, | 4058 result = re_match_2_internal (bufp, string1, size1, string2, size2, |
4059 pos, regs, stop); | 4059 pos, regs, stop); |
4060 alloca (0); | 4060 alloca (0); |
4061 return result; | 4061 return result; |
4062 } | 4062 } |
4063 | 4063 |
4064 /* This is a separate function so that we can force an alloca cleanup | 4064 /* This is a separate function so that we can force an alloca cleanup |
4065 afterwards. */ | 4065 afterwards. */ |
4066 static int | 4066 static int |
4067 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop) | 4067 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop) |
4068 struct re_pattern_buffer *bufp; | 4068 struct re_pattern_buffer *bufp; |
4069 const char *string1, *string2; | 4069 const char *string1, *string2; |
4070 int size1, size2; | 4070 int size1, size2; |
4078 | 4078 |
4079 /* Just past the end of the corresponding string. */ | 4079 /* Just past the end of the corresponding string. */ |
4080 const char *end1, *end2; | 4080 const char *end1, *end2; |
4081 | 4081 |
4082 /* Pointers into string1 and string2, just past the last characters in | 4082 /* Pointers into string1 and string2, just past the last characters in |
4083 each to consider matching. */ | 4083 each to consider matching. */ |
4084 const char *end_match_1, *end_match_2; | 4084 const char *end_match_1, *end_match_2; |
4085 | 4085 |
4086 /* Where we are in the data, and the end of the current string. */ | 4086 /* Where we are in the data, and the end of the current string. */ |
4087 const char *d, *dend; | 4087 const char *d, *dend; |
4088 | 4088 |
4092 | 4092 |
4093 /* Mark the opcode just after a start_memory, so we can test for an | 4093 /* Mark the opcode just after a start_memory, so we can test for an |
4094 empty subpattern when we get to the stop_memory. */ | 4094 empty subpattern when we get to the stop_memory. */ |
4095 unsigned char *just_past_start_mem = 0; | 4095 unsigned char *just_past_start_mem = 0; |
4096 | 4096 |
4097 /* We use this to map every character in the string. */ | 4097 /* We use this to map every character in the string. */ |
4098 RE_TRANSLATE_TYPE translate = bufp->translate; | 4098 RE_TRANSLATE_TYPE translate = bufp->translate; |
4099 | 4099 |
4100 /* Nonzero if we have to concern multibyte character. */ | 4100 /* Nonzero if we have to concern multibyte character. */ |
4101 int multibyte = bufp->multibyte; | 4101 int multibyte = bufp->multibyte; |
4102 | 4102 |
4103 /* Failure point stack. Each place that can handle a failure further | 4103 /* Failure point stack. Each place that can handle a failure further |
4104 down the line pushes a failure point on this stack. It consists of | 4104 down the line pushes a failure point on this stack. It consists of |
4105 restart, regend, and reg_info for all registers corresponding to | 4105 restart, regend, and reg_info for all registers corresponding to |
4106 the subexpressions we're currently inside, plus the number of such | 4106 the subexpressions we're currently inside, plus the number of such |
4107 registers, and, finally, two char *'s. The first char * is where | 4107 registers, and, finally, two char *'s. The first char * is where |
4108 to resume scanning the pattern; the second one is where to resume | 4108 to resume scanning the pattern; the second one is where to resume |
4109 scanning the strings. If the latter is zero, the failure point is | 4109 scanning the strings. If the latter is zero, the failure point is |
4110 a ``dummy''; if a failure happens and the failure point is a dummy, | 4110 a ``dummy''; if a failure happens and the failure point is a dummy, |
4111 it gets discarded and the next next one is tried. */ | 4111 it gets discarded and the next next one is tried. */ |
4112 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ | 4112 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ |
4113 fail_stack_type fail_stack; | 4113 fail_stack_type fail_stack; |
4114 #endif | 4114 #endif |
4115 #ifdef DEBUG | 4115 #ifdef DEBUG |
4116 static unsigned failure_id = 0; | 4116 static unsigned failure_id = 0; |
4117 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0; | 4117 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0; |
4120 /* This holds the pointer to the failure stack, when | 4120 /* This holds the pointer to the failure stack, when |
4121 it is allocated relocatably. */ | 4121 it is allocated relocatably. */ |
4122 fail_stack_elt_t *failure_stack_ptr; | 4122 fail_stack_elt_t *failure_stack_ptr; |
4123 | 4123 |
4124 /* We fill all the registers internally, independent of what we | 4124 /* We fill all the registers internally, independent of what we |
4125 return, for use in backreferences. The number here includes | 4125 return, for use in backreferences. The number here includes |
4126 an element for register zero. */ | 4126 an element for register zero. */ |
4127 unsigned num_regs = bufp->re_nsub + 1; | 4127 unsigned num_regs = bufp->re_nsub + 1; |
4128 | 4128 |
4129 /* The currently active registers. */ | 4129 /* The currently active registers. */ |
4130 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG; | 4130 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
4153 /* The is_active field of reg_info helps us keep track of which (possibly | 4153 /* The is_active field of reg_info helps us keep track of which (possibly |
4154 nested) subexpressions we are currently in. The matched_something | 4154 nested) subexpressions we are currently in. The matched_something |
4155 field of reg_info[reg_num] helps us tell whether or not we have | 4155 field of reg_info[reg_num] helps us tell whether or not we have |
4156 matched any of the pattern so far this time through the reg_num-th | 4156 matched any of the pattern so far this time through the reg_num-th |
4157 subexpression. These two fields get reset each time through any | 4157 subexpression. These two fields get reset each time through any |
4158 loop their register is in. */ | 4158 loop their register is in. */ |
4159 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ | 4159 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ |
4160 register_info_type *reg_info; | 4160 register_info_type *reg_info; |
4161 #endif | 4161 #endif |
4162 | 4162 |
4163 /* The following record the register info as found in the above | 4163 /* The following record the register info as found in the above |
4164 variables when we find a match better than any we've seen before. | 4164 variables when we find a match better than any we've seen before. |
4169 const char **best_regstart, **best_regend; | 4169 const char **best_regstart, **best_regend; |
4170 #endif | 4170 #endif |
4171 | 4171 |
4172 /* Logically, this is `best_regend[0]'. But we don't want to have to | 4172 /* Logically, this is `best_regend[0]'. But we don't want to have to |
4173 allocate space for that if we're not allocating space for anything | 4173 allocate space for that if we're not allocating space for anything |
4174 else (see below). Also, we never need info about register 0 for | 4174 else (see below). Also, we never need info about register 0 for |
4175 any of the other register vectors, and it seems rather a kludge to | 4175 any of the other register vectors, and it seems rather a kludge to |
4176 treat `best_regend' differently than the rest. So we keep track of | 4176 treat `best_regend' differently than the rest. So we keep track of |
4177 the end of the best match so far in a separate variable. We | 4177 the end of the best match so far in a separate variable. We |
4178 initialize this to NULL so that when we backtrack the first time | 4178 initialize this to NULL so that when we backtrack the first time |
4179 and need to test it, it's not garbage. */ | 4179 and need to test it, it's not garbage. */ |
4214 reg_info = REGEX_TALLOC (num_regs, register_info_type); | 4214 reg_info = REGEX_TALLOC (num_regs, register_info_type); |
4215 reg_dummy = REGEX_TALLOC (num_regs, const char *); | 4215 reg_dummy = REGEX_TALLOC (num_regs, const char *); |
4216 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); | 4216 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); |
4217 | 4217 |
4218 if (!(regstart && regend && old_regstart && old_regend && reg_info | 4218 if (!(regstart && regend && old_regstart && old_regend && reg_info |
4219 && best_regstart && best_regend && reg_dummy && reg_info_dummy)) | 4219 && best_regstart && best_regend && reg_dummy && reg_info_dummy)) |
4220 { | 4220 { |
4221 FREE_VARIABLES (); | 4221 FREE_VARIABLES (); |
4222 return -2; | 4222 return -2; |
4223 } | 4223 } |
4224 } | 4224 } |
4225 else | 4225 else |
4226 { | 4226 { |
4227 /* We must initialize all our variables to NULL, so that | 4227 /* We must initialize all our variables to NULL, so that |
4228 `FREE_VARIABLES' doesn't try to free them. */ | 4228 `FREE_VARIABLES' doesn't try to free them. */ |
4229 regstart = regend = old_regstart = old_regend = best_regstart | 4229 regstart = regend = old_regstart = old_regend = best_regstart |
4230 = best_regend = reg_dummy = NULL; | 4230 = best_regend = reg_dummy = NULL; |
4231 reg_info = reg_info_dummy = (register_info_type *) NULL; | 4231 reg_info = reg_info_dummy = (register_info_type *) NULL; |
4232 } | 4232 } |
4233 #endif /* MATCH_MAY_ALLOCATE */ | 4233 #endif /* MATCH_MAY_ALLOCATE */ |
4234 | 4234 |
4235 /* The starting position is bogus. */ | 4235 /* The starting position is bogus. */ |
4243 start_memory/stop_memory has been seen for. Also initialize the | 4243 start_memory/stop_memory has been seen for. Also initialize the |
4244 register information struct. */ | 4244 register information struct. */ |
4245 for (mcnt = 1; mcnt < num_regs; mcnt++) | 4245 for (mcnt = 1; mcnt < num_regs; mcnt++) |
4246 { | 4246 { |
4247 regstart[mcnt] = regend[mcnt] | 4247 regstart[mcnt] = regend[mcnt] |
4248 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; | 4248 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; |
4249 | 4249 |
4250 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; | 4250 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; |
4251 IS_ACTIVE (reg_info[mcnt]) = 0; | 4251 IS_ACTIVE (reg_info[mcnt]) = 0; |
4252 MATCHED_SOMETHING (reg_info[mcnt]) = 0; | 4252 MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
4253 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; | 4253 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
4254 } | 4254 } |
4255 | 4255 |
4256 /* We move `string1' into `string2' if the latter's empty -- but not if | 4256 /* We move `string1' into `string2' if the latter's empty -- but not if |
4257 `string1' is null. */ | 4257 `string1' is null. */ |
4258 if (size2 == 0 && string1 != NULL) | 4258 if (size2 == 0 && string1 != NULL) |
4259 { | 4259 { |
4260 string2 = string1; | 4260 string2 = string1; |
4261 size2 = size1; | 4261 size2 = size1; |
4262 string1 = 0; | 4262 string1 = 0; |
4298 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); | 4298 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); |
4299 DEBUG_PRINT1 ("The string to match is: `"); | 4299 DEBUG_PRINT1 ("The string to match is: `"); |
4300 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); | 4300 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); |
4301 DEBUG_PRINT1 ("'\n"); | 4301 DEBUG_PRINT1 ("'\n"); |
4302 | 4302 |
4303 /* This loops over pattern commands. It exits by returning from the | 4303 /* This loops over pattern commands. It exits by returning from the |
4304 function if the match is complete, or it drops through if the match | 4304 function if the match is complete, or it drops through if the match |
4305 fails at this starting point in the input data. */ | 4305 fails at this starting point in the input data. */ |
4306 for (;;) | 4306 for (;;) |
4307 { | 4307 { |
4308 DEBUG_PRINT2 ("\n0x%x: ", p); | 4308 DEBUG_PRINT2 ("\n0x%x: ", p); |
4309 | 4309 |
4310 if (p == pend) | 4310 if (p == pend) |
4311 { /* End of pattern means we might have succeeded. */ | 4311 { /* End of pattern means we might have succeeded. */ |
4312 DEBUG_PRINT1 ("end of pattern ... "); | 4312 DEBUG_PRINT1 ("end of pattern ... "); |
4313 | 4313 |
4314 /* If we haven't matched the entire string, and we want the | 4314 /* If we haven't matched the entire string, and we want the |
4315 longest match, try backtracking. */ | 4315 longest match, try backtracking. */ |
4316 if (d != end_match_2) | 4316 if (d != end_match_2) |
4317 { | 4317 { |
4318 /* 1 if this match ends in the same string (string1 or string2) | 4318 /* 1 if this match ends in the same string (string1 or string2) |
4319 as the best previous match. */ | 4319 as the best previous match. */ |
4320 boolean same_str_p = (FIRST_STRING_P (match_end) | 4320 boolean same_str_p = (FIRST_STRING_P (match_end) |
4321 == MATCHING_IN_FIRST_STRING); | 4321 == MATCHING_IN_FIRST_STRING); |
4322 /* 1 if this match is the best seen so far. */ | 4322 /* 1 if this match is the best seen so far. */ |
4323 boolean best_match_p; | 4323 boolean best_match_p; |
4324 | 4324 |
4325 /* AIX compiler got confused when this was combined | 4325 /* AIX compiler got confused when this was combined |
4326 with the previous declaration. */ | 4326 with the previous declaration. */ |
4327 if (same_str_p) | 4327 if (same_str_p) |
4328 best_match_p = d > match_end; | 4328 best_match_p = d > match_end; |
4329 else | 4329 else |
4330 best_match_p = !MATCHING_IN_FIRST_STRING; | 4330 best_match_p = !MATCHING_IN_FIRST_STRING; |
4331 | 4331 |
4332 DEBUG_PRINT1 ("backtracking.\n"); | 4332 DEBUG_PRINT1 ("backtracking.\n"); |
4333 | 4333 |
4334 if (!FAIL_STACK_EMPTY ()) | 4334 if (!FAIL_STACK_EMPTY ()) |
4335 { /* More failure points to try. */ | 4335 { /* More failure points to try. */ |
4336 | 4336 |
4337 /* If exceeds best match so far, save it. */ | 4337 /* If exceeds best match so far, save it. */ |
4338 if (!best_regs_set || best_match_p) | 4338 if (!best_regs_set || best_match_p) |
4339 { | 4339 { |
4340 best_regs_set = true; | 4340 best_regs_set = true; |
4341 match_end = d; | 4341 match_end = d; |
4342 | 4342 |
4343 DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); | 4343 DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); |
4344 | 4344 |
4345 for (mcnt = 1; mcnt < num_regs; mcnt++) | 4345 for (mcnt = 1; mcnt < num_regs; mcnt++) |
4346 { | 4346 { |
4347 best_regstart[mcnt] = regstart[mcnt]; | 4347 best_regstart[mcnt] = regstart[mcnt]; |
4348 best_regend[mcnt] = regend[mcnt]; | 4348 best_regend[mcnt] = regend[mcnt]; |
4349 } | 4349 } |
4350 } | 4350 } |
4351 goto fail; | 4351 goto fail; |
4352 } | 4352 } |
4353 | 4353 |
4354 /* If no failure points, don't restore garbage. And if | 4354 /* If no failure points, don't restore garbage. And if |
4355 last match is real best match, don't restore second | 4355 last match is real best match, don't restore second |
4356 best one. */ | 4356 best one. */ |
4357 else if (best_regs_set && !best_match_p) | 4357 else if (best_regs_set && !best_match_p) |
4358 { | 4358 { |
4359 restore_best_regs: | 4359 restore_best_regs: |
4360 /* Restore best match. It may happen that `dend == | 4360 /* Restore best match. It may happen that `dend == |
4361 end_match_1' while the restored d is in string2. | 4361 end_match_1' while the restored d is in string2. |
4362 For example, the pattern `x.*y.*z' against the | 4362 For example, the pattern `x.*y.*z' against the |
4363 strings `x-' and `y-z-', if the two strings are | 4363 strings `x-' and `y-z-', if the two strings are |
4364 not consecutive in memory. */ | 4364 not consecutive in memory. */ |
4365 DEBUG_PRINT1 ("Restoring best registers.\n"); | 4365 DEBUG_PRINT1 ("Restoring best registers.\n"); |
4366 | 4366 |
4367 d = match_end; | 4367 d = match_end; |
4368 dend = ((d >= string1 && d <= end1) | 4368 dend = ((d >= string1 && d <= end1) |
4369 ? end_match_1 : end_match_2); | 4369 ? end_match_1 : end_match_2); |
4370 | 4370 |
4371 for (mcnt = 1; mcnt < num_regs; mcnt++) | 4371 for (mcnt = 1; mcnt < num_regs; mcnt++) |
4372 { | 4372 { |
4373 regstart[mcnt] = best_regstart[mcnt]; | 4373 regstart[mcnt] = best_regstart[mcnt]; |
4374 regend[mcnt] = best_regend[mcnt]; | 4374 regend[mcnt] = best_regend[mcnt]; |
4375 } | 4375 } |
4376 } | 4376 } |
4377 } /* d != end_match_2 */ | 4377 } /* d != end_match_2 */ |
4378 | 4378 |
4379 succeed_label: | 4379 succeed_label: |
4380 DEBUG_PRINT1 ("Accepting match.\n"); | 4380 DEBUG_PRINT1 ("Accepting match.\n"); |
4381 | 4381 |
4382 /* If caller wants register contents data back, do it. */ | 4382 /* If caller wants register contents data back, do it. */ |
4383 if (regs && !bufp->no_sub) | 4383 if (regs && !bufp->no_sub) |
4384 { | 4384 { |
4385 /* Have the register data arrays been allocated? */ | 4385 /* Have the register data arrays been allocated? */ |
4386 if (bufp->regs_allocated == REGS_UNALLOCATED) | 4386 if (bufp->regs_allocated == REGS_UNALLOCATED) |
4387 { /* No. So allocate them with malloc. We need one | 4387 { /* No. So allocate them with malloc. We need one |
4388 extra element beyond `num_regs' for the `-1' marker | 4388 extra element beyond `num_regs' for the `-1' marker |
4389 GNU code uses. */ | 4389 GNU code uses. */ |
4390 regs->num_regs = MAX (RE_NREGS, num_regs + 1); | 4390 regs->num_regs = MAX (RE_NREGS, num_regs + 1); |
4391 regs->start = TALLOC (regs->num_regs, regoff_t); | 4391 regs->start = TALLOC (regs->num_regs, regoff_t); |
4392 regs->end = TALLOC (regs->num_regs, regoff_t); | 4392 regs->end = TALLOC (regs->num_regs, regoff_t); |
4393 if (regs->start == NULL || regs->end == NULL) | 4393 if (regs->start == NULL || regs->end == NULL) |
4394 { | 4394 { |
4395 FREE_VARIABLES (); | 4395 FREE_VARIABLES (); |
4396 return -2; | 4396 return -2; |
4397 } | 4397 } |
4398 bufp->regs_allocated = REGS_REALLOCATE; | 4398 bufp->regs_allocated = REGS_REALLOCATE; |
4399 } | 4399 } |
4400 else if (bufp->regs_allocated == REGS_REALLOCATE) | 4400 else if (bufp->regs_allocated == REGS_REALLOCATE) |
4401 { /* Yes. If we need more elements than were already | 4401 { /* Yes. If we need more elements than were already |
4402 allocated, reallocate them. If we need fewer, just | 4402 allocated, reallocate them. If we need fewer, just |
4403 leave it alone. */ | 4403 leave it alone. */ |
4404 if (regs->num_regs < num_regs + 1) | 4404 if (regs->num_regs < num_regs + 1) |
4405 { | 4405 { |
4406 regs->num_regs = num_regs + 1; | 4406 regs->num_regs = num_regs + 1; |
4407 RETALLOC (regs->start, regs->num_regs, regoff_t); | 4407 RETALLOC (regs->start, regs->num_regs, regoff_t); |
4408 RETALLOC (regs->end, regs->num_regs, regoff_t); | 4408 RETALLOC (regs->end, regs->num_regs, regoff_t); |
4409 if (regs->start == NULL || regs->end == NULL) | 4409 if (regs->start == NULL || regs->end == NULL) |
4410 { | 4410 { |
4411 FREE_VARIABLES (); | 4411 FREE_VARIABLES (); |
4412 return -2; | 4412 return -2; |
4413 } | 4413 } |
4414 } | 4414 } |
4415 } | 4415 } |
4416 else | 4416 else |
4417 { | 4417 { |
4418 /* These braces fend off a "empty body in an else-statement" | 4418 /* These braces fend off a "empty body in an else-statement" |
4419 warning under GCC when assert expands to nothing. */ | 4419 warning under GCC when assert expands to nothing. */ |
4420 assert (bufp->regs_allocated == REGS_FIXED); | 4420 assert (bufp->regs_allocated == REGS_FIXED); |
4421 } | 4421 } |
4422 | 4422 |
4423 /* Convert the pointer data in `regstart' and `regend' to | 4423 /* Convert the pointer data in `regstart' and `regend' to |
4424 indices. Register zero has to be set differently, | 4424 indices. Register zero has to be set differently, |
4425 since we haven't kept track of any info for it. */ | 4425 since we haven't kept track of any info for it. */ |
4426 if (regs->num_regs > 0) | 4426 if (regs->num_regs > 0) |
4427 { | 4427 { |
4428 regs->start[0] = pos; | 4428 regs->start[0] = pos; |
4429 regs->end[0] = (MATCHING_IN_FIRST_STRING | 4429 regs->end[0] = (MATCHING_IN_FIRST_STRING |
4430 ? ((regoff_t) (d - string1)) | 4430 ? ((regoff_t) (d - string1)) |
4431 : ((regoff_t) (d - string2 + size1))); | 4431 : ((regoff_t) (d - string2 + size1))); |
4432 } | 4432 } |
4433 | 4433 |
4434 /* Go through the first `min (num_regs, regs->num_regs)' | 4434 /* Go through the first `min (num_regs, regs->num_regs)' |
4435 registers, since that is all we initialized. */ | 4435 registers, since that is all we initialized. */ |
4436 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++) | 4436 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++) |
4437 { | 4437 { |
4438 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt])) | 4438 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt])) |
4439 regs->start[mcnt] = regs->end[mcnt] = -1; | 4439 regs->start[mcnt] = regs->end[mcnt] = -1; |
4440 else | 4440 else |
4441 { | 4441 { |
4442 regs->start[mcnt] | 4442 regs->start[mcnt] |
4443 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]); | 4443 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]); |
4444 regs->end[mcnt] | 4444 regs->end[mcnt] |
4445 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]); | 4445 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]); |
4446 } | 4446 } |
4447 } | 4447 } |
4448 | 4448 |
4449 /* If the regs structure we return has more elements than | 4449 /* If the regs structure we return has more elements than |
4450 were in the pattern, set the extra elements to -1. If | 4450 were in the pattern, set the extra elements to -1. If |
4451 we (re)allocated the registers, this is the case, | 4451 we (re)allocated the registers, this is the case, |
4452 because we always allocate enough to have at least one | 4452 because we always allocate enough to have at least one |
4453 -1 at the end. */ | 4453 -1 at the end. */ |
4454 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++) | 4454 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++) |
4455 regs->start[mcnt] = regs->end[mcnt] = -1; | 4455 regs->start[mcnt] = regs->end[mcnt] = -1; |
4456 } /* regs && !bufp->no_sub */ | 4456 } /* regs && !bufp->no_sub */ |
4457 | 4457 |
4458 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n", | 4458 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n", |
4459 nfailure_points_pushed, nfailure_points_popped, | 4459 nfailure_points_pushed, nfailure_points_popped, |
4460 nfailure_points_pushed - nfailure_points_popped); | 4460 nfailure_points_pushed - nfailure_points_popped); |
4461 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); | 4461 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); |
4462 | 4462 |
4463 mcnt = d - pos - (MATCHING_IN_FIRST_STRING | 4463 mcnt = d - pos - (MATCHING_IN_FIRST_STRING |
4464 ? string1 | 4464 ? string1 |
4465 : string2 - size1); | 4465 : string2 - size1); |
4466 | 4466 |
4467 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); | 4467 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); |
4468 | 4468 |
4469 FREE_VARIABLES (); | 4469 FREE_VARIABLES (); |
4470 return mcnt; | 4470 return mcnt; |
4471 } | 4471 } |
4472 | 4472 |
4473 /* Otherwise match next pattern command. */ | 4473 /* Otherwise match next pattern command. */ |
4474 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) | 4474 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) |
4475 { | 4475 { |
4476 /* Ignore these. Used to ignore the n of succeed_n's which | 4476 /* Ignore these. Used to ignore the n of succeed_n's which |
4477 currently have n == 0. */ | 4477 currently have n == 0. */ |
4478 case no_op: | 4478 case no_op: |
4479 DEBUG_PRINT1 ("EXECUTING no_op.\n"); | 4479 DEBUG_PRINT1 ("EXECUTING no_op.\n"); |
4480 break; | 4480 break; |
4481 | 4481 |
4482 case succeed: | 4482 case succeed: |
4483 DEBUG_PRINT1 ("EXECUTING succeed.\n"); | 4483 DEBUG_PRINT1 ("EXECUTING succeed.\n"); |
4484 goto succeed_label; | 4484 goto succeed_label; |
4485 | 4485 |
4486 /* Match the next n pattern characters exactly. The following | 4486 /* Match the next n pattern characters exactly. The following |
4487 byte in the pattern defines n, and the n bytes after that | 4487 byte in the pattern defines n, and the n bytes after that |
4488 are the characters to match. */ | 4488 are the characters to match. */ |
4489 case exactn: | 4489 case exactn: |
4490 mcnt = *p++; | 4490 mcnt = *p++; |
4491 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt); | 4491 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt); |
4492 | 4492 |
4493 /* This is written out as an if-else so we don't waste time | 4493 /* This is written out as an if-else so we don't waste time |
4494 testing `translate' inside the loop. */ | 4494 testing `translate' inside the loop. */ |
4495 if (translate) | 4495 if (translate) |
4496 { | 4496 { |
4497 do | 4497 do |
4498 { | 4498 { |
4499 PREFETCH (); | 4499 PREFETCH (); |
4500 if ((unsigned char) translate[(unsigned char) *d++] | 4500 if ((unsigned char) translate[(unsigned char) *d++] |
4501 != (unsigned char) *p++) | 4501 != (unsigned char) *p++) |
4502 goto fail; | 4502 goto fail; |
4503 } | 4503 } |
4504 while (--mcnt); | 4504 while (--mcnt); |
4505 } | 4505 } |
4506 else | 4506 else |
4507 { | 4507 { |
4511 if (*d++ != (char) *p++) goto fail; | 4511 if (*d++ != (char) *p++) goto fail; |
4512 } | 4512 } |
4513 while (--mcnt); | 4513 while (--mcnt); |
4514 } | 4514 } |
4515 SET_REGS_MATCHED (); | 4515 SET_REGS_MATCHED (); |
4516 break; | 4516 break; |
4517 | 4517 |
4518 | 4518 |
4519 /* Match any character except possibly a newline or a null. */ | 4519 /* Match any character except possibly a newline or a null. */ |
4520 case anychar: | 4520 case anychar: |
4521 DEBUG_PRINT1 ("EXECUTING anychar.\n"); | 4521 DEBUG_PRINT1 ("EXECUTING anychar.\n"); |
4522 | 4522 |
4523 PREFETCH (); | 4523 PREFETCH (); |
4524 | 4524 |
4525 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n') | 4525 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n') |
4526 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000')) | 4526 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000')) |
4527 goto fail; | 4527 goto fail; |
4528 | 4528 |
4529 SET_REGS_MATCHED (); | 4529 SET_REGS_MATCHED (); |
4530 DEBUG_PRINT2 (" Matched `%d'.\n", *d); | 4530 DEBUG_PRINT2 (" Matched `%d'.\n", *d); |
4531 d += multibyte ? MULTIBYTE_FORM_LENGTH (d, dend - d) : 1; | 4531 d += multibyte ? MULTIBYTE_FORM_LENGTH (d, dend - d) : 1; |
4532 break; | 4532 break; |
4533 | 4533 |
4534 | 4534 |
4535 case charset: | 4535 case charset: |
4541 | 4541 |
4542 /* Start of actual range_table, or end of bitmap if there is no | 4542 /* Start of actual range_table, or end of bitmap if there is no |
4543 range table. */ | 4543 range table. */ |
4544 unsigned char *range_table; | 4544 unsigned char *range_table; |
4545 | 4545 |
4546 /* Nonzero if there is range table. */ | 4546 /* Nonzero if there is range table. */ |
4547 int range_table_exists; | 4547 int range_table_exists; |
4548 | 4548 |
4549 /* Number of ranges of range table. Not in bytes. */ | 4549 /* Number of ranges of range table. Not in bytes. */ |
4550 int count; | 4550 int count; |
4551 | 4551 |
4552 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : ""); | 4552 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : ""); |
4553 | 4553 |
4554 PREFETCH (); | 4554 PREFETCH (); |
4555 c = (unsigned char) *d; | 4555 c = (unsigned char) *d; |
4556 | 4556 |
4557 range_table = CHARSET_RANGE_TABLE (&p[-1]); /* Past the bitmap. */ | 4557 range_table = CHARSET_RANGE_TABLE (&p[-1]); /* Past the bitmap. */ |
4586 d += len; | 4586 d += len; |
4587 break; | 4587 break; |
4588 } | 4588 } |
4589 | 4589 |
4590 | 4590 |
4591 /* The beginning of a group is represented by start_memory. | 4591 /* The beginning of a group is represented by start_memory. |
4592 The arguments are the register number in the next byte, and the | 4592 The arguments are the register number in the next byte, and the |
4593 number of groups inner to this one in the next. The text | 4593 number of groups inner to this one in the next. The text |
4594 matched within the group is recorded (in the internal | 4594 matched within the group is recorded (in the internal |
4595 registers data structure) under the register number. */ | 4595 registers data structure) under the register number. */ |
4596 case start_memory: | 4596 case start_memory: |
4597 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); | 4597 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); |
4598 | 4598 |
4599 /* Find out if this group can match the empty string. */ | 4599 /* Find out if this group can match the empty string. */ |
4600 p1 = p; /* To send to group_match_null_string_p. */ | 4600 p1 = p; /* To send to group_match_null_string_p. */ |
4601 | 4601 |
4602 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) | 4602 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) |
4603 REG_MATCH_NULL_STRING_P (reg_info[*p]) | 4603 REG_MATCH_NULL_STRING_P (reg_info[*p]) |
4604 = group_match_null_string_p (&p1, pend, reg_info); | 4604 = group_match_null_string_p (&p1, pend, reg_info); |
4605 | 4605 |
4606 /* Save the position in the string where we were the last time | 4606 /* Save the position in the string where we were the last time |
4607 we were at this open-group operator in case the group is | 4607 we were at this open-group operator in case the group is |
4608 operated upon by a repetition operator, e.g., with `(a*)*b' | 4608 operated upon by a repetition operator, e.g., with `(a*)*b' |
4609 against `ab'; then we want to ignore where we are now in | 4609 against `ab'; then we want to ignore where we are now in |
4610 the string in case this attempt to match fails. */ | 4610 the string in case this attempt to match fails. */ |
4611 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) | 4611 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
4612 ? REG_UNSET (regstart[*p]) ? d : regstart[*p] | 4612 ? REG_UNSET (regstart[*p]) ? d : regstart[*p] |
4613 : regstart[*p]; | 4613 : regstart[*p]; |
4614 DEBUG_PRINT2 (" old_regstart: %d\n", | 4614 DEBUG_PRINT2 (" old_regstart: %d\n", |
4615 POINTER_TO_OFFSET (old_regstart[*p])); | 4615 POINTER_TO_OFFSET (old_regstart[*p])); |
4616 | 4616 |
4617 regstart[*p] = d; | 4617 regstart[*p] = d; |
4618 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); | 4618 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); |
4619 | 4619 |
4620 IS_ACTIVE (reg_info[*p]) = 1; | 4620 IS_ACTIVE (reg_info[*p]) = 1; |
4621 MATCHED_SOMETHING (reg_info[*p]) = 0; | 4621 MATCHED_SOMETHING (reg_info[*p]) = 0; |
4622 | 4622 |
4623 /* Clear this whenever we change the register activity status. */ | 4623 /* Clear this whenever we change the register activity status. */ |
4624 set_regs_matched_done = 0; | 4624 set_regs_matched_done = 0; |
4625 | 4625 |
4626 /* This is the new highest active register. */ | 4626 /* This is the new highest active register. */ |
4627 highest_active_reg = *p; | 4627 highest_active_reg = *p; |
4628 | 4628 |
4629 /* If nothing was active before, this is the new lowest active | 4629 /* If nothing was active before, this is the new lowest active |
4630 register. */ | 4630 register. */ |
4631 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) | 4631 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
4632 lowest_active_reg = *p; | 4632 lowest_active_reg = *p; |
4633 | 4633 |
4634 /* Move past the register number and inner group count. */ | 4634 /* Move past the register number and inner group count. */ |
4635 p += 2; | 4635 p += 2; |
4636 just_past_start_mem = p; | 4636 just_past_start_mem = p; |
4637 | 4637 |
4638 break; | 4638 break; |
4639 | 4639 |
4640 | 4640 |
4641 /* The stop_memory opcode represents the end of a group. Its | 4641 /* The stop_memory opcode represents the end of a group. Its |
4642 arguments are the same as start_memory's: the register | 4642 arguments are the same as start_memory's: the register |
4643 number, and the number of inner groups. */ | 4643 number, and the number of inner groups. */ |
4644 case stop_memory: | 4644 case stop_memory: |
4645 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); | 4645 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); |
4646 | 4646 |
4647 /* We need to save the string position the last time we were at | 4647 /* We need to save the string position the last time we were at |
4648 this close-group operator in case the group is operated | 4648 this close-group operator in case the group is operated |
4649 upon by a repetition operator, e.g., with `((a*)*(b*)*)*' | 4649 upon by a repetition operator, e.g., with `((a*)*(b*)*)*' |
4650 against `aba'; then we want to ignore where we are now in | 4650 against `aba'; then we want to ignore where we are now in |
4651 the string in case this attempt to match fails. */ | 4651 the string in case this attempt to match fails. */ |
4652 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) | 4652 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
4653 ? REG_UNSET (regend[*p]) ? d : regend[*p] | 4653 ? REG_UNSET (regend[*p]) ? d : regend[*p] |
4654 : regend[*p]; | 4654 : regend[*p]; |
4655 DEBUG_PRINT2 (" old_regend: %d\n", | 4655 DEBUG_PRINT2 (" old_regend: %d\n", |
4656 POINTER_TO_OFFSET (old_regend[*p])); | 4656 POINTER_TO_OFFSET (old_regend[*p])); |
4657 | 4657 |
4658 regend[*p] = d; | 4658 regend[*p] = d; |
4659 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); | 4659 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); |
4660 | 4660 |
4661 /* This register isn't active anymore. */ | 4661 /* This register isn't active anymore. */ |
4662 IS_ACTIVE (reg_info[*p]) = 0; | 4662 IS_ACTIVE (reg_info[*p]) = 0; |
4663 | 4663 |
4664 /* Clear this whenever we change the register activity status. */ | 4664 /* Clear this whenever we change the register activity status. */ |
4665 set_regs_matched_done = 0; | 4665 set_regs_matched_done = 0; |
4666 | 4666 |
4667 /* If this was the only register active, nothing is active | 4667 /* If this was the only register active, nothing is active |
4668 anymore. */ | 4668 anymore. */ |
4669 if (lowest_active_reg == highest_active_reg) | 4669 if (lowest_active_reg == highest_active_reg) |
4670 { | 4670 { |
4671 lowest_active_reg = NO_LOWEST_ACTIVE_REG; | 4671 lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
4672 highest_active_reg = NO_HIGHEST_ACTIVE_REG; | 4672 highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
4673 } | 4673 } |
4674 else | 4674 else |
4675 { /* We must scan for the new highest active register, since | 4675 { /* We must scan for the new highest active register, since |
4676 it isn't necessarily one less than now: consider | 4676 it isn't necessarily one less than now: consider |
4677 (a(b)c(d(e)f)g). When group 3 ends, after the f), the | 4677 (a(b)c(d(e)f)g). When group 3 ends, after the f), the |
4678 new highest active register is 1. */ | 4678 new highest active register is 1. */ |
4679 unsigned char r = *p - 1; | 4679 unsigned char r = *p - 1; |
4680 while (r > 0 && !IS_ACTIVE (reg_info[r])) | 4680 while (r > 0 && !IS_ACTIVE (reg_info[r])) |
4681 r--; | 4681 r--; |
4682 | 4682 |
4683 /* If we end up at register zero, that means that we saved | 4683 /* If we end up at register zero, that means that we saved |
4684 the registers as the result of an `on_failure_jump', not | 4684 the registers as the result of an `on_failure_jump', not |
4685 a `start_memory', and we jumped to past the innermost | 4685 a `start_memory', and we jumped to past the innermost |
4686 `stop_memory'. For example, in ((.)*) we save | 4686 `stop_memory'. For example, in ((.)*) we save |
4687 registers 1 and 2 as a result of the *, but when we pop | 4687 registers 1 and 2 as a result of the *, but when we pop |
4688 back to the second ), we are at the stop_memory 1. | 4688 back to the second ), we are at the stop_memory 1. |
4689 Thus, nothing is active. */ | 4689 Thus, nothing is active. */ |
4690 if (r == 0) | 4690 if (r == 0) |
4691 { | 4691 { |
4692 lowest_active_reg = NO_LOWEST_ACTIVE_REG; | 4692 lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
4693 highest_active_reg = NO_HIGHEST_ACTIVE_REG; | 4693 highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
4694 } | 4694 } |
4695 else | 4695 else |
4696 highest_active_reg = r; | 4696 highest_active_reg = r; |
4697 } | 4697 } |
4698 | 4698 |
4699 /* If just failed to match something this time around with a | 4699 /* If just failed to match something this time around with a |
4700 group that's operated on by a repetition operator, try to | 4700 group that's operated on by a repetition operator, try to |
4701 force exit from the ``loop'', and restore the register | 4701 force exit from the ``loop'', and restore the register |
4702 information for this group that we had before trying this | 4702 information for this group that we had before trying this |
4703 last match. */ | 4703 last match. */ |
4704 if ((!MATCHED_SOMETHING (reg_info[*p]) | 4704 if ((!MATCHED_SOMETHING (reg_info[*p]) |
4705 || just_past_start_mem == p - 1) | 4705 || just_past_start_mem == p - 1) |
4706 && (p + 2) < pend) | 4706 && (p + 2) < pend) |
4707 { | 4707 { |
4708 boolean is_a_jump_n = false; | 4708 boolean is_a_jump_n = false; |
4709 | 4709 |
4710 p1 = p + 2; | 4710 p1 = p + 2; |
4711 mcnt = 0; | 4711 mcnt = 0; |
4712 switch ((re_opcode_t) *p1++) | 4712 switch ((re_opcode_t) *p1++) |
4713 { | 4713 { |
4714 case jump_n: | 4714 case jump_n: |
4715 is_a_jump_n = true; | 4715 is_a_jump_n = true; |
4716 case pop_failure_jump: | 4716 case pop_failure_jump: |
4717 case maybe_pop_jump: | 4717 case maybe_pop_jump: |
4718 case jump: | 4718 case jump: |
4719 case dummy_failure_jump: | 4719 case dummy_failure_jump: |
4720 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | 4720 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4721 if (is_a_jump_n) | 4721 if (is_a_jump_n) |
4722 p1 += 2; | 4722 p1 += 2; |
4723 break; | 4723 break; |
4724 | 4724 |
4725 default: | 4725 default: |
4726 /* do nothing */ ; | 4726 /* do nothing */ ; |
4727 } | 4727 } |
4728 p1 += mcnt; | 4728 p1 += mcnt; |
4729 | 4729 |
4730 /* If the next operation is a jump backwards in the pattern | 4730 /* If the next operation is a jump backwards in the pattern |
4731 to an on_failure_jump right before the start_memory | 4731 to an on_failure_jump right before the start_memory |
4732 corresponding to this stop_memory, exit from the loop | 4732 corresponding to this stop_memory, exit from the loop |
4733 by forcing a failure after pushing on the stack the | 4733 by forcing a failure after pushing on the stack the |
4734 on_failure_jump's jump in the pattern, and d. */ | 4734 on_failure_jump's jump in the pattern, and d. */ |
4735 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump | 4735 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump |
4736 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p) | 4736 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p) |
4737 { | 4737 { |
4738 /* If this group ever matched anything, then restore | 4738 /* If this group ever matched anything, then restore |
4739 what its registers were before trying this last | 4739 what its registers were before trying this last |
4740 failed match, e.g., with `(a*)*b' against `ab' for | 4740 failed match, e.g., with `(a*)*b' against `ab' for |
4741 regstart[1], and, e.g., with `((a*)*(b*)*)*' | 4741 regstart[1], and, e.g., with `((a*)*(b*)*)*' |
4742 against `aba' for regend[3]. | 4742 against `aba' for regend[3]. |
4743 | 4743 |
4744 Also restore the registers for inner groups for, | 4744 Also restore the registers for inner groups for, |
4745 e.g., `((a*)(b*))*' against `aba' (register 3 would | 4745 e.g., `((a*)(b*))*' against `aba' (register 3 would |
4746 otherwise get trashed). */ | 4746 otherwise get trashed). */ |
4747 | 4747 |
4748 if (EVER_MATCHED_SOMETHING (reg_info[*p])) | 4748 if (EVER_MATCHED_SOMETHING (reg_info[*p])) |
4749 { | 4749 { |
4750 unsigned r; | 4750 unsigned r; |
4751 | 4751 |
4752 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; | 4752 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; |
4753 | 4753 |
4754 /* Restore this and inner groups' (if any) registers. */ | 4754 /* Restore this and inner groups' (if any) registers. */ |
4755 for (r = *p; r < *p + *(p + 1); r++) | 4755 for (r = *p; r < *p + *(p + 1); r++) |
4756 { | 4756 { |
4757 regstart[r] = old_regstart[r]; | 4757 regstart[r] = old_regstart[r]; |
4758 | 4758 |
4759 /* xx why this test? */ | 4759 /* xx why this test? */ |
4760 if (old_regend[r] >= regstart[r]) | 4760 if (old_regend[r] >= regstart[r]) |
4761 regend[r] = old_regend[r]; | 4761 regend[r] = old_regend[r]; |
4762 } | 4762 } |
4763 } | 4763 } |
4764 p1++; | 4764 p1++; |
4765 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | 4765 EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4766 PUSH_FAILURE_POINT (p1 + mcnt, d, -2); | 4766 PUSH_FAILURE_POINT (p1 + mcnt, d, -2); |
4767 | 4767 |
4768 goto fail; | 4768 goto fail; |
4769 } | 4769 } |
4770 } | 4770 } |
4771 | 4771 |
4772 /* Move past the register number and the inner group count. */ | 4772 /* Move past the register number and the inner group count. */ |
4773 p += 2; | 4773 p += 2; |
4774 break; | 4774 break; |
4775 | 4775 |
4776 | 4776 |
4777 /* \<digit> has been turned into a `duplicate' command which is | 4777 /* \<digit> has been turned into a `duplicate' command which is |
4778 followed by the numeric value of <digit> as the register number. */ | 4778 followed by the numeric value of <digit> as the register number. */ |
4779 case duplicate: | 4779 case duplicate: |
4780 { | 4780 { |
4781 register const char *d2, *dend2; | 4781 register const char *d2, *dend2; |
4782 int regno = *p++; /* Get which register to match against. */ | 4782 int regno = *p++; /* Get which register to match against. */ |
4783 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno); | 4783 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno); |
4784 | 4784 |
4785 /* Can't back reference a group which we've never matched. */ | 4785 /* Can't back reference a group which we've never matched. */ |
4786 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) | 4786 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) |
4787 goto fail; | 4787 goto fail; |
4788 | 4788 |
4789 /* Where in input to try to start matching. */ | 4789 /* Where in input to try to start matching. */ |
4790 d2 = regstart[regno]; | 4790 d2 = regstart[regno]; |
4791 | 4791 |
4792 /* Where to stop matching; if both the place to start and | 4792 /* Where to stop matching; if both the place to start and |
4793 the place to stop matching are in the same string, then | 4793 the place to stop matching are in the same string, then |
4794 set to the place to stop, otherwise, for now have to use | 4794 set to the place to stop, otherwise, for now have to use |
4795 the end of the first string. */ | 4795 the end of the first string. */ |
4796 | 4796 |
4797 dend2 = ((FIRST_STRING_P (regstart[regno]) | 4797 dend2 = ((FIRST_STRING_P (regstart[regno]) |
4798 == FIRST_STRING_P (regend[regno])) | 4798 == FIRST_STRING_P (regend[regno])) |
4799 ? regend[regno] : end_match_1); | 4799 ? regend[regno] : end_match_1); |
4800 for (;;) | 4800 for (;;) |
4801 { | 4801 { |
4802 /* If necessary, advance to next segment in register | 4802 /* If necessary, advance to next segment in register |
4803 contents. */ | 4803 contents. */ |
4804 while (d2 == dend2) | 4804 while (d2 == dend2) |
4805 { | 4805 { |
4806 if (dend2 == end_match_2) break; | 4806 if (dend2 == end_match_2) break; |
4807 if (dend2 == regend[regno]) break; | 4807 if (dend2 == regend[regno]) break; |
4808 | 4808 |
4809 /* End of string1 => advance to string2. */ | 4809 /* End of string1 => advance to string2. */ |
4810 d2 = string2; | 4810 d2 = string2; |
4811 dend2 = regend[regno]; | 4811 dend2 = regend[regno]; |
4812 } | 4812 } |
4813 /* At end of register contents => success */ | 4813 /* At end of register contents => success */ |
4814 if (d2 == dend2) break; | 4814 if (d2 == dend2) break; |
4815 | 4815 |
4816 /* If necessary, advance to next segment in data. */ | 4816 /* If necessary, advance to next segment in data. */ |
4818 | 4818 |
4819 /* How many characters left in this segment to match. */ | 4819 /* How many characters left in this segment to match. */ |
4820 mcnt = dend - d; | 4820 mcnt = dend - d; |
4821 | 4821 |
4822 /* Want how many consecutive characters we can match in | 4822 /* Want how many consecutive characters we can match in |
4823 one shot, so, if necessary, adjust the count. */ | 4823 one shot, so, if necessary, adjust the count. */ |
4824 if (mcnt > dend2 - d2) | 4824 if (mcnt > dend2 - d2) |
4825 mcnt = dend2 - d2; | 4825 mcnt = dend2 - d2; |
4826 | 4826 |
4827 /* Compare that many; failure if mismatch, else move | 4827 /* Compare that many; failure if mismatch, else move |
4828 past them. */ | 4828 past them. */ |
4829 if (translate | 4829 if (translate |
4830 ? bcmp_translate (d, d2, mcnt, translate) | 4830 ? bcmp_translate (d, d2, mcnt, translate) |
4831 : bcmp (d, d2, mcnt)) | 4831 : bcmp (d, d2, mcnt)) |
4832 goto fail; | 4832 goto fail; |
4833 d += mcnt, d2 += mcnt; | 4833 d += mcnt, d2 += mcnt; |
4834 | 4834 |
4835 /* Do this because we've match some characters. */ | 4835 /* Do this because we've match some characters. */ |
4836 SET_REGS_MATCHED (); | 4836 SET_REGS_MATCHED (); |
4837 } | 4837 } |
4838 } | 4838 } |
4839 break; | 4839 break; |
4840 | 4840 |
4841 | 4841 |
4842 /* begline matches the empty string at the beginning of the string | 4842 /* begline matches the empty string at the beginning of the string |
4843 (unless `not_bol' is set in `bufp'), and, if | 4843 (unless `not_bol' is set in `bufp'), and, if |
4844 `newline_anchor' is set, after newlines. */ | 4844 `newline_anchor' is set, after newlines. */ |
4845 case begline: | 4845 case begline: |
4846 DEBUG_PRINT1 ("EXECUTING begline.\n"); | 4846 DEBUG_PRINT1 ("EXECUTING begline.\n"); |
4847 | 4847 |
4848 if (AT_STRINGS_BEG (d)) | 4848 if (AT_STRINGS_BEG (d)) |
4849 { | 4849 { |
4850 if (!bufp->not_bol) break; | 4850 if (!bufp->not_bol) break; |
4851 } | 4851 } |
4852 else if (d[-1] == '\n' && bufp->newline_anchor) | 4852 else if (d[-1] == '\n' && bufp->newline_anchor) |
4853 { | 4853 { |
4854 break; | 4854 break; |
4855 } | 4855 } |
4856 /* In all other cases, we fail. */ | 4856 /* In all other cases, we fail. */ |
4857 goto fail; | 4857 goto fail; |
4858 | 4858 |
4859 | 4859 |
4860 /* endline is the dual of begline. */ | 4860 /* endline is the dual of begline. */ |
4861 case endline: | 4861 case endline: |
4862 DEBUG_PRINT1 ("EXECUTING endline.\n"); | 4862 DEBUG_PRINT1 ("EXECUTING endline.\n"); |
4863 | 4863 |
4864 if (AT_STRINGS_END (d)) | 4864 if (AT_STRINGS_END (d)) |
4865 { | 4865 { |
4866 if (!bufp->not_eol) break; | 4866 if (!bufp->not_eol) break; |
4867 } | 4867 } |
4868 | 4868 |
4869 /* We have to ``prefetch'' the next character. */ | 4869 /* We have to ``prefetch'' the next character. */ |
4870 else if ((d == end1 ? *string2 : *d) == '\n' | 4870 else if ((d == end1 ? *string2 : *d) == '\n' |
4871 && bufp->newline_anchor) | 4871 && bufp->newline_anchor) |
4872 { | 4872 { |
4873 break; | 4873 break; |
4874 } | 4874 } |
4875 goto fail; | 4875 goto fail; |
4876 | 4876 |
4877 | 4877 |
4878 /* Match at the very beginning of the data. */ | 4878 /* Match at the very beginning of the data. */ |
4879 case begbuf: | 4879 case begbuf: |
4880 DEBUG_PRINT1 ("EXECUTING begbuf.\n"); | 4880 DEBUG_PRINT1 ("EXECUTING begbuf.\n"); |
4881 if (AT_STRINGS_BEG (d)) | 4881 if (AT_STRINGS_BEG (d)) |
4882 break; | 4882 break; |
4883 goto fail; | 4883 goto fail; |
4884 | 4884 |
4885 | 4885 |
4886 /* Match at the very end of the data. */ | 4886 /* Match at the very end of the data. */ |
4887 case endbuf: | 4887 case endbuf: |
4888 DEBUG_PRINT1 ("EXECUTING endbuf.\n"); | 4888 DEBUG_PRINT1 ("EXECUTING endbuf.\n"); |
4889 if (AT_STRINGS_END (d)) | 4889 if (AT_STRINGS_END (d)) |
4890 break; | 4890 break; |
4891 goto fail; | 4891 goto fail; |
4892 | 4892 |
4893 | 4893 |
4894 /* on_failure_keep_string_jump is used to optimize `.*\n'. It | 4894 /* on_failure_keep_string_jump is used to optimize `.*\n'. It |
4895 pushes NULL as the value for the string on the stack. Then | 4895 pushes NULL as the value for the string on the stack. Then |
4896 `pop_failure_point' will keep the current value for the | 4896 `pop_failure_point' will keep the current value for the |
4897 string, instead of restoring it. To see why, consider | 4897 string, instead of restoring it. To see why, consider |
4898 matching `foo\nbar' against `.*\n'. The .* matches the foo; | 4898 matching `foo\nbar' against `.*\n'. The .* matches the foo; |
4899 then the . fails against the \n. But the next thing we want | 4899 then the . fails against the \n. But the next thing we want |
4900 to do is match the \n against the \n; if we restored the | 4900 to do is match the \n against the \n; if we restored the |
4901 string value, we would be back at the foo. | 4901 string value, we would be back at the foo. |
4902 | 4902 |
4903 Because this is used only in specific cases, we don't need to | 4903 Because this is used only in specific cases, we don't need to |
4904 check all the things that `on_failure_jump' does, to make | 4904 check all the things that `on_failure_jump' does, to make |
4905 sure the right things get saved on the stack. Hence we don't | 4905 sure the right things get saved on the stack. Hence we don't |
4906 share its code. The only reason to push anything on the | 4906 share its code. The only reason to push anything on the |
4907 stack at all is that otherwise we would have to change | 4907 stack at all is that otherwise we would have to change |
4908 `anychar's code to do something besides goto fail in this | 4908 `anychar's code to do something besides goto fail in this |
4909 case; that seems worse than this. */ | 4909 case; that seems worse than this. */ |
4910 case on_failure_keep_string_jump: | 4910 case on_failure_keep_string_jump: |
4911 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); | 4911 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); |
4912 | 4912 |
4913 EXTRACT_NUMBER_AND_INCR (mcnt, p); | 4913 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
4914 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); | 4914 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); |
4915 | 4915 |
4916 PUSH_FAILURE_POINT (p + mcnt, NULL, -2); | 4916 PUSH_FAILURE_POINT (p + mcnt, NULL, -2); |
4917 break; | 4917 break; |
4918 | 4918 |
4919 | 4919 |
4920 /* Uses of on_failure_jump: | 4920 /* Uses of on_failure_jump: |
4921 | 4921 |
4922 Each alternative starts with an on_failure_jump that points | 4922 Each alternative starts with an on_failure_jump that points |
4923 to the beginning of the next alternative. Each alternative | 4923 to the beginning of the next alternative. Each alternative |
4924 except the last ends with a jump that in effect jumps past | 4924 except the last ends with a jump that in effect jumps past |
4925 the rest of the alternatives. (They really jump to the | 4925 the rest of the alternatives. (They really jump to the |
4926 ending jump of the following alternative, because tensioning | 4926 ending jump of the following alternative, because tensioning |
4927 these jumps is a hassle.) | 4927 these jumps is a hassle.) |
4928 | 4928 |
4929 Repeats start with an on_failure_jump that points past both | 4929 Repeats start with an on_failure_jump that points past both |
4930 the repetition text and either the following jump or | 4930 the repetition text and either the following jump or |
4931 pop_failure_jump back to this on_failure_jump. */ | 4931 pop_failure_jump back to this on_failure_jump. */ |
4932 case on_failure_jump: | 4932 case on_failure_jump: |
4933 on_failure: | 4933 on_failure: |
4934 DEBUG_PRINT1 ("EXECUTING on_failure_jump"); | 4934 DEBUG_PRINT1 ("EXECUTING on_failure_jump"); |
4935 | 4935 |
4936 EXTRACT_NUMBER_AND_INCR (mcnt, p); | 4936 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
4937 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt); | 4937 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt); |
4938 | 4938 |
4939 /* If this on_failure_jump comes right before a group (i.e., | 4939 /* If this on_failure_jump comes right before a group (i.e., |
4940 the original * applied to a group), save the information | 4940 the original * applied to a group), save the information |
4941 for that group and all inner ones, so that if we fail back | 4941 for that group and all inner ones, so that if we fail back |
4942 to this point, the group's information will be correct. | 4942 to this point, the group's information will be correct. |
4943 For example, in \(a*\)*\1, we need the preceding group, | 4943 For example, in \(a*\)*\1, we need the preceding group, |
4944 and in \(zz\(a*\)b*\)\2, we need the inner group. */ | 4944 and in \(zz\(a*\)b*\)\2, we need the inner group. */ |
4945 | 4945 |
4946 /* We can't use `p' to check ahead because we push | 4946 /* We can't use `p' to check ahead because we push |
4947 a failure point to `p + mcnt' after we do this. */ | 4947 a failure point to `p + mcnt' after we do this. */ |
4948 p1 = p; | 4948 p1 = p; |
4949 | 4949 |
4950 /* We need to skip no_op's before we look for the | 4950 /* We need to skip no_op's before we look for the |
4951 start_memory in case this on_failure_jump is happening as | 4951 start_memory in case this on_failure_jump is happening as |
4952 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 | 4952 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 |
4953 against aba. */ | 4953 against aba. */ |
4954 while (p1 < pend && (re_opcode_t) *p1 == no_op) | 4954 while (p1 < pend && (re_opcode_t) *p1 == no_op) |
4955 p1++; | 4955 p1++; |
4956 | 4956 |
4957 if (p1 < pend && (re_opcode_t) *p1 == start_memory) | 4957 if (p1 < pend && (re_opcode_t) *p1 == start_memory) |
4958 { | 4958 { |
4959 /* We have a new highest active register now. This will | 4959 /* We have a new highest active register now. This will |
4960 get reset at the start_memory we are about to get to, | 4960 get reset at the start_memory we are about to get to, |
4961 but we will have saved all the registers relevant to | 4961 but we will have saved all the registers relevant to |
4962 this repetition op, as described above. */ | 4962 this repetition op, as described above. */ |
4963 highest_active_reg = *(p1 + 1) + *(p1 + 2); | 4963 highest_active_reg = *(p1 + 1) + *(p1 + 2); |
4964 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) | 4964 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
4965 lowest_active_reg = *(p1 + 1); | 4965 lowest_active_reg = *(p1 + 1); |
4966 } | 4966 } |
4967 | 4967 |
4968 DEBUG_PRINT1 (":\n"); | 4968 DEBUG_PRINT1 (":\n"); |
4969 PUSH_FAILURE_POINT (p + mcnt, d, -2); | 4969 PUSH_FAILURE_POINT (p + mcnt, d, -2); |
4970 break; | 4970 break; |
4971 | 4971 |
4972 | 4972 |
4973 /* A smart repeat ends with `maybe_pop_jump'. | 4973 /* A smart repeat ends with `maybe_pop_jump'. |
4974 We change it to either `pop_failure_jump' or `jump'. */ | 4974 We change it to either `pop_failure_jump' or `jump'. */ |
4975 case maybe_pop_jump: | 4975 case maybe_pop_jump: |
4976 EXTRACT_NUMBER_AND_INCR (mcnt, p); | 4976 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
4977 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); | 4977 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); |
4978 { | 4978 { |
4979 register unsigned char *p2 = p; | 4979 register unsigned char *p2 = p; |
4980 | 4980 |
4981 /* Compare the beginning of the repeat with what in the | 4981 /* Compare the beginning of the repeat with what in the |
4982 pattern follows its end. If we can establish that there | 4982 pattern follows its end. If we can establish that there |
4983 is nothing that they would both match, i.e., that we | 4983 is nothing that they would both match, i.e., that we |
4984 would have to backtrack because of (as in, e.g., `a*a') | 4984 would have to backtrack because of (as in, e.g., `a*a') |
4985 then we can change to pop_failure_jump, because we'll | 4985 then we can change to pop_failure_jump, because we'll |
4986 never have to backtrack. | 4986 never have to backtrack. |
4987 | 4987 |
4988 This is not true in the case of alternatives: in | 4988 This is not true in the case of alternatives: in |
4989 `(a|ab)*' we do need to backtrack to the `ab' alternative | 4989 `(a|ab)*' we do need to backtrack to the `ab' alternative |
4990 (e.g., if the string was `ab'). But instead of trying to | 4990 (e.g., if the string was `ab'). But instead of trying to |
4991 detect that here, the alternative has put on a dummy | 4991 detect that here, the alternative has put on a dummy |
4992 failure point which is what we will end up popping. */ | 4992 failure point which is what we will end up popping. */ |
4993 | 4993 |
4994 /* Skip over open/close-group commands. | 4994 /* Skip over open/close-group commands. |
4995 If what follows this loop is a ...+ construct, | 4995 If what follows this loop is a ...+ construct, |
4996 look at what begins its body, since we will have to | 4996 look at what begins its body, since we will have to |
4997 match at least one of that. */ | 4997 match at least one of that. */ |
5009 } | 5009 } |
5010 | 5010 |
5011 p1 = p + mcnt; | 5011 p1 = p + mcnt; |
5012 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding | 5012 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding |
5013 to the `maybe_finalize_jump' of this case. Examine what | 5013 to the `maybe_finalize_jump' of this case. Examine what |
5014 follows. */ | 5014 follows. */ |
5015 | 5015 |
5016 /* If we're at the end of the pattern, we can change. */ | 5016 /* If we're at the end of the pattern, we can change. */ |
5017 if (p2 == pend) | 5017 if (p2 == pend) |
5018 { | 5018 { |
5019 /* Consider what happens when matching ":\(.*\)" | 5019 /* Consider what happens when matching ":\(.*\)" |
5020 against ":/". I don't really understand this code | 5020 against ":/". I don't really understand this code |
5021 yet. */ | 5021 yet. */ |
5022 p[-3] = (unsigned char) pop_failure_jump; | 5022 p[-3] = (unsigned char) pop_failure_jump; |
5023 DEBUG_PRINT1 | 5023 DEBUG_PRINT1 |
5024 (" End of pattern: change to `pop_failure_jump'.\n"); | 5024 (" End of pattern: change to `pop_failure_jump'.\n"); |
5025 } | 5025 } |
5026 | 5026 |
5027 else if ((re_opcode_t) *p2 == exactn | 5027 else if ((re_opcode_t) *p2 == exactn |
5028 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) | 5028 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) |
5029 { | 5029 { |
5030 register unsigned int c | 5030 register unsigned int c |
5031 = *p2 == (unsigned char) endline ? '\n' : p2[2]; | 5031 = *p2 == (unsigned char) endline ? '\n' : p2[2]; |
5032 | 5032 |
5033 if ((re_opcode_t) p1[3] == exactn) | 5033 if ((re_opcode_t) p1[3] == exactn) |
5034 { | 5034 { |
5035 if (!(multibyte /* && (c != '\n') */ | 5035 if (!(multibyte /* && (c != '\n') */ |
5036 && BASE_LEADING_CODE_P (c)) | 5036 && BASE_LEADING_CODE_P (c)) |
5037 ? c != p1[5] | 5037 ? c != p1[5] |
5038 : (STRING_CHAR (&p2[2], pend - &p2[2]) | 5038 : (STRING_CHAR (&p2[2], pend - &p2[2]) |
5039 != STRING_CHAR (&p1[5], pend - &p1[5]))) | 5039 != STRING_CHAR (&p1[5], pend - &p1[5]))) |
5040 { | 5040 { |
5041 p[-3] = (unsigned char) pop_failure_jump; | 5041 p[-3] = (unsigned char) pop_failure_jump; |
5042 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", | 5042 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", |
5043 c, p1[5]); | 5043 c, p1[5]); |
5044 } | 5044 } |
5045 } | 5045 } |
5046 | 5046 |
5047 else if ((re_opcode_t) p1[3] == charset | 5047 else if ((re_opcode_t) p1[3] == charset |
5048 || (re_opcode_t) p1[3] == charset_not) | 5048 || (re_opcode_t) p1[3] == charset_not) |
5049 { | 5049 { |
5062 not = !not; | 5062 not = !not; |
5063 } | 5063 } |
5064 else if (CHARSET_RANGE_TABLE_EXISTS_P (&p1[3])) | 5064 else if (CHARSET_RANGE_TABLE_EXISTS_P (&p1[3])) |
5065 CHARSET_LOOKUP_RANGE_TABLE (not, c, &p1[3]); | 5065 CHARSET_LOOKUP_RANGE_TABLE (not, c, &p1[3]); |
5066 | 5066 |
5067 /* `not' is equal to 1 if c would match, which means | 5067 /* `not' is equal to 1 if c would match, which means |
5068 that we can't change to pop_failure_jump. */ | 5068 that we can't change to pop_failure_jump. */ |
5069 if (!not) | 5069 if (!not) |
5070 { | 5070 { |
5071 p[-3] = (unsigned char) pop_failure_jump; | 5071 p[-3] = (unsigned char) pop_failure_jump; |
5072 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); | 5072 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); |
5073 } | 5073 } |
5074 } | 5074 } |
5075 } | 5075 } |
5076 else if ((re_opcode_t) *p2 == charset) | 5076 else if ((re_opcode_t) *p2 == charset) |
5077 { | 5077 { |
5078 if ((re_opcode_t) p1[3] == exactn) | 5078 if ((re_opcode_t) p1[3] == exactn) |
5079 { | 5079 { |
5080 register unsigned int c = p1[5]; | 5080 register unsigned int c = p1[5]; |
5081 int not = 0; | 5081 int not = 0; |
5082 | 5082 |
5083 if (multibyte && BASE_LEADING_CODE_P (c)) | 5083 if (multibyte && BASE_LEADING_CODE_P (c)) |
5084 c = STRING_CHAR (&p1[5], pend - &p1[5]); | 5084 c = STRING_CHAR (&p1[5], pend - &p1[5]); |
5085 | 5085 |
5086 /* Test if C is listed in charset at `p2'. */ | 5086 /* Test if C is listed in charset at `p2'. */ |
5087 if (SINGLE_BYTE_CHAR_P (c)) | 5087 if (SINGLE_BYTE_CHAR_P (c)) |
5088 { | 5088 { |
5089 if (c < CHARSET_BITMAP_SIZE (p2) * BYTEWIDTH | 5089 if (c < CHARSET_BITMAP_SIZE (p2) * BYTEWIDTH |
5090 && (p2[2 + c / BYTEWIDTH] | 5090 && (p2[2 + c / BYTEWIDTH] |
5091 & (1 << (c % BYTEWIDTH)))) | 5091 & (1 << (c % BYTEWIDTH)))) |
5093 } | 5093 } |
5094 else if (CHARSET_RANGE_TABLE_EXISTS_P (p2)) | 5094 else if (CHARSET_RANGE_TABLE_EXISTS_P (p2)) |
5095 CHARSET_LOOKUP_RANGE_TABLE (not, c, p2); | 5095 CHARSET_LOOKUP_RANGE_TABLE (not, c, p2); |
5096 | 5096 |
5097 if (!not) | 5097 if (!not) |
5098 { | 5098 { |
5099 p[-3] = (unsigned char) pop_failure_jump; | 5099 p[-3] = (unsigned char) pop_failure_jump; |
5100 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); | 5100 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); |
5101 } | 5101 } |
5102 } | 5102 } |
5103 | 5103 |
5104 /* It is hard to list up all the character in charset | 5104 /* It is hard to list up all the character in charset |
5105 P2 if it includes multibyte character. Give up in | 5105 P2 if it includes multibyte character. Give up in |
5106 such case. */ | 5106 such case. */ |
5107 else if (!multibyte || !CHARSET_RANGE_TABLE_EXISTS_P (p2)) | 5107 else if (!multibyte || !CHARSET_RANGE_TABLE_EXISTS_P (p2)) |
5108 { | 5108 { |
5109 /* Now, we are sure that P2 has no range table. | 5109 /* Now, we are sure that P2 has no range table. |
5110 So, for the size of bitmap in P2, `p2[1]' is | 5110 So, for the size of bitmap in P2, `p2[1]' is |
5111 enough. But P1 may have range table, so the | 5111 enough. But P1 may have range table, so the |
5112 size of bitmap table of P1 is extracted by | 5112 size of bitmap table of P1 is extracted by |
5113 using macro `CHARSET_BITMAP_SIZE'. | 5113 using macro `CHARSET_BITMAP_SIZE'. |
5114 | 5114 |
5115 Since we know that all the character listed in | 5115 Since we know that all the character listed in |
5116 P2 is ASCII, it is enough to test only bitmap | 5116 P2 is ASCII, it is enough to test only bitmap |
5117 table of P1. */ | 5117 table of P1. */ |
5118 | 5118 |
5119 if ((re_opcode_t) p1[3] == charset_not) | 5119 if ((re_opcode_t) p1[3] == charset_not) |
5120 { | 5120 { |
5121 int idx; | 5121 int idx; |
5122 /* We win if the charset_not inside the loop lists | 5122 /* We win if the charset_not inside the loop lists |
5123 every character listed in the charset after. */ | 5123 every character listed in the charset after. */ |
5124 for (idx = 0; idx < (int) p2[1]; idx++) | 5124 for (idx = 0; idx < (int) p2[1]; idx++) |
5125 if (! (p2[2 + idx] == 0 | 5125 if (! (p2[2 + idx] == 0 |
5126 || (idx < CHARSET_BITMAP_SIZE (&p1[3]) | 5126 || (idx < CHARSET_BITMAP_SIZE (&p1[3]) |
5127 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0)))) | 5127 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0)))) |
5128 break; | 5128 break; |
5129 | 5129 |
5130 if (idx == p2[1]) | 5130 if (idx == p2[1]) |
5131 { | 5131 { |
5132 p[-3] = (unsigned char) pop_failure_jump; | 5132 p[-3] = (unsigned char) pop_failure_jump; |
5133 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); | 5133 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); |
5134 } | 5134 } |
5135 } | 5135 } |
5136 else if ((re_opcode_t) p1[3] == charset) | 5136 else if ((re_opcode_t) p1[3] == charset) |
5137 { | 5137 { |
5138 int idx; | 5138 int idx; |
5139 /* We win if the charset inside the loop | 5139 /* We win if the charset inside the loop |
5145 if ((p2[2 + idx] & p1[5 + idx]) != 0) | 5145 if ((p2[2 + idx] & p1[5 + idx]) != 0) |
5146 break; | 5146 break; |
5147 | 5147 |
5148 if (idx == p2[1] | 5148 if (idx == p2[1] |
5149 || idx == CHARSET_BITMAP_SIZE (&p1[3])) | 5149 || idx == CHARSET_BITMAP_SIZE (&p1[3])) |
5150 { | 5150 { |
5151 p[-3] = (unsigned char) pop_failure_jump; | 5151 p[-3] = (unsigned char) pop_failure_jump; |
5152 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); | 5152 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); |
5153 } | 5153 } |
5154 } | 5154 } |
5155 } | 5155 } |
5156 } | 5156 } |
5157 } | 5157 } |
5158 p -= 2; /* Point at relative address again. */ | 5158 p -= 2; /* Point at relative address again. */ |
5159 if ((re_opcode_t) p[-1] != pop_failure_jump) | 5159 if ((re_opcode_t) p[-1] != pop_failure_jump) |
5160 { | 5160 { |
5161 p[-1] = (unsigned char) jump; | 5161 p[-1] = (unsigned char) jump; |
5162 DEBUG_PRINT1 (" Match => jump.\n"); | 5162 DEBUG_PRINT1 (" Match => jump.\n"); |
5163 goto unconditional_jump; | 5163 goto unconditional_jump; |
5164 } | 5164 } |
5165 /* Note fall through. */ | 5165 /* Note fall through. */ |
5166 | 5166 |
5167 | 5167 |
5168 /* The end of a simple repeat has a pop_failure_jump back to | 5168 /* The end of a simple repeat has a pop_failure_jump back to |
5169 its matching on_failure_jump, where the latter will push a | 5169 its matching on_failure_jump, where the latter will push a |
5170 failure point. The pop_failure_jump takes off failure | 5170 failure point. The pop_failure_jump takes off failure |
5171 points put on by this pop_failure_jump's matching | 5171 points put on by this pop_failure_jump's matching |
5172 on_failure_jump; we got through the pattern to here from the | 5172 on_failure_jump; we got through the pattern to here from the |
5173 matching on_failure_jump, so didn't fail. */ | 5173 matching on_failure_jump, so didn't fail. */ |
5174 case pop_failure_jump: | 5174 case pop_failure_jump: |
5175 { | 5175 { |
5176 /* We need to pass separate storage for the lowest and | 5176 /* We need to pass separate storage for the lowest and |
5177 highest registers, even though we don't care about the | 5177 highest registers, even though we don't care about the |
5178 actual values. Otherwise, we will restore only one | 5178 actual values. Otherwise, we will restore only one |
5179 register from the stack, since lowest will == highest in | 5179 register from the stack, since lowest will == highest in |
5180 `pop_failure_point'. */ | 5180 `pop_failure_point'. */ |
5181 unsigned dummy_low_reg, dummy_high_reg; | 5181 unsigned dummy_low_reg, dummy_high_reg; |
5182 unsigned char *pdummy; | 5182 unsigned char *pdummy; |
5183 const char *sdummy; | 5183 const char *sdummy; |
5184 | 5184 |
5185 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n"); | 5185 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n"); |
5186 POP_FAILURE_POINT (sdummy, pdummy, | 5186 POP_FAILURE_POINT (sdummy, pdummy, |
5187 dummy_low_reg, dummy_high_reg, | 5187 dummy_low_reg, dummy_high_reg, |
5188 reg_dummy, reg_dummy, reg_info_dummy); | 5188 reg_dummy, reg_dummy, reg_info_dummy); |
5189 } | 5189 } |
5190 /* Note fall through. */ | 5190 /* Note fall through. */ |
5191 | 5191 |
5192 | 5192 |
5193 /* Unconditionally jump (without popping any failure points). */ | 5193 /* Unconditionally jump (without popping any failure points). */ |
5194 case jump: | 5194 case jump: |
5195 unconditional_jump: | 5195 unconditional_jump: |
5196 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ | 5196 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ |
5197 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt); | 5197 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt); |
5198 p += mcnt; /* Do the jump. */ | 5198 p += mcnt; /* Do the jump. */ |
5199 DEBUG_PRINT2 ("(to 0x%x).\n", p); | 5199 DEBUG_PRINT2 ("(to 0x%x).\n", p); |
5200 break; | 5200 break; |
5201 | 5201 |
5202 | 5202 |
5203 /* We need this opcode so we can detect where alternatives end | 5203 /* We need this opcode so we can detect where alternatives end |
5204 in `group_match_null_string_p' et al. */ | 5204 in `group_match_null_string_p' et al. */ |
5205 case jump_past_alt: | 5205 case jump_past_alt: |
5206 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n"); | 5206 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n"); |
5207 goto unconditional_jump; | 5207 goto unconditional_jump; |
5208 | 5208 |
5209 | 5209 |
5210 /* Normally, the on_failure_jump pushes a failure point, which | 5210 /* Normally, the on_failure_jump pushes a failure point, which |
5211 then gets popped at pop_failure_jump. We will end up at | 5211 then gets popped at pop_failure_jump. We will end up at |
5212 pop_failure_jump, also, and with a pattern of, say, `a+', we | 5212 pop_failure_jump, also, and with a pattern of, say, `a+', we |
5213 are skipping over the on_failure_jump, so we have to push | 5213 are skipping over the on_failure_jump, so we have to push |
5214 something meaningless for pop_failure_jump to pop. */ | 5214 something meaningless for pop_failure_jump to pop. */ |
5215 case dummy_failure_jump: | 5215 case dummy_failure_jump: |
5216 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n"); | 5216 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n"); |
5217 /* It doesn't matter what we push for the string here. What | 5217 /* It doesn't matter what we push for the string here. What |
5218 the code at `fail' tests is the value for the pattern. */ | 5218 the code at `fail' tests is the value for the pattern. */ |
5219 PUSH_FAILURE_POINT (0, 0, -2); | 5219 PUSH_FAILURE_POINT (0, 0, -2); |
5220 goto unconditional_jump; | 5220 goto unconditional_jump; |
5221 | 5221 |
5222 | 5222 |
5223 /* At the end of an alternative, we need to push a dummy failure | 5223 /* At the end of an alternative, we need to push a dummy failure |
5224 point in case we are followed by a `pop_failure_jump', because | 5224 point in case we are followed by a `pop_failure_jump', because |
5225 we don't want the failure point for the alternative to be | 5225 we don't want the failure point for the alternative to be |
5226 popped. For example, matching `(a|ab)*' against `aab' | 5226 popped. For example, matching `(a|ab)*' against `aab' |
5227 requires that we match the `ab' alternative. */ | 5227 requires that we match the `ab' alternative. */ |
5228 case push_dummy_failure: | 5228 case push_dummy_failure: |
5229 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n"); | 5229 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n"); |
5230 /* See comments just above at `dummy_failure_jump' about the | 5230 /* See comments just above at `dummy_failure_jump' about the |
5231 two zeroes. */ | 5231 two zeroes. */ |
5232 PUSH_FAILURE_POINT (0, 0, -2); | 5232 PUSH_FAILURE_POINT (0, 0, -2); |
5233 break; | 5233 break; |
5234 | 5234 |
5235 /* Have to succeed matching what follows at least n times. | 5235 /* Have to succeed matching what follows at least n times. |
5236 After that, handle like `on_failure_jump'. */ | 5236 After that, handle like `on_failure_jump'. */ |
5237 case succeed_n: | 5237 case succeed_n: |
5238 EXTRACT_NUMBER (mcnt, p + 2); | 5238 EXTRACT_NUMBER (mcnt, p + 2); |
5239 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); | 5239 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); |
5240 | 5240 |
5241 assert (mcnt >= 0); | 5241 assert (mcnt >= 0); |
5242 /* Originally, this is how many times we HAVE to succeed. */ | 5242 /* Originally, this is how many times we HAVE to succeed. */ |
5243 if (mcnt > 0) | 5243 if (mcnt > 0) |
5244 { | 5244 { |
5245 mcnt--; | 5245 mcnt--; |
5246 p += 2; | 5246 p += 2; |
5247 STORE_NUMBER_AND_INCR (p, mcnt); | 5247 STORE_NUMBER_AND_INCR (p, mcnt); |
5248 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt); | 5248 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt); |
5249 } | 5249 } |
5250 else if (mcnt == 0) | 5250 else if (mcnt == 0) |
5251 { | 5251 { |
5252 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2); | 5252 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2); |
5253 p[2] = (unsigned char) no_op; | 5253 p[2] = (unsigned char) no_op; |
5254 p[3] = (unsigned char) no_op; | 5254 p[3] = (unsigned char) no_op; |
5255 goto on_failure; | 5255 goto on_failure; |
5256 } | 5256 } |
5257 break; | 5257 break; |
5258 | 5258 |
5259 case jump_n: | 5259 case jump_n: |
5260 EXTRACT_NUMBER (mcnt, p + 2); | 5260 EXTRACT_NUMBER (mcnt, p + 2); |
5261 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); | 5261 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); |
5262 | 5262 |
5263 /* Originally, this is how many times we CAN jump. */ | 5263 /* Originally, this is how many times we CAN jump. */ |
5264 if (mcnt) | 5264 if (mcnt) |
5265 { | 5265 { |
5266 mcnt--; | 5266 mcnt--; |
5267 STORE_NUMBER (p + 2, mcnt); | 5267 STORE_NUMBER (p + 2, mcnt); |
5268 goto unconditional_jump; | 5268 goto unconditional_jump; |
5269 } | 5269 } |
5270 /* If don't have to jump any more, skip over the rest of command. */ | 5270 /* If don't have to jump any more, skip over the rest of command. */ |
5271 else | 5271 else |
5272 p += 4; | 5272 p += 4; |
5273 break; | 5273 break; |
5274 | 5274 |
5275 case set_number_at: | 5275 case set_number_at: |
5276 { | 5276 { |
5277 DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); | 5277 DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); |
5278 | 5278 |
5279 EXTRACT_NUMBER_AND_INCR (mcnt, p); | 5279 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
5280 p1 = p + mcnt; | 5280 p1 = p + mcnt; |
5281 EXTRACT_NUMBER_AND_INCR (mcnt, p); | 5281 EXTRACT_NUMBER_AND_INCR (mcnt, p); |
5282 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt); | 5282 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt); |
5283 STORE_NUMBER (p1, mcnt); | 5283 STORE_NUMBER (p1, mcnt); |
5284 break; | 5284 break; |
5285 } | 5285 } |
5286 | 5286 |
5287 case wordbound: | 5287 case wordbound: |
5288 DEBUG_PRINT1 ("EXECUTING wordbound.\n"); | 5288 DEBUG_PRINT1 ("EXECUTING wordbound.\n"); |
5289 | 5289 |
5290 /* We SUCCEED in one of the following cases: */ | 5290 /* We SUCCEED in one of the following cases: */ |
5301 | 5301 |
5302 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); | 5302 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); |
5303 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); | 5303 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); |
5304 #ifdef emacs | 5304 #ifdef emacs |
5305 UPDATE_SYNTAX_TABLE (pos1 ? pos1 : 1); | 5305 UPDATE_SYNTAX_TABLE (pos1 ? pos1 : 1); |
5306 #endif | 5306 #endif |
5307 s1 = SYNTAX (c1); | 5307 s1 = SYNTAX (c1); |
5308 #ifdef emacs | 5308 #ifdef emacs |
5309 UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1); | 5309 UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1); |
5310 #endif | 5310 #endif |
5311 s2 = SYNTAX (c2); | 5311 s2 = SYNTAX (c2); |
5312 | 5312 |
5313 if (/* Case 2: Only one of S1 and S2 is Sword. */ | 5313 if (/* Case 2: Only one of S1 and S2 is Sword. */ |
5314 ((s1 == Sword) != (s2 == Sword)) | 5314 ((s1 == Sword) != (s2 == Sword)) |
5315 /* Case 3: Both of S1 and S2 are Sword, and macro | 5315 /* Case 3: Both of S1 and S2 are Sword, and macro |
5316 WORD_BOUNDARY_P (C1, C2) returns nonzero. */ | 5316 WORD_BOUNDARY_P (C1, C2) returns nonzero. */ |
5317 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2))) | 5317 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2))) |
5318 break; | 5318 break; |
5319 } | 5319 } |
5320 goto fail; | 5320 goto fail; |
5321 | 5321 |
5336 | 5336 |
5337 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); | 5337 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); |
5338 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); | 5338 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); |
5339 #ifdef emacs | 5339 #ifdef emacs |
5340 UPDATE_SYNTAX_TABLE (pos1); | 5340 UPDATE_SYNTAX_TABLE (pos1); |
5341 #endif | 5341 #endif |
5342 s1 = SYNTAX (c1); | 5342 s1 = SYNTAX (c1); |
5343 #ifdef emacs | 5343 #ifdef emacs |
5344 UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1); | 5344 UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1); |
5345 #endif | 5345 #endif |
5346 s2 = SYNTAX (c2); | 5346 s2 = SYNTAX (c2); |
5347 | 5347 |
5348 if (/* Case 2: Only one of S1 and S2 is Sword. */ | 5348 if (/* Case 2: Only one of S1 and S2 is Sword. */ |
5349 ((s1 == Sword) != (s2 == Sword)) | 5349 ((s1 == Sword) != (s2 == Sword)) |
5350 /* Case 3: Both of S1 and S2 are Sword, and macro | 5350 /* Case 3: Both of S1 and S2 are Sword, and macro |
5351 WORD_BOUNDARY_P (C1, C2) returns nonzero. */ | 5351 WORD_BOUNDARY_P (C1, C2) returns nonzero. */ |
5352 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2))) | 5352 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2))) |
5353 goto fail; | 5353 goto fail; |
5354 } | 5354 } |
5355 break; | 5355 break; |
5356 | 5356 |
5357 case wordbeg: | 5357 case wordbeg: |
5358 DEBUG_PRINT1 ("EXECUTING wordbeg.\n"); | 5358 DEBUG_PRINT1 ("EXECUTING wordbeg.\n"); |
5359 | 5359 |
5360 /* We FAIL in one of the following cases: */ | 5360 /* We FAIL in one of the following cases: */ |
5361 | 5361 |
5362 /* Case 1: D is at the end of string. */ | 5362 /* Case 1: D is at the end of string. */ |
5363 if (AT_STRINGS_END (d)) | 5363 if (AT_STRINGS_END (d)) |
5364 goto fail; | 5364 goto fail; |
5365 else | 5365 else |
5366 { | 5366 { |
5367 /* C1 is the character before D, S1 is the syntax of C1, C2 | 5367 /* C1 is the character before D, S1 is the syntax of C1, C2 |
5368 is the character at D, and S2 is the syntax of C2. */ | 5368 is the character at D, and S2 is the syntax of C2. */ |
5369 int c1, c2, s1, s2; | 5369 int c1, c2, s1, s2; |
5370 int pos1 = PTR_TO_OFFSET (d); | 5370 int pos1 = PTR_TO_OFFSET (d); |
5371 | 5371 |
5372 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); | 5372 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); |
5373 #ifdef emacs | 5373 #ifdef emacs |
5374 UPDATE_SYNTAX_TABLE (pos1); | 5374 UPDATE_SYNTAX_TABLE (pos1); |
5375 #endif | 5375 #endif |
5376 s2 = SYNTAX (c2); | 5376 s2 = SYNTAX (c2); |
5377 | 5377 |
5378 /* Case 2: S2 is not Sword. */ | 5378 /* Case 2: S2 is not Sword. */ |
5379 if (s2 != Sword) | 5379 if (s2 != Sword) |
5380 goto fail; | 5380 goto fail; |
5381 | 5381 |
5382 /* Case 3: D is not at the beginning of string ... */ | 5382 /* Case 3: D is not at the beginning of string ... */ |
5383 if (!AT_STRINGS_BEG (d)) | 5383 if (!AT_STRINGS_BEG (d)) |
5384 { | 5384 { |
5385 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); | 5385 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2); |
5386 #ifdef emacs | 5386 #ifdef emacs |
5387 UPDATE_SYNTAX_TABLE_BACKWARD (pos1 - 1); | 5387 UPDATE_SYNTAX_TABLE_BACKWARD (pos1 - 1); |
5388 #endif | 5388 #endif |
5389 s1 = SYNTAX (c1); | 5389 s1 = SYNTAX (c1); |
5390 | 5390 |
5391 /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2) | 5391 /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2) |
5392 returns 0. */ | 5392 returns 0. */ |
5393 if ((s1 == Sword) && !WORD_BOUNDARY_P (c1, c2)) | 5393 if ((s1 == Sword) && !WORD_BOUNDARY_P (c1, c2)) |
5394 goto fail; | 5394 goto fail; |
5395 } | 5395 } |
5396 } | 5396 } |
5397 break; | 5397 break; |
5398 | 5398 |
5399 case wordend: | 5399 case wordend: |
5400 DEBUG_PRINT1 ("EXECUTING wordend.\n"); | 5400 DEBUG_PRINT1 ("EXECUTING wordend.\n"); |
5401 | 5401 |
5402 /* We FAIL in one of the following cases: */ | 5402 /* We FAIL in one of the following cases: */ |
5403 | 5403 |
5404 /* Case 1: D is at the beginning of string. */ | 5404 /* Case 1: D is at the beginning of string. */ |
5405 if (AT_STRINGS_BEG (d)) | 5405 if (AT_STRINGS_BEG (d)) |
5422 { | 5422 { |
5423 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); | 5423 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2); |
5424 s2 = SYNTAX (c2); | 5424 s2 = SYNTAX (c2); |
5425 | 5425 |
5426 /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2) | 5426 /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2) |
5427 returns 0. */ | 5427 returns 0. */ |
5428 if ((s2 == Sword) && !WORD_BOUNDARY_P (c1, c2)) | 5428 if ((s2 == Sword) && !WORD_BOUNDARY_P (c1, c2)) |
5429 goto fail; | 5429 goto fail; |
5430 } | 5430 } |
5431 } | 5431 } |
5432 break; | 5432 break; |
5433 | 5433 |
5434 #ifdef emacs | 5434 #ifdef emacs |
5435 case before_dot: | 5435 case before_dot: |
5436 DEBUG_PRINT1 ("EXECUTING before_dot.\n"); | 5436 DEBUG_PRINT1 ("EXECUTING before_dot.\n"); |
5437 if (PTR_CHAR_POS ((unsigned char *) d) >= PT) | 5437 if (PTR_CHAR_POS ((unsigned char *) d) >= PT) |
5438 goto fail; | 5438 goto fail; |
5439 break; | 5439 break; |
5440 | 5440 |
5441 case at_dot: | 5441 case at_dot: |
5442 DEBUG_PRINT1 ("EXECUTING at_dot.\n"); | 5442 DEBUG_PRINT1 ("EXECUTING at_dot.\n"); |
5443 if (PTR_CHAR_POS ((unsigned char *) d) != PT) | 5443 if (PTR_CHAR_POS ((unsigned char *) d) != PT) |
5444 goto fail; | 5444 goto fail; |
5445 break; | 5445 break; |
5446 | 5446 |
5447 case after_dot: | 5447 case after_dot: |
5448 DEBUG_PRINT1 ("EXECUTING after_dot.\n"); | 5448 DEBUG_PRINT1 ("EXECUTING after_dot.\n"); |
5449 if (PTR_CHAR_POS ((unsigned char *) d) <= PT) | 5449 if (PTR_CHAR_POS ((unsigned char *) d) <= PT) |
5450 goto fail; | 5450 goto fail; |
5451 break; | 5451 break; |
5452 | 5452 |
5453 case syntaxspec: | 5453 case syntaxspec: |
5454 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); | 5454 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); |
5455 mcnt = *p++; | 5455 mcnt = *p++; |
5456 goto matchsyntax; | 5456 goto matchsyntax; |
5457 | 5457 |
5458 case wordchar: | 5458 case wordchar: |
5459 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n"); | 5459 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n"); |
5460 mcnt = (int) Sword; | 5460 mcnt = (int) Sword; |
5461 matchsyntax: | 5461 matchsyntax: |
5462 PREFETCH (); | 5462 PREFETCH (); |
5463 #ifdef emacs | 5463 #ifdef emacs |
5464 { | 5464 { |
5465 int pos1 = PTR_TO_OFFSET (d); | 5465 int pos1 = PTR_TO_OFFSET (d); |
5466 UPDATE_SYNTAX_TABLE (pos1); | 5466 UPDATE_SYNTAX_TABLE (pos1); |
5467 } | 5467 } |
5468 #endif | 5468 #endif |
5469 { | 5469 { |
5470 int c, len; | 5470 int c, len; |
5471 | 5471 |
5472 if (multibyte) | 5472 if (multibyte) |
5473 /* we must concern about multibyte form, ... */ | 5473 /* we must concern about multibyte form, ... */ |
5479 | 5479 |
5480 if (SYNTAX (c) != (enum syntaxcode) mcnt) | 5480 if (SYNTAX (c) != (enum syntaxcode) mcnt) |
5481 goto fail; | 5481 goto fail; |
5482 d += len; | 5482 d += len; |
5483 } | 5483 } |
5484 SET_REGS_MATCHED (); | 5484 SET_REGS_MATCHED (); |
5485 break; | 5485 break; |
5486 | 5486 |
5487 case notsyntaxspec: | 5487 case notsyntaxspec: |
5488 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); | 5488 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); |
5489 mcnt = *p++; | 5489 mcnt = *p++; |
5490 goto matchnotsyntax; | 5490 goto matchnotsyntax; |
5491 | 5491 |
5492 case notwordchar: | 5492 case notwordchar: |
5493 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n"); | 5493 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n"); |
5494 mcnt = (int) Sword; | 5494 mcnt = (int) Sword; |
5495 matchnotsyntax: | 5495 matchnotsyntax: |
5496 PREFETCH (); | 5496 PREFETCH (); |
5497 #ifdef emacs | 5497 #ifdef emacs |
5498 { | 5498 { |
5499 int pos1 = PTR_TO_OFFSET (d); | 5499 int pos1 = PTR_TO_OFFSET (d); |
5500 UPDATE_SYNTAX_TABLE (pos1); | 5500 UPDATE_SYNTAX_TABLE (pos1); |
5501 } | 5501 } |
5502 #endif | 5502 #endif |
5503 { | 5503 { |
5504 int c, len; | 5504 int c, len; |
5505 | 5505 |
5506 if (multibyte) | 5506 if (multibyte) |
5507 c = STRING_CHAR_AND_LENGTH (d, dend - d, len); | 5507 c = STRING_CHAR_AND_LENGTH (d, dend - d, len); |