Mercurial > hg > xemacs-beta
annotate src/regex.c @ 4750:b5f21bb36684
Fix crash in regex.c (closes issue630).
| author | Stephen J. Turnbull <stephen@xemacs.org> |
|---|---|
| date | Tue, 17 Nov 2009 23:41:39 +0900 |
| parents | 8418d1ad4944 |
| children | aa5ed11f473b |
| rev | line source |
|---|---|
| 428 | 1 /* Extended regular expression matching and search library, |
| 2 version 0.12, extended for XEmacs. | |
| 3 (Implements POSIX draft P10003.2/D11.2, except for | |
| 4 internationalization features.) | |
| 5 | |
| 6 Copyright (C) 1993, 1994, 1995 Free Software Foundation, Inc. | |
| 7 Copyright (C) 1995 Sun Microsystems, Inc. | |
| 1333 | 8 Copyright (C) 1995, 2001, 2002, 2003 Ben Wing. |
| 428 | 9 |
| 10 This program is free software; you can redistribute it and/or modify | |
| 11 it under the terms of the GNU General Public License as published by | |
| 12 the Free Software Foundation; either version 2, or (at your option) | |
| 13 any later version. | |
| 14 | |
| 15 This program is distributed in the hope that it will be useful, | |
| 16 but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 17 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
| 18 GNU General Public License for more details. | |
| 19 | |
| 20 You should have received a copy of the GNU General Public License | |
| 21 along with this program; see the file COPYING. If not, write to | |
| 22 the Free Software Foundation, Inc., 59 Temple Place - Suite 330, | |
| 23 Boston, MA 02111-1307, USA. */ | |
| 24 | |
| 25 /* Synched up with: FSF 19.29. */ | |
| 26 | |
| 27 #ifdef HAVE_CONFIG_H | |
| 28 #include <config.h> | |
| 29 #endif | |
| 30 | |
| 31 #ifndef _GNU_SOURCE | |
| 32 #define _GNU_SOURCE 1 | |
| 33 #endif | |
| 34 | |
| 35 /* We assume non-Mule if emacs isn't defined. */ | |
| 36 #ifndef emacs | |
| 37 #undef MULE | |
| 38 #endif | |
| 39 | |
| 771 | 40 /* XEmacs addition */ |
| 41 #ifdef REL_ALLOC | |
| 42 #define REGEX_REL_ALLOC /* may be undefined below */ | |
| 43 #endif | |
| 44 | |
| 428 | 45 /* XEmacs: define this to add in a speedup for patterns anchored at |
| 46 the beginning of a line. Keep the ifdefs so that it's easier to | |
| 47 tell where/why this code has diverged from v19. */ | |
| 48 #define REGEX_BEGLINE_CHECK | |
| 49 | |
| 50 /* XEmacs: the current mmap-based ralloc handles small blocks very | |
| 51 poorly, so we disable it here. */ | |
| 52 | |
| 771 | 53 #if defined (HAVE_MMAP) || defined (DOUG_LEA_MALLOC) |
| 54 # undef REGEX_REL_ALLOC | |
| 428 | 55 #endif |
| 56 | |
| 57 /* The `emacs' switch turns on certain matching commands | |
| 58 that make sense only in Emacs. */ | |
| 59 #ifdef emacs | |
| 60 | |
| 61 #include "lisp.h" | |
| 62 #include "buffer.h" | |
| 63 #include "syntax.h" | |
| 64 | |
| 65 #if (defined (DEBUG_XEMACS) && !defined (DEBUG)) | |
| 66 #define DEBUG | |
| 67 #endif | |
| 68 | |
| 867 | 69 #define RE_TRANSLATE_1(ch) TRT_TABLE_OF (translate, (Ichar) ch) |
| 446 | 70 #define TRANSLATE_P(tr) (!NILP (tr)) |
| 428 | 71 |
| 826 | 72 /* Converts the pointer to the char to BEG-based offset from the start. */ |
| 73 #define PTR_TO_OFFSET(d) (MATCHING_IN_FIRST_STRING \ | |
| 74 ? (d) - string1 : (d) - (string2 - size1)) | |
| 75 | |
| 428 | 76 #else /* not emacs */ |
| 77 | |
| 2367 | 78 #include <stdlib.h> |
| 79 #include <sys/types.h> | |
| 80 #include <stddef.h> /* needed for ptrdiff_t under Solaris */ | |
| 81 #include <string.h> | |
| 82 | |
| 2286 | 83 #include "compiler.h" /* Get compiler-specific definitions like UNUSED */ |
| 84 | |
| 2500 | 85 #define ABORT abort |
| 86 | |
| 428 | 87 /* If we are not linking with Emacs proper, |
| 88 we can't use the relocating allocator | |
| 89 even if config.h says that we can. */ | |
| 771 | 90 #undef REGEX_REL_ALLOC |
| 428 | 91 |
| 544 | 92 /* defined in lisp.h */ |
| 93 #ifdef REGEX_MALLOC | |
| 94 #ifndef DECLARE_NOTHING | |
| 95 #define DECLARE_NOTHING struct nosuchstruct | |
| 96 #endif | |
| 97 #endif | |
| 98 | |
| 867 | 99 #define itext_ichar(str) ((Ichar) (str)[0]) |
| 100 #define itext_ichar_fmt(str, fmt, object) ((Ichar) (str)[0]) | |
| 101 #define itext_ichar_ascii_fmt(str, fmt, object) ((Ichar) (str)[0]) | |
| 428 | 102 |
| 103 #if (LONGBITS > INTBITS) | |
| 104 # define EMACS_INT long | |
| 105 #else | |
| 106 # define EMACS_INT int | |
| 107 #endif | |
| 108 | |
| 867 | 109 typedef int Ichar; |
| 110 | |
| 111 #define INC_IBYTEPTR(p) ((p)++) | |
| 112 #define INC_IBYTEPTR_FMT(p, fmt) ((p)++) | |
| 113 #define DEC_IBYTEPTR(p) ((p)--) | |
| 114 #define DEC_IBYTEPTR_FMT(p, fmt) ((p)--) | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
115 #define MAX_ICHAR_LEN 1 |
| 867 | 116 #define itext_ichar_len(ptr) 1 |
| 117 #define itext_ichar_len_fmt(ptr, fmt) 1 | |
| 428 | 118 |
| 119 /* Define the syntax stuff for \<, \>, etc. */ | |
| 120 | |
| 121 /* This must be nonzero for the wordchar and notwordchar pattern | |
| 122 commands in re_match_2. */ | |
| 123 #ifndef Sword | |
| 124 #define Sword 1 | |
| 125 #endif | |
| 126 | |
| 127 #ifdef SYNTAX_TABLE | |
| 128 | |
| 129 extern char *re_syntax_table; | |
| 130 | |
| 131 #else /* not SYNTAX_TABLE */ | |
| 132 | |
| 133 /* How many characters in the character set. */ | |
| 134 #define CHAR_SET_SIZE 256 | |
| 135 | |
| 136 static char re_syntax_table[CHAR_SET_SIZE]; | |
| 137 | |
| 138 static void | |
| 139 init_syntax_once (void) | |
| 140 { | |
| 141 static int done = 0; | |
| 142 | |
| 143 if (!done) | |
| 144 { | |
| 442 | 145 const char *word_syntax_chars = |
| 428 | 146 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789_"; |
| 147 | |
| 148 memset (re_syntax_table, 0, sizeof (re_syntax_table)); | |
| 149 | |
| 150 while (*word_syntax_chars) | |
| 647 | 151 re_syntax_table[(unsigned int) (*word_syntax_chars++)] = Sword; |
| 428 | 152 |
| 153 done = 1; | |
| 154 } | |
| 155 } | |
| 156 | |
| 446 | 157 #endif /* SYNTAX_TABLE */ |
| 428 | 158 |
| 826 | 159 #define SYNTAX(ignored, c) re_syntax_table[c] |
| 460 | 160 #undef SYNTAX_FROM_CACHE |
| 826 | 161 #define SYNTAX_FROM_CACHE SYNTAX |
| 162 | |
| 163 #define RE_TRANSLATE_1(c) translate[(unsigned char) (c)] | |
| 446 | 164 #define TRANSLATE_P(tr) tr |
| 165 | |
| 166 #endif /* emacs */ | |
| 428 | 167 |
| 2201 | 168 /* This is for other GNU distributions with internationalized messages. */ |
| 169 #if defined (I18N3) && (defined (HAVE_LIBINTL_H) || defined (_LIBC)) | |
| 170 # include <libintl.h> | |
| 171 #else | |
| 172 # define gettext(msgid) (msgid) | |
| 173 #endif | |
| 174 | |
| 428 | 175 /* Under XEmacs, this is needed because we don't define it elsewhere. */ |
| 176 #ifdef SWITCH_ENUM_BUG | |
| 177 #define SWITCH_ENUM_CAST(x) ((int)(x)) | |
| 178 #else | |
| 179 #define SWITCH_ENUM_CAST(x) (x) | |
| 180 #endif | |
| 181 | |
| 182 | |
| 183 /* Get the interface, including the syntax bits. */ | |
| 184 #include "regex.h" | |
| 185 | |
| 186 /* isalpha etc. are used for the character classes. */ | |
| 187 #include <ctype.h> | |
| 188 | |
| 189 /* Jim Meyering writes: | |
| 190 | |
| 191 "... Some ctype macros are valid only for character codes that | |
| 192 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when | |
| 193 using /bin/cc or gcc but without giving an ansi option). So, all | |
| 194 ctype uses should be through macros like ISPRINT... If | |
| 195 STDC_HEADERS is defined, then autoconf has verified that the ctype | |
| 196 macros don't need to be guarded with references to isascii. ... | |
| 197 Defining isascii to 1 should let any compiler worth its salt | |
| 198 eliminate the && through constant folding." */ | |
| 199 | |
| 200 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII)) | |
| 201 #define ISASCII_1(c) 1 | |
| 202 #else | |
| 203 #define ISASCII_1(c) isascii(c) | |
| 204 #endif | |
| 205 | |
| 206 #ifdef MULE | |
| 207 /* The IS*() macros can be passed any character, including an extended | |
| 208 one. We need to make sure there are no crashes, which would occur | |
| 209 otherwise due to out-of-bounds array references. */ | |
| 210 #define ISASCII(c) (((EMACS_UINT) (c)) < 0x100 && ISASCII_1 (c)) | |
| 211 #else | |
| 212 #define ISASCII(c) ISASCII_1 (c) | |
| 213 #endif /* MULE */ | |
| 214 | |
| 215 #ifdef isblank | |
| 216 #define ISBLANK(c) (ISASCII (c) && isblank (c)) | |
| 217 #else | |
| 218 #define ISBLANK(c) ((c) == ' ' || (c) == '\t') | |
| 219 #endif | |
| 220 #ifdef isgraph | |
| 221 #define ISGRAPH(c) (ISASCII (c) && isgraph (c)) | |
| 222 #else | |
| 223 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c)) | |
| 224 #endif | |
| 225 | |
| 226 #define ISPRINT(c) (ISASCII (c) && isprint (c)) | |
| 227 #define ISDIGIT(c) (ISASCII (c) && isdigit (c)) | |
| 228 #define ISALNUM(c) (ISASCII (c) && isalnum (c)) | |
| 229 #define ISALPHA(c) (ISASCII (c) && isalpha (c)) | |
| 230 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c)) | |
| 231 #define ISLOWER(c) (ISASCII (c) && islower (c)) | |
| 232 #define ISPUNCT(c) (ISASCII (c) && ispunct (c)) | |
| 233 #define ISSPACE(c) (ISASCII (c) && isspace (c)) | |
| 234 #define ISUPPER(c) (ISASCII (c) && isupper (c)) | |
| 235 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c)) | |
| 236 | |
| 237 #ifndef NULL | |
| 238 #define NULL (void *)0 | |
| 239 #endif | |
| 240 | |
| 241 /* We remove any previous definition of `SIGN_EXTEND_CHAR', | |
| 242 since ours (we hope) works properly with all combinations of | |
| 243 machines, compilers, `char' and `unsigned char' argument types. | |
| 244 (Per Bothner suggested the basic approach.) */ | |
| 245 #undef SIGN_EXTEND_CHAR | |
| 246 #if __STDC__ | |
| 247 #define SIGN_EXTEND_CHAR(c) ((signed char) (c)) | |
| 248 #else /* not __STDC__ */ | |
| 249 /* As in Harbison and Steele. */ | |
| 250 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128) | |
| 251 #endif | |
| 252 | |
| 253 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we | |
| 254 use `alloca' instead of `malloc'. This is because using malloc in | |
| 255 re_search* or re_match* could cause memory leaks when C-g is used in | |
| 256 Emacs; also, malloc is slower and causes storage fragmentation. On | |
| 257 the other hand, malloc is more portable, and easier to debug. | |
| 258 | |
| 259 Because we sometimes use alloca, some routines have to be macros, | |
| 260 not functions -- `alloca'-allocated space disappears at the end of the | |
| 261 function it is called in. */ | |
| 262 | |
| 1333 | 263 #ifndef emacs |
| 264 #define ALLOCA alloca | |
| 265 #define xmalloc malloc | |
| 266 #define xrealloc realloc | |
| 1726 | 267 #define xfree(x,type) free (x) |
| 1333 | 268 #endif |
| 269 | |
| 270 #ifdef emacs | |
| 271 #define ALLOCA_GARBAGE_COLLECT() \ | |
| 272 do \ | |
| 273 { \ | |
| 274 if (need_to_check_c_alloca) \ | |
| 275 xemacs_c_alloca (0); \ | |
| 276 } while (0) | |
| 277 #elif defined (C_ALLOCA) | |
| 278 #define ALLOCA_GARBAGE_COLLECT() alloca (0) | |
| 279 #else | |
| 280 #define ALLOCA_GARBAGE_COLLECT() | |
| 281 #endif | |
| 282 | |
| 283 #ifndef emacs | |
| 284 /* So we can use just it to conditionalize on */ | |
| 285 #undef ERROR_CHECK_MALLOC | |
| 286 #endif | |
| 287 | |
| 288 #ifdef ERROR_CHECK_MALLOC | |
| 289 /* When REL_ALLOC, malloc() is problematic because it could potentially | |
| 290 cause all rel-alloc()ed data -- including buffer text -- to be relocated. | |
| 291 We deal with this by checking for such relocation whenever we have | |
| 292 executed a statement that may call malloc() -- or alloca(), which may | |
| 293 end up calling malloc() in some circumstances -- and recomputing all | |
| 294 of our string pointers in re_match_2_internal() and re_search_2(). | |
| 295 However, if malloc() or alloca() happens and we don't know about it, | |
| 296 we could still be screwed. So we set up a system where we indicate all | |
| 297 places where we are prepared for malloc() or alloca(), and in any | |
| 298 other circumstances, calls to those functions (from anywhere inside of | |
| 2500 | 299 XEmacs!) will ABORT(). We do this even when REL_ALLOC is not defined |
| 1333 | 300 so that we catch these problems sooner, since many developers and beta |
| 301 testers will not be running with REL_ALLOC. */ | |
| 302 int regex_malloc_disallowed; | |
| 303 #define BEGIN_REGEX_MALLOC_OK() regex_malloc_disallowed = 0 | |
| 304 #define END_REGEX_MALLOC_OK() regex_malloc_disallowed = 1 | |
| 305 #define UNBIND_REGEX_MALLOC_CHECK() unbind_to (depth) | |
| 306 #else | |
| 307 #define BEGIN_REGEX_MALLOC_OK() | |
| 308 #define END_REGEX_MALLOC_OK() | |
| 309 #define UNBIND_REGEX_MALLOC_CHECK() | |
| 310 #endif | |
| 311 | |
| 312 | |
| 428 | 313 #ifdef REGEX_MALLOC |
| 314 | |
| 1333 | 315 #define REGEX_ALLOCATE xmalloc |
| 316 #define REGEX_REALLOCATE(source, osize, nsize) xrealloc (source, nsize) | |
| 317 #define REGEX_FREE xfree | |
| 428 | 318 |
| 319 #else /* not REGEX_MALLOC */ | |
| 320 | |
| 321 /* Emacs already defines alloca, sometimes. */ | |
| 322 #ifndef alloca | |
| 323 | |
| 324 /* Make alloca work the best possible way. */ | |
| 325 #ifdef __GNUC__ | |
| 326 #define alloca __builtin_alloca | |
| 771 | 327 #elif defined (__DECC) /* XEmacs: added next 3 lines, similar to config.h.in */ |
| 328 #include <alloca.h> | |
| 329 #pragma intrinsic(alloca) | |
| 428 | 330 #else /* not __GNUC__ */ |
| 331 #if HAVE_ALLOCA_H | |
| 332 #include <alloca.h> | |
| 333 #else /* not __GNUC__ or HAVE_ALLOCA_H */ | |
| 334 #ifndef _AIX /* Already did AIX, up at the top. */ | |
| 444 | 335 void *alloca (); |
| 428 | 336 #endif /* not _AIX */ |
| 446 | 337 #endif /* HAVE_ALLOCA_H */ |
| 338 #endif /* __GNUC__ */ | |
| 428 | 339 |
| 340 #endif /* not alloca */ | |
| 341 | |
| 1333 | 342 #define REGEX_ALLOCATE ALLOCA |
| 428 | 343 |
| 2367 | 344 /* !!#### Needs review */ |
| 428 | 345 /* Assumes a `char *destination' variable. */ |
| 346 #define REGEX_REALLOCATE(source, osize, nsize) \ | |
| 1333 | 347 (destination = (char *) ALLOCA (nsize), \ |
| 428 | 348 memmove (destination, source, osize), \ |
| 349 destination) | |
| 350 | |
| 1726 | 351 /* No need to do anything to free, after alloca. |
| 352 Do nothing! But inhibit gcc warning. */ | |
| 353 #define REGEX_FREE(arg,type) ((void)0) | |
| 428 | 354 |
| 446 | 355 #endif /* REGEX_MALLOC */ |
| 428 | 356 |
| 357 /* Define how to allocate the failure stack. */ | |
| 358 | |
| 771 | 359 #ifdef REGEX_REL_ALLOC |
| 428 | 360 #define REGEX_ALLOCATE_STACK(size) \ |
| 1346 | 361 r_alloc ((unsigned char **) &failure_stack_ptr, (size)) |
| 428 | 362 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \ |
| 1346 | 363 r_re_alloc ((unsigned char **) &failure_stack_ptr, (nsize)) |
| 428 | 364 #define REGEX_FREE_STACK(ptr) \ |
| 1346 | 365 r_alloc_free ((unsigned char **) &failure_stack_ptr) |
| 428 | 366 |
| 771 | 367 #else /* not REGEX_REL_ALLOC */ |
| 428 | 368 |
| 369 #ifdef REGEX_MALLOC | |
| 370 | |
| 1333 | 371 #define REGEX_ALLOCATE_STACK xmalloc |
| 372 #define REGEX_REALLOCATE_STACK(source, osize, nsize) xrealloc (source, nsize) | |
| 1726 | 373 #define REGEX_FREE_STACK(arg) xfree (arg, fail_stack_elt_t *) |
| 428 | 374 |
| 375 #else /* not REGEX_MALLOC */ | |
| 376 | |
| 1333 | 377 #define REGEX_ALLOCATE_STACK ALLOCA |
| 428 | 378 |
| 379 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \ | |
| 380 REGEX_REALLOCATE (source, osize, nsize) | |
| 381 /* No need to explicitly free anything. */ | |
| 382 #define REGEX_FREE_STACK(arg) | |
| 383 | |
| 446 | 384 #endif /* REGEX_MALLOC */ |
| 771 | 385 #endif /* REGEX_REL_ALLOC */ |
| 428 | 386 |
| 387 | |
| 388 /* True if `size1' is non-NULL and PTR is pointing anywhere inside | |
| 389 `string1' or just past its end. This works if PTR is NULL, which is | |
| 390 a good thing. */ | |
| 391 #define FIRST_STRING_P(ptr) \ | |
| 392 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) | |
| 393 | |
| 394 /* (Re)Allocate N items of type T using malloc, or fail. */ | |
| 1333 | 395 #define TALLOC(n, t) ((t *) xmalloc ((n) * sizeof (t))) |
| 396 #define RETALLOC(addr, n, t) ((addr) = (t *) xrealloc (addr, (n) * sizeof (t))) | |
| 428 | 397 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) |
| 398 | |
| 399 #define BYTEWIDTH 8 /* In bits. */ | |
| 400 | |
| 434 | 401 #define STREQ(s1, s2) (strcmp (s1, s2) == 0) |
| 428 | 402 |
| 403 #undef MAX | |
| 404 #undef MIN | |
| 405 #define MAX(a, b) ((a) > (b) ? (a) : (b)) | |
| 406 #define MIN(a, b) ((a) < (b) ? (a) : (b)) | |
| 407 | |
| 446 | 408 /* Type of source-pattern and string chars. */ |
| 409 typedef const unsigned char re_char; | |
| 410 | |
| 460 | 411 typedef char re_bool; |
| 428 | 412 #define false 0 |
| 413 #define true 1 | |
| 414 | |
| 415 | |
| 1346 | 416 #ifdef emacs |
| 417 | |
| 418 #ifdef MULE | |
| 419 | |
| 420 Lisp_Object Vthe_lisp_rangetab; | |
| 421 | |
| 422 void | |
| 423 vars_of_regex (void) | |
| 424 { | |
| 2421 | 425 Vthe_lisp_rangetab = Fmake_range_table (Qstart_closed_end_closed); |
| 1346 | 426 staticpro (&Vthe_lisp_rangetab); |
| 427 } | |
| 428 | |
| 429 #else /* not MULE */ | |
| 430 | |
| 431 void | |
| 432 vars_of_regex (void) | |
| 433 { | |
| 434 } | |
| 435 | |
| 436 #endif /* MULE */ | |
| 437 | |
| 438 /* Convert an offset from the start of the logical text string formed by | |
| 439 concatenating the two strings together into a character position in the | |
| 440 Lisp buffer or string that the text represents. Knows that | |
| 441 when handling buffer text, the "string" we're passed in is always | |
| 442 BEGV - ZV. */ | |
| 443 | |
| 444 static Charxpos | |
| 445 offset_to_charxpos (Lisp_Object lispobj, int off) | |
| 446 { | |
| 447 if (STRINGP (lispobj)) | |
| 448 return string_index_byte_to_char (lispobj, off); | |
| 449 else if (BUFFERP (lispobj)) | |
| 450 return bytebpos_to_charbpos (XBUFFER (lispobj), | |
| 451 off + BYTE_BUF_BEGV (XBUFFER (lispobj))); | |
| 452 else | |
| 453 return 0; | |
| 454 } | |
| 455 | |
| 456 #ifdef REL_ALLOC | |
| 457 | |
| 458 /* STRING1 is the value of STRING1 given to re_match_2(). LISPOBJ is | |
| 459 the Lisp object (if any) from which the string is taken. If LISPOBJ | |
| 460 is a buffer, return a relocation offset to be added to all pointers to | |
| 461 string data so that they will be accurate again, after an allocation or | |
| 462 reallocation that potentially relocated the buffer data. | |
| 463 */ | |
| 464 static Bytecount | |
| 465 offset_post_relocation (Lisp_Object lispobj, Ibyte *orig_buftext) | |
| 466 { | |
| 467 if (!BUFFERP (lispobj)) | |
| 468 return 0; | |
| 469 return (BYTE_BUF_BYTE_ADDRESS (XBUFFER (lispobj), | |
| 470 BYTE_BUF_BEGV (XBUFFER (lispobj))) - | |
| 471 orig_buftext); | |
| 472 } | |
| 473 | |
| 474 #endif /* REL_ALLOC */ | |
| 475 | |
| 476 #ifdef ERROR_CHECK_MALLOC | |
| 477 | |
| 478 /* NOTE that this can run malloc() so you need to adjust afterwards. */ | |
| 479 | |
| 480 static int | |
| 481 bind_regex_malloc_disallowed (int value) | |
| 482 { | |
| 483 /* Tricky, because the act of binding can run malloc(). */ | |
| 484 int old_regex_malloc_disallowed = regex_malloc_disallowed; | |
| 485 int depth; | |
| 486 regex_malloc_disallowed = 0; | |
| 487 depth = record_unwind_protect_restoring_int (®ex_malloc_disallowed, | |
| 488 old_regex_malloc_disallowed); | |
| 489 regex_malloc_disallowed = value; | |
| 490 return depth; | |
| 491 } | |
| 492 | |
| 493 #endif /* ERROR_CHECK_MALLOC */ | |
| 494 | |
| 495 #endif /* emacs */ | |
| 496 | |
| 497 | |
| 428 | 498 /* These are the command codes that appear in compiled regular |
| 499 expressions. Some opcodes are followed by argument bytes. A | |
| 500 command code can specify any interpretation whatsoever for its | |
| 501 arguments. Zero bytes may appear in the compiled regular expression. */ | |
| 502 | |
| 503 typedef enum | |
| 504 { | |
| 505 no_op = 0, | |
| 506 | |
| 507 /* Succeed right away--no more backtracking. */ | |
| 508 succeed, | |
| 509 | |
| 510 /* Followed by one byte giving n, then by n literal bytes. */ | |
| 511 exactn, | |
| 512 | |
| 513 /* Matches any (more or less) character. */ | |
| 514 anychar, | |
| 515 | |
| 516 /* Matches any one char belonging to specified set. First | |
| 517 following byte is number of bitmap bytes. Then come bytes | |
| 518 for a bitmap saying which chars are in. Bits in each byte | |
| 519 are ordered low-bit-first. A character is in the set if its | |
| 520 bit is 1. A character too large to have a bit in the map is | |
| 521 automatically not in the set. */ | |
| 522 charset, | |
| 523 | |
| 524 /* Same parameters as charset, but match any character that is | |
| 525 not one of those specified. */ | |
| 526 charset_not, | |
| 527 | |
| 528 /* Start remembering the text that is matched, for storing in a | |
| 529 register. Followed by one byte with the register number, in | |
| 502 | 530 the range 1 to the pattern buffer's re_ngroups |
| 428 | 531 field. Then followed by one byte with the number of groups |
| 532 inner to this one. (This last has to be part of the | |
| 533 start_memory only because we need it in the on_failure_jump | |
| 534 of re_match_2.) */ | |
| 535 start_memory, | |
| 536 | |
| 537 /* Stop remembering the text that is matched and store it in a | |
| 538 memory register. Followed by one byte with the register | |
| 502 | 539 number, in the range 1 to `re_ngroups' in the |
| 428 | 540 pattern buffer, and one byte with the number of inner groups, |
| 541 just like `start_memory'. (We need the number of inner | |
| 542 groups here because we don't have any easy way of finding the | |
| 543 corresponding start_memory when we're at a stop_memory.) */ | |
| 544 stop_memory, | |
| 545 | |
| 546 /* Match a duplicate of something remembered. Followed by one | |
| 547 byte containing the register number. */ | |
| 548 duplicate, | |
| 549 | |
| 550 /* Fail unless at beginning of line. */ | |
| 551 begline, | |
| 552 | |
| 553 /* Fail unless at end of line. */ | |
| 554 endline, | |
| 555 | |
| 556 /* Succeeds if at beginning of buffer (if emacs) or at beginning | |
| 557 of string to be matched (if not). */ | |
| 558 begbuf, | |
| 559 | |
| 560 /* Analogously, for end of buffer/string. */ | |
| 561 endbuf, | |
| 562 | |
| 563 /* Followed by two byte relative address to which to jump. */ | |
| 564 jump, | |
| 565 | |
| 566 /* Same as jump, but marks the end of an alternative. */ | |
| 567 jump_past_alt, | |
| 568 | |
| 569 /* Followed by two-byte relative address of place to resume at | |
| 570 in case of failure. */ | |
| 571 on_failure_jump, | |
| 572 | |
| 573 /* Like on_failure_jump, but pushes a placeholder instead of the | |
| 574 current string position when executed. */ | |
| 575 on_failure_keep_string_jump, | |
| 576 | |
| 577 /* Throw away latest failure point and then jump to following | |
| 578 two-byte relative address. */ | |
| 579 pop_failure_jump, | |
| 580 | |
| 581 /* Change to pop_failure_jump if know won't have to backtrack to | |
| 582 match; otherwise change to jump. This is used to jump | |
| 583 back to the beginning of a repeat. If what follows this jump | |
| 584 clearly won't match what the repeat does, such that we can be | |
| 585 sure that there is no use backtracking out of repetitions | |
| 586 already matched, then we change it to a pop_failure_jump. | |
| 587 Followed by two-byte address. */ | |
| 588 maybe_pop_jump, | |
| 589 | |
| 590 /* Jump to following two-byte address, and push a dummy failure | |
| 591 point. This failure point will be thrown away if an attempt | |
| 592 is made to use it for a failure. A `+' construct makes this | |
| 593 before the first repeat. Also used as an intermediary kind | |
| 594 of jump when compiling an alternative. */ | |
| 595 dummy_failure_jump, | |
| 596 | |
| 597 /* Push a dummy failure point and continue. Used at the end of | |
| 598 alternatives. */ | |
| 599 push_dummy_failure, | |
| 600 | |
| 601 /* Followed by two-byte relative address and two-byte number n. | |
| 602 After matching N times, jump to the address upon failure. */ | |
| 603 succeed_n, | |
| 604 | |
| 605 /* Followed by two-byte relative address, and two-byte number n. | |
| 606 Jump to the address N times, then fail. */ | |
| 607 jump_n, | |
| 608 | |
| 609 /* Set the following two-byte relative address to the | |
| 610 subsequent two-byte number. The address *includes* the two | |
| 611 bytes of number. */ | |
| 612 set_number_at, | |
| 613 | |
| 614 wordchar, /* Matches any word-constituent character. */ | |
| 615 notwordchar, /* Matches any char that is not a word-constituent. */ | |
| 616 | |
| 617 wordbeg, /* Succeeds if at word beginning. */ | |
| 618 wordend, /* Succeeds if at word end. */ | |
| 619 | |
| 620 wordbound, /* Succeeds if at a word boundary. */ | |
| 621 notwordbound /* Succeeds if not at a word boundary. */ | |
| 622 | |
| 623 #ifdef emacs | |
| 624 ,before_dot, /* Succeeds if before point. */ | |
| 625 at_dot, /* Succeeds if at point. */ | |
| 626 after_dot, /* Succeeds if after point. */ | |
| 627 | |
| 628 /* Matches any character whose syntax is specified. Followed by | |
| 629 a byte which contains a syntax code, e.g., Sword. */ | |
| 630 syntaxspec, | |
| 631 | |
| 632 /* Matches any character whose syntax is not that specified. */ | |
| 633 notsyntaxspec | |
| 634 | |
| 635 #endif /* emacs */ | |
| 636 | |
| 637 #ifdef MULE | |
| 638 /* need extra stuff to be able to properly work with XEmacs/Mule | |
| 639 characters (which may take up more than one byte) */ | |
| 640 | |
| 641 ,charset_mule, /* Matches any character belonging to specified set. | |
| 642 The set is stored in "unified range-table | |
| 643 format"; see rangetab.c. Unlike the `charset' | |
| 644 opcode, this can handle arbitrary characters. */ | |
| 645 | |
| 646 charset_mule_not /* Same parameters as charset_mule, but match any | |
| 647 character that is not one of those specified. */ | |
| 648 | |
| 649 /* 97/2/17 jhod: The following two were merged back in from the Mule | |
| 650 2.3 code to enable some language specific processing */ | |
| 651 ,categoryspec, /* Matches entries in the character category tables */ | |
| 652 notcategoryspec /* The opposite of the above */ | |
| 653 #endif /* MULE */ | |
| 654 | |
| 655 } re_opcode_t; | |
| 656 | |
| 657 /* Common operations on the compiled pattern. */ | |
| 658 | |
| 659 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */ | |
| 660 | |
| 661 #define STORE_NUMBER(destination, number) \ | |
| 662 do { \ | |
| 663 (destination)[0] = (number) & 0377; \ | |
| 664 (destination)[1] = (number) >> 8; \ | |
| 665 } while (0) | |
| 666 | |
| 667 /* Same as STORE_NUMBER, except increment DESTINATION to | |
| 668 the byte after where the number is stored. Therefore, DESTINATION | |
| 669 must be an lvalue. */ | |
| 670 | |
| 671 #define STORE_NUMBER_AND_INCR(destination, number) \ | |
| 672 do { \ | |
| 673 STORE_NUMBER (destination, number); \ | |
| 674 (destination) += 2; \ | |
| 675 } while (0) | |
| 676 | |
| 677 /* Put into DESTINATION a number stored in two contiguous bytes starting | |
| 678 at SOURCE. */ | |
| 679 | |
| 680 #define EXTRACT_NUMBER(destination, source) \ | |
| 681 do { \ | |
| 682 (destination) = *(source) & 0377; \ | |
| 683 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \ | |
| 684 } while (0) | |
| 685 | |
| 686 #ifdef DEBUG | |
| 687 static void | |
| 446 | 688 extract_number (int *dest, re_char *source) |
| 428 | 689 { |
| 690 int temp = SIGN_EXTEND_CHAR (*(source + 1)); | |
| 691 *dest = *source & 0377; | |
| 692 *dest += temp << 8; | |
| 693 } | |
| 694 | |
| 695 #ifndef EXTRACT_MACROS /* To debug the macros. */ | |
| 696 #undef EXTRACT_NUMBER | |
| 697 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) | |
| 698 #endif /* not EXTRACT_MACROS */ | |
| 699 | |
| 700 #endif /* DEBUG */ | |
| 701 | |
| 702 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number. | |
| 703 SOURCE must be an lvalue. */ | |
| 704 | |
| 705 #define EXTRACT_NUMBER_AND_INCR(destination, source) \ | |
| 706 do { \ | |
| 707 EXTRACT_NUMBER (destination, source); \ | |
| 708 (source) += 2; \ | |
| 709 } while (0) | |
| 710 | |
| 711 #ifdef DEBUG | |
| 712 static void | |
| 713 extract_number_and_incr (int *destination, unsigned char **source) | |
| 714 { | |
| 715 extract_number (destination, *source); | |
| 716 *source += 2; | |
| 717 } | |
| 718 | |
| 719 #ifndef EXTRACT_MACROS | |
| 720 #undef EXTRACT_NUMBER_AND_INCR | |
| 721 #define EXTRACT_NUMBER_AND_INCR(dest, src) \ | |
| 722 extract_number_and_incr (&dest, &src) | |
| 723 #endif /* not EXTRACT_MACROS */ | |
| 724 | |
| 725 #endif /* DEBUG */ | |
| 726 | |
| 727 /* If DEBUG is defined, Regex prints many voluminous messages about what | |
| 728 it is doing (if the variable `debug' is nonzero). If linked with the | |
| 729 main program in `iregex.c', you can enter patterns and strings | |
| 730 interactively. And if linked with the main program in `main.c' and | |
| 731 the other test files, you can run the already-written tests. */ | |
| 732 | |
| 733 #if defined (DEBUG) | |
| 734 | |
| 735 /* We use standard I/O for debugging. */ | |
| 736 #include <stdio.h> | |
| 737 | |
| 738 #ifndef emacs | |
| 739 /* XEmacs provides its own version of assert() */ | |
| 740 /* It is useful to test things that ``must'' be true when debugging. */ | |
| 741 #include <assert.h> | |
| 742 #endif | |
| 743 | |
| 744 static int debug = 0; | |
| 745 | |
| 746 #define DEBUG_STATEMENT(e) e | |
| 747 #define DEBUG_PRINT1(x) if (debug) printf (x) | |
| 748 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2) | |
| 749 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3) | |
| 750 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4) | |
| 751 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ | |
| 752 if (debug) print_partial_compiled_pattern (s, e) | |
| 753 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ | |
| 754 if (debug) print_double_string (w, s1, sz1, s2, sz2) | |
| 755 | |
| 756 | |
| 757 /* Print the fastmap in human-readable form. */ | |
| 758 | |
| 759 static void | |
| 760 print_fastmap (char *fastmap) | |
| 761 { | |
| 647 | 762 int was_a_range = 0; |
| 763 int i = 0; | |
| 428 | 764 |
| 765 while (i < (1 << BYTEWIDTH)) | |
| 766 { | |
| 767 if (fastmap[i++]) | |
| 768 { | |
| 769 was_a_range = 0; | |
| 770 putchar (i - 1); | |
| 771 while (i < (1 << BYTEWIDTH) && fastmap[i]) | |
| 772 { | |
| 773 was_a_range = 1; | |
| 774 i++; | |
| 775 } | |
| 776 if (was_a_range) | |
| 777 { | |
| 778 putchar ('-'); | |
| 779 putchar (i - 1); | |
| 780 } | |
| 781 } | |
| 782 } | |
| 783 putchar ('\n'); | |
| 784 } | |
| 785 | |
| 786 | |
| 787 /* Print a compiled pattern string in human-readable form, starting at | |
| 788 the START pointer into it and ending just before the pointer END. */ | |
| 789 | |
| 790 static void | |
| 446 | 791 print_partial_compiled_pattern (re_char *start, re_char *end) |
| 428 | 792 { |
| 793 int mcnt, mcnt2; | |
| 446 | 794 unsigned char *p = (unsigned char *) start; |
| 795 re_char *pend = end; | |
| 428 | 796 |
| 797 if (start == NULL) | |
| 798 { | |
| 799 puts ("(null)"); | |
| 800 return; | |
| 801 } | |
| 802 | |
| 803 /* Loop over pattern commands. */ | |
| 804 while (p < pend) | |
| 805 { | |
| 806 printf ("%ld:\t", (long)(p - start)); | |
| 807 | |
| 808 switch ((re_opcode_t) *p++) | |
| 809 { | |
| 810 case no_op: | |
| 811 printf ("/no_op"); | |
| 812 break; | |
| 813 | |
| 814 case exactn: | |
| 815 mcnt = *p++; | |
| 816 printf ("/exactn/%d", mcnt); | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
817 while (mcnt--) |
| 428 | 818 { |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
819 putchar ('/'); |
| 428 | 820 putchar (*p++); |
| 821 } | |
| 822 break; | |
| 823 | |
| 824 case start_memory: | |
| 825 mcnt = *p++; | |
| 826 printf ("/start_memory/%d/%d", mcnt, *p++); | |
| 827 break; | |
| 828 | |
| 829 case stop_memory: | |
| 830 mcnt = *p++; | |
| 831 printf ("/stop_memory/%d/%d", mcnt, *p++); | |
| 832 break; | |
| 833 | |
| 834 case duplicate: | |
| 835 printf ("/duplicate/%d", *p++); | |
| 836 break; | |
| 837 | |
| 838 case anychar: | |
| 839 printf ("/anychar"); | |
| 840 break; | |
| 841 | |
| 842 case charset: | |
| 843 case charset_not: | |
| 844 { | |
| 845 REGISTER int c, last = -100; | |
| 846 REGISTER int in_range = 0; | |
| 847 | |
| 848 printf ("/charset [%s", | |
| 849 (re_opcode_t) *(p - 1) == charset_not ? "^" : ""); | |
| 850 | |
| 851 assert (p + *p < pend); | |
| 852 | |
| 853 for (c = 0; c < 256; c++) | |
| 854 if (((unsigned char) (c / 8) < *p) | |
| 855 && (p[1 + (c/8)] & (1 << (c % 8)))) | |
| 856 { | |
| 857 /* Are we starting a range? */ | |
| 858 if (last + 1 == c && ! in_range) | |
| 859 { | |
| 860 putchar ('-'); | |
| 861 in_range = 1; | |
| 862 } | |
| 863 /* Have we broken a range? */ | |
| 864 else if (last + 1 != c && in_range) | |
| 865 { | |
| 866 putchar (last); | |
| 867 in_range = 0; | |
| 868 } | |
| 869 | |
| 870 if (! in_range) | |
| 871 putchar (c); | |
| 872 | |
| 873 last = c; | |
| 874 } | |
| 875 | |
| 876 if (in_range) | |
| 877 putchar (last); | |
| 878 | |
| 879 putchar (']'); | |
| 880 | |
| 881 p += 1 + *p; | |
| 882 } | |
| 883 break; | |
| 884 | |
| 885 #ifdef MULE | |
| 886 case charset_mule: | |
| 887 case charset_mule_not: | |
| 888 { | |
| 889 int nentries, i; | |
| 890 | |
| 891 printf ("/charset_mule [%s", | |
| 892 (re_opcode_t) *(p - 1) == charset_mule_not ? "^" : ""); | |
| 893 nentries = unified_range_table_nentries (p); | |
| 894 for (i = 0; i < nentries; i++) | |
| 895 { | |
| 896 EMACS_INT first, last; | |
| 897 Lisp_Object dummy_val; | |
| 898 | |
| 899 unified_range_table_get_range (p, i, &first, &last, | |
| 900 &dummy_val); | |
| 901 if (first < 0x100) | |
| 902 putchar (first); | |
| 903 else | |
| 904 printf ("(0x%lx)", (long)first); | |
| 905 if (first != last) | |
| 906 { | |
| 907 putchar ('-'); | |
| 908 if (last < 0x100) | |
| 909 putchar (last); | |
| 910 else | |
| 911 printf ("(0x%lx)", (long)last); | |
| 912 } | |
| 913 } | |
| 914 putchar (']'); | |
| 915 p += unified_range_table_bytes_used (p); | |
| 916 } | |
| 917 break; | |
| 918 #endif | |
| 919 | |
| 920 case begline: | |
| 921 printf ("/begline"); | |
| 922 break; | |
| 923 | |
| 924 case endline: | |
| 925 printf ("/endline"); | |
| 926 break; | |
| 927 | |
| 928 case on_failure_jump: | |
| 929 extract_number_and_incr (&mcnt, &p); | |
| 930 printf ("/on_failure_jump to %ld", (long)(p + mcnt - start)); | |
| 931 break; | |
| 932 | |
| 933 case on_failure_keep_string_jump: | |
| 934 extract_number_and_incr (&mcnt, &p); | |
| 935 printf ("/on_failure_keep_string_jump to %ld", (long)(p + mcnt - start)); | |
| 936 break; | |
| 937 | |
| 938 case dummy_failure_jump: | |
| 939 extract_number_and_incr (&mcnt, &p); | |
| 940 printf ("/dummy_failure_jump to %ld", (long)(p + mcnt - start)); | |
| 941 break; | |
| 942 | |
| 943 case push_dummy_failure: | |
| 944 printf ("/push_dummy_failure"); | |
| 945 break; | |
| 946 | |
| 947 case maybe_pop_jump: | |
| 948 extract_number_and_incr (&mcnt, &p); | |
| 949 printf ("/maybe_pop_jump to %ld", (long)(p + mcnt - start)); | |
| 950 break; | |
| 951 | |
| 952 case pop_failure_jump: | |
| 953 extract_number_and_incr (&mcnt, &p); | |
| 954 printf ("/pop_failure_jump to %ld", (long)(p + mcnt - start)); | |
| 955 break; | |
| 956 | |
| 957 case jump_past_alt: | |
| 958 extract_number_and_incr (&mcnt, &p); | |
| 959 printf ("/jump_past_alt to %ld", (long)(p + mcnt - start)); | |
| 960 break; | |
| 961 | |
| 962 case jump: | |
| 963 extract_number_and_incr (&mcnt, &p); | |
| 964 printf ("/jump to %ld", (long)(p + mcnt - start)); | |
| 965 break; | |
| 966 | |
| 967 case succeed_n: | |
| 968 extract_number_and_incr (&mcnt, &p); | |
| 969 extract_number_and_incr (&mcnt2, &p); | |
| 970 printf ("/succeed_n to %ld, %d times", (long)(p + mcnt - start), mcnt2); | |
| 971 break; | |
| 972 | |
| 973 case jump_n: | |
| 974 extract_number_and_incr (&mcnt, &p); | |
| 975 extract_number_and_incr (&mcnt2, &p); | |
| 976 printf ("/jump_n to %ld, %d times", (long)(p + mcnt - start), mcnt2); | |
| 977 break; | |
| 978 | |
| 979 case set_number_at: | |
| 980 extract_number_and_incr (&mcnt, &p); | |
| 981 extract_number_and_incr (&mcnt2, &p); | |
| 982 printf ("/set_number_at location %ld to %d", (long)(p + mcnt - start), mcnt2); | |
| 983 break; | |
| 984 | |
| 985 case wordbound: | |
| 986 printf ("/wordbound"); | |
| 987 break; | |
| 988 | |
| 989 case notwordbound: | |
| 990 printf ("/notwordbound"); | |
| 991 break; | |
| 992 | |
| 993 case wordbeg: | |
| 994 printf ("/wordbeg"); | |
| 995 break; | |
| 996 | |
| 997 case wordend: | |
| 998 printf ("/wordend"); | |
| 999 | |
| 1000 #ifdef emacs | |
| 1001 case before_dot: | |
| 1002 printf ("/before_dot"); | |
| 1003 break; | |
| 1004 | |
| 1005 case at_dot: | |
| 1006 printf ("/at_dot"); | |
| 1007 break; | |
| 1008 | |
| 1009 case after_dot: | |
| 1010 printf ("/after_dot"); | |
| 1011 break; | |
| 1012 | |
| 1013 case syntaxspec: | |
| 1014 printf ("/syntaxspec"); | |
| 1015 mcnt = *p++; | |
| 1016 printf ("/%d", mcnt); | |
| 1017 break; | |
| 1018 | |
| 1019 case notsyntaxspec: | |
| 1020 printf ("/notsyntaxspec"); | |
| 1021 mcnt = *p++; | |
| 1022 printf ("/%d", mcnt); | |
| 1023 break; | |
| 1024 | |
| 1025 #ifdef MULE | |
| 1026 /* 97/2/17 jhod Mule category patch */ | |
| 1027 case categoryspec: | |
| 1028 printf ("/categoryspec"); | |
| 1029 mcnt = *p++; | |
| 1030 printf ("/%d", mcnt); | |
| 1031 break; | |
| 1032 | |
| 1033 case notcategoryspec: | |
| 1034 printf ("/notcategoryspec"); | |
| 1035 mcnt = *p++; | |
| 1036 printf ("/%d", mcnt); | |
| 1037 break; | |
| 1038 /* end of category patch */ | |
| 1039 #endif /* MULE */ | |
| 1040 #endif /* emacs */ | |
| 1041 | |
| 1042 case wordchar: | |
| 1043 printf ("/wordchar"); | |
| 1044 break; | |
| 1045 | |
| 1046 case notwordchar: | |
| 1047 printf ("/notwordchar"); | |
| 1048 break; | |
| 1049 | |
| 1050 case begbuf: | |
| 1051 printf ("/begbuf"); | |
| 1052 break; | |
| 1053 | |
| 1054 case endbuf: | |
| 1055 printf ("/endbuf"); | |
| 1056 break; | |
| 1057 | |
| 1058 default: | |
| 1059 printf ("?%d", *(p-1)); | |
| 1060 } | |
| 1061 | |
| 1062 putchar ('\n'); | |
| 1063 } | |
| 1064 | |
| 1065 printf ("%ld:\tend of pattern.\n", (long)(p - start)); | |
| 1066 } | |
| 1067 | |
| 1068 | |
| 1069 static void | |
| 1070 print_compiled_pattern (struct re_pattern_buffer *bufp) | |
| 1071 { | |
| 446 | 1072 re_char *buffer = bufp->buffer; |
| 428 | 1073 |
| 1074 print_partial_compiled_pattern (buffer, buffer + bufp->used); | |
| 1075 printf ("%ld bytes used/%ld bytes allocated.\n", bufp->used, | |
| 1076 bufp->allocated); | |
| 1077 | |
| 1078 if (bufp->fastmap_accurate && bufp->fastmap) | |
| 1079 { | |
| 1080 printf ("fastmap: "); | |
| 1081 print_fastmap (bufp->fastmap); | |
| 1082 } | |
| 1083 | |
| 1084 printf ("re_nsub: %ld\t", (long)bufp->re_nsub); | |
| 502 | 1085 printf ("re_ngroups: %ld\t", (long)bufp->re_ngroups); |
| 428 | 1086 printf ("regs_alloc: %d\t", bufp->regs_allocated); |
| 1087 printf ("can_be_null: %d\t", bufp->can_be_null); | |
| 1088 printf ("newline_anchor: %d\n", bufp->newline_anchor); | |
| 1089 printf ("no_sub: %d\t", bufp->no_sub); | |
| 1090 printf ("not_bol: %d\t", bufp->not_bol); | |
| 1091 printf ("not_eol: %d\t", bufp->not_eol); | |
| 1092 printf ("syntax: %d\n", bufp->syntax); | |
| 1093 /* Perhaps we should print the translate table? */ | |
| 1094 /* and maybe the category table? */ | |
| 502 | 1095 |
| 1096 if (bufp->external_to_internal_register) | |
| 1097 { | |
| 1098 int i; | |
| 1099 | |
| 1100 printf ("external_to_internal_register:\n"); | |
| 1101 for (i = 0; i <= bufp->re_nsub; i++) | |
| 1102 { | |
| 1103 if (i > 0) | |
| 1104 printf (", "); | |
| 1105 printf ("%d -> %d", i, bufp->external_to_internal_register[i]); | |
| 1106 } | |
| 1107 printf ("\n"); | |
| 1108 } | |
| 428 | 1109 } |
| 1110 | |
| 1111 | |
| 1112 static void | |
| 446 | 1113 print_double_string (re_char *where, re_char *string1, int size1, |
| 1114 re_char *string2, int size2) | |
| 428 | 1115 { |
| 1116 if (where == NULL) | |
| 1117 printf ("(null)"); | |
| 1118 else | |
| 1119 { | |
| 647 | 1120 int this_char; |
| 428 | 1121 |
| 1122 if (FIRST_STRING_P (where)) | |
| 1123 { | |
| 1124 for (this_char = where - string1; this_char < size1; this_char++) | |
| 1125 putchar (string1[this_char]); | |
| 1126 | |
| 1127 where = string2; | |
| 1128 } | |
| 1129 | |
| 1130 for (this_char = where - string2; this_char < size2; this_char++) | |
| 1131 putchar (string2[this_char]); | |
| 1132 } | |
| 1133 } | |
| 1134 | |
| 1135 #else /* not DEBUG */ | |
| 1136 | |
| 771 | 1137 #ifndef emacs |
| 428 | 1138 #undef assert |
| 771 | 1139 #define assert(e) ((void) (1)) |
| 1140 #endif | |
| 428 | 1141 |
| 1142 #define DEBUG_STATEMENT(e) | |
| 1143 #define DEBUG_PRINT1(x) | |
| 1144 #define DEBUG_PRINT2(x1, x2) | |
| 1145 #define DEBUG_PRINT3(x1, x2, x3) | |
| 1146 #define DEBUG_PRINT4(x1, x2, x3, x4) | |
| 1147 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) | |
| 1148 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) | |
| 1149 | |
| 446 | 1150 #endif /* DEBUG */ |
| 428 | 1151 |
| 1152 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can | |
| 1153 also be assigned to arbitrarily: each pattern buffer stores its own | |
| 1154 syntax, so it can be changed between regex compilations. */ | |
| 1155 /* This has no initializer because initialized variables in Emacs | |
| 1156 become read-only after dumping. */ | |
| 1157 reg_syntax_t re_syntax_options; | |
| 1158 | |
| 1159 | |
| 1160 /* Specify the precise syntax of regexps for compilation. This provides | |
| 1161 for compatibility for various utilities which historically have | |
| 1162 different, incompatible syntaxes. | |
| 1163 | |
| 1164 The argument SYNTAX is a bit mask comprised of the various bits | |
| 1165 defined in regex.h. We return the old syntax. */ | |
| 1166 | |
| 1167 reg_syntax_t | |
| 1168 re_set_syntax (reg_syntax_t syntax) | |
| 1169 { | |
| 1170 reg_syntax_t ret = re_syntax_options; | |
| 1171 | |
| 1172 re_syntax_options = syntax; | |
| 1173 return ret; | |
| 1174 } | |
| 1175 | |
| 1176 /* This table gives an error message for each of the error codes listed | |
| 1177 in regex.h. Obviously the order here has to be same as there. | |
| 1178 POSIX doesn't require that we do anything for REG_NOERROR, | |
| 1179 but why not be nice? */ | |
| 1180 | |
| 442 | 1181 static const char *re_error_msgid[] = |
| 428 | 1182 { |
| 1183 "Success", /* REG_NOERROR */ | |
| 1184 "No match", /* REG_NOMATCH */ | |
| 1185 "Invalid regular expression", /* REG_BADPAT */ | |
| 1186 "Invalid collation character", /* REG_ECOLLATE */ | |
| 1187 "Invalid character class name", /* REG_ECTYPE */ | |
| 1188 "Trailing backslash", /* REG_EESCAPE */ | |
| 1189 "Invalid back reference", /* REG_ESUBREG */ | |
| 1190 "Unmatched [ or [^", /* REG_EBRACK */ | |
| 1191 "Unmatched ( or \\(", /* REG_EPAREN */ | |
| 1192 "Unmatched \\{", /* REG_EBRACE */ | |
| 1193 "Invalid content of \\{\\}", /* REG_BADBR */ | |
| 1194 "Invalid range end", /* REG_ERANGE */ | |
| 1195 "Memory exhausted", /* REG_ESPACE */ | |
| 1196 "Invalid preceding regular expression", /* REG_BADRPT */ | |
| 1197 "Premature end of regular expression", /* REG_EEND */ | |
| 1198 "Regular expression too big", /* REG_ESIZE */ | |
| 1199 "Unmatched ) or \\)", /* REG_ERPAREN */ | |
| 1200 #ifdef emacs | |
| 1201 "Invalid syntax designator", /* REG_ESYNTAX */ | |
| 1202 #endif | |
| 1203 #ifdef MULE | |
| 1204 "Ranges may not span charsets", /* REG_ERANGESPAN */ | |
| 1205 "Invalid category designator", /* REG_ECATEGORY */ | |
| 1206 #endif | |
| 1207 }; | |
| 1208 | |
| 1209 /* Avoiding alloca during matching, to placate r_alloc. */ | |
| 1210 | |
| 1333 | 1211 /* About these various flags: |
| 1212 | |
| 1213 MATCH_MAY_ALLOCATE indicates that it's OK to do allocation in the | |
| 1214 searching and matching functions. In this case, we use local variables | |
| 1215 to hold the values allocated. If not, we use *global* variables, which | |
| 1216 are pre-allocated. NOTE: XEmacs ***MUST*** run with MATCH_MAY_ALLOCATE, | |
| 1217 because the regexp routines may get called reentrantly as a result of | |
| 1218 QUIT processing (e.g. under Windows: re_match -> QUIT -> quit_p -> drain | |
| 1219 events -> process WM_INITMENU -> call filter -> re_match; see stack | |
| 1220 trace in signal.c), so we cannot have any global variables (unless we do | |
| 1221 lots of trickiness including some unwind-protects, which isn't worth it | |
| 1222 at this point). | |
| 1223 | |
| 1224 REL_ALLOC means that the relocating allocator is in use, for buffers | |
| 1225 and such. REGEX_REL_ALLOC means that we use rel-alloc to manage the | |
| 1226 fail stack, which may grow quite large. REGEX_MALLOC means we use | |
| 1227 malloc() in place of alloca() to allocate the fail stack -- only | |
| 1228 applicable if REGEX_REL_ALLOC is not defined. | |
| 1229 */ | |
| 1230 | |
| 428 | 1231 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the |
| 1232 searching and matching functions should not call alloca. On some | |
| 1233 systems, alloca is implemented in terms of malloc, and if we're | |
| 1234 using the relocating allocator routines, then malloc could cause a | |
| 1235 relocation, which might (if the strings being searched are in the | |
| 1236 ralloc heap) shift the data out from underneath the regexp | |
| 771 | 1237 routines. [To clarify: The purpose of rel-alloc is to allow data to |
| 1238 be moved in memory from one place to another so that all data | |
| 1239 blocks can be consolidated together and excess memory released back | |
| 1240 to the operating system. This requires that all the blocks that | |
| 1241 are managed by rel-alloc go at the very end of the program's heap, | |
| 1242 after all regularly malloc()ed data. malloc(), however, is used to | |
| 1243 owning the end of the heap, so that when more memory is needed, it | |
| 1244 just expands the heap using sbrk(). This is reconciled by using a | |
| 1245 malloc() (such as malloc.c, gmalloc.c, or recent versions of | |
| 1246 malloc() in libc) where the sbrk() call can be replaced with a | |
| 1247 user-specified call -- in this case, to rel-alloc's r_alloc_sbrk() | |
| 1248 routine. This routine calls the real sbrk(), but then shifts all | |
| 1249 the rel-alloc-managed blocks forward to the end of the heap again, | |
| 1250 so that malloc() gets the memory it needs in the location it needs | |
| 1251 it at. The regex routines may well have pointers to buffer data as | |
| 1252 their arguments, and buffers are managed by rel-alloc if rel-alloc | |
| 1253 has been enabled, so calling malloc() may potentially screw things | |
| 1254 up badly if it runs out of space and asks for more from the OS.] | |
| 1255 | |
| 1256 [[Here's another reason to avoid allocation: Emacs processes input | |
| 1257 from X in a signal handler; processing X input may call malloc; if | |
| 1258 input arrives while a matching routine is calling malloc, then | |
| 1259 we're scrod. But Emacs can't just block input while calling | |
| 1260 matching routines; then we don't notice interrupts when they come | |
| 1261 in. So, Emacs blocks input around all regexp calls except the | |
| 1262 matching calls, which it leaves unprotected, in the faith that they | |
| 1333 | 1263 will not malloc.]] This previous paragraph is irrelevant under XEmacs, |
| 1264 as we *do not* do anything so stupid as process input from within a | |
| 1265 signal handler. | |
| 1266 | |
| 1267 However, the regexp routines may get called reentrantly as a result of | |
| 1268 QUIT processing (e.g. under Windows: re_match -> QUIT -> quit_p -> drain | |
| 1269 events -> process WM_INITMENU -> call filter -> re_match; see stack | |
| 1270 trace in signal.c), so we cannot have any global variables (unless we do | |
| 1271 lots of trickiness including some unwind-protects, which isn't worth it | |
| 1272 at this point). Hence we MUST have MATCH_MAY_ALLOCATE defined. | |
| 1273 | |
| 1274 Also, the first paragraph does not make complete sense to me -- what | |
| 1275 about the use of rel-alloc to handle the fail stacks? Shouldn't these | |
| 1276 reallocations potentially cause buffer data to be relocated as well? I | |
| 826 | 1277 must be missing something, though -- perhaps the writer above is |
| 1278 assuming that the failure stack(s) will always be allocated after the | |
| 1279 buffer data, and thus reallocating them with rel-alloc won't move buffer | |
| 1333 | 1280 data. (In fact, a cursory glance at the code in ralloc.c seems to |
| 1281 confirm this.) --ben */ | |
| 428 | 1282 |
| 1283 /* Normally, this is fine. */ | |
| 1284 #define MATCH_MAY_ALLOCATE | |
| 1285 | |
| 1286 /* When using GNU C, we are not REALLY using the C alloca, no matter | |
| 1287 what config.h may say. So don't take precautions for it. */ | |
| 1288 #ifdef __GNUC__ | |
| 1289 #undef C_ALLOCA | |
| 1290 #endif | |
| 1291 | |
| 1292 /* The match routines may not allocate if (1) they would do it with malloc | |
| 1293 and (2) it's not safe for them to use malloc. | |
| 1294 Note that if REL_ALLOC is defined, matching would not use malloc for the | |
| 1295 failure stack, but we would still use it for the register vectors; | |
| 1296 so REL_ALLOC should not affect this. */ | |
| 771 | 1297 |
| 1333 | 1298 /* XEmacs can handle REL_ALLOC and malloc() OK */ |
| 1299 #if !defined (emacs) && (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (REL_ALLOC) | |
| 428 | 1300 #undef MATCH_MAY_ALLOCATE |
| 1301 #endif | |
| 1302 | |
| 1333 | 1303 #if !defined (MATCH_MAY_ALLOCATE) && defined (emacs) |
| 771 | 1304 #error regex must be handle reentrancy; MATCH_MAY_ALLOCATE must be defined |
| 1305 #endif | |
| 1306 | |
| 428 | 1307 |
| 1308 /* Failure stack declarations and macros; both re_compile_fastmap and | |
| 1309 re_match_2 use a failure stack. These have to be macros because of | |
| 1310 REGEX_ALLOCATE_STACK. */ | |
| 1311 | |
| 1312 | |
| 1313 /* Number of failure points for which to initially allocate space | |
| 1314 when matching. If this number is exceeded, we allocate more | |
| 1315 space, so it is not a hard limit. */ | |
| 1316 #ifndef INIT_FAILURE_ALLOC | |
| 3300 | 1317 #define INIT_FAILURE_ALLOC 20 |
| 428 | 1318 #endif |
| 1319 | |
| 1320 /* Roughly the maximum number of failure points on the stack. Would be | |
| 1321 exactly that if always used MAX_FAILURE_SPACE each time we failed. | |
| 1322 This is a variable only so users of regex can assign to it; we never | |
| 1323 change it ourselves. */ | |
| 1324 #if defined (MATCH_MAY_ALLOCATE) | |
| 1325 /* 4400 was enough to cause a crash on Alpha OSF/1, | |
| 1326 whose default stack limit is 2mb. */ | |
| 3300 | 1327 int re_max_failures = 40000; |
| 428 | 1328 #else |
| 3300 | 1329 int re_max_failures = 4000; |
| 428 | 1330 #endif |
| 1331 | |
| 1332 union fail_stack_elt | |
| 1333 { | |
| 446 | 1334 re_char *pointer; |
| 428 | 1335 int integer; |
| 1336 }; | |
| 1337 | |
| 1338 typedef union fail_stack_elt fail_stack_elt_t; | |
| 1339 | |
| 1340 typedef struct | |
| 1341 { | |
| 1342 fail_stack_elt_t *stack; | |
| 665 | 1343 Elemcount size; |
| 1344 Elemcount avail; /* Offset of next open position. */ | |
| 428 | 1345 } fail_stack_type; |
| 1346 | |
| 1347 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0) | |
| 1348 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0) | |
| 1349 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size) | |
| 1350 | |
| 1351 | |
| 1352 /* Define macros to initialize and free the failure stack. | |
| 1353 Do `return -2' if the alloc fails. */ | |
| 1354 | |
| 1355 #ifdef MATCH_MAY_ALLOCATE | |
| 1333 | 1356 #define INIT_FAIL_STACK() \ |
| 1357 do { \ | |
| 1358 fail_stack.stack = (fail_stack_elt_t *) \ | |
| 1359 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * \ | |
| 1360 sizeof (fail_stack_elt_t)); \ | |
| 1361 \ | |
| 1362 if (fail_stack.stack == NULL) \ | |
| 1363 { \ | |
| 1364 UNBIND_REGEX_MALLOC_CHECK (); \ | |
| 1365 return -2; \ | |
| 1366 } \ | |
| 1367 \ | |
| 1368 fail_stack.size = INIT_FAILURE_ALLOC; \ | |
| 1369 fail_stack.avail = 0; \ | |
| 428 | 1370 } while (0) |
| 1371 | |
| 1372 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack) | |
| 1373 #else | |
| 1374 #define INIT_FAIL_STACK() \ | |
| 1375 do { \ | |
| 1376 fail_stack.avail = 0; \ | |
| 1377 } while (0) | |
| 1378 | |
| 1379 #define RESET_FAIL_STACK() | |
| 1380 #endif | |
| 1381 | |
| 1382 | |
| 1383 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. | |
| 1384 | |
| 1385 Return 1 if succeeds, and 0 if either ran out of memory | |
| 1386 allocating space for it or it was already too large. | |
| 1387 | |
| 1388 REGEX_REALLOCATE_STACK requires `destination' be declared. */ | |
| 1389 | |
| 1390 #define DOUBLE_FAIL_STACK(fail_stack) \ | |
| 1391 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \ | |
| 1392 ? 0 \ | |
| 1393 : ((fail_stack).stack = (fail_stack_elt_t *) \ | |
| 1394 REGEX_REALLOCATE_STACK ((fail_stack).stack, \ | |
| 1395 (fail_stack).size * sizeof (fail_stack_elt_t), \ | |
| 1396 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ | |
| 1397 \ | |
| 1398 (fail_stack).stack == NULL \ | |
| 1399 ? 0 \ | |
| 1400 : ((fail_stack).size <<= 1, \ | |
| 1401 1))) | |
| 1402 | |
| 1333 | 1403 #if !defined (emacs) || !defined (REL_ALLOC) |
| 1404 #define RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS() | |
| 1405 #else | |
| 1406 /* Don't change NULL pointers */ | |
| 1407 #define ADD_IF_NZ(val) if (val) val += rmdp_offset | |
| 1346 | 1408 #define RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS() \ |
| 1409 do \ | |
| 1410 { \ | |
| 1411 Bytecount rmdp_offset = offset_post_relocation (lispobj, orig_buftext); \ | |
| 1412 \ | |
| 1413 if (rmdp_offset) \ | |
| 1414 { \ | |
| 1415 int i; \ | |
| 1416 \ | |
| 1417 ADD_IF_NZ (string1); \ | |
| 1418 ADD_IF_NZ (string2); \ | |
| 1419 ADD_IF_NZ (d); \ | |
| 1420 ADD_IF_NZ (dend); \ | |
| 1421 ADD_IF_NZ (end1); \ | |
| 1422 ADD_IF_NZ (end2); \ | |
| 1423 ADD_IF_NZ (end_match_1); \ | |
| 1424 ADD_IF_NZ (end_match_2); \ | |
| 1425 \ | |
| 1426 if (bufp->re_ngroups) \ | |
| 1427 { \ | |
| 1428 for (i = 0; i < num_regs; i++) \ | |
| 1429 { \ | |
| 1430 ADD_IF_NZ (regstart[i]); \ | |
| 1431 ADD_IF_NZ (regend[i]); \ | |
| 1432 ADD_IF_NZ (old_regstart[i]); \ | |
| 1433 ADD_IF_NZ (old_regend[i]); \ | |
| 1434 ADD_IF_NZ (best_regstart[i]); \ | |
| 1435 ADD_IF_NZ (best_regend[i]); \ | |
| 1436 ADD_IF_NZ (reg_dummy[i]); \ | |
| 1437 } \ | |
| 1438 } \ | |
| 1439 \ | |
| 1440 ADD_IF_NZ (match_end); \ | |
| 1441 } \ | |
| 1333 | 1442 } while (0) |
| 1443 #endif /* !defined (emacs) || !defined (REL_ALLOC) */ | |
| 1444 | |
| 1445 #if !defined (emacs) || !defined (REL_ALLOC) | |
| 1446 #define RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS() | |
| 1447 #else | |
| 1346 | 1448 #define RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS() \ |
| 1449 do \ | |
| 1450 { \ | |
| 1451 Bytecount rmdp_offset = offset_post_relocation (lispobj, orig_buftext); \ | |
| 1452 \ | |
| 1453 if (rmdp_offset) \ | |
| 1454 { \ | |
| 1455 ADD_IF_NZ (str1); \ | |
| 1456 ADD_IF_NZ (str2); \ | |
| 1457 ADD_IF_NZ (string1); \ | |
| 1458 ADD_IF_NZ (string2); \ | |
| 1459 ADD_IF_NZ (d); \ | |
| 1460 } \ | |
| 1333 | 1461 } while (0) |
| 1462 | |
| 1463 #endif /* emacs */ | |
| 428 | 1464 |
| 1465 /* Push pointer POINTER on FAIL_STACK. | |
| 1466 Return 1 if was able to do so and 0 if ran out of memory allocating | |
| 1467 space to do so. */ | |
| 1468 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \ | |
| 1469 ((FAIL_STACK_FULL () \ | |
| 1470 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \ | |
| 1471 ? 0 \ | |
| 1472 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \ | |
| 1473 1)) | |
| 1474 | |
| 1475 /* Push a pointer value onto the failure stack. | |
| 1476 Assumes the variable `fail_stack'. Probably should only | |
| 1477 be called from within `PUSH_FAILURE_POINT'. */ | |
| 1478 #define PUSH_FAILURE_POINTER(item) \ | |
| 1479 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item) | |
| 1480 | |
| 1481 /* This pushes an integer-valued item onto the failure stack. | |
| 1482 Assumes the variable `fail_stack'. Probably should only | |
| 1483 be called from within `PUSH_FAILURE_POINT'. */ | |
| 1484 #define PUSH_FAILURE_INT(item) \ | |
| 1485 fail_stack.stack[fail_stack.avail++].integer = (item) | |
| 1486 | |
| 1487 /* Push a fail_stack_elt_t value onto the failure stack. | |
| 1488 Assumes the variable `fail_stack'. Probably should only | |
| 1489 be called from within `PUSH_FAILURE_POINT'. */ | |
| 1490 #define PUSH_FAILURE_ELT(item) \ | |
| 1491 fail_stack.stack[fail_stack.avail++] = (item) | |
| 1492 | |
| 1493 /* These three POP... operations complement the three PUSH... operations. | |
| 1494 All assume that `fail_stack' is nonempty. */ | |
| 1495 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer | |
| 1496 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer | |
| 1497 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail] | |
| 1498 | |
| 1499 /* Used to omit pushing failure point id's when we're not debugging. */ | |
| 1500 #ifdef DEBUG | |
| 1501 #define DEBUG_PUSH PUSH_FAILURE_INT | |
| 1502 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT () | |
| 1503 #else | |
| 1504 #define DEBUG_PUSH(item) | |
| 1505 #define DEBUG_POP(item_addr) | |
| 1506 #endif | |
| 1507 | |
| 1508 | |
| 1509 /* Push the information about the state we will need | |
| 1510 if we ever fail back to it. | |
| 1511 | |
| 1512 Requires variables fail_stack, regstart, regend, reg_info, and | |
| 1513 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be | |
| 1514 declared. | |
| 1515 | |
| 1516 Does `return FAILURE_CODE' if runs out of memory. */ | |
| 1517 | |
| 771 | 1518 #if !defined (REGEX_MALLOC) && !defined (REGEX_REL_ALLOC) |
| 456 | 1519 #define DECLARE_DESTINATION char *destination |
| 428 | 1520 #else |
| 456 | 1521 #define DECLARE_DESTINATION DECLARE_NOTHING |
| 428 | 1522 #endif |
| 1523 | |
| 1524 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ | |
| 456 | 1525 do { \ |
| 1526 DECLARE_DESTINATION; \ | |
| 1527 /* Must be int, so when we don't save any registers, the arithmetic \ | |
| 1528 of 0 + -1 isn't done as unsigned. */ \ | |
| 1529 int this_reg; \ | |
| 428 | 1530 \ |
| 456 | 1531 DEBUG_STATEMENT (failure_id++); \ |
| 1532 DEBUG_STATEMENT (nfailure_points_pushed++); \ | |
| 647 | 1533 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%d:\n", failure_id); \ |
| 1534 DEBUG_PRINT2 (" Before push, next avail: %ld\n", \ | |
| 1535 (long) (fail_stack).avail); \ | |
| 1536 DEBUG_PRINT2 (" size: %ld\n", \ | |
| 1537 (long) (fail_stack).size); \ | |
| 456 | 1538 \ |
| 1539 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \ | |
| 1540 DEBUG_PRINT2 (" available: %ld\n", \ | |
| 1541 (long) REMAINING_AVAIL_SLOTS); \ | |
| 428 | 1542 \ |
| 456 | 1543 /* Ensure we have enough space allocated for what we will push. */ \ |
| 1544 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ | |
| 1545 { \ | |
| 1333 | 1546 BEGIN_REGEX_MALLOC_OK (); \ |
| 456 | 1547 if (!DOUBLE_FAIL_STACK (fail_stack)) \ |
| 1333 | 1548 { \ |
| 1549 END_REGEX_MALLOC_OK (); \ | |
| 1550 UNBIND_REGEX_MALLOC_CHECK (); \ | |
| 1551 return failure_code; \ | |
| 1552 } \ | |
| 1553 END_REGEX_MALLOC_OK (); \ | |
| 647 | 1554 DEBUG_PRINT2 ("\n Doubled stack; size now: %ld\n", \ |
| 1555 (long) (fail_stack).size); \ | |
| 456 | 1556 DEBUG_PRINT2 (" slots available: %ld\n", \ |
| 1557 (long) REMAINING_AVAIL_SLOTS); \ | |
| 1333 | 1558 \ |
| 1559 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); \ | |
| 456 | 1560 } \ |
| 428 | 1561 \ |
| 456 | 1562 /* Push the info, starting with the registers. */ \ |
| 1563 DEBUG_PRINT1 ("\n"); \ | |
| 428 | 1564 \ |
| 456 | 1565 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \ |
| 1566 this_reg++) \ | |
| 1567 { \ | |
| 1568 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \ | |
| 1569 DEBUG_STATEMENT (num_regs_pushed++); \ | |
| 428 | 1570 \ |
| 456 | 1571 DEBUG_PRINT2 (" start: 0x%lx\n", (long) regstart[this_reg]); \ |
| 1572 PUSH_FAILURE_POINTER (regstart[this_reg]); \ | |
| 1573 \ | |
| 1574 DEBUG_PRINT2 (" end: 0x%lx\n", (long) regend[this_reg]); \ | |
| 1575 PUSH_FAILURE_POINTER (regend[this_reg]); \ | |
| 428 | 1576 \ |
| 456 | 1577 DEBUG_PRINT2 (" info: 0x%lx\n ", \ |
| 1578 * (long *) (®_info[this_reg])); \ | |
| 1579 DEBUG_PRINT2 (" match_null=%d", \ | |
| 1580 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \ | |
| 1581 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \ | |
| 1582 DEBUG_PRINT2 (" matched_something=%d", \ | |
| 1583 MATCHED_SOMETHING (reg_info[this_reg])); \ | |
| 1584 DEBUG_PRINT2 (" ever_matched_something=%d", \ | |
| 1585 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \ | |
| 1586 DEBUG_PRINT1 ("\n"); \ | |
| 1587 PUSH_FAILURE_ELT (reg_info[this_reg].word); \ | |
| 1588 } \ | |
| 428 | 1589 \ |
| 456 | 1590 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg); \ |
| 1591 PUSH_FAILURE_INT (lowest_active_reg); \ | |
| 428 | 1592 \ |
| 456 | 1593 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg); \ |
| 1594 PUSH_FAILURE_INT (highest_active_reg); \ | |
| 428 | 1595 \ |
| 456 | 1596 DEBUG_PRINT2 (" Pushing pattern 0x%lx: \n", (long) pattern_place); \ |
| 1597 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ | |
| 1598 PUSH_FAILURE_POINTER (pattern_place); \ | |
| 428 | 1599 \ |
| 456 | 1600 DEBUG_PRINT2 (" Pushing string 0x%lx: `", (long) string_place); \ |
| 1601 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ | |
| 1602 size2); \ | |
| 1603 DEBUG_PRINT1 ("'\n"); \ | |
| 1604 PUSH_FAILURE_POINTER (string_place); \ | |
| 428 | 1605 \ |
| 456 | 1606 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \ |
| 1607 DEBUG_PUSH (failure_id); \ | |
| 1608 } while (0) | |
| 428 | 1609 |
| 1610 /* This is the number of items that are pushed and popped on the stack | |
| 1611 for each register. */ | |
| 1612 #define NUM_REG_ITEMS 3 | |
| 1613 | |
| 1614 /* Individual items aside from the registers. */ | |
| 1615 #ifdef DEBUG | |
| 1616 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */ | |
| 1617 #else | |
| 1618 #define NUM_NONREG_ITEMS 4 | |
| 1619 #endif | |
| 1620 | |
| 1621 /* We push at most this many items on the stack. */ | |
| 1622 /* We used to use (num_regs - 1), which is the number of registers | |
| 1623 this regexp will save; but that was changed to 5 | |
| 1624 to avoid stack overflow for a regexp with lots of parens. */ | |
| 1625 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS) | |
| 1626 | |
| 1627 /* We actually push this many items. */ | |
| 1628 #define NUM_FAILURE_ITEMS \ | |
| 1629 ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \ | |
| 1630 + NUM_NONREG_ITEMS) | |
| 1631 | |
| 1632 /* How many items can still be added to the stack without overflowing it. */ | |
| 1633 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail) | |
| 1634 | |
| 1635 | |
| 1636 /* Pops what PUSH_FAIL_STACK pushes. | |
| 1637 | |
| 1638 We restore into the parameters, all of which should be lvalues: | |
| 1639 STR -- the saved data position. | |
| 1640 PAT -- the saved pattern position. | |
| 1641 LOW_REG, HIGH_REG -- the highest and lowest active registers. | |
| 1642 REGSTART, REGEND -- arrays of string positions. | |
| 1643 REG_INFO -- array of information about each subexpression. | |
| 1644 | |
| 1645 Also assumes the variables `fail_stack' and (if debugging), `bufp', | |
| 1646 `pend', `string1', `size1', `string2', and `size2'. */ | |
| 1647 | |
| 456 | 1648 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, \ |
| 1649 regstart, regend, reg_info) \ | |
| 1650 do { \ | |
| 428 | 1651 DEBUG_STATEMENT (fail_stack_elt_t ffailure_id;) \ |
| 1652 int this_reg; \ | |
| 442 | 1653 const unsigned char *string_temp; \ |
| 428 | 1654 \ |
| 1655 assert (!FAIL_STACK_EMPTY ()); \ | |
| 1656 \ | |
| 1657 /* Remove failure points and point to how many regs pushed. */ \ | |
| 1658 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \ | |
| 647 | 1659 DEBUG_PRINT2 (" Before pop, next avail: %ld\n", \ |
| 1660 (long) fail_stack.avail); \ | |
| 1661 DEBUG_PRINT2 (" size: %ld\n", \ | |
| 1662 (long) fail_stack.size); \ | |
| 428 | 1663 \ |
| 1664 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ | |
| 1665 \ | |
| 1666 DEBUG_POP (&ffailure_id.integer); \ | |
| 647 | 1667 DEBUG_PRINT2 (" Popping failure id: %d\n", \ |
| 1668 * (int *) &ffailure_id); \ | |
| 428 | 1669 \ |
| 1670 /* If the saved string location is NULL, it came from an \ | |
| 1671 on_failure_keep_string_jump opcode, and we want to throw away the \ | |
| 1672 saved NULL, thus retaining our current position in the string. */ \ | |
| 1673 string_temp = POP_FAILURE_POINTER (); \ | |
| 1674 if (string_temp != NULL) \ | |
| 446 | 1675 str = string_temp; \ |
| 428 | 1676 \ |
| 1677 DEBUG_PRINT2 (" Popping string 0x%lx: `", (long) str); \ | |
| 1678 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \ | |
| 1679 DEBUG_PRINT1 ("'\n"); \ | |
| 1680 \ | |
| 1681 pat = (unsigned char *) POP_FAILURE_POINTER (); \ | |
| 1682 DEBUG_PRINT2 (" Popping pattern 0x%lx: ", (long) pat); \ | |
| 1683 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \ | |
| 1684 \ | |
| 1685 /* Restore register info. */ \ | |
| 647 | 1686 high_reg = POP_FAILURE_INT (); \ |
| 428 | 1687 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \ |
| 1688 \ | |
| 647 | 1689 low_reg = POP_FAILURE_INT (); \ |
| 428 | 1690 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \ |
| 1691 \ | |
| 1692 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ | |
| 1693 { \ | |
| 1694 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \ | |
| 1695 \ | |
| 1696 reg_info[this_reg].word = POP_FAILURE_ELT (); \ | |
| 1697 DEBUG_PRINT2 (" info: 0x%lx\n", \ | |
| 1698 * (long *) ®_info[this_reg]); \ | |
| 1699 \ | |
| 446 | 1700 regend[this_reg] = POP_FAILURE_POINTER (); \ |
| 428 | 1701 DEBUG_PRINT2 (" end: 0x%lx\n", (long) regend[this_reg]); \ |
| 1702 \ | |
| 446 | 1703 regstart[this_reg] = POP_FAILURE_POINTER (); \ |
| 428 | 1704 DEBUG_PRINT2 (" start: 0x%lx\n", (long) regstart[this_reg]); \ |
| 1705 } \ | |
| 1706 \ | |
| 1707 set_regs_matched_done = 0; \ | |
| 1708 DEBUG_STATEMENT (nfailure_points_popped++); \ | |
| 456 | 1709 } while (0) /* POP_FAILURE_POINT */ |
| 428 | 1710 |
| 1711 | |
| 1712 | |
| 1713 /* Structure for per-register (a.k.a. per-group) information. | |
| 1714 Other register information, such as the | |
| 1715 starting and ending positions (which are addresses), and the list of | |
| 1716 inner groups (which is a bits list) are maintained in separate | |
| 1717 variables. | |
| 1718 | |
| 1719 We are making a (strictly speaking) nonportable assumption here: that | |
| 1720 the compiler will pack our bit fields into something that fits into | |
| 1721 the type of `word', i.e., is something that fits into one item on the | |
| 1722 failure stack. */ | |
| 1723 | |
| 1724 typedef union | |
| 1725 { | |
| 1726 fail_stack_elt_t word; | |
| 1727 struct | |
| 1728 { | |
| 1729 /* This field is one if this group can match the empty string, | |
| 1730 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ | |
| 1731 #define MATCH_NULL_UNSET_VALUE 3 | |
| 647 | 1732 unsigned int match_null_string_p : 2; |
| 1733 unsigned int is_active : 1; | |
| 1734 unsigned int matched_something : 1; | |
| 1735 unsigned int ever_matched_something : 1; | |
| 428 | 1736 } bits; |
| 1737 } register_info_type; | |
| 1738 | |
| 1739 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p) | |
| 1740 #define IS_ACTIVE(R) ((R).bits.is_active) | |
| 1741 #define MATCHED_SOMETHING(R) ((R).bits.matched_something) | |
| 1742 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something) | |
| 1743 | |
| 1744 | |
| 1745 /* Call this when have matched a real character; it sets `matched' flags | |
| 1746 for the subexpressions which we are currently inside. Also records | |
| 1747 that those subexprs have matched. */ | |
| 1748 #define SET_REGS_MATCHED() \ | |
| 1749 do \ | |
| 1750 { \ | |
| 1751 if (!set_regs_matched_done) \ | |
| 1752 { \ | |
| 647 | 1753 int r; \ |
| 428 | 1754 set_regs_matched_done = 1; \ |
| 1755 for (r = lowest_active_reg; r <= highest_active_reg; r++) \ | |
| 1756 { \ | |
| 1757 MATCHED_SOMETHING (reg_info[r]) \ | |
| 1758 = EVER_MATCHED_SOMETHING (reg_info[r]) \ | |
| 1759 = 1; \ | |
| 1760 } \ | |
| 1761 } \ | |
| 1762 } \ | |
| 1763 while (0) | |
| 1764 | |
| 1765 /* Registers are set to a sentinel when they haven't yet matched. */ | |
| 446 | 1766 static unsigned char reg_unset_dummy; |
| 428 | 1767 #define REG_UNSET_VALUE (®_unset_dummy) |
| 1768 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE) | |
| 1769 | |
| 1770 /* Subroutine declarations and macros for regex_compile. */ | |
| 1771 | |
| 1772 /* Fetch the next character in the uncompiled pattern---translating it | |
| 826 | 1773 if necessary. */ |
| 428 | 1774 #define PATFETCH(c) \ |
| 446 | 1775 do { \ |
| 1776 PATFETCH_RAW (c); \ | |
| 826 | 1777 c = RE_TRANSLATE (c); \ |
| 428 | 1778 } while (0) |
| 1779 | |
| 1780 /* Fetch the next character in the uncompiled pattern, with no | |
| 1781 translation. */ | |
| 1782 #define PATFETCH_RAW(c) \ | |
| 1783 do {if (p == pend) return REG_EEND; \ | |
| 1784 assert (p < pend); \ | |
| 867 | 1785 c = itext_ichar (p); \ |
| 1786 INC_IBYTEPTR (p); \ | |
| 428 | 1787 } while (0) |
| 1788 | |
| 1789 /* Go backwards one character in the pattern. */ | |
| 867 | 1790 #define PATUNFETCH DEC_IBYTEPTR (p) |
| 428 | 1791 |
| 1792 /* If `translate' is non-null, return translate[D], else just D. We | |
| 1793 cast the subscript to translate because some data is declared as | |
| 1794 `char *', to avoid warnings when a string constant is passed. But | |
| 1795 when we use a character as a subscript we must make it unsigned. */ | |
| 826 | 1796 #define RE_TRANSLATE(d) \ |
| 1797 (TRANSLATE_P (translate) ? RE_TRANSLATE_1 (d) : (d)) | |
| 428 | 1798 |
| 1799 /* Macros for outputting the compiled pattern into `buffer'. */ | |
| 1800 | |
| 1801 /* If the buffer isn't allocated when it comes in, use this. */ | |
| 1802 #define INIT_BUF_SIZE 32 | |
| 1803 | |
| 1804 /* Make sure we have at least N more bytes of space in buffer. */ | |
| 1805 #define GET_BUFFER_SPACE(n) \ | |
| 647 | 1806 while (buf_end - bufp->buffer + (n) > (ptrdiff_t) bufp->allocated) \ |
| 428 | 1807 EXTEND_BUFFER () |
| 1808 | |
| 1809 /* Make sure we have one more byte of buffer space and then add C to it. */ | |
| 1810 #define BUF_PUSH(c) \ | |
| 1811 do { \ | |
| 1812 GET_BUFFER_SPACE (1); \ | |
| 446 | 1813 *buf_end++ = (unsigned char) (c); \ |
| 428 | 1814 } while (0) |
| 1815 | |
| 1816 | |
| 1817 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */ | |
| 1818 #define BUF_PUSH_2(c1, c2) \ | |
| 1819 do { \ | |
| 1820 GET_BUFFER_SPACE (2); \ | |
| 446 | 1821 *buf_end++ = (unsigned char) (c1); \ |
| 1822 *buf_end++ = (unsigned char) (c2); \ | |
| 428 | 1823 } while (0) |
| 1824 | |
| 1825 | |
| 1826 /* As with BUF_PUSH_2, except for three bytes. */ | |
| 1827 #define BUF_PUSH_3(c1, c2, c3) \ | |
| 1828 do { \ | |
| 1829 GET_BUFFER_SPACE (3); \ | |
| 446 | 1830 *buf_end++ = (unsigned char) (c1); \ |
| 1831 *buf_end++ = (unsigned char) (c2); \ | |
| 1832 *buf_end++ = (unsigned char) (c3); \ | |
| 428 | 1833 } while (0) |
| 1834 | |
| 1835 | |
| 1836 /* Store a jump with opcode OP at LOC to location TO. We store a | |
| 1837 relative address offset by the three bytes the jump itself occupies. */ | |
| 1838 #define STORE_JUMP(op, loc, to) \ | |
| 1839 store_op1 (op, loc, (to) - (loc) - 3) | |
| 1840 | |
| 1841 /* Likewise, for a two-argument jump. */ | |
| 1842 #define STORE_JUMP2(op, loc, to, arg) \ | |
| 1843 store_op2 (op, loc, (to) - (loc) - 3, arg) | |
| 1844 | |
| 446 | 1845 /* Like `STORE_JUMP', but for inserting. Assume `buf_end' is the |
| 1846 buffer end. */ | |
| 428 | 1847 #define INSERT_JUMP(op, loc, to) \ |
| 446 | 1848 insert_op1 (op, loc, (to) - (loc) - 3, buf_end) |
| 1849 | |
| 1850 /* Like `STORE_JUMP2', but for inserting. Assume `buf_end' is the | |
| 1851 buffer end. */ | |
| 428 | 1852 #define INSERT_JUMP2(op, loc, to, arg) \ |
| 446 | 1853 insert_op2 (op, loc, (to) - (loc) - 3, arg, buf_end) |
| 428 | 1854 |
| 1855 | |
| 1856 /* This is not an arbitrary limit: the arguments which represent offsets | |
| 1857 into the pattern are two bytes long. So if 2^16 bytes turns out to | |
| 1858 be too small, many things would have to change. */ | |
| 1859 #define MAX_BUF_SIZE (1L << 16) | |
| 1860 | |
| 1861 | |
| 1862 /* Extend the buffer by twice its current size via realloc and | |
| 1863 reset the pointers that pointed into the old block to point to the | |
| 1864 correct places in the new one. If extending the buffer results in it | |
| 1865 being larger than MAX_BUF_SIZE, then flag memory exhausted. */ | |
| 1333 | 1866 #define EXTEND_BUFFER() \ |
| 1867 do { \ | |
| 1868 re_char *old_buffer = bufp->buffer; \ | |
| 1869 if (bufp->allocated == MAX_BUF_SIZE) \ | |
| 1870 return REG_ESIZE; \ | |
| 1871 bufp->allocated <<= 1; \ | |
| 1872 if (bufp->allocated > MAX_BUF_SIZE) \ | |
| 1873 bufp->allocated = MAX_BUF_SIZE; \ | |
| 1874 bufp->buffer = \ | |
| 1875 (unsigned char *) xrealloc (bufp->buffer, bufp->allocated); \ | |
| 1876 if (bufp->buffer == NULL) \ | |
| 1877 return REG_ESPACE; \ | |
| 1878 /* If the buffer moved, move all the pointers into it. */ \ | |
| 1879 if (old_buffer != bufp->buffer) \ | |
| 1880 { \ | |
| 1881 buf_end = (buf_end - old_buffer) + bufp->buffer; \ | |
| 1882 begalt = (begalt - old_buffer) + bufp->buffer; \ | |
| 1883 if (fixup_alt_jump) \ | |
| 1884 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer; \ | |
| 1885 if (laststart) \ | |
| 1886 laststart = (laststart - old_buffer) + bufp->buffer; \ | |
| 1887 if (pending_exact) \ | |
| 1888 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ | |
| 1889 } \ | |
| 428 | 1890 } while (0) |
| 1891 | |
| 1892 | |
| 1893 /* Since we have one byte reserved for the register number argument to | |
| 1894 {start,stop}_memory, the maximum number of groups we can report | |
| 1895 things about is what fits in that byte. */ | |
| 1896 #define MAX_REGNUM 255 | |
| 1897 | |
| 1898 /* But patterns can have more than `MAX_REGNUM' registers. We just | |
| 502 | 1899 ignore the excess. |
| 1900 #### not true! groups past this will fail in lots of ways, if we | |
| 1901 ever have to backtrack. | |
| 1902 */ | |
| 647 | 1903 typedef int regnum_t; |
| 428 | 1904 |
| 502 | 1905 #define INIT_REG_TRANSLATE_SIZE 5 |
| 428 | 1906 |
| 1907 /* Macros for the compile stack. */ | |
| 1908 | |
| 1909 /* Since offsets can go either forwards or backwards, this type needs to | |
| 1910 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ | |
| 1911 typedef int pattern_offset_t; | |
| 1912 | |
| 1913 typedef struct | |
| 1914 { | |
| 1915 pattern_offset_t begalt_offset; | |
| 1916 pattern_offset_t fixup_alt_jump; | |
| 1917 pattern_offset_t inner_group_offset; | |
| 1918 pattern_offset_t laststart_offset; | |
| 1919 regnum_t regnum; | |
| 1920 } compile_stack_elt_t; | |
| 1921 | |
| 1922 | |
| 1923 typedef struct | |
| 1924 { | |
| 1925 compile_stack_elt_t *stack; | |
| 647 | 1926 int size; |
| 1927 int avail; /* Offset of next open position. */ | |
| 428 | 1928 } compile_stack_type; |
| 1929 | |
| 1930 | |
| 1931 #define INIT_COMPILE_STACK_SIZE 32 | |
| 1932 | |
| 1933 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0) | |
| 1934 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) | |
| 1935 | |
| 1936 /* The next available element. */ | |
| 1937 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) | |
| 1938 | |
| 1939 | |
| 1940 /* Set the bit for character C in a bit vector. */ | |
| 1941 #define SET_LIST_BIT(c) \ | |
| 446 | 1942 (buf_end[((unsigned char) (c)) / BYTEWIDTH] \ |
| 428 | 1943 |= 1 << (((unsigned char) c) % BYTEWIDTH)) |
| 1944 | |
| 1945 #ifdef MULE | |
| 1946 | |
| 1947 /* Set the "bit" for character C in a range table. */ | |
| 1948 #define SET_RANGETAB_BIT(c) put_range_table (rtab, c, c, Qt) | |
| 1949 | |
| 1950 /* Set the "bit" for character c in the appropriate table. */ | |
| 1951 #define SET_EITHER_BIT(c) \ | |
| 1952 do { \ | |
| 1953 if (has_extended_chars) \ | |
| 1954 SET_RANGETAB_BIT (c); \ | |
| 1955 else \ | |
| 1956 SET_LIST_BIT (c); \ | |
| 1957 } while (0) | |
| 1958 | |
| 1959 #else /* not MULE */ | |
| 1960 | |
| 1961 #define SET_EITHER_BIT(c) SET_LIST_BIT (c) | |
| 1962 | |
| 1963 #endif | |
| 1964 | |
| 1965 | |
| 1966 /* Get the next unsigned number in the uncompiled pattern. */ | |
| 1967 #define GET_UNSIGNED_NUMBER(num) \ | |
| 1968 { if (p != pend) \ | |
| 1969 { \ | |
| 1970 PATFETCH (c); \ | |
| 1971 while (ISDIGIT (c)) \ | |
| 1972 { \ | |
| 1973 if (num < 0) \ | |
| 1974 num = 0; \ | |
| 1975 num = num * 10 + c - '0'; \ | |
| 1976 if (p == pend) \ | |
| 1977 break; \ | |
| 1978 PATFETCH (c); \ | |
| 1979 } \ | |
| 1980 } \ | |
| 1981 } | |
| 1982 | |
| 1983 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ | |
| 1984 | |
| 1985 #define IS_CHAR_CLASS(string) \ | |
| 1986 (STREQ (string, "alpha") || STREQ (string, "upper") \ | |
| 1987 || STREQ (string, "lower") || STREQ (string, "digit") \ | |
| 1988 || STREQ (string, "alnum") || STREQ (string, "xdigit") \ | |
| 1989 || STREQ (string, "space") || STREQ (string, "print") \ | |
| 1990 || STREQ (string, "punct") || STREQ (string, "graph") \ | |
| 1991 || STREQ (string, "cntrl") || STREQ (string, "blank")) | |
| 1992 | |
| 1993 static void store_op1 (re_opcode_t op, unsigned char *loc, int arg); | |
| 1994 static void store_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2); | |
| 1995 static void insert_op1 (re_opcode_t op, unsigned char *loc, int arg, | |
| 1996 unsigned char *end); | |
| 1997 static void insert_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2, | |
| 1998 unsigned char *end); | |
| 460 | 1999 static re_bool at_begline_loc_p (re_char *pattern, re_char *p, |
| 428 | 2000 reg_syntax_t syntax); |
| 460 | 2001 static re_bool at_endline_loc_p (re_char *p, re_char *pend, int syntax); |
| 2002 static re_bool group_in_compile_stack (compile_stack_type compile_stack, | |
| 428 | 2003 regnum_t regnum); |
| 446 | 2004 static reg_errcode_t compile_range (re_char **p_ptr, re_char *pend, |
| 2005 RE_TRANSLATE_TYPE translate, | |
| 2006 reg_syntax_t syntax, | |
| 428 | 2007 unsigned char *b); |
| 2008 #ifdef MULE | |
| 446 | 2009 static reg_errcode_t compile_extended_range (re_char **p_ptr, |
| 2010 re_char *pend, | |
| 2011 RE_TRANSLATE_TYPE translate, | |
| 428 | 2012 reg_syntax_t syntax, |
| 2013 Lisp_Object rtab); | |
| 2014 #endif /* MULE */ | |
| 460 | 2015 static re_bool group_match_null_string_p (unsigned char **p, |
| 428 | 2016 unsigned char *end, |
| 2017 register_info_type *reg_info); | |
| 460 | 2018 static re_bool alt_match_null_string_p (unsigned char *p, unsigned char *end, |
| 428 | 2019 register_info_type *reg_info); |
| 460 | 2020 static re_bool common_op_match_null_string_p (unsigned char **p, |
| 428 | 2021 unsigned char *end, |
| 2022 register_info_type *reg_info); | |
| 826 | 2023 static int bcmp_translate (re_char *s1, re_char *s2, |
| 2024 REGISTER int len, RE_TRANSLATE_TYPE translate | |
| 2025 #ifdef emacs | |
| 2026 , Internal_Format fmt, Lisp_Object lispobj | |
| 2027 #endif | |
| 2028 ); | |
| 428 | 2029 static int re_match_2_internal (struct re_pattern_buffer *bufp, |
| 446 | 2030 re_char *string1, int size1, |
| 2031 re_char *string2, int size2, int pos, | |
| 826 | 2032 struct re_registers *regs, int stop |
| 2033 RE_LISP_CONTEXT_ARGS_DECL); | |
| 428 | 2034 |
| 2035 #ifndef MATCH_MAY_ALLOCATE | |
| 2036 | |
| 2037 /* If we cannot allocate large objects within re_match_2_internal, | |
| 2038 we make the fail stack and register vectors global. | |
| 2039 The fail stack, we grow to the maximum size when a regexp | |
| 2040 is compiled. | |
| 2041 The register vectors, we adjust in size each time we | |
| 2042 compile a regexp, according to the number of registers it needs. */ | |
| 2043 | |
| 2044 static fail_stack_type fail_stack; | |
| 2045 | |
| 2046 /* Size with which the following vectors are currently allocated. | |
| 2047 That is so we can make them bigger as needed, | |
| 2048 but never make them smaller. */ | |
| 2049 static int regs_allocated_size; | |
| 2050 | |
| 446 | 2051 static re_char ** regstart, ** regend; |
| 2052 static re_char ** old_regstart, ** old_regend; | |
| 2053 static re_char **best_regstart, **best_regend; | |
| 428 | 2054 static register_info_type *reg_info; |
| 446 | 2055 static re_char **reg_dummy; |
| 428 | 2056 static register_info_type *reg_info_dummy; |
| 2057 | |
| 2058 /* Make the register vectors big enough for NUM_REGS registers, | |
| 2059 but don't make them smaller. */ | |
| 2060 | |
| 2061 static | |
| 2062 regex_grow_registers (int num_regs) | |
| 2063 { | |
| 2064 if (num_regs > regs_allocated_size) | |
| 2065 { | |
| 551 | 2066 RETALLOC (regstart, num_regs, re_char *); |
| 2067 RETALLOC (regend, num_regs, re_char *); | |
| 2068 RETALLOC (old_regstart, num_regs, re_char *); | |
| 2069 RETALLOC (old_regend, num_regs, re_char *); | |
| 2070 RETALLOC (best_regstart, num_regs, re_char *); | |
| 2071 RETALLOC (best_regend, num_regs, re_char *); | |
| 2072 RETALLOC (reg_info, num_regs, register_info_type); | |
| 2073 RETALLOC (reg_dummy, num_regs, re_char *); | |
| 2074 RETALLOC (reg_info_dummy, num_regs, register_info_type); | |
| 428 | 2075 |
| 2076 regs_allocated_size = num_regs; | |
| 2077 } | |
| 2078 } | |
| 2079 | |
| 2080 #endif /* not MATCH_MAY_ALLOCATE */ | |
| 2081 | |
| 2082 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX. | |
| 2083 Returns one of error codes defined in `regex.h', or zero for success. | |
| 2084 | |
| 2085 Assumes the `allocated' (and perhaps `buffer') and `translate' | |
| 2086 fields are set in BUFP on entry. | |
| 2087 | |
| 2088 If it succeeds, results are put in BUFP (if it returns an error, the | |
| 2089 contents of BUFP are undefined): | |
| 2090 `buffer' is the compiled pattern; | |
| 2091 `syntax' is set to SYNTAX; | |
| 2092 `used' is set to the length of the compiled pattern; | |
| 2093 `fastmap_accurate' is zero; | |
| 502 | 2094 `re_ngroups' is the number of groups/subexpressions (including shy |
| 2095 groups) in PATTERN; | |
| 2096 `re_nsub' is the number of non-shy groups in PATTERN; | |
| 428 | 2097 `not_bol' and `not_eol' are zero; |
| 2098 | |
| 2099 The `fastmap' and `newline_anchor' fields are neither | |
| 2100 examined nor set. */ | |
| 2101 | |
| 2102 /* Return, freeing storage we allocated. */ | |
| 1726 | 2103 #define FREE_STACK_RETURN(value) \ |
| 2104 do \ | |
| 2105 { \ | |
| 2106 xfree (compile_stack.stack, compile_stack_elt_t *); \ | |
| 2107 return value; \ | |
| 1333 | 2108 } while (0) |
| 428 | 2109 |
| 2110 static reg_errcode_t | |
| 446 | 2111 regex_compile (re_char *pattern, int size, reg_syntax_t syntax, |
| 428 | 2112 struct re_pattern_buffer *bufp) |
| 2113 { | |
| 2114 /* We fetch characters from PATTERN here. We declare these as int | |
| 2115 (or possibly long) so that chars above 127 can be used as | |
| 2116 array indices. The macros that fetch a character from the pattern | |
| 2117 make sure to coerce to unsigned char before assigning, so we won't | |
| 2118 get bitten by negative numbers here. */ | |
| 2119 /* XEmacs change: used to be unsigned char. */ | |
| 2120 REGISTER EMACS_INT c, c1; | |
| 2121 | |
| 2122 /* A random temporary spot in PATTERN. */ | |
| 446 | 2123 re_char *p1; |
| 428 | 2124 |
| 2125 /* Points to the end of the buffer, where we should append. */ | |
| 446 | 2126 REGISTER unsigned char *buf_end; |
| 428 | 2127 |
| 2128 /* Keeps track of unclosed groups. */ | |
| 2129 compile_stack_type compile_stack; | |
| 2130 | |
| 2131 /* Points to the current (ending) position in the pattern. */ | |
| 446 | 2132 re_char *p = pattern; |
| 2133 re_char *pend = pattern + size; | |
| 428 | 2134 |
| 2135 /* How to translate the characters in the pattern. */ | |
| 446 | 2136 RE_TRANSLATE_TYPE translate = bufp->translate; |
| 428 | 2137 |
| 2138 /* Address of the count-byte of the most recently inserted `exactn' | |
| 2139 command. This makes it possible to tell if a new exact-match | |
| 2140 character can be added to that command or if the character requires | |
| 2141 a new `exactn' command. */ | |
| 2142 unsigned char *pending_exact = 0; | |
| 2143 | |
| 2144 /* Address of start of the most recently finished expression. | |
| 2145 This tells, e.g., postfix * where to find the start of its | |
| 2146 operand. Reset at the beginning of groups and alternatives. */ | |
| 2147 unsigned char *laststart = 0; | |
| 2148 | |
| 2149 /* Address of beginning of regexp, or inside of last group. */ | |
| 2150 unsigned char *begalt; | |
| 2151 | |
| 2152 /* Place in the uncompiled pattern (i.e., the {) to | |
| 2153 which to go back if the interval is invalid. */ | |
| 446 | 2154 re_char *beg_interval; |
| 428 | 2155 |
| 2156 /* Address of the place where a forward jump should go to the end of | |
| 2157 the containing expression. Each alternative of an `or' -- except the | |
| 2158 last -- ends with a forward jump of this sort. */ | |
| 2159 unsigned char *fixup_alt_jump = 0; | |
| 2160 | |
| 2161 /* Counts open-groups as they are encountered. Remembered for the | |
| 2162 matching close-group on the compile stack, so the same register | |
| 2163 number is put in the stop_memory as the start_memory. */ | |
| 2164 regnum_t regnum = 0; | |
| 2165 | |
| 2166 #ifdef DEBUG | |
| 2167 DEBUG_PRINT1 ("\nCompiling pattern: "); | |
| 2168 if (debug) | |
| 2169 { | |
| 647 | 2170 int debug_count; |
| 428 | 2171 |
| 2172 for (debug_count = 0; debug_count < size; debug_count++) | |
| 2173 putchar (pattern[debug_count]); | |
| 2174 putchar ('\n'); | |
| 2175 } | |
| 2176 #endif /* DEBUG */ | |
| 2177 | |
| 2178 /* Initialize the compile stack. */ | |
| 2179 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t); | |
| 2180 if (compile_stack.stack == NULL) | |
| 2181 return REG_ESPACE; | |
| 2182 | |
| 2183 compile_stack.size = INIT_COMPILE_STACK_SIZE; | |
| 2184 compile_stack.avail = 0; | |
| 2185 | |
| 2186 /* Initialize the pattern buffer. */ | |
| 2187 bufp->syntax = syntax; | |
| 2188 bufp->fastmap_accurate = 0; | |
| 2189 bufp->not_bol = bufp->not_eol = 0; | |
| 2190 | |
| 2191 /* Set `used' to zero, so that if we return an error, the pattern | |
| 2192 printer (for debugging) will think there's no pattern. We reset it | |
| 2193 at the end. */ | |
| 2194 bufp->used = 0; | |
| 2195 | |
| 2196 /* Always count groups, whether or not bufp->no_sub is set. */ | |
| 2197 bufp->re_nsub = 0; | |
| 502 | 2198 bufp->re_ngroups = 0; |
| 2199 | |
| 2200 bufp->warned_about_incompatible_back_references = 0; | |
| 2201 | |
| 2202 if (bufp->external_to_internal_register == 0) | |
| 2203 { | |
| 2204 bufp->external_to_internal_register_size = INIT_REG_TRANSLATE_SIZE; | |
| 2205 RETALLOC (bufp->external_to_internal_register, | |
| 2206 bufp->external_to_internal_register_size, | |
| 2207 int); | |
| 2208 } | |
| 2209 | |
| 2210 { | |
| 2211 int i; | |
| 2212 | |
| 2213 bufp->external_to_internal_register[0] = 0; | |
| 2214 for (i = 1; i < bufp->external_to_internal_register_size; i++) | |
| 2215 bufp->external_to_internal_register[i] = (int) 0xDEADBEEF; | |
| 2216 } | |
| 428 | 2217 |
| 2218 #if !defined (emacs) && !defined (SYNTAX_TABLE) | |
| 2219 /* Initialize the syntax table. */ | |
| 2220 init_syntax_once (); | |
| 2221 #endif | |
| 2222 | |
| 2223 if (bufp->allocated == 0) | |
| 2224 { | |
| 2225 if (bufp->buffer) | |
| 2226 { /* If zero allocated, but buffer is non-null, try to realloc | |
| 2227 enough space. This loses if buffer's address is bogus, but | |
| 2228 that is the user's responsibility. */ | |
| 2229 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); | |
| 2230 } | |
| 2231 else | |
| 2232 { /* Caller did not allocate a buffer. Do it for them. */ | |
| 2233 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); | |
| 2234 } | |
| 2235 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE); | |
| 2236 | |
| 2237 bufp->allocated = INIT_BUF_SIZE; | |
| 2238 } | |
| 2239 | |
| 446 | 2240 begalt = buf_end = bufp->buffer; |
| 428 | 2241 |
| 2242 /* Loop through the uncompiled pattern until we're at the end. */ | |
| 2243 while (p != pend) | |
| 2244 { | |
| 2245 PATFETCH (c); | |
| 2246 | |
| 2247 switch (c) | |
| 2248 { | |
| 2249 case '^': | |
| 2250 { | |
| 2251 if ( /* If at start of pattern, it's an operator. */ | |
| 2252 p == pattern + 1 | |
| 2253 /* If context independent, it's an operator. */ | |
| 2254 || syntax & RE_CONTEXT_INDEP_ANCHORS | |
| 2255 /* Otherwise, depends on what's come before. */ | |
| 2256 || at_begline_loc_p (pattern, p, syntax)) | |
| 2257 BUF_PUSH (begline); | |
| 2258 else | |
| 2259 goto normal_char; | |
| 2260 } | |
| 2261 break; | |
| 2262 | |
| 2263 | |
| 2264 case '$': | |
| 2265 { | |
| 2266 if ( /* If at end of pattern, it's an operator. */ | |
| 2267 p == pend | |
| 2268 /* If context independent, it's an operator. */ | |
| 2269 || syntax & RE_CONTEXT_INDEP_ANCHORS | |
| 2270 /* Otherwise, depends on what's next. */ | |
| 2271 || at_endline_loc_p (p, pend, syntax)) | |
| 2272 BUF_PUSH (endline); | |
| 2273 else | |
| 2274 goto normal_char; | |
| 2275 } | |
| 2276 break; | |
| 2277 | |
| 2278 | |
| 2279 case '+': | |
| 2280 case '?': | |
| 2281 if ((syntax & RE_BK_PLUS_QM) | |
| 2282 || (syntax & RE_LIMITED_OPS)) | |
| 2283 goto normal_char; | |
| 2284 handle_plus: | |
| 2285 case '*': | |
| 2286 /* If there is no previous pattern... */ | |
| 2287 if (!laststart) | |
| 2288 { | |
| 2289 if (syntax & RE_CONTEXT_INVALID_OPS) | |
| 2290 FREE_STACK_RETURN (REG_BADRPT); | |
| 2291 else if (!(syntax & RE_CONTEXT_INDEP_OPS)) | |
| 2292 goto normal_char; | |
| 2293 } | |
| 2294 | |
| 2295 { | |
| 2296 /* true means zero/many matches are allowed. */ | |
| 460 | 2297 re_bool zero_times_ok = c != '+'; |
| 2298 re_bool many_times_ok = c != '?'; | |
| 428 | 2299 |
| 2300 /* true means match shortest string possible. */ | |
| 460 | 2301 re_bool minimal = false; |
| 428 | 2302 |
| 2303 /* If there is a sequence of repetition chars, collapse it | |
| 2304 down to just one (the right one). We can't combine | |
| 2305 interval operators with these because of, e.g., `a{2}*', | |
| 2306 which should only match an even number of `a's. */ | |
| 2307 while (p != pend) | |
| 2308 { | |
| 2309 PATFETCH (c); | |
| 2310 | |
| 2311 if (c == '*' || (!(syntax & RE_BK_PLUS_QM) | |
| 2312 && (c == '+' || c == '?'))) | |
| 2313 ; | |
| 2314 | |
| 2315 else if (syntax & RE_BK_PLUS_QM && c == '\\') | |
| 2316 { | |
| 2317 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | |
| 2318 | |
| 2319 PATFETCH (c1); | |
| 2320 if (!(c1 == '+' || c1 == '?')) | |
| 2321 { | |
| 2322 PATUNFETCH; | |
| 2323 PATUNFETCH; | |
| 2324 break; | |
| 2325 } | |
| 2326 | |
| 2327 c = c1; | |
| 2328 } | |
| 2329 else | |
| 2330 { | |
| 2331 PATUNFETCH; | |
| 2332 break; | |
| 2333 } | |
| 2334 | |
| 2335 /* If we get here, we found another repeat character. */ | |
| 2336 if (!(syntax & RE_NO_MINIMAL_MATCHING)) | |
| 2337 { | |
| 440 | 2338 /* "*?" and "+?" and "??" are okay (and mean match |
| 2339 minimally), but other sequences (such as "*??" and | |
| 2340 "+++") are rejected (reserved for future use). */ | |
| 428 | 2341 if (minimal || c != '?') |
| 2342 FREE_STACK_RETURN (REG_BADRPT); | |
| 2343 minimal = true; | |
| 2344 } | |
| 2345 else | |
| 2346 { | |
| 2347 zero_times_ok |= c != '+'; | |
| 2348 many_times_ok |= c != '?'; | |
| 2349 } | |
| 2350 } | |
| 2351 | |
| 2352 /* Star, etc. applied to an empty pattern is equivalent | |
| 2353 to an empty pattern. */ | |
| 2354 if (!laststart) | |
| 2355 break; | |
| 2356 | |
| 2357 /* Now we know whether zero matches is allowed | |
| 2358 and whether two or more matches is allowed | |
| 2359 and whether we want minimal or maximal matching. */ | |
| 2360 if (minimal) | |
| 2361 { | |
| 2362 if (!many_times_ok) | |
| 2363 { | |
| 2364 /* "a??" becomes: | |
| 2365 0: /on_failure_jump to 6 | |
| 2366 3: /jump to 9 | |
| 2367 6: /exactn/1/A | |
| 2368 9: end of pattern. | |
| 2369 */ | |
| 2370 GET_BUFFER_SPACE (6); | |
| 446 | 2371 INSERT_JUMP (jump, laststart, buf_end + 3); |
| 2372 buf_end += 3; | |
| 428 | 2373 INSERT_JUMP (on_failure_jump, laststart, laststart + 6); |
| 446 | 2374 buf_end += 3; |
| 428 | 2375 } |
| 2376 else if (zero_times_ok) | |
| 2377 { | |
| 2378 /* "a*?" becomes: | |
| 2379 0: /jump to 6 | |
| 2380 3: /exactn/1/A | |
| 2381 6: /on_failure_jump to 3 | |
| 2382 9: end of pattern. | |
| 2383 */ | |
| 2384 GET_BUFFER_SPACE (6); | |
| 446 | 2385 INSERT_JUMP (jump, laststart, buf_end + 3); |
| 2386 buf_end += 3; | |
| 2387 STORE_JUMP (on_failure_jump, buf_end, laststart + 3); | |
| 2388 buf_end += 3; | |
| 428 | 2389 } |
| 2390 else | |
| 2391 { | |
| 2392 /* "a+?" becomes: | |
| 2393 0: /exactn/1/A | |
| 2394 3: /on_failure_jump to 0 | |
| 2395 6: end of pattern. | |
| 2396 */ | |
| 2397 GET_BUFFER_SPACE (3); | |
| 446 | 2398 STORE_JUMP (on_failure_jump, buf_end, laststart); |
| 2399 buf_end += 3; | |
| 428 | 2400 } |
| 2401 } | |
| 2402 else | |
| 2403 { | |
| 2404 /* Are we optimizing this jump? */ | |
| 460 | 2405 re_bool keep_string_p = false; |
| 428 | 2406 |
| 2407 if (many_times_ok) | |
| 446 | 2408 { /* More than one repetition is allowed, so put in |
| 2409 at the end a backward relative jump from | |
| 2410 `buf_end' to before the next jump we're going | |
| 2411 to put in below (which jumps from laststart to | |
| 2412 after this jump). | |
| 428 | 2413 |
| 2414 But if we are at the `*' in the exact sequence `.*\n', | |
| 2415 insert an unconditional jump backwards to the ., | |
| 2416 instead of the beginning of the loop. This way we only | |
| 2417 push a failure point once, instead of every time | |
| 2418 through the loop. */ | |
| 2419 assert (p - 1 > pattern); | |
| 2420 | |
| 2421 /* Allocate the space for the jump. */ | |
| 2422 GET_BUFFER_SPACE (3); | |
| 2423 | |
| 2424 /* We know we are not at the first character of the | |
| 2425 pattern, because laststart was nonzero. And we've | |
| 2426 already incremented `p', by the way, to be the | |
| 2427 character after the `*'. Do we have to do something | |
| 2428 analogous here for null bytes, because of | |
| 2429 RE_DOT_NOT_NULL? */ | |
| 446 | 2430 if (*(p - 2) == '.' |
| 428 | 2431 && zero_times_ok |
| 446 | 2432 && p < pend && *p == '\n' |
| 428 | 2433 && !(syntax & RE_DOT_NEWLINE)) |
| 2434 { /* We have .*\n. */ | |
| 446 | 2435 STORE_JUMP (jump, buf_end, laststart); |
| 428 | 2436 keep_string_p = true; |
| 2437 } | |
| 2438 else | |
| 2439 /* Anything else. */ | |
| 446 | 2440 STORE_JUMP (maybe_pop_jump, buf_end, laststart - 3); |
| 428 | 2441 |
| 2442 /* We've added more stuff to the buffer. */ | |
| 446 | 2443 buf_end += 3; |
| 428 | 2444 } |
| 2445 | |
| 446 | 2446 /* On failure, jump from laststart to buf_end + 3, |
| 2447 which will be the end of the buffer after this jump | |
| 2448 is inserted. */ | |
| 428 | 2449 GET_BUFFER_SPACE (3); |
| 2450 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump | |
| 2451 : on_failure_jump, | |
| 446 | 2452 laststart, buf_end + 3); |
| 2453 buf_end += 3; | |
| 428 | 2454 |
| 2455 if (!zero_times_ok) | |
| 2456 { | |
| 2457 /* At least one repetition is required, so insert a | |
| 2458 `dummy_failure_jump' before the initial | |
| 2459 `on_failure_jump' instruction of the loop. This | |
| 2460 effects a skip over that instruction the first time | |
| 2461 we hit that loop. */ | |
| 2462 GET_BUFFER_SPACE (3); | |
| 2463 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); | |
| 446 | 2464 buf_end += 3; |
| 428 | 2465 } |
| 2466 } | |
| 2467 pending_exact = 0; | |
| 2468 } | |
| 2469 break; | |
| 2470 | |
| 2471 | |
| 2472 case '.': | |
| 446 | 2473 laststart = buf_end; |
| 428 | 2474 BUF_PUSH (anychar); |
| 2475 break; | |
| 2476 | |
| 2477 | |
| 2478 case '[': | |
| 2479 { | |
| 2480 /* XEmacs change: this whole section */ | |
| 460 | 2481 re_bool had_char_class = false; |
| 428 | 2482 #ifdef MULE |
| 460 | 2483 re_bool has_extended_chars = false; |
| 428 | 2484 REGISTER Lisp_Object rtab = Qnil; |
| 2485 #endif | |
| 2486 | |
| 2487 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2488 | |
| 2489 /* Ensure that we have enough space to push a charset: the | |
| 2490 opcode, the length count, and the bitset; 34 bytes in all. */ | |
| 2491 GET_BUFFER_SPACE (34); | |
| 2492 | |
| 446 | 2493 laststart = buf_end; |
| 428 | 2494 |
| 2495 /* We test `*p == '^' twice, instead of using an if | |
| 2496 statement, so we only need one BUF_PUSH. */ | |
| 2497 BUF_PUSH (*p == '^' ? charset_not : charset); | |
| 2498 if (*p == '^') | |
| 2499 p++; | |
| 2500 | |
| 2501 /* Remember the first position in the bracket expression. */ | |
| 2502 p1 = p; | |
| 2503 | |
| 2504 /* Push the number of bytes in the bitmap. */ | |
| 2505 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); | |
| 2506 | |
| 2507 /* Clear the whole map. */ | |
| 446 | 2508 memset (buf_end, 0, (1 << BYTEWIDTH) / BYTEWIDTH); |
| 428 | 2509 |
| 2510 /* charset_not matches newline according to a syntax bit. */ | |
| 446 | 2511 if ((re_opcode_t) buf_end[-2] == charset_not |
| 428 | 2512 && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) |
| 2513 SET_LIST_BIT ('\n'); | |
| 2514 | |
| 2515 #ifdef MULE | |
| 2516 start_over_with_extended: | |
| 2517 if (has_extended_chars) | |
| 2518 { | |
| 2519 /* There are extended chars here, which means we need to start | |
| 2520 over and shift to unified range-table format. */ | |
| 446 | 2521 if (buf_end[-2] == charset) |
| 2522 buf_end[-2] = charset_mule; | |
| 428 | 2523 else |
| 446 | 2524 buf_end[-2] = charset_mule_not; |
| 2525 buf_end--; | |
| 428 | 2526 p = p1; /* go back to the beginning of the charset, after |
| 2527 a possible ^. */ | |
| 2528 rtab = Vthe_lisp_rangetab; | |
| 2529 Fclear_range_table (rtab); | |
| 2530 | |
| 2531 /* charset_not matches newline according to a syntax bit. */ | |
| 446 | 2532 if ((re_opcode_t) buf_end[-1] == charset_mule_not |
| 428 | 2533 && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) |
| 2534 SET_EITHER_BIT ('\n'); | |
| 2535 } | |
| 2536 #endif /* MULE */ | |
| 2537 | |
| 2538 /* Read in characters and ranges, setting map bits. */ | |
| 2539 for (;;) | |
| 2540 { | |
| 2541 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2542 | |
| 446 | 2543 PATFETCH (c); |
| 428 | 2544 |
| 2545 #ifdef MULE | |
| 2546 if (c >= 0x80 && !has_extended_chars) | |
| 2547 { | |
| 2548 has_extended_chars = 1; | |
| 2549 /* Frumble-bumble, we've found some extended chars. | |
| 2550 Need to start over, process everything using | |
| 2551 the general extended-char mechanism, and need | |
| 2552 to use charset_mule and charset_mule_not instead | |
| 2553 of charset and charset_not. */ | |
| 2554 goto start_over_with_extended; | |
| 2555 } | |
| 2556 #endif /* MULE */ | |
| 2557 /* \ might escape characters inside [...] and [^...]. */ | |
| 2558 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') | |
| 2559 { | |
| 2560 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | |
| 2561 | |
| 446 | 2562 PATFETCH (c1); |
| 428 | 2563 #ifdef MULE |
| 2564 if (c1 >= 0x80 && !has_extended_chars) | |
| 2565 { | |
| 2566 has_extended_chars = 1; | |
| 2567 goto start_over_with_extended; | |
| 2568 } | |
| 2569 #endif /* MULE */ | |
| 2570 SET_EITHER_BIT (c1); | |
| 2571 continue; | |
| 2572 } | |
| 2573 | |
| 2574 /* Could be the end of the bracket expression. If it's | |
| 2575 not (i.e., when the bracket expression is `[]' so | |
| 2576 far), the ']' character bit gets set way below. */ | |
| 2577 if (c == ']' && p != p1 + 1) | |
| 2578 break; | |
| 2579 | |
| 2580 /* Look ahead to see if it's a range when the last thing | |
| 2581 was a character class. */ | |
| 2582 if (had_char_class && c == '-' && *p != ']') | |
| 2583 FREE_STACK_RETURN (REG_ERANGE); | |
| 2584 | |
| 2585 /* Look ahead to see if it's a range when the last thing | |
| 2586 was a character: if this is a hyphen not at the | |
| 2587 beginning or the end of a list, then it's the range | |
| 2588 operator. */ | |
| 2589 if (c == '-' | |
| 2590 && !(p - 2 >= pattern && p[-2] == '[') | |
| 446 | 2591 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') |
| 428 | 2592 && *p != ']') |
| 2593 { | |
| 2594 reg_errcode_t ret; | |
| 2595 | |
| 2596 #ifdef MULE | |
| 2597 if (* (unsigned char *) p >= 0x80 && !has_extended_chars) | |
| 2598 { | |
| 2599 has_extended_chars = 1; | |
| 2600 goto start_over_with_extended; | |
| 2601 } | |
| 2602 if (has_extended_chars) | |
| 2603 ret = compile_extended_range (&p, pend, translate, | |
| 2604 syntax, rtab); | |
| 2605 else | |
| 2606 #endif /* MULE */ | |
| 446 | 2607 ret = compile_range (&p, pend, translate, syntax, buf_end); |
| 428 | 2608 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); |
| 2609 } | |
| 2610 | |
| 2611 else if (p[0] == '-' && p[1] != ']') | |
| 2612 { /* This handles ranges made up of characters only. */ | |
| 2613 reg_errcode_t ret; | |
| 2614 | |
| 2615 /* Move past the `-'. */ | |
| 2616 PATFETCH (c1); | |
| 2617 | |
| 2618 #ifdef MULE | |
| 2619 if (* (unsigned char *) p >= 0x80 && !has_extended_chars) | |
| 2620 { | |
| 2621 has_extended_chars = 1; | |
| 2622 goto start_over_with_extended; | |
| 2623 } | |
| 2624 if (has_extended_chars) | |
| 2625 ret = compile_extended_range (&p, pend, translate, | |
| 2626 syntax, rtab); | |
| 2627 else | |
| 2628 #endif /* MULE */ | |
| 446 | 2629 ret = compile_range (&p, pend, translate, syntax, buf_end); |
| 428 | 2630 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); |
| 2631 } | |
| 2632 | |
| 2633 /* See if we're at the beginning of a possible character | |
| 2634 class. */ | |
| 2635 | |
| 2636 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') | |
| 2637 { /* Leave room for the null. */ | |
| 2638 char str[CHAR_CLASS_MAX_LENGTH + 1]; | |
| 2639 | |
| 2640 PATFETCH (c); | |
| 2641 c1 = 0; | |
| 2642 | |
| 2643 /* If pattern is `[[:'. */ | |
| 2644 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2645 | |
| 2646 for (;;) | |
| 2647 { | |
| 446 | 2648 /* #### This code is unused. |
| 2649 Correctness is not checked after TRT | |
| 2650 table change. */ | |
| 428 | 2651 PATFETCH (c); |
| 2652 if (c == ':' || c == ']' || p == pend | |
| 2653 || c1 == CHAR_CLASS_MAX_LENGTH) | |
| 2654 break; | |
| 442 | 2655 str[c1++] = (char) c; |
| 428 | 2656 } |
| 2657 str[c1] = '\0'; | |
| 2658 | |
| 446 | 2659 /* If isn't a word bracketed by `[:' and `:]': |
| 428 | 2660 undo the ending character, the letters, and leave |
| 2661 the leading `:' and `[' (but set bits for them). */ | |
| 2662 if (c == ':' && *p == ']') | |
| 2663 { | |
| 2664 int ch; | |
| 460 | 2665 re_bool is_alnum = STREQ (str, "alnum"); |
| 2666 re_bool is_alpha = STREQ (str, "alpha"); | |
| 2667 re_bool is_blank = STREQ (str, "blank"); | |
| 2668 re_bool is_cntrl = STREQ (str, "cntrl"); | |
| 2669 re_bool is_digit = STREQ (str, "digit"); | |
| 2670 re_bool is_graph = STREQ (str, "graph"); | |
| 2671 re_bool is_lower = STREQ (str, "lower"); | |
| 2672 re_bool is_print = STREQ (str, "print"); | |
| 2673 re_bool is_punct = STREQ (str, "punct"); | |
| 2674 re_bool is_space = STREQ (str, "space"); | |
| 2675 re_bool is_upper = STREQ (str, "upper"); | |
| 2676 re_bool is_xdigit = STREQ (str, "xdigit"); | |
| 428 | 2677 |
| 2678 if (!IS_CHAR_CLASS (str)) | |
| 2679 FREE_STACK_RETURN (REG_ECTYPE); | |
| 2680 | |
| 2681 /* Throw away the ] at the end of the character | |
| 2682 class. */ | |
| 2683 PATFETCH (c); | |
| 2684 | |
| 2685 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2686 | |
| 2687 for (ch = 0; ch < 1 << BYTEWIDTH; ch++) | |
| 2688 { | |
| 2689 /* This was split into 3 if's to | |
| 2690 avoid an arbitrary limit in some compiler. */ | |
| 2691 if ( (is_alnum && ISALNUM (ch)) | |
| 2692 || (is_alpha && ISALPHA (ch)) | |
| 2693 || (is_blank && ISBLANK (ch)) | |
| 2694 || (is_cntrl && ISCNTRL (ch))) | |
| 2695 SET_EITHER_BIT (ch); | |
| 2696 if ( (is_digit && ISDIGIT (ch)) | |
| 2697 || (is_graph && ISGRAPH (ch)) | |
| 2698 || (is_lower && ISLOWER (ch)) | |
| 2699 || (is_print && ISPRINT (ch))) | |
| 2700 SET_EITHER_BIT (ch); | |
| 2701 if ( (is_punct && ISPUNCT (ch)) | |
| 2702 || (is_space && ISSPACE (ch)) | |
| 2703 || (is_upper && ISUPPER (ch)) | |
| 2704 || (is_xdigit && ISXDIGIT (ch))) | |
| 2705 SET_EITHER_BIT (ch); | |
| 2706 } | |
| 2707 had_char_class = true; | |
| 2708 } | |
| 2709 else | |
| 2710 { | |
| 2711 c1++; | |
| 2712 while (c1--) | |
| 2713 PATUNFETCH; | |
| 2714 SET_EITHER_BIT ('['); | |
| 2715 SET_EITHER_BIT (':'); | |
| 2716 had_char_class = false; | |
| 2717 } | |
| 2718 } | |
| 2719 else | |
| 2720 { | |
| 2721 had_char_class = false; | |
| 2722 SET_EITHER_BIT (c); | |
| 2723 } | |
| 2724 } | |
| 2725 | |
| 2726 #ifdef MULE | |
| 2727 if (has_extended_chars) | |
| 2728 { | |
| 2729 /* We have a range table, not a bit vector. */ | |
| 2730 int bytes_needed = | |
| 2731 unified_range_table_bytes_needed (rtab); | |
| 2732 GET_BUFFER_SPACE (bytes_needed); | |
| 446 | 2733 unified_range_table_copy_data (rtab, buf_end); |
| 2734 buf_end += unified_range_table_bytes_used (buf_end); | |
| 428 | 2735 break; |
| 2736 } | |
| 2737 #endif /* MULE */ | |
| 2738 /* Discard any (non)matching list bytes that are all 0 at the | |
| 2739 end of the map. Decrease the map-length byte too. */ | |
| 446 | 2740 while ((int) buf_end[-1] > 0 && buf_end[buf_end[-1] - 1] == 0) |
| 2741 buf_end[-1]--; | |
| 2742 buf_end += buf_end[-1]; | |
| 428 | 2743 } |
| 2744 break; | |
| 2745 | |
| 2746 | |
| 2747 case '(': | |
| 2748 if (syntax & RE_NO_BK_PARENS) | |
| 2749 goto handle_open; | |
| 2750 else | |
| 2751 goto normal_char; | |
| 2752 | |
| 2753 | |
| 2754 case ')': | |
| 2755 if (syntax & RE_NO_BK_PARENS) | |
| 2756 goto handle_close; | |
| 2757 else | |
| 2758 goto normal_char; | |
| 2759 | |
| 2760 | |
| 2761 case '\n': | |
| 2762 if (syntax & RE_NEWLINE_ALT) | |
| 2763 goto handle_alt; | |
| 2764 else | |
| 2765 goto normal_char; | |
| 2766 | |
| 2767 | |
| 2768 case '|': | |
| 2769 if (syntax & RE_NO_BK_VBAR) | |
| 2770 goto handle_alt; | |
| 2771 else | |
| 2772 goto normal_char; | |
| 2773 | |
| 2774 | |
| 2775 case '{': | |
| 2776 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) | |
| 2777 goto handle_interval; | |
| 2778 else | |
| 2779 goto normal_char; | |
| 2780 | |
| 2781 | |
| 2782 case '\\': | |
| 2783 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | |
| 2784 | |
| 2785 /* Do not translate the character after the \, so that we can | |
| 2786 distinguish, e.g., \B from \b, even if we normally would | |
| 2787 translate, e.g., B to b. */ | |
| 2788 PATFETCH_RAW (c); | |
| 2789 | |
| 2790 switch (c) | |
| 2791 { | |
| 2792 case '(': | |
| 2793 if (syntax & RE_NO_BK_PARENS) | |
| 2794 goto normal_backslash; | |
| 2795 | |
| 2796 handle_open: | |
| 2797 { | |
| 2798 regnum_t r; | |
| 502 | 2799 int shy = 0; |
| 428 | 2800 |
| 2801 if (!(syntax & RE_NO_SHY_GROUPS) | |
| 2802 && p != pend | |
| 446 | 2803 && *p == '?') |
| 428 | 2804 { |
| 2805 p++; | |
| 446 | 2806 PATFETCH (c); |
| 428 | 2807 switch (c) |
| 2808 { | |
| 2809 case ':': /* shy groups */ | |
| 502 | 2810 shy = 1; |
| 428 | 2811 break; |
| 2812 | |
| 2813 /* All others are reserved for future constructs. */ | |
| 2814 default: | |
| 2815 FREE_STACK_RETURN (REG_BADPAT); | |
| 2816 } | |
| 2817 } | |
| 502 | 2818 |
| 2819 r = ++regnum; | |
| 2820 bufp->re_ngroups++; | |
| 2821 if (!shy) | |
| 2822 { | |
| 2823 bufp->re_nsub++; | |
| 2824 while (bufp->external_to_internal_register_size <= | |
| 2825 bufp->re_nsub) | |
| 2826 { | |
| 2827 int i; | |
| 2828 int old_size = | |
| 2829 bufp->external_to_internal_register_size; | |
| 2830 bufp->external_to_internal_register_size += 5; | |
| 2831 RETALLOC (bufp->external_to_internal_register, | |
| 2832 bufp->external_to_internal_register_size, | |
| 2833 int); | |
| 2834 /* debugging */ | |
| 2835 for (i = old_size; | |
| 2836 i < bufp->external_to_internal_register_size; i++) | |
| 2837 bufp->external_to_internal_register[i] = | |
| 2838 (int) 0xDEADBEEF; | |
| 2839 } | |
| 2840 | |
| 2841 bufp->external_to_internal_register[bufp->re_nsub] = | |
| 2842 bufp->re_ngroups; | |
| 2843 } | |
| 428 | 2844 |
| 2845 if (COMPILE_STACK_FULL) | |
| 2846 { | |
| 2847 RETALLOC (compile_stack.stack, compile_stack.size << 1, | |
| 2848 compile_stack_elt_t); | |
| 2849 if (compile_stack.stack == NULL) return REG_ESPACE; | |
| 2850 | |
| 2851 compile_stack.size <<= 1; | |
| 2852 } | |
| 2853 | |
| 2854 /* These are the values to restore when we hit end of this | |
| 2855 group. They are all relative offsets, so that if the | |
| 2856 whole pattern moves because of realloc, they will still | |
| 2857 be valid. */ | |
| 2858 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; | |
| 2859 COMPILE_STACK_TOP.fixup_alt_jump | |
| 2860 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; | |
| 446 | 2861 COMPILE_STACK_TOP.laststart_offset = buf_end - bufp->buffer; |
| 428 | 2862 COMPILE_STACK_TOP.regnum = r; |
| 2863 | |
| 2864 /* We will eventually replace the 0 with the number of | |
| 2865 groups inner to this one. But do not push a | |
| 2866 start_memory for groups beyond the last one we can | |
| 502 | 2867 represent in the compiled pattern. |
| 2868 #### bad bad bad. this will fail in lots of ways, if we | |
| 2869 ever have to backtrack for these groups. | |
| 2870 */ | |
| 428 | 2871 if (r <= MAX_REGNUM) |
| 2872 { | |
| 2873 COMPILE_STACK_TOP.inner_group_offset | |
| 446 | 2874 = buf_end - bufp->buffer + 2; |
| 428 | 2875 BUF_PUSH_3 (start_memory, r, 0); |
| 2876 } | |
| 2877 | |
| 2878 compile_stack.avail++; | |
| 2879 | |
| 2880 fixup_alt_jump = 0; | |
| 2881 laststart = 0; | |
| 446 | 2882 begalt = buf_end; |
| 428 | 2883 /* If we've reached MAX_REGNUM groups, then this open |
| 2884 won't actually generate any code, so we'll have to | |
| 2885 clear pending_exact explicitly. */ | |
| 2886 pending_exact = 0; | |
| 2887 } | |
| 2888 break; | |
| 2889 | |
| 2890 | |
| 2891 case ')': | |
| 2892 if (syntax & RE_NO_BK_PARENS) goto normal_backslash; | |
| 2893 | |
| 2894 if (COMPILE_STACK_EMPTY) { | |
| 2895 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) | |
| 2896 goto normal_backslash; | |
| 2897 else | |
| 2898 FREE_STACK_RETURN (REG_ERPAREN); | |
| 2899 } | |
| 2900 | |
| 2901 handle_close: | |
| 2902 if (fixup_alt_jump) | |
| 2903 { /* Push a dummy failure point at the end of the | |
| 2904 alternative for a possible future | |
| 2905 `pop_failure_jump' to pop. See comments at | |
| 2906 `push_dummy_failure' in `re_match_2'. */ | |
| 2907 BUF_PUSH (push_dummy_failure); | |
| 2908 | |
| 2909 /* We allocated space for this jump when we assigned | |
| 2910 to `fixup_alt_jump', in the `handle_alt' case below. */ | |
| 446 | 2911 STORE_JUMP (jump_past_alt, fixup_alt_jump, buf_end - 1); |
| 428 | 2912 } |
| 2913 | |
| 2914 /* See similar code for backslashed left paren above. */ | |
| 2915 if (COMPILE_STACK_EMPTY) { | |
| 2916 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) | |
| 2917 goto normal_char; | |
| 2918 else | |
| 2919 FREE_STACK_RETURN (REG_ERPAREN); | |
| 2920 } | |
| 2921 | |
| 2922 /* Since we just checked for an empty stack above, this | |
| 2923 ``can't happen''. */ | |
| 2924 assert (compile_stack.avail != 0); | |
| 2925 { | |
| 2926 /* We don't just want to restore into `regnum', because | |
| 2927 later groups should continue to be numbered higher, | |
| 2928 as in `(ab)c(de)' -- the second group is #2. */ | |
| 2929 regnum_t this_group_regnum; | |
| 2930 | |
| 2931 compile_stack.avail--; | |
| 2932 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; | |
| 2933 fixup_alt_jump | |
| 2934 = COMPILE_STACK_TOP.fixup_alt_jump | |
| 2935 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 | |
| 2936 : 0; | |
| 2937 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; | |
| 2938 this_group_regnum = COMPILE_STACK_TOP.regnum; | |
| 2939 /* If we've reached MAX_REGNUM groups, then this open | |
| 2940 won't actually generate any code, so we'll have to | |
| 2941 clear pending_exact explicitly. */ | |
| 2942 pending_exact = 0; | |
| 2943 | |
| 2944 /* We're at the end of the group, so now we know how many | |
| 2945 groups were inside this one. */ | |
| 2946 if (this_group_regnum <= MAX_REGNUM) | |
| 2947 { | |
| 2948 unsigned char *inner_group_loc | |
| 2949 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; | |
| 2950 | |
| 2951 *inner_group_loc = regnum - this_group_regnum; | |
| 2952 BUF_PUSH_3 (stop_memory, this_group_regnum, | |
| 2953 regnum - this_group_regnum); | |
| 2954 } | |
| 2955 } | |
| 2956 break; | |
| 2957 | |
| 2958 | |
| 2959 case '|': /* `\|'. */ | |
| 2960 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) | |
| 2961 goto normal_backslash; | |
| 2962 handle_alt: | |
| 2963 if (syntax & RE_LIMITED_OPS) | |
| 2964 goto normal_char; | |
| 2965 | |
| 2966 /* Insert before the previous alternative a jump which | |
| 2967 jumps to this alternative if the former fails. */ | |
| 2968 GET_BUFFER_SPACE (3); | |
| 446 | 2969 INSERT_JUMP (on_failure_jump, begalt, buf_end + 6); |
| 428 | 2970 pending_exact = 0; |
| 446 | 2971 buf_end += 3; |
| 428 | 2972 |
| 2973 /* The alternative before this one has a jump after it | |
| 2974 which gets executed if it gets matched. Adjust that | |
| 2975 jump so it will jump to this alternative's analogous | |
| 2976 jump (put in below, which in turn will jump to the next | |
| 2977 (if any) alternative's such jump, etc.). The last such | |
| 2978 jump jumps to the correct final destination. A picture: | |
| 2979 _____ _____ | |
| 2980 | | | | | |
| 2981 | v | v | |
| 2982 a | b | c | |
| 2983 | |
| 2984 If we are at `b', then fixup_alt_jump right now points to a | |
| 2985 three-byte space after `a'. We'll put in the jump, set | |
| 2986 fixup_alt_jump to right after `b', and leave behind three | |
| 2987 bytes which we'll fill in when we get to after `c'. */ | |
| 2988 | |
| 2989 if (fixup_alt_jump) | |
| 446 | 2990 STORE_JUMP (jump_past_alt, fixup_alt_jump, buf_end); |
| 428 | 2991 |
| 2992 /* Mark and leave space for a jump after this alternative, | |
| 2993 to be filled in later either by next alternative or | |
| 2994 when know we're at the end of a series of alternatives. */ | |
| 446 | 2995 fixup_alt_jump = buf_end; |
| 428 | 2996 GET_BUFFER_SPACE (3); |
| 446 | 2997 buf_end += 3; |
| 428 | 2998 |
| 2999 laststart = 0; | |
| 446 | 3000 begalt = buf_end; |
| 428 | 3001 break; |
| 3002 | |
| 3003 | |
| 3004 case '{': | |
| 3005 /* If \{ is a literal. */ | |
| 3006 if (!(syntax & RE_INTERVALS) | |
| 3007 /* If we're at `\{' and it's not the open-interval | |
| 3008 operator. */ | |
| 3009 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) | |
| 3010 || (p - 2 == pattern && p == pend)) | |
| 3011 goto normal_backslash; | |
| 3012 | |
| 3013 handle_interval: | |
| 3014 { | |
| 3015 /* If got here, then the syntax allows intervals. */ | |
| 3016 | |
| 3017 /* At least (most) this many matches must be made. */ | |
| 3018 int lower_bound = -1, upper_bound = -1; | |
| 3019 | |
| 3020 beg_interval = p - 1; | |
| 3021 | |
| 3022 if (p == pend) | |
| 3023 { | |
| 3024 if (syntax & RE_NO_BK_BRACES) | |
| 3025 goto unfetch_interval; | |
| 3026 else | |
| 3027 FREE_STACK_RETURN (REG_EBRACE); | |
| 3028 } | |
| 3029 | |
| 3030 GET_UNSIGNED_NUMBER (lower_bound); | |
| 3031 | |
| 3032 if (c == ',') | |
| 3033 { | |
| 3034 GET_UNSIGNED_NUMBER (upper_bound); | |
| 3035 if (upper_bound < 0) upper_bound = RE_DUP_MAX; | |
| 3036 } | |
| 3037 else | |
| 3038 /* Interval such as `{1}' => match exactly once. */ | |
| 3039 upper_bound = lower_bound; | |
| 3040 | |
| 3041 if (lower_bound < 0 || upper_bound > RE_DUP_MAX | |
| 3042 || lower_bound > upper_bound) | |
| 3043 { | |
| 3044 if (syntax & RE_NO_BK_BRACES) | |
| 3045 goto unfetch_interval; | |
| 3046 else | |
| 3047 FREE_STACK_RETURN (REG_BADBR); | |
| 3048 } | |
| 3049 | |
| 3050 if (!(syntax & RE_NO_BK_BRACES)) | |
| 3051 { | |
| 3052 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE); | |
| 3053 | |
| 3054 PATFETCH (c); | |
| 3055 } | |
| 3056 | |
| 3057 if (c != '}') | |
| 3058 { | |
| 3059 if (syntax & RE_NO_BK_BRACES) | |
| 3060 goto unfetch_interval; | |
| 3061 else | |
| 3062 FREE_STACK_RETURN (REG_BADBR); | |
| 3063 } | |
| 3064 | |
| 3065 /* We just parsed a valid interval. */ | |
| 3066 | |
| 3067 /* If it's invalid to have no preceding re. */ | |
| 3068 if (!laststart) | |
| 3069 { | |
| 3070 if (syntax & RE_CONTEXT_INVALID_OPS) | |
| 3071 FREE_STACK_RETURN (REG_BADRPT); | |
| 3072 else if (syntax & RE_CONTEXT_INDEP_OPS) | |
| 446 | 3073 laststart = buf_end; |
| 428 | 3074 else |
| 3075 goto unfetch_interval; | |
| 3076 } | |
| 3077 | |
| 3078 /* If the upper bound is zero, don't want to succeed at | |
| 3079 all; jump from `laststart' to `b + 3', which will be | |
| 3080 the end of the buffer after we insert the jump. */ | |
| 3081 if (upper_bound == 0) | |
| 3082 { | |
| 3083 GET_BUFFER_SPACE (3); | |
| 446 | 3084 INSERT_JUMP (jump, laststart, buf_end + 3); |
| 3085 buf_end += 3; | |
| 428 | 3086 } |
| 3087 | |
| 3088 /* Otherwise, we have a nontrivial interval. When | |
| 3089 we're all done, the pattern will look like: | |
| 3090 set_number_at <jump count> <upper bound> | |
| 3091 set_number_at <succeed_n count> <lower bound> | |
| 3092 succeed_n <after jump addr> <succeed_n count> | |
| 3093 <body of loop> | |
| 3094 jump_n <succeed_n addr> <jump count> | |
| 3095 (The upper bound and `jump_n' are omitted if | |
| 3096 `upper_bound' is 1, though.) */ | |
| 3097 else | |
| 3098 { /* If the upper bound is > 1, we need to insert | |
| 3099 more at the end of the loop. */ | |
| 647 | 3100 int nbytes = 10 + (upper_bound > 1) * 10; |
| 428 | 3101 |
| 3102 GET_BUFFER_SPACE (nbytes); | |
| 3103 | |
| 3104 /* Initialize lower bound of the `succeed_n', even | |
| 3105 though it will be set during matching by its | |
| 3106 attendant `set_number_at' (inserted next), | |
| 3107 because `re_compile_fastmap' needs to know. | |
| 3108 Jump to the `jump_n' we might insert below. */ | |
| 3109 INSERT_JUMP2 (succeed_n, laststart, | |
| 446 | 3110 buf_end + 5 + (upper_bound > 1) * 5, |
| 428 | 3111 lower_bound); |
| 446 | 3112 buf_end += 5; |
| 428 | 3113 |
| 3114 /* Code to initialize the lower bound. Insert | |
| 3115 before the `succeed_n'. The `5' is the last two | |
| 3116 bytes of this `set_number_at', plus 3 bytes of | |
| 3117 the following `succeed_n'. */ | |
| 446 | 3118 insert_op2 (set_number_at, laststart, 5, lower_bound, buf_end); |
| 3119 buf_end += 5; | |
| 428 | 3120 |
| 3121 if (upper_bound > 1) | |
| 3122 { /* More than one repetition is allowed, so | |
| 3123 append a backward jump to the `succeed_n' | |
| 3124 that starts this interval. | |
| 3125 | |
| 3126 When we've reached this during matching, | |
| 3127 we'll have matched the interval once, so | |
| 3128 jump back only `upper_bound - 1' times. */ | |
| 446 | 3129 STORE_JUMP2 (jump_n, buf_end, laststart + 5, |
| 428 | 3130 upper_bound - 1); |
| 446 | 3131 buf_end += 5; |
| 428 | 3132 |
| 3133 /* The location we want to set is the second | |
| 3134 parameter of the `jump_n'; that is `b-2' as | |
| 3135 an absolute address. `laststart' will be | |
| 3136 the `set_number_at' we're about to insert; | |
| 3137 `laststart+3' the number to set, the source | |
| 3138 for the relative address. But we are | |
| 3139 inserting into the middle of the pattern -- | |
| 3140 so everything is getting moved up by 5. | |
| 3141 Conclusion: (b - 2) - (laststart + 3) + 5, | |
| 3142 i.e., b - laststart. | |
| 3143 | |
| 3144 We insert this at the beginning of the loop | |
| 3145 so that if we fail during matching, we'll | |
| 3146 reinitialize the bounds. */ | |
| 446 | 3147 insert_op2 (set_number_at, laststart, |
| 3148 buf_end - laststart, | |
| 3149 upper_bound - 1, buf_end); | |
| 3150 buf_end += 5; | |
| 428 | 3151 } |
| 3152 } | |
| 3153 pending_exact = 0; | |
| 3154 beg_interval = NULL; | |
| 3155 } | |
| 3156 break; | |
| 3157 | |
| 3158 unfetch_interval: | |
| 3159 /* If an invalid interval, match the characters as literals. */ | |
| 3160 assert (beg_interval); | |
| 3161 p = beg_interval; | |
| 3162 beg_interval = NULL; | |
| 3163 | |
| 3164 /* normal_char and normal_backslash need `c'. */ | |
| 3165 PATFETCH (c); | |
| 3166 | |
| 3167 if (!(syntax & RE_NO_BK_BRACES)) | |
| 3168 { | |
| 3169 if (p > pattern && p[-1] == '\\') | |
| 3170 goto normal_backslash; | |
| 3171 } | |
| 3172 goto normal_char; | |
| 3173 | |
| 3174 #ifdef emacs | |
| 3175 /* There is no way to specify the before_dot and after_dot | |
| 3176 operators. rms says this is ok. --karl */ | |
| 3177 case '=': | |
| 3178 BUF_PUSH (at_dot); | |
| 3179 break; | |
| 3180 | |
| 3181 case 's': | |
| 446 | 3182 laststart = buf_end; |
| 428 | 3183 PATFETCH (c); |
| 3184 /* XEmacs addition */ | |
| 3185 if (c >= 0x80 || syntax_spec_code[c] == 0377) | |
| 3186 FREE_STACK_RETURN (REG_ESYNTAX); | |
| 3187 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); | |
| 3188 break; | |
| 3189 | |
| 3190 case 'S': | |
| 446 | 3191 laststart = buf_end; |
| 428 | 3192 PATFETCH (c); |
| 3193 /* XEmacs addition */ | |
| 3194 if (c >= 0x80 || syntax_spec_code[c] == 0377) | |
| 3195 FREE_STACK_RETURN (REG_ESYNTAX); | |
| 3196 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); | |
| 3197 break; | |
| 3198 | |
| 3199 #ifdef MULE | |
| 3200 /* 97.2.17 jhod merged in to XEmacs from mule-2.3 */ | |
| 3201 case 'c': | |
| 446 | 3202 laststart = buf_end; |
| 428 | 3203 PATFETCH_RAW (c); |
| 3204 if (c < 32 || c > 127) | |
| 3205 FREE_STACK_RETURN (REG_ECATEGORY); | |
| 3206 BUF_PUSH_2 (categoryspec, c); | |
| 3207 break; | |
| 3208 | |
| 3209 case 'C': | |
| 446 | 3210 laststart = buf_end; |
| 428 | 3211 PATFETCH_RAW (c); |
| 3212 if (c < 32 || c > 127) | |
| 3213 FREE_STACK_RETURN (REG_ECATEGORY); | |
| 3214 BUF_PUSH_2 (notcategoryspec, c); | |
| 3215 break; | |
| 3216 /* end of category patch */ | |
| 3217 #endif /* MULE */ | |
| 3218 #endif /* emacs */ | |
| 3219 | |
| 3220 | |
| 3221 case 'w': | |
| 446 | 3222 laststart = buf_end; |
| 428 | 3223 BUF_PUSH (wordchar); |
| 3224 break; | |
| 3225 | |
| 3226 | |
| 3227 case 'W': | |
| 446 | 3228 laststart = buf_end; |
| 428 | 3229 BUF_PUSH (notwordchar); |
| 3230 break; | |
| 3231 | |
| 3232 | |
| 3233 case '<': | |
| 3234 BUF_PUSH (wordbeg); | |
| 3235 break; | |
| 3236 | |
| 3237 case '>': | |
| 3238 BUF_PUSH (wordend); | |
| 3239 break; | |
| 3240 | |
| 3241 case 'b': | |
| 3242 BUF_PUSH (wordbound); | |
| 3243 break; | |
| 3244 | |
| 3245 case 'B': | |
| 3246 BUF_PUSH (notwordbound); | |
| 3247 break; | |
| 3248 | |
| 3249 case '`': | |
| 3250 BUF_PUSH (begbuf); | |
| 3251 break; | |
| 3252 | |
| 3253 case '\'': | |
| 3254 BUF_PUSH (endbuf); | |
| 3255 break; | |
| 3256 | |
| 3257 case '1': case '2': case '3': case '4': case '5': | |
| 3258 case '6': case '7': case '8': case '9': | |
| 446 | 3259 { |
| 502 | 3260 regnum_t reg, regint; |
| 3261 int may_need_to_unfetch = 0; | |
| 446 | 3262 if (syntax & RE_NO_BK_REFS) |
| 3263 goto normal_char; | |
| 3264 | |
| 502 | 3265 /* This only goes up to 99. It could be extended to work |
| 3266 up to 255 (the maximum number of registers that can be | |
| 3267 handled by the current regexp engine, because it stores | |
| 3268 its register numbers in the compiled pattern as one byte, | |
| 3269 ugh). Doing that's a bit trickier, because you might | |
| 3270 have the case where \25 a back-ref but \255 is not, ... */ | |
| 446 | 3271 reg = c - '0'; |
| 502 | 3272 if (p < pend) |
| 3273 { | |
| 3274 PATFETCH (c); | |
| 3275 if (c >= '0' && c <= '9') | |
| 3276 { | |
| 3277 regnum_t new_reg = reg * 10 + c - '0'; | |
| 3278 if (new_reg <= bufp->re_nsub) | |
| 3279 { | |
| 3280 reg = new_reg; | |
| 3281 may_need_to_unfetch = 1; | |
| 3282 } | |
| 3283 else | |
| 3284 PATUNFETCH; | |
| 3285 } | |
| 523 | 3286 else |
| 3287 PATUNFETCH; | |
| 502 | 3288 } |
| 3289 | |
| 3290 if (reg > bufp->re_nsub) | |
| 446 | 3291 FREE_STACK_RETURN (REG_ESUBREG); |
| 3292 | |
| 502 | 3293 regint = bufp->external_to_internal_register[reg]; |
| 446 | 3294 /* Can't back reference to a subexpression if inside of it. */ |
| 502 | 3295 if (group_in_compile_stack (compile_stack, regint)) |
| 3296 { | |
| 3297 if (may_need_to_unfetch) | |
| 3298 PATUNFETCH; | |
| 3299 goto normal_char; | |
| 3300 } | |
| 3301 | |
| 3302 #ifdef emacs | |
| 3303 if (reg > 9 && | |
| 3304 bufp->warned_about_incompatible_back_references == 0) | |
| 3305 { | |
| 3306 bufp->warned_about_incompatible_back_references = 1; | |
| 3307 warn_when_safe (intern ("regex"), Qinfo, | |
| 3308 "Back reference \\%d now has new " | |
| 3309 "semantics in %s", reg, pattern); | |
| 3310 } | |
| 3311 #endif | |
| 446 | 3312 |
| 3313 laststart = buf_end; | |
| 502 | 3314 BUF_PUSH_2 (duplicate, regint); |
| 446 | 3315 } |
| 428 | 3316 break; |
| 3317 | |
| 3318 | |
| 3319 case '+': | |
| 3320 case '?': | |
| 3321 if (syntax & RE_BK_PLUS_QM) | |
| 3322 goto handle_plus; | |
| 3323 else | |
| 3324 goto normal_backslash; | |
| 3325 | |
| 3326 default: | |
| 3327 normal_backslash: | |
| 3328 /* You might think it would be useful for \ to mean | |
| 3329 not to translate; but if we don't translate it, | |
| 3330 it will never match anything. */ | |
| 826 | 3331 c = RE_TRANSLATE (c); |
| 428 | 3332 goto normal_char; |
| 3333 } | |
| 3334 break; | |
| 3335 | |
| 3336 | |
| 3337 default: | |
| 3338 /* Expects the character in `c'. */ | |
| 3339 /* `p' points to the location after where `c' came from. */ | |
| 3340 normal_char: | |
| 3341 { | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3342 /* The following conditional synced to GNU Emacs 22.1. */ |
| 428 | 3343 /* If no exactn currently being built. */ |
| 3344 if (!pending_exact | |
| 3345 | |
| 3346 /* If last exactn not at current position. */ | |
| 446 | 3347 || pending_exact + *pending_exact + 1 != buf_end |
| 428 | 3348 |
| 3349 /* We have only one byte following the exactn for the count. */ | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3350 || *pending_exact >= (1 << BYTEWIDTH) - MAX_ICHAR_LEN |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3351 |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3352 /* If followed by a repetition operator. |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3353 If the lookahead fails because of end of pattern, any |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3354 trailing backslash will get caught later. */ |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3355 || (p != pend && (*p == '*' || *p == '^')) |
| 428 | 3356 || ((syntax & RE_BK_PLUS_QM) |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3357 ? p + 1 < pend && *p == '\\' && (p[1] == '+' || p[1] == '?') |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3358 : p != pend && (*p == '+' || *p == '?')) |
| 428 | 3359 || ((syntax & RE_INTERVALS) |
| 3360 && ((syntax & RE_NO_BK_BRACES) | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3361 ? p != pend && *p == '{' |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3362 : p + 1 < pend && (p[0] == '\\' && p[1] == '{')))) |
| 428 | 3363 { |
| 3364 /* Start building a new exactn. */ | |
| 3365 | |
| 446 | 3366 laststart = buf_end; |
| 428 | 3367 |
| 3368 BUF_PUSH_2 (exactn, 0); | |
| 446 | 3369 pending_exact = buf_end - 1; |
| 428 | 3370 } |
| 3371 | |
| 446 | 3372 #ifndef MULE |
| 428 | 3373 BUF_PUSH (c); |
| 3374 (*pending_exact)++; | |
| 446 | 3375 #else |
| 3376 { | |
| 3377 Bytecount bt_count; | |
| 867 | 3378 Ibyte tmp_buf[MAX_ICHAR_LEN]; |
| 446 | 3379 int i; |
| 3380 | |
| 867 | 3381 bt_count = set_itext_ichar (tmp_buf, c); |
| 446 | 3382 |
| 3383 for (i = 0; i < bt_count; i++) | |
| 3384 { | |
| 3385 BUF_PUSH (tmp_buf[i]); | |
| 3386 (*pending_exact)++; | |
| 3387 } | |
| 3388 } | |
| 3389 #endif | |
| 428 | 3390 break; |
| 3391 } | |
| 3392 } /* switch (c) */ | |
| 3393 } /* while p != pend */ | |
| 3394 | |
| 3395 | |
| 3396 /* Through the pattern now. */ | |
| 3397 | |
| 3398 if (fixup_alt_jump) | |
| 446 | 3399 STORE_JUMP (jump_past_alt, fixup_alt_jump, buf_end); |
| 428 | 3400 |
| 3401 if (!COMPILE_STACK_EMPTY) | |
| 3402 FREE_STACK_RETURN (REG_EPAREN); | |
| 3403 | |
| 3404 /* If we don't want backtracking, force success | |
| 3405 the first time we reach the end of the compiled pattern. */ | |
| 3406 if (syntax & RE_NO_POSIX_BACKTRACKING) | |
| 3407 BUF_PUSH (succeed); | |
| 3408 | |
| 1726 | 3409 xfree (compile_stack.stack, compile_stack_elt_t *); |
| 428 | 3410 |
| 3411 /* We have succeeded; set the length of the buffer. */ | |
| 446 | 3412 bufp->used = buf_end - bufp->buffer; |
| 428 | 3413 |
| 3414 #ifdef DEBUG | |
| 3415 if (debug) | |
| 3416 { | |
| 3417 DEBUG_PRINT1 ("\nCompiled pattern: \n"); | |
| 3418 print_compiled_pattern (bufp); | |
| 3419 } | |
| 3420 #endif /* DEBUG */ | |
| 3421 | |
| 3422 #ifndef MATCH_MAY_ALLOCATE | |
| 3423 /* Initialize the failure stack to the largest possible stack. This | |
| 3424 isn't necessary unless we're trying to avoid calling alloca in | |
| 3425 the search and match routines. */ | |
| 3426 { | |
| 502 | 3427 int num_regs = bufp->re_ngroups + 1; |
| 428 | 3428 |
| 3429 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size | |
| 3430 is strictly greater than re_max_failures, the largest possible stack | |
| 3431 is 2 * re_max_failures failure points. */ | |
| 3432 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS)) | |
| 3433 { | |
| 3434 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS); | |
| 3435 | |
| 3436 if (! fail_stack.stack) | |
| 3437 fail_stack.stack | |
| 3438 = (fail_stack_elt_t *) xmalloc (fail_stack.size | |
| 3439 * sizeof (fail_stack_elt_t)); | |
| 3440 else | |
| 3441 fail_stack.stack | |
| 3442 = (fail_stack_elt_t *) xrealloc (fail_stack.stack, | |
| 3443 (fail_stack.size | |
| 3444 * sizeof (fail_stack_elt_t))); | |
| 3445 } | |
| 3446 | |
| 3447 regex_grow_registers (num_regs); | |
| 3448 } | |
| 3449 #endif /* not MATCH_MAY_ALLOCATE */ | |
| 3450 | |
| 3451 return REG_NOERROR; | |
| 3452 } /* regex_compile */ | |
| 3453 | |
| 3454 /* Subroutines for `regex_compile'. */ | |
| 3455 | |
| 3456 /* Store OP at LOC followed by two-byte integer parameter ARG. */ | |
| 3457 | |
| 3458 static void | |
| 3459 store_op1 (re_opcode_t op, unsigned char *loc, int arg) | |
| 3460 { | |
| 3461 *loc = (unsigned char) op; | |
| 3462 STORE_NUMBER (loc + 1, arg); | |
| 3463 } | |
| 3464 | |
| 3465 | |
| 3466 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */ | |
| 3467 | |
| 3468 static void | |
| 3469 store_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2) | |
| 3470 { | |
| 3471 *loc = (unsigned char) op; | |
| 3472 STORE_NUMBER (loc + 1, arg1); | |
| 3473 STORE_NUMBER (loc + 3, arg2); | |
| 3474 } | |
| 3475 | |
| 3476 | |
| 3477 /* Copy the bytes from LOC to END to open up three bytes of space at LOC | |
| 3478 for OP followed by two-byte integer parameter ARG. */ | |
| 3479 | |
| 3480 static void | |
| 3481 insert_op1 (re_opcode_t op, unsigned char *loc, int arg, unsigned char *end) | |
| 3482 { | |
| 3483 REGISTER unsigned char *pfrom = end; | |
| 3484 REGISTER unsigned char *pto = end + 3; | |
| 3485 | |
| 3486 while (pfrom != loc) | |
| 3487 *--pto = *--pfrom; | |
| 3488 | |
| 3489 store_op1 (op, loc, arg); | |
| 3490 } | |
| 3491 | |
| 3492 | |
| 3493 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */ | |
| 3494 | |
| 3495 static void | |
| 3496 insert_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2, | |
| 3497 unsigned char *end) | |
| 3498 { | |
| 3499 REGISTER unsigned char *pfrom = end; | |
| 3500 REGISTER unsigned char *pto = end + 5; | |
| 3501 | |
| 3502 while (pfrom != loc) | |
| 3503 *--pto = *--pfrom; | |
| 3504 | |
| 3505 store_op2 (op, loc, arg1, arg2); | |
| 3506 } | |
| 3507 | |
| 3508 | |
| 3509 /* P points to just after a ^ in PATTERN. Return true if that ^ comes | |
| 3510 after an alternative or a begin-subexpression. We assume there is at | |
| 3511 least one character before the ^. */ | |
| 3512 | |
| 460 | 3513 static re_bool |
| 446 | 3514 at_begline_loc_p (re_char *pattern, re_char *p, reg_syntax_t syntax) |
| 428 | 3515 { |
| 446 | 3516 re_char *prev = p - 2; |
| 460 | 3517 re_bool prev_prev_backslash = prev > pattern && prev[-1] == '\\'; |
| 428 | 3518 |
| 3519 return | |
| 3520 /* After a subexpression? */ | |
| 3521 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) | |
| 3522 /* After an alternative? */ | |
| 3523 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); | |
| 3524 } | |
| 3525 | |
| 3526 | |
| 3527 /* The dual of at_begline_loc_p. This one is for $. We assume there is | |
| 3528 at least one character after the $, i.e., `P < PEND'. */ | |
| 3529 | |
| 460 | 3530 static re_bool |
| 446 | 3531 at_endline_loc_p (re_char *p, re_char *pend, int syntax) |
| 428 | 3532 { |
| 446 | 3533 re_char *next = p; |
| 460 | 3534 re_bool next_backslash = *next == '\\'; |
| 446 | 3535 re_char *next_next = p + 1 < pend ? p + 1 : 0; |
| 428 | 3536 |
| 3537 return | |
| 3538 /* Before a subexpression? */ | |
| 3539 (syntax & RE_NO_BK_PARENS ? *next == ')' | |
| 3540 : next_backslash && next_next && *next_next == ')') | |
| 3541 /* Before an alternative? */ | |
| 3542 || (syntax & RE_NO_BK_VBAR ? *next == '|' | |
| 3543 : next_backslash && next_next && *next_next == '|'); | |
| 3544 } | |
| 3545 | |
| 3546 | |
| 3547 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and | |
| 3548 false if it's not. */ | |
| 3549 | |
| 460 | 3550 static re_bool |
| 428 | 3551 group_in_compile_stack (compile_stack_type compile_stack, regnum_t regnum) |
| 3552 { | |
| 3553 int this_element; | |
| 3554 | |
| 3555 for (this_element = compile_stack.avail - 1; | |
| 3556 this_element >= 0; | |
| 3557 this_element--) | |
| 3558 if (compile_stack.stack[this_element].regnum == regnum) | |
| 3559 return true; | |
| 3560 | |
| 3561 return false; | |
| 3562 } | |
| 3563 | |
| 3564 | |
| 3565 /* Read the ending character of a range (in a bracket expression) from the | |
| 3566 uncompiled pattern *P_PTR (which ends at PEND). We assume the | |
| 3567 starting character is in `P[-2]'. (`P[-1]' is the character `-'.) | |
| 3568 Then we set the translation of all bits between the starting and | |
| 3569 ending characters (inclusive) in the compiled pattern B. | |
| 3570 | |
| 3571 Return an error code. | |
| 3572 | |
| 3573 We use these short variable names so we can use the same macros as | |
| 826 | 3574 `regex_compile' itself. |
| 3575 | |
| 3576 Under Mule, this is only called when both chars of the range are | |
| 3577 ASCII. */ | |
| 428 | 3578 |
| 3579 static reg_errcode_t | |
| 446 | 3580 compile_range (re_char **p_ptr, re_char *pend, RE_TRANSLATE_TYPE translate, |
| 3581 reg_syntax_t syntax, unsigned char *buf_end) | |
| 428 | 3582 { |
| 867 | 3583 Ichar this_char; |
| 428 | 3584 |
| 446 | 3585 re_char *p = *p_ptr; |
| 428 | 3586 int range_start, range_end; |
| 3587 | |
| 3588 if (p == pend) | |
| 3589 return REG_ERANGE; | |
| 3590 | |
| 3591 /* Even though the pattern is a signed `char *', we need to fetch | |
| 3592 with unsigned char *'s; if the high bit of the pattern character | |
| 3593 is set, the range endpoints will be negative if we fetch using a | |
| 3594 signed char *. | |
| 3595 | |
| 3596 We also want to fetch the endpoints without translating them; the | |
| 3597 appropriate translation is done in the bit-setting loop below. */ | |
| 442 | 3598 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */ |
| 3599 range_start = ((const unsigned char *) p)[-2]; | |
| 3600 range_end = ((const unsigned char *) p)[0]; | |
| 428 | 3601 |
| 3602 /* Have to increment the pointer into the pattern string, so the | |
| 3603 caller isn't still at the ending character. */ | |
| 3604 (*p_ptr)++; | |
| 3605 | |
| 3606 /* If the start is after the end, the range is empty. */ | |
| 3607 if (range_start > range_end) | |
| 3608 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; | |
| 3609 | |
| 3610 /* Here we see why `this_char' has to be larger than an `unsigned | |
| 3611 char' -- the range is inclusive, so if `range_end' == 0xff | |
| 3612 (assuming 8-bit characters), we would otherwise go into an infinite | |
| 3613 loop, since all characters <= 0xff. */ | |
| 3614 for (this_char = range_start; this_char <= range_end; this_char++) | |
| 3615 { | |
| 826 | 3616 SET_LIST_BIT (RE_TRANSLATE (this_char)); |
| 428 | 3617 } |
| 3618 | |
| 3619 return REG_NOERROR; | |
| 3620 } | |
| 3621 | |
| 3622 #ifdef MULE | |
| 3623 | |
| 3624 static reg_errcode_t | |
| 446 | 3625 compile_extended_range (re_char **p_ptr, re_char *pend, |
| 3626 RE_TRANSLATE_TYPE translate, | |
| 428 | 3627 reg_syntax_t syntax, Lisp_Object rtab) |
| 3628 { | |
| 867 | 3629 Ichar this_char, range_start, range_end; |
| 3630 const Ibyte *p; | |
| 428 | 3631 |
| 3632 if (*p_ptr == pend) | |
| 3633 return REG_ERANGE; | |
| 3634 | |
| 867 | 3635 p = (const Ibyte *) *p_ptr; |
| 3636 range_end = itext_ichar (p); | |
| 428 | 3637 p--; /* back to '-' */ |
| 867 | 3638 DEC_IBYTEPTR (p); /* back to start of range */ |
| 428 | 3639 /* We also want to fetch the endpoints without translating them; the |
| 3640 appropriate translation is done in the bit-setting loop below. */ | |
| 867 | 3641 range_start = itext_ichar (p); |
| 3642 INC_IBYTEPTR (*p_ptr); | |
| 428 | 3643 |
| 3644 /* If the start is after the end, the range is empty. */ | |
| 3645 if (range_start > range_end) | |
| 3646 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; | |
| 3647 | |
| 3648 /* Can't have ranges spanning different charsets, except maybe for | |
| 3649 ranges entirely within the first 256 chars. */ | |
| 3650 | |
| 3651 if ((range_start >= 0x100 || range_end >= 0x100) | |
| 867 | 3652 && ichar_leading_byte (range_start) != |
| 3653 ichar_leading_byte (range_end)) | |
| 428 | 3654 return REG_ERANGESPAN; |
| 3655 | |
| 826 | 3656 /* #### This might be way inefficient if the range encompasses 10,000 |
| 3657 chars or something. To be efficient, you'd have to do something like | |
| 3658 this: | |
| 428 | 3659 |
| 3660 range_table a; | |
| 3661 range_table b; | |
| 3662 map over translation table in [range_start, range_end] of | |
| 3663 (put the mapped range in a; | |
| 3664 put the translation in b) | |
| 3665 invert the range in a and truncate to [range_start, range_end] | |
| 3666 compute the union of a, b | |
| 3667 union the result into rtab | |
| 3668 */ | |
| 826 | 3669 for (this_char = range_start; this_char <= range_end; this_char++) |
| 428 | 3670 { |
| 826 | 3671 SET_RANGETAB_BIT (RE_TRANSLATE (this_char)); |
| 428 | 3672 } |
| 3673 | |
| 3674 if (this_char <= range_end) | |
| 3675 put_range_table (rtab, this_char, range_end, Qt); | |
| 3676 | |
| 3677 return REG_NOERROR; | |
| 3678 } | |
| 3679 | |
| 3680 #endif /* MULE */ | |
| 3681 | |
| 3682 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in | |
| 3683 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible | |
| 3684 characters can start a string that matches the pattern. This fastmap | |
| 3685 is used by re_search to skip quickly over impossible starting points. | |
| 3686 | |
| 3687 The caller must supply the address of a (1 << BYTEWIDTH)-byte data | |
| 3688 area as BUFP->fastmap. | |
| 3689 | |
| 3690 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in | |
| 3691 the pattern buffer. | |
| 3692 | |
| 3693 Returns 0 if we succeed, -2 if an internal error. */ | |
| 3694 | |
| 3695 int | |
| 826 | 3696 re_compile_fastmap (struct re_pattern_buffer *bufp |
| 3697 RE_LISP_SHORT_CONTEXT_ARGS_DECL) | |
| 428 | 3698 { |
| 3699 int j, k; | |
| 3700 #ifdef MATCH_MAY_ALLOCATE | |
| 3701 fail_stack_type fail_stack; | |
| 3702 #endif | |
| 456 | 3703 DECLARE_DESTINATION; |
| 428 | 3704 /* We don't push any register information onto the failure stack. */ |
| 3705 | |
| 826 | 3706 /* &&#### this should be changed for 8-bit-fixed, for efficiency. see |
| 3707 comment marked with &&#### in re_search_2. */ | |
| 3708 | |
| 428 | 3709 REGISTER char *fastmap = bufp->fastmap; |
| 3710 unsigned char *pattern = bufp->buffer; | |
| 647 | 3711 long size = bufp->used; |
| 428 | 3712 unsigned char *p = pattern; |
| 3713 REGISTER unsigned char *pend = pattern + size; | |
| 3714 | |
| 771 | 3715 #ifdef REGEX_REL_ALLOC |
| 428 | 3716 /* This holds the pointer to the failure stack, when |
| 3717 it is allocated relocatably. */ | |
| 3718 fail_stack_elt_t *failure_stack_ptr; | |
| 3719 #endif | |
| 3720 | |
| 3721 /* Assume that each path through the pattern can be null until | |
| 3722 proven otherwise. We set this false at the bottom of switch | |
| 3723 statement, to which we get only if a particular path doesn't | |
| 3724 match the empty string. */ | |
| 460 | 3725 re_bool path_can_be_null = true; |
| 428 | 3726 |
| 3727 /* We aren't doing a `succeed_n' to begin with. */ | |
| 460 | 3728 re_bool succeed_n_p = false; |
| 428 | 3729 |
| 1333 | 3730 #ifdef ERROR_CHECK_MALLOC |
| 3731 /* The pattern comes from string data, not buffer data. We don't access | |
| 3732 any buffer data, so we don't have to worry about malloc() (but the | |
| 3733 disallowed flag may have been set by a caller). */ | |
| 3734 int depth = bind_regex_malloc_disallowed (0); | |
| 3735 #endif | |
| 3736 | |
| 428 | 3737 assert (fastmap != NULL && p != NULL); |
| 3738 | |
| 3739 INIT_FAIL_STACK (); | |
| 3740 memset (fastmap, 0, 1 << BYTEWIDTH); /* Assume nothing's valid. */ | |
| 3741 bufp->fastmap_accurate = 1; /* It will be when we're done. */ | |
| 3742 bufp->can_be_null = 0; | |
| 3743 | |
| 3744 while (1) | |
| 3745 { | |
| 3746 if (p == pend || *p == succeed) | |
| 3747 { | |
| 3748 /* We have reached the (effective) end of pattern. */ | |
| 3749 if (!FAIL_STACK_EMPTY ()) | |
| 3750 { | |
| 3751 bufp->can_be_null |= path_can_be_null; | |
| 3752 | |
| 3753 /* Reset for next path. */ | |
| 3754 path_can_be_null = true; | |
| 3755 | |
| 446 | 3756 p = (unsigned char *) fail_stack.stack[--fail_stack.avail].pointer; |
| 428 | 3757 |
| 3758 continue; | |
| 3759 } | |
| 3760 else | |
| 3761 break; | |
| 3762 } | |
| 3763 | |
| 3764 /* We should never be about to go beyond the end of the pattern. */ | |
| 3765 assert (p < pend); | |
| 3766 | |
| 3767 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) | |
| 3768 { | |
| 3769 | |
| 3770 /* I guess the idea here is to simply not bother with a fastmap | |
| 3771 if a backreference is used, since it's too hard to figure out | |
| 3772 the fastmap for the corresponding group. Setting | |
| 3773 `can_be_null' stops `re_search_2' from using the fastmap, so | |
| 3774 that is all we do. */ | |
| 3775 case duplicate: | |
| 3776 bufp->can_be_null = 1; | |
| 3777 goto done; | |
| 3778 | |
| 3779 | |
| 3780 /* Following are the cases which match a character. These end | |
| 3781 with `break'. */ | |
| 3782 | |
| 3783 case exactn: | |
| 3784 fastmap[p[1]] = 1; | |
| 3785 break; | |
| 3786 | |
| 3787 | |
| 3788 case charset: | |
| 3789 /* XEmacs: Under Mule, these bit vectors will | |
| 3790 only contain values for characters below 0x80. */ | |
| 3791 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
| 3792 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) | |
| 3793 fastmap[j] = 1; | |
| 3794 break; | |
| 3795 | |
| 3796 | |
| 3797 case charset_not: | |
| 3798 /* Chars beyond end of map must be allowed. */ | |
| 3799 #ifdef MULE | |
| 3800 for (j = *p * BYTEWIDTH; j < 0x80; j++) | |
| 3801 fastmap[j] = 1; | |
| 3802 /* And all extended characters must be allowed, too. */ | |
| 3803 for (j = 0x80; j < 0xA0; j++) | |
| 3804 fastmap[j] = 1; | |
| 446 | 3805 #else /* not MULE */ |
| 428 | 3806 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) |
| 3807 fastmap[j] = 1; | |
| 446 | 3808 #endif /* MULE */ |
| 428 | 3809 |
| 3810 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
| 3811 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) | |
| 3812 fastmap[j] = 1; | |
| 3813 break; | |
| 3814 | |
| 3815 #ifdef MULE | |
| 3816 case charset_mule: | |
| 3817 { | |
| 3818 int nentries; | |
| 3819 int i; | |
| 3820 | |
| 3821 nentries = unified_range_table_nentries (p); | |
| 3822 for (i = 0; i < nentries; i++) | |
| 3823 { | |
| 3824 EMACS_INT first, last; | |
| 3825 Lisp_Object dummy_val; | |
| 3826 int jj; | |
| 867 | 3827 Ibyte strr[MAX_ICHAR_LEN]; |
| 428 | 3828 |
| 3829 unified_range_table_get_range (p, i, &first, &last, | |
| 3830 &dummy_val); | |
| 3831 for (jj = first; jj <= last && jj < 0x80; jj++) | |
| 3832 fastmap[jj] = 1; | |
| 3833 /* Ranges below 0x100 can span charsets, but there | |
| 3834 are only two (Control-1 and Latin-1), and | |
| 3835 either first or last has to be in them. */ | |
| 867 | 3836 set_itext_ichar (strr, first); |
| 428 | 3837 fastmap[*strr] = 1; |
| 3838 if (last < 0x100) | |
| 3839 { | |
| 867 | 3840 set_itext_ichar (strr, last); |
| 428 | 3841 fastmap[*strr] = 1; |
| 3842 } | |
| 3843 } | |
| 3844 } | |
| 3845 break; | |
| 3846 | |
| 3847 case charset_mule_not: | |
| 3848 { | |
| 3849 int nentries; | |
| 3850 int i; | |
| 3851 | |
| 3852 nentries = unified_range_table_nentries (p); | |
| 3853 for (i = 0; i < nentries; i++) | |
| 3854 { | |
| 3855 EMACS_INT first, last; | |
| 3856 Lisp_Object dummy_val; | |
| 3857 int jj; | |
| 3858 int smallest_prev = 0; | |
| 3859 | |
| 3860 unified_range_table_get_range (p, i, &first, &last, | |
| 3861 &dummy_val); | |
| 3862 for (jj = smallest_prev; jj < first && jj < 0x80; jj++) | |
| 3863 fastmap[jj] = 1; | |
| 3864 smallest_prev = last + 1; | |
| 3865 if (smallest_prev >= 0x80) | |
| 3866 break; | |
| 3867 } | |
| 3868 /* Calculating which leading bytes are actually allowed | |
| 3869 here is rather difficult, so we just punt and allow | |
| 3870 all of them. */ | |
| 3871 for (i = 0x80; i < 0xA0; i++) | |
| 3872 fastmap[i] = 1; | |
| 3873 } | |
| 3874 break; | |
| 3875 #endif /* MULE */ | |
| 3876 | |
| 3877 | |
| 3878 case anychar: | |
| 3879 { | |
| 3880 int fastmap_newline = fastmap['\n']; | |
| 3881 | |
| 3882 /* `.' matches anything ... */ | |
| 3883 #ifdef MULE | |
| 3884 /* "anything" only includes bytes that can be the | |
| 3885 first byte of a character. */ | |
| 3886 for (j = 0; j < 0xA0; j++) | |
| 3887 fastmap[j] = 1; | |
| 3888 #else | |
| 3889 for (j = 0; j < (1 << BYTEWIDTH); j++) | |
| 3890 fastmap[j] = 1; | |
| 3891 #endif | |
| 3892 | |
| 3893 /* ... except perhaps newline. */ | |
| 3894 if (!(bufp->syntax & RE_DOT_NEWLINE)) | |
| 3895 fastmap['\n'] = fastmap_newline; | |
| 3896 | |
| 3897 /* Return if we have already set `can_be_null'; if we have, | |
| 3898 then the fastmap is irrelevant. Something's wrong here. */ | |
| 3899 else if (bufp->can_be_null) | |
| 3900 goto done; | |
| 3901 | |
| 3902 /* Otherwise, have to check alternative paths. */ | |
| 3903 break; | |
| 3904 } | |
| 3905 | |
| 826 | 3906 #ifndef emacs |
| 3907 case wordchar: | |
| 3908 for (j = 0; j < (1 << BYTEWIDTH); j++) | |
| 3909 if (SYNTAX (ignored, j) == Sword) | |
| 3910 fastmap[j] = 1; | |
| 3911 break; | |
| 3912 | |
| 3913 case notwordchar: | |
| 3914 for (j = 0; j < (1 << BYTEWIDTH); j++) | |
| 3915 if (SYNTAX (ignored, j) != Sword) | |
| 3916 fastmap[j] = 1; | |
| 3917 break; | |
| 3918 #else /* emacs */ | |
| 3919 case wordchar: | |
| 3920 case notwordchar: | |
| 460 | 3921 case wordbound: |
| 3922 case notwordbound: | |
| 3923 case wordbeg: | |
| 3924 case wordend: | |
| 3925 case notsyntaxspec: | |
| 3926 case syntaxspec: | |
| 3927 /* This match depends on text properties. These end with | |
| 3928 aborting optimizations. */ | |
| 3929 bufp->can_be_null = 1; | |
| 3930 goto done; | |
| 826 | 3931 #if 0 /* all of the following code is unused now that the `syntax-table' |
| 3932 property exists -- it's trickier to do this than just look in | |
| 3933 the buffer. &&#### but we could just use the syntax-cache stuff | |
| 3934 instead; why don't we? --ben */ | |
| 3935 case wordchar: | |
| 3936 k = (int) Sword; | |
| 3937 goto matchsyntax; | |
| 3938 | |
| 3939 case notwordchar: | |
| 3940 k = (int) Sword; | |
| 3941 goto matchnotsyntax; | |
| 3942 | |
| 428 | 3943 case syntaxspec: |
| 3944 k = *p++; | |
| 826 | 3945 matchsyntax: |
| 428 | 3946 #ifdef MULE |
| 3947 for (j = 0; j < 0x80; j++) | |
| 826 | 3948 if (SYNTAX |
| 3949 (XCHAR_TABLE (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) == | |
| 428 | 3950 (enum syntaxcode) k) |
| 3951 fastmap[j] = 1; | |
| 3952 for (j = 0x80; j < 0xA0; j++) | |
| 3953 { | |
| 826 | 3954 if (leading_byte_prefix_p ((unsigned char) j)) |
| 428 | 3955 /* too complicated to calculate this right */ |
| 3956 fastmap[j] = 1; | |
| 3957 else | |
| 3958 { | |
| 3959 int multi_p; | |
| 3960 Lisp_Object cset; | |
| 3961 | |
| 826 | 3962 cset = charset_by_leading_byte (j); |
| 428 | 3963 if (CHARSETP (cset)) |
| 3964 { | |
| 826 | 3965 if (charset_syntax (lispbuf, cset, &multi_p) |
| 428 | 3966 == Sword || multi_p) |
| 3967 fastmap[j] = 1; | |
| 3968 } | |
| 3969 } | |
| 3970 } | |
| 446 | 3971 #else /* not MULE */ |
| 428 | 3972 for (j = 0; j < (1 << BYTEWIDTH); j++) |
| 826 | 3973 if (SYNTAX |
| 3974 (XCHAR_TABLE (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) == | |
| 428 | 3975 (enum syntaxcode) k) |
| 3976 fastmap[j] = 1; | |
| 446 | 3977 #endif /* MULE */ |
| 428 | 3978 break; |
| 3979 | |
| 3980 | |
| 3981 case notsyntaxspec: | |
| 3982 k = *p++; | |
| 826 | 3983 matchnotsyntax: |
| 428 | 3984 #ifdef MULE |
| 3985 for (j = 0; j < 0x80; j++) | |
| 826 | 3986 if (SYNTAX |
| 428 | 3987 (XCHAR_TABLE |
| 826 | 3988 (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) != |
| 428 | 3989 (enum syntaxcode) k) |
| 3990 fastmap[j] = 1; | |
| 3991 for (j = 0x80; j < 0xA0; j++) | |
| 3992 { | |
| 826 | 3993 if (leading_byte_prefix_p ((unsigned char) j)) |
| 428 | 3994 /* too complicated to calculate this right */ |
| 3995 fastmap[j] = 1; | |
| 3996 else | |
| 3997 { | |
| 3998 int multi_p; | |
| 3999 Lisp_Object cset; | |
| 4000 | |
| 826 | 4001 cset = charset_by_leading_byte (j); |
| 428 | 4002 if (CHARSETP (cset)) |
| 4003 { | |
| 826 | 4004 if (charset_syntax (lispbuf, cset, &multi_p) |
| 428 | 4005 != Sword || multi_p) |
| 4006 fastmap[j] = 1; | |
| 4007 } | |
| 4008 } | |
| 4009 } | |
| 446 | 4010 #else /* not MULE */ |
| 428 | 4011 for (j = 0; j < (1 << BYTEWIDTH); j++) |
| 826 | 4012 if (SYNTAX |
| 428 | 4013 (XCHAR_TABLE |
| 826 | 4014 (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) != |
| 428 | 4015 (enum syntaxcode) k) |
| 4016 fastmap[j] = 1; | |
| 446 | 4017 #endif /* MULE */ |
| 428 | 4018 break; |
| 826 | 4019 #endif /* 0 */ |
| 428 | 4020 |
| 4021 #ifdef MULE | |
| 4022 /* 97/2/17 jhod category patch */ | |
| 4023 case categoryspec: | |
| 4024 case notcategoryspec: | |
| 4025 bufp->can_be_null = 1; | |
| 1333 | 4026 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4027 return 0; |
| 4028 /* end if category patch */ | |
| 4029 #endif /* MULE */ | |
| 4030 | |
| 4031 /* All cases after this match the empty string. These end with | |
| 4032 `continue'. */ | |
| 4033 case before_dot: | |
| 4034 case at_dot: | |
| 4035 case after_dot: | |
| 4036 continue; | |
| 826 | 4037 #endif /* emacs */ |
| 428 | 4038 |
| 4039 | |
| 4040 case no_op: | |
| 4041 case begline: | |
| 4042 case endline: | |
| 4043 case begbuf: | |
| 4044 case endbuf: | |
| 460 | 4045 #ifndef emacs |
| 428 | 4046 case wordbound: |
| 4047 case notwordbound: | |
| 4048 case wordbeg: | |
| 4049 case wordend: | |
| 460 | 4050 #endif |
| 428 | 4051 case push_dummy_failure: |
| 4052 continue; | |
| 4053 | |
| 4054 | |
| 4055 case jump_n: | |
| 4056 case pop_failure_jump: | |
| 4057 case maybe_pop_jump: | |
| 4058 case jump: | |
| 4059 case jump_past_alt: | |
| 4060 case dummy_failure_jump: | |
| 4061 EXTRACT_NUMBER_AND_INCR (j, p); | |
| 4062 p += j; | |
| 4063 if (j > 0) | |
| 4064 continue; | |
| 4065 | |
| 4066 /* Jump backward implies we just went through the body of a | |
| 4067 loop and matched nothing. Opcode jumped to should be | |
| 4068 `on_failure_jump' or `succeed_n'. Just treat it like an | |
| 4069 ordinary jump. For a * loop, it has pushed its failure | |
| 4070 point already; if so, discard that as redundant. */ | |
| 4071 if ((re_opcode_t) *p != on_failure_jump | |
| 4072 && (re_opcode_t) *p != succeed_n) | |
| 4073 continue; | |
| 4074 | |
| 4075 p++; | |
| 4076 EXTRACT_NUMBER_AND_INCR (j, p); | |
| 4077 p += j; | |
| 4078 | |
| 4079 /* If what's on the stack is where we are now, pop it. */ | |
| 4080 if (!FAIL_STACK_EMPTY () | |
| 4081 && fail_stack.stack[fail_stack.avail - 1].pointer == p) | |
| 4082 fail_stack.avail--; | |
| 4083 | |
| 4084 continue; | |
| 4085 | |
| 4086 | |
| 4087 case on_failure_jump: | |
| 4088 case on_failure_keep_string_jump: | |
| 4089 handle_on_failure_jump: | |
| 4090 EXTRACT_NUMBER_AND_INCR (j, p); | |
| 4091 | |
| 4092 /* For some patterns, e.g., `(a?)?', `p+j' here points to the | |
| 4093 end of the pattern. We don't want to push such a point, | |
| 4094 since when we restore it above, entering the switch will | |
| 4095 increment `p' past the end of the pattern. We don't need | |
| 4096 to push such a point since we obviously won't find any more | |
| 4097 fastmap entries beyond `pend'. Such a pattern can match | |
| 4098 the null string, though. */ | |
| 4099 if (p + j < pend) | |
| 4100 { | |
| 4101 if (!PUSH_PATTERN_OP (p + j, fail_stack)) | |
| 4102 { | |
| 4103 RESET_FAIL_STACK (); | |
| 1333 | 4104 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4105 return -2; |
| 4106 } | |
| 4107 } | |
| 4108 else | |
| 4109 bufp->can_be_null = 1; | |
| 4110 | |
| 4111 if (succeed_n_p) | |
| 4112 { | |
| 4113 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ | |
| 4114 succeed_n_p = false; | |
| 4115 } | |
| 4116 | |
| 4117 continue; | |
| 4118 | |
| 4119 | |
| 4120 case succeed_n: | |
| 4121 /* Get to the number of times to succeed. */ | |
| 4122 p += 2; | |
| 4123 | |
| 4124 /* Increment p past the n for when k != 0. */ | |
| 4125 EXTRACT_NUMBER_AND_INCR (k, p); | |
| 4126 if (k == 0) | |
| 4127 { | |
| 4128 p -= 4; | |
| 4129 succeed_n_p = true; /* Spaghetti code alert. */ | |
| 4130 goto handle_on_failure_jump; | |
| 4131 } | |
| 4132 continue; | |
| 4133 | |
| 4134 | |
| 4135 case set_number_at: | |
| 4136 p += 4; | |
| 4137 continue; | |
| 4138 | |
| 4139 | |
| 4140 case start_memory: | |
| 4141 case stop_memory: | |
| 4142 p += 2; | |
| 4143 continue; | |
| 4144 | |
| 4145 | |
| 4146 default: | |
| 2500 | 4147 ABORT (); /* We have listed all the cases. */ |
| 428 | 4148 } /* switch *p++ */ |
| 4149 | |
| 4150 /* Getting here means we have found the possible starting | |
| 4151 characters for one path of the pattern -- and that the empty | |
| 4152 string does not match. We need not follow this path further. | |
| 4153 Instead, look at the next alternative (remembered on the | |
| 4154 stack), or quit if no more. The test at the top of the loop | |
| 4155 does these things. */ | |
| 4156 path_can_be_null = false; | |
| 4157 p = pend; | |
| 4158 } /* while p */ | |
| 4159 | |
| 4160 /* Set `can_be_null' for the last path (also the first path, if the | |
| 4161 pattern is empty). */ | |
| 4162 bufp->can_be_null |= path_can_be_null; | |
| 4163 | |
| 4164 done: | |
| 4165 RESET_FAIL_STACK (); | |
| 1333 | 4166 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4167 return 0; |
| 4168 } /* re_compile_fastmap */ | |
| 4169 | |
| 4170 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and | |
| 4171 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use | |
| 4172 this memory for recording register information. STARTS and ENDS | |
| 4173 must be allocated using the malloc library routine, and must each | |
| 4174 be at least NUM_REGS * sizeof (regoff_t) bytes long. | |
| 4175 | |
| 4176 If NUM_REGS == 0, then subsequent matches should allocate their own | |
| 4177 register data. | |
| 4178 | |
| 4179 Unless this function is called, the first search or match using | |
| 4180 PATTERN_BUFFER will allocate its own register data, without | |
| 4181 freeing the old data. */ | |
| 4182 | |
| 4183 void | |
| 4184 re_set_registers (struct re_pattern_buffer *bufp, struct re_registers *regs, | |
| 647 | 4185 int num_regs, regoff_t *starts, regoff_t *ends) |
| 428 | 4186 { |
| 4187 if (num_regs) | |
| 4188 { | |
| 4189 bufp->regs_allocated = REGS_REALLOCATE; | |
| 4190 regs->num_regs = num_regs; | |
| 4191 regs->start = starts; | |
| 4192 regs->end = ends; | |
| 4193 } | |
| 4194 else | |
| 4195 { | |
| 4196 bufp->regs_allocated = REGS_UNALLOCATED; | |
| 4197 regs->num_regs = 0; | |
| 4198 regs->start = regs->end = (regoff_t *) 0; | |
| 4199 } | |
| 4200 } | |
| 4201 | |
| 4202 /* Searching routines. */ | |
| 4203 | |
| 4204 /* Like re_search_2, below, but only one string is specified, and | |
| 4205 doesn't let you say where to stop matching. */ | |
| 4206 | |
| 4207 int | |
| 442 | 4208 re_search (struct re_pattern_buffer *bufp, const char *string, int size, |
| 826 | 4209 int startpos, int range, struct re_registers *regs |
| 4210 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4211 { |
| 4212 return re_search_2 (bufp, NULL, 0, string, size, startpos, range, | |
| 826 | 4213 regs, size RE_LISP_CONTEXT_ARGS); |
| 428 | 4214 } |
| 4215 | |
| 4216 /* Using the compiled pattern in BUFP->buffer, first tries to match the | |
| 4217 virtual concatenation of STRING1 and STRING2, starting first at index | |
| 4218 STARTPOS, then at STARTPOS + 1, and so on. | |
| 4219 | |
| 4220 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. | |
| 4221 | |
| 4222 RANGE is how far to scan while trying to match. RANGE = 0 means try | |
| 4223 only at STARTPOS; in general, the last start tried is STARTPOS + | |
| 4224 RANGE. | |
| 4225 | |
| 826 | 4226 All sizes and positions refer to bytes (not chars); under Mule, the code |
| 4227 knows about the format of the text and will only check at positions | |
| 4228 where a character starts. | |
| 4229 | |
| 428 | 4230 With MULE, RANGE is a byte position, not a char position. The last |
| 4231 start tried is the character starting <= STARTPOS + RANGE. | |
| 4232 | |
| 4233 In REGS, return the indices of the virtual concatenation of STRING1 | |
| 4234 and STRING2 that matched the entire BUFP->buffer and its contained | |
| 4235 subexpressions. | |
| 4236 | |
| 4237 Do not consider matching one past the index STOP in the virtual | |
| 4238 concatenation of STRING1 and STRING2. | |
| 4239 | |
| 4240 We return either the position in the strings at which the match was | |
| 4241 found, -1 if no match, or -2 if error (such as failure | |
| 4242 stack overflow). */ | |
| 4243 | |
| 4244 int | |
| 446 | 4245 re_search_2 (struct re_pattern_buffer *bufp, const char *str1, |
| 4246 int size1, const char *str2, int size2, int startpos, | |
| 826 | 4247 int range, struct re_registers *regs, int stop |
| 4248 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4249 { |
| 4250 int val; | |
| 446 | 4251 re_char *string1 = (re_char *) str1; |
| 4252 re_char *string2 = (re_char *) str2; | |
| 428 | 4253 REGISTER char *fastmap = bufp->fastmap; |
| 446 | 4254 REGISTER RE_TRANSLATE_TYPE translate = bufp->translate; |
| 428 | 4255 int total_size = size1 + size2; |
| 4256 int endpos = startpos + range; | |
| 4257 #ifdef REGEX_BEGLINE_CHECK | |
| 4258 int anchored_at_begline = 0; | |
| 4259 #endif | |
| 446 | 4260 re_char *d; |
| 826 | 4261 #ifdef emacs |
| 4262 Internal_Format fmt = buffer_or_other_internal_format (lispobj); | |
| 1346 | 4263 #ifdef REL_ALLOC |
| 4264 Ibyte *orig_buftext = | |
| 4265 BUFFERP (lispobj) ? | |
| 4266 BYTE_BUF_BYTE_ADDRESS (XBUFFER (lispobj), | |
| 4267 BYTE_BUF_BEGV (XBUFFER (lispobj))) : | |
| 4268 0; | |
| 4269 #endif | |
| 1333 | 4270 #ifdef ERROR_CHECK_MALLOC |
| 4271 int depth; | |
| 4272 #endif | |
| 826 | 4273 #endif /* emacs */ |
| 4274 #if 1 | |
| 4275 int forward_search_p; | |
| 4276 #endif | |
| 428 | 4277 |
| 4278 /* Check for out-of-range STARTPOS. */ | |
| 4279 if (startpos < 0 || startpos > total_size) | |
| 4280 return -1; | |
| 4281 | |
| 4282 /* Fix up RANGE if it might eventually take us outside | |
| 4283 the virtual concatenation of STRING1 and STRING2. */ | |
| 4284 if (endpos < 0) | |
| 4285 range = 0 - startpos; | |
| 4286 else if (endpos > total_size) | |
| 4287 range = total_size - startpos; | |
| 4288 | |
| 826 | 4289 #if 1 |
| 4290 forward_search_p = range > 0; | |
| 4291 #endif | |
| 4292 | |
| 428 | 4293 /* If the search isn't to be a backwards one, don't waste time in a |
| 4294 search for a pattern that must be anchored. */ | |
| 4295 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0) | |
| 4296 { | |
| 4297 if (startpos > 0) | |
| 4298 return -1; | |
| 4299 else | |
| 4300 { | |
| 442 | 4301 d = ((const unsigned char *) |
| 428 | 4302 (startpos >= size1 ? string2 - size1 : string1) + startpos); |
| 867 | 4303 range = itext_ichar_len_fmt (d, fmt); |
| 428 | 4304 } |
| 4305 } | |
| 4306 | |
| 460 | 4307 #ifdef emacs |
| 4308 /* In a forward search for something that starts with \=. | |
| 4309 don't keep searching past point. */ | |
| 4310 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0) | |
| 4311 { | |
| 826 | 4312 if (!BUFFERP (lispobj)) |
| 4313 return -1; | |
|
4527
8418d1ad4944
Fix at_dot regex under Mule. <87hc6rv53v.fsf@uwakimon.sk.tsukuba.ac.jp>
Stephen J. Turnbull <stephen@xemacs.org>
parents:
3300
diff
changeset
|
4314 range = (BYTE_BUF_PT (XBUFFER (lispobj)) |
|
8418d1ad4944
Fix at_dot regex under Mule. <87hc6rv53v.fsf@uwakimon.sk.tsukuba.ac.jp>
Stephen J. Turnbull <stephen@xemacs.org>
parents:
3300
diff
changeset
|
4315 - BYTE_BUF_BEGV (XBUFFER (lispobj)) - startpos); |
| 460 | 4316 if (range < 0) |
| 4317 return -1; | |
| 4318 } | |
| 4319 #endif /* emacs */ | |
| 4320 | |
| 1333 | 4321 #ifdef ERROR_CHECK_MALLOC |
| 4322 /* Do this after the above return()s. */ | |
| 4323 depth = bind_regex_malloc_disallowed (1); | |
| 4324 #endif | |
| 4325 | |
| 428 | 4326 /* Update the fastmap now if not correct already. */ |
| 1333 | 4327 BEGIN_REGEX_MALLOC_OK (); |
| 428 | 4328 if (fastmap && !bufp->fastmap_accurate) |
| 826 | 4329 if (re_compile_fastmap (bufp RE_LISP_SHORT_CONTEXT_ARGS) == -2) |
| 1333 | 4330 { |
| 4331 END_REGEX_MALLOC_OK (); | |
| 4332 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4333 return -2; | |
| 4334 } | |
| 4335 | |
| 4336 END_REGEX_MALLOC_OK (); | |
| 4337 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 428 | 4338 |
| 4339 #ifdef REGEX_BEGLINE_CHECK | |
| 4340 { | |
| 647 | 4341 long i = 0; |
| 428 | 4342 |
| 4343 while (i < bufp->used) | |
| 4344 { | |
| 4345 if (bufp->buffer[i] == start_memory || | |
| 4346 bufp->buffer[i] == stop_memory) | |
| 4347 i += 2; | |
| 4348 else | |
| 4349 break; | |
| 4350 } | |
| 4351 anchored_at_begline = i < bufp->used && bufp->buffer[i] == begline; | |
| 4352 } | |
| 4353 #endif | |
| 4354 | |
| 460 | 4355 #ifdef emacs |
| 1333 | 4356 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 4357 scache = setup_syntax_cache (scache, lispobj, lispbuf, |
| 4358 offset_to_charxpos (lispobj, startpos), | |
| 4359 1); | |
| 1333 | 4360 END_REGEX_MALLOC_OK (); |
| 4361 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 460 | 4362 #endif |
| 4363 | |
| 428 | 4364 /* Loop through the string, looking for a place to start matching. */ |
| 4365 for (;;) | |
| 4366 { | |
| 4367 #ifdef REGEX_BEGLINE_CHECK | |
| 826 | 4368 /* If the regex is anchored at the beginning of a line (i.e. with a |
| 4369 ^), then we can speed things up by skipping to the next | |
| 4370 beginning-of-line. However, to determine "beginning of line" we | |
| 4371 need to look at the previous char, so can't do this check if at | |
| 4372 beginning of either string. (Well, we could if at the beginning of | |
| 4373 the second string, but it would require additional code, and this | |
| 4374 is just an optimization.) */ | |
| 4375 if (anchored_at_begline && startpos > 0 && startpos != size1) | |
| 428 | 4376 { |
| 826 | 4377 if (range > 0) |
| 4378 { | |
| 4379 /* whose stupid idea was it anyway to make this | |
| 4380 function take two strings to match?? */ | |
| 4381 int lim = 0; | |
| 4382 re_char *orig_d; | |
| 4383 re_char *stop_d; | |
| 4384 | |
| 4385 /* Compute limit as below in fastmap code, so we are guaranteed | |
| 4386 to remain within a single string. */ | |
| 4387 if (startpos < size1 && startpos + range >= size1) | |
| 4388 lim = range - (size1 - startpos); | |
| 4389 | |
| 4390 d = ((const unsigned char *) | |
| 4391 (startpos >= size1 ? string2 - size1 : string1) + startpos); | |
| 4392 orig_d = d; | |
| 4393 stop_d = d + range - lim; | |
| 4394 | |
| 4395 /* We want to find the next location (including the current | |
| 4396 one) where the previous char is a newline, so back up one | |
| 4397 and search forward for a newline. */ | |
| 867 | 4398 DEC_IBYTEPTR_FMT (d, fmt); /* Ok, since startpos != size1. */ |
| 826 | 4399 |
| 4400 /* Written out as an if-else to avoid testing `translate' | |
| 4401 inside the loop. */ | |
| 4402 if (TRANSLATE_P (translate)) | |
| 4403 while (d < stop_d && | |
| 867 | 4404 RE_TRANSLATE_1 (itext_ichar_fmt (d, fmt, lispobj)) |
| 826 | 4405 != '\n') |
| 867 | 4406 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4407 else |
| 4408 while (d < stop_d && | |
| 867 | 4409 itext_ichar_ascii_fmt (d, fmt, lispobj) != '\n') |
| 4410 INC_IBYTEPTR_FMT (d, fmt); | |
| 826 | 4411 |
| 4412 /* If we were stopped by a newline, skip forward over it. | |
| 4413 Otherwise we will get in an infloop when our start position | |
| 4414 was at begline. */ | |
| 4415 if (d < stop_d) | |
| 867 | 4416 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4417 range -= d - orig_d; |
| 4418 startpos += d - orig_d; | |
| 4419 #if 1 | |
| 4420 assert (!forward_search_p || range >= 0); | |
| 4421 #endif | |
| 4422 } | |
| 4423 else if (range < 0) | |
| 4424 { | |
| 4425 /* We're lazy, like in the fastmap code below */ | |
| 867 | 4426 Ichar c; |
| 826 | 4427 |
| 4428 d = ((const unsigned char *) | |
| 4429 (startpos >= size1 ? string2 - size1 : string1) + startpos); | |
| 867 | 4430 DEC_IBYTEPTR_FMT (d, fmt); |
| 4431 c = itext_ichar_fmt (d, fmt, lispobj); | |
| 826 | 4432 c = RE_TRANSLATE (c); |
| 4433 if (c != '\n') | |
| 4434 goto advance; | |
| 4435 } | |
| 428 | 4436 } |
| 4437 #endif /* REGEX_BEGLINE_CHECK */ | |
| 4438 | |
| 4439 /* If a fastmap is supplied, skip quickly over characters that | |
| 4440 cannot be the start of a match. If the pattern can match the | |
| 4441 null string, however, we don't need to skip characters; we want | |
| 4442 the first null string. */ | |
| 4443 if (fastmap && startpos < total_size && !bufp->can_be_null) | |
| 4444 { | |
| 826 | 4445 /* For the moment, fastmap always works as if buffer |
| 4446 is in default format, so convert chars in the search strings | |
| 4447 into default format as we go along, if necessary. | |
| 4448 | |
| 4449 &&#### fastmap needs rethinking for 8-bit-fixed so | |
| 4450 it's faster. We need it to reflect the raw | |
| 4451 8-bit-fixed values. That isn't so hard if we assume | |
| 4452 that the top 96 bytes represent a single 1-byte | |
| 4453 charset. For 16-bit/32-bit stuff it's probably not | |
| 4454 worth it to make the fastmap represent the raw, due to | |
| 4455 its nature -- we'd have to use the LSB for the | |
| 4456 fastmap, and that causes lots of problems with Mule | |
| 4457 chars, where it essentially wipes out the usefulness | |
| 4458 of the fastmap entirely. */ | |
| 428 | 4459 if (range > 0) /* Searching forwards. */ |
| 4460 { | |
| 4461 int lim = 0; | |
| 4462 int irange = range; | |
| 4463 | |
| 4464 if (startpos < size1 && startpos + range >= size1) | |
| 4465 lim = range - (size1 - startpos); | |
| 4466 | |
| 442 | 4467 d = ((const unsigned char *) |
| 428 | 4468 (startpos >= size1 ? string2 - size1 : string1) + startpos); |
| 4469 | |
| 4470 /* Written out as an if-else to avoid testing `translate' | |
| 4471 inside the loop. */ | |
| 446 | 4472 if (TRANSLATE_P (translate)) |
| 826 | 4473 { |
| 4474 while (range > lim) | |
| 4475 { | |
| 4476 re_char *old_d = d; | |
| 428 | 4477 #ifdef MULE |
| 867 | 4478 Ibyte tempch[MAX_ICHAR_LEN]; |
| 4479 Ichar buf_ch = | |
| 4480 RE_TRANSLATE_1 (itext_ichar_fmt (d, fmt, lispobj)); | |
| 4481 set_itext_ichar (tempch, buf_ch); | |
| 826 | 4482 if (fastmap[*tempch]) |
| 4483 break; | |
| 446 | 4484 #else |
| 826 | 4485 if (fastmap[(unsigned char) RE_TRANSLATE_1 (*d)]) |
| 4486 break; | |
| 446 | 4487 #endif /* MULE */ |
| 867 | 4488 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4489 range -= (d - old_d); |
| 4490 #if 1 | |
| 1333 | 4491 assert (!forward_search_p || range >= 0); |
| 826 | 4492 #endif |
| 4493 } | |
| 4494 } | |
| 4495 #ifdef MULE | |
| 4496 else if (fmt != FORMAT_DEFAULT) | |
| 4497 { | |
| 4498 while (range > lim) | |
| 4499 { | |
| 4500 re_char *old_d = d; | |
| 867 | 4501 Ibyte tempch[MAX_ICHAR_LEN]; |
| 4502 Ichar buf_ch = itext_ichar_fmt (d, fmt, lispobj); | |
| 4503 set_itext_ichar (tempch, buf_ch); | |
| 826 | 4504 if (fastmap[*tempch]) |
| 4505 break; | |
| 867 | 4506 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4507 range -= (d - old_d); |
| 4508 #if 1 | |
| 1333 | 4509 assert (!forward_search_p || range >= 0); |
| 826 | 4510 #endif |
| 4511 } | |
| 4512 } | |
| 4513 #endif /* MULE */ | |
| 428 | 4514 else |
| 826 | 4515 { |
| 4516 while (range > lim && !fastmap[*d]) | |
| 4517 { | |
| 4518 re_char *old_d = d; | |
| 867 | 4519 INC_IBYTEPTR (d); |
| 826 | 4520 range -= (d - old_d); |
| 4521 #if 1 | |
| 4522 assert (!forward_search_p || range >= 0); | |
| 4523 #endif | |
| 4524 } | |
| 4525 } | |
| 428 | 4526 |
| 4527 startpos += irange - range; | |
| 4528 } | |
| 4529 else /* Searching backwards. */ | |
| 4530 { | |
| 826 | 4531 /* #### It's not clear why we don't just write a loop, like |
| 4532 for the moving-forward case. Perhaps the writer got lazy, | |
| 4533 since backward searches aren't so common. */ | |
| 4534 d = ((const unsigned char *) | |
| 4535 (startpos >= size1 ? string2 - size1 : string1) + startpos); | |
| 428 | 4536 #ifdef MULE |
| 826 | 4537 { |
| 867 | 4538 Ibyte tempch[MAX_ICHAR_LEN]; |
| 4539 Ichar buf_ch = | |
| 4540 RE_TRANSLATE (itext_ichar_fmt (d, fmt, lispobj)); | |
| 4541 set_itext_ichar (tempch, buf_ch); | |
| 826 | 4542 if (!fastmap[*tempch]) |
| 4543 goto advance; | |
| 4544 } | |
| 428 | 4545 #else |
| 826 | 4546 if (!fastmap[(unsigned char) RE_TRANSLATE (*d)]) |
| 446 | 4547 goto advance; |
| 826 | 4548 #endif /* MULE */ |
| 428 | 4549 } |
| 4550 } | |
| 4551 | |
| 4552 /* If can't match the null string, and that's all we have left, fail. */ | |
| 4553 if (range >= 0 && startpos == total_size && fastmap | |
| 4554 && !bufp->can_be_null) | |
| 1333 | 4555 { |
| 4556 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4557 return -1; | |
| 4558 } | |
| 428 | 4559 |
| 4560 #ifdef emacs /* XEmacs added, w/removal of immediate_quit */ | |
| 4561 if (!no_quit_in_re_search) | |
| 1333 | 4562 { |
| 4563 BEGIN_REGEX_MALLOC_OK (); | |
| 4564 QUIT; | |
| 4565 END_REGEX_MALLOC_OK (); | |
| 4566 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 4567 } | |
| 4568 | |
| 428 | 4569 #endif |
| 1333 | 4570 BEGIN_REGEX_MALLOC_OK (); |
| 428 | 4571 val = re_match_2_internal (bufp, string1, size1, string2, size2, |
| 826 | 4572 startpos, regs, stop |
| 4573 RE_LISP_CONTEXT_ARGS); | |
| 428 | 4574 #ifndef REGEX_MALLOC |
| 1333 | 4575 ALLOCA_GARBAGE_COLLECT (); |
| 428 | 4576 #endif |
| 1333 | 4577 END_REGEX_MALLOC_OK (); |
| 4578 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 428 | 4579 |
| 4580 if (val >= 0) | |
| 1333 | 4581 { |
| 4582 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4583 return startpos; | |
| 4584 } | |
| 428 | 4585 |
| 4586 if (val == -2) | |
| 1333 | 4587 { |
| 4588 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4589 return -2; | |
| 4590 } | |
| 4591 | |
| 4592 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 428 | 4593 advance: |
| 4594 if (!range) | |
| 4595 break; | |
| 4596 else if (range > 0) | |
| 4597 { | |
| 826 | 4598 Bytecount d_size; |
| 442 | 4599 d = ((const unsigned char *) |
| 428 | 4600 (startpos >= size1 ? string2 - size1 : string1) + startpos); |
| 867 | 4601 d_size = itext_ichar_len_fmt (d, fmt); |
| 428 | 4602 range -= d_size; |
| 826 | 4603 #if 1 |
| 4604 assert (!forward_search_p || range >= 0); | |
| 4605 #endif | |
| 428 | 4606 startpos += d_size; |
| 4607 } | |
| 4608 else | |
| 4609 { | |
| 826 | 4610 Bytecount d_size; |
| 428 | 4611 /* Note startpos > size1 not >=. If we are on the |
| 4612 string1/string2 boundary, we want to backup into string1. */ | |
| 442 | 4613 d = ((const unsigned char *) |
| 428 | 4614 (startpos > size1 ? string2 - size1 : string1) + startpos); |
| 867 | 4615 DEC_IBYTEPTR_FMT (d, fmt); |
| 4616 d_size = itext_ichar_len_fmt (d, fmt); | |
| 428 | 4617 range += d_size; |
| 826 | 4618 #if 1 |
| 4619 assert (!forward_search_p || range >= 0); | |
| 4620 #endif | |
| 428 | 4621 startpos -= d_size; |
| 4622 } | |
| 4623 } | |
| 1333 | 4624 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4625 return -1; |
| 4626 } /* re_search_2 */ | |
| 826 | 4627 |
| 428 | 4628 |
| 4629 /* Declarations and macros for re_match_2. */ | |
| 4630 | |
| 4631 /* This converts PTR, a pointer into one of the search strings `string1' | |
| 4632 and `string2' into an offset from the beginning of that string. */ | |
| 4633 #define POINTER_TO_OFFSET(ptr) \ | |
| 4634 (FIRST_STRING_P (ptr) \ | |
| 4635 ? ((regoff_t) ((ptr) - string1)) \ | |
| 4636 : ((regoff_t) ((ptr) - string2 + size1))) | |
| 4637 | |
| 4638 /* Macros for dealing with the split strings in re_match_2. */ | |
| 4639 | |
| 4640 #define MATCHING_IN_FIRST_STRING (dend == end_match_1) | |
| 4641 | |
| 4642 /* Call before fetching a character with *d. This switches over to | |
| 4643 string2 if necessary. */ | |
| 826 | 4644 #define REGEX_PREFETCH() \ |
| 428 | 4645 while (d == dend) \ |
| 4646 { \ | |
| 4647 /* End of string2 => fail. */ \ | |
| 4648 if (dend == end_match_2) \ | |
| 4649 goto fail; \ | |
| 4650 /* End of string1 => advance to string2. */ \ | |
| 4651 d = string2; \ | |
| 4652 dend = end_match_2; \ | |
| 4653 } | |
| 4654 | |
| 4655 | |
| 4656 /* Test if at very beginning or at very end of the virtual concatenation | |
| 4657 of `string1' and `string2'. If only one string, it's `string2'. */ | |
| 4658 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) | |
| 4659 #define AT_STRINGS_END(d) ((d) == end2) | |
| 4660 | |
| 4661 /* XEmacs change: | |
| 4662 If the given position straddles the string gap, return the equivalent | |
| 4663 position that is before or after the gap, respectively; otherwise, | |
| 4664 return the same position. */ | |
| 4665 #define POS_BEFORE_GAP_UNSAFE(d) ((d) == string2 ? end1 : (d)) | |
| 4666 #define POS_AFTER_GAP_UNSAFE(d) ((d) == end1 ? string2 : (d)) | |
| 4667 | |
| 4668 /* Test if CH is a word-constituent character. (XEmacs change) */ | |
| 826 | 4669 #define WORDCHAR_P(ch) \ |
| 4670 (SYNTAX (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf), ch) == Sword) | |
| 428 | 4671 |
| 4672 /* Free everything we malloc. */ | |
| 4673 #ifdef MATCH_MAY_ALLOCATE | |
| 1726 | 4674 #define FREE_VAR(var,type) if (var) REGEX_FREE (var, type); var = NULL |
| 428 | 4675 #define FREE_VARIABLES() \ |
| 4676 do { \ | |
| 1333 | 4677 UNBIND_REGEX_MALLOC_CHECK (); \ |
| 428 | 4678 REGEX_FREE_STACK (fail_stack.stack); \ |
| 1726 | 4679 FREE_VAR (regstart, re_char **); \ |
| 4680 FREE_VAR (regend, re_char **); \ | |
| 4681 FREE_VAR (old_regstart, re_char **); \ | |
| 4682 FREE_VAR (old_regend, re_char **); \ | |
| 4683 FREE_VAR (best_regstart, re_char **); \ | |
| 4684 FREE_VAR (best_regend, re_char **); \ | |
| 4685 FREE_VAR (reg_info, register_info_type *); \ | |
| 4686 FREE_VAR (reg_dummy, re_char **); \ | |
| 4687 FREE_VAR (reg_info_dummy, register_info_type *); \ | |
| 428 | 4688 } while (0) |
| 446 | 4689 #else /* not MATCH_MAY_ALLOCATE */ |
| 1333 | 4690 #define FREE_VARIABLES() \ |
| 4691 do { \ | |
| 4692 UNBIND_REGEX_MALLOC_CHECK (); \ | |
| 4693 } while (0) | |
| 446 | 4694 #endif /* MATCH_MAY_ALLOCATE */ |
| 428 | 4695 |
| 4696 /* These values must meet several constraints. They must not be valid | |
| 4697 register values; since we have a limit of 255 registers (because | |
| 4698 we use only one byte in the pattern for the register number), we can | |
| 4699 use numbers larger than 255. They must differ by 1, because of | |
| 4700 NUM_FAILURE_ITEMS above. And the value for the lowest register must | |
| 4701 be larger than the value for the highest register, so we do not try | |
| 4702 to actually save any registers when none are active. */ | |
| 4703 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) | |
| 4704 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) | |
| 4705 | |
| 4706 /* Matching routines. */ | |
| 4707 | |
| 826 | 4708 #ifndef emacs /* XEmacs never uses this. */ |
| 428 | 4709 /* re_match is like re_match_2 except it takes only a single string. */ |
| 4710 | |
| 4711 int | |
| 442 | 4712 re_match (struct re_pattern_buffer *bufp, const char *string, int size, |
| 826 | 4713 int pos, struct re_registers *regs |
| 4714 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4715 { |
| 446 | 4716 int result = re_match_2_internal (bufp, NULL, 0, (re_char *) string, size, |
| 826 | 4717 pos, regs, size |
| 4718 RE_LISP_CONTEXT_ARGS); | |
| 1333 | 4719 ALLOCA_GARBAGE_COLLECT (); |
| 428 | 4720 return result; |
| 4721 } | |
| 4722 #endif /* not emacs */ | |
| 4723 | |
| 4724 /* re_match_2 matches the compiled pattern in BUFP against the | |
| 4725 (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 and | |
| 4726 SIZE2, respectively). We start matching at POS, and stop matching | |
| 4727 at STOP. | |
| 4728 | |
| 4729 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we | |
| 4730 store offsets for the substring each group matched in REGS. See the | |
| 4731 documentation for exactly how many groups we fill. | |
| 4732 | |
| 4733 We return -1 if no match, -2 if an internal error (such as the | |
| 4734 failure stack overflowing). Otherwise, we return the length of the | |
| 4735 matched substring. */ | |
| 4736 | |
| 4737 int | |
| 442 | 4738 re_match_2 (struct re_pattern_buffer *bufp, const char *string1, |
| 4739 int size1, const char *string2, int size2, int pos, | |
| 826 | 4740 struct re_registers *regs, int stop |
| 4741 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4742 { |
| 460 | 4743 int result; |
| 4744 | |
| 4745 #ifdef emacs | |
| 826 | 4746 scache = setup_syntax_cache (scache, lispobj, lispbuf, |
| 4747 offset_to_charxpos (lispobj, pos), | |
| 4748 1); | |
| 460 | 4749 #endif |
| 4750 | |
| 4751 result = re_match_2_internal (bufp, (re_char *) string1, size1, | |
| 4752 (re_char *) string2, size2, | |
| 826 | 4753 pos, regs, stop |
| 4754 RE_LISP_CONTEXT_ARGS); | |
| 460 | 4755 |
| 1333 | 4756 ALLOCA_GARBAGE_COLLECT (); |
| 428 | 4757 return result; |
| 4758 } | |
| 4759 | |
| 4760 /* This is a separate function so that we can force an alloca cleanup | |
| 4761 afterwards. */ | |
| 4762 static int | |
| 446 | 4763 re_match_2_internal (struct re_pattern_buffer *bufp, re_char *string1, |
| 4764 int size1, re_char *string2, int size2, int pos, | |
| 826 | 4765 struct re_registers *regs, int stop |
| 2333 | 4766 RE_LISP_CONTEXT_ARGS_MULE_DECL) |
| 428 | 4767 { |
| 4768 /* General temporaries. */ | |
| 4769 int mcnt; | |
| 4770 unsigned char *p1; | |
| 4771 int should_succeed; /* XEmacs change */ | |
| 4772 | |
| 4773 /* Just past the end of the corresponding string. */ | |
| 446 | 4774 re_char *end1, *end2; |
| 428 | 4775 |
| 4776 /* Pointers into string1 and string2, just past the last characters in | |
| 4777 each to consider matching. */ | |
| 446 | 4778 re_char *end_match_1, *end_match_2; |
| 428 | 4779 |
| 4780 /* Where we are in the data, and the end of the current string. */ | |
| 446 | 4781 re_char *d, *dend; |
| 428 | 4782 |
| 4783 /* Where we are in the pattern, and the end of the pattern. */ | |
| 4784 unsigned char *p = bufp->buffer; | |
| 4785 REGISTER unsigned char *pend = p + bufp->used; | |
| 4786 | |
| 4787 /* Mark the opcode just after a start_memory, so we can test for an | |
| 4788 empty subpattern when we get to the stop_memory. */ | |
| 446 | 4789 re_char *just_past_start_mem = 0; |
| 428 | 4790 |
| 4791 /* We use this to map every character in the string. */ | |
| 446 | 4792 RE_TRANSLATE_TYPE translate = bufp->translate; |
| 428 | 4793 |
| 4794 /* Failure point stack. Each place that can handle a failure further | |
| 4795 down the line pushes a failure point on this stack. It consists of | |
| 4796 restart, regend, and reg_info for all registers corresponding to | |
| 4797 the subexpressions we're currently inside, plus the number of such | |
| 4798 registers, and, finally, two char *'s. The first char * is where | |
| 4799 to resume scanning the pattern; the second one is where to resume | |
| 4800 scanning the strings. If the latter is zero, the failure point is | |
| 4801 a ``dummy''; if a failure happens and the failure point is a dummy, | |
| 4802 it gets discarded and the next one is tried. */ | |
| 4803 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ | |
| 4804 fail_stack_type fail_stack; | |
| 4805 #endif | |
| 4806 #ifdef DEBUG | |
| 647 | 4807 static int failure_id; |
| 4808 int nfailure_points_pushed = 0, nfailure_points_popped = 0; | |
| 428 | 4809 #endif |
| 4810 | |
| 771 | 4811 #ifdef REGEX_REL_ALLOC |
| 428 | 4812 /* This holds the pointer to the failure stack, when |
| 4813 it is allocated relocatably. */ | |
| 4814 fail_stack_elt_t *failure_stack_ptr; | |
| 4815 #endif | |
| 4816 | |
| 4817 /* We fill all the registers internally, independent of what we | |
| 4818 return, for use in backreferences. The number here includes | |
| 4819 an element for register zero. */ | |
| 647 | 4820 int num_regs = bufp->re_ngroups + 1; |
| 428 | 4821 |
| 4822 /* The currently active registers. */ | |
| 647 | 4823 int lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
| 4824 int highest_active_reg = NO_HIGHEST_ACTIVE_REG; | |
| 428 | 4825 |
| 4826 /* Information on the contents of registers. These are pointers into | |
| 4827 the input strings; they record just what was matched (on this | |
| 4828 attempt) by a subexpression part of the pattern, that is, the | |
| 4829 regnum-th regstart pointer points to where in the pattern we began | |
| 4830 matching and the regnum-th regend points to right after where we | |
| 4831 stopped matching the regnum-th subexpression. (The zeroth register | |
| 4832 keeps track of what the whole pattern matches.) */ | |
| 4833 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ | |
| 446 | 4834 re_char **regstart, **regend; |
| 428 | 4835 #endif |
| 4836 | |
| 4837 /* If a group that's operated upon by a repetition operator fails to | |
| 4838 match anything, then the register for its start will need to be | |
| 4839 restored because it will have been set to wherever in the string we | |
| 4840 are when we last see its open-group operator. Similarly for a | |
| 4841 register's end. */ | |
| 4842 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ | |
| 446 | 4843 re_char **old_regstart, **old_regend; |
| 428 | 4844 #endif |
| 4845 | |
| 4846 /* The is_active field of reg_info helps us keep track of which (possibly | |
| 4847 nested) subexpressions we are currently in. The matched_something | |
| 4848 field of reg_info[reg_num] helps us tell whether or not we have | |
| 4849 matched any of the pattern so far this time through the reg_num-th | |
| 4850 subexpression. These two fields get reset each time through any | |
| 4851 loop their register is in. */ | |
| 4852 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ | |
| 4853 register_info_type *reg_info; | |
| 4854 #endif | |
| 4855 | |
| 4856 /* The following record the register info as found in the above | |
| 4857 variables when we find a match better than any we've seen before. | |
| 4858 This happens as we backtrack through the failure points, which in | |
| 4859 turn happens only if we have not yet matched the entire string. */ | |
| 647 | 4860 int best_regs_set = false; |
| 428 | 4861 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ |
| 446 | 4862 re_char **best_regstart, **best_regend; |
| 428 | 4863 #endif |
| 4864 | |
| 4865 /* Logically, this is `best_regend[0]'. But we don't want to have to | |
| 4866 allocate space for that if we're not allocating space for anything | |
| 4867 else (see below). Also, we never need info about register 0 for | |
| 4868 any of the other register vectors, and it seems rather a kludge to | |
| 4869 treat `best_regend' differently than the rest. So we keep track of | |
| 4870 the end of the best match so far in a separate variable. We | |
| 4871 initialize this to NULL so that when we backtrack the first time | |
| 4872 and need to test it, it's not garbage. */ | |
| 446 | 4873 re_char *match_end = NULL; |
| 428 | 4874 |
| 4875 /* This helps SET_REGS_MATCHED avoid doing redundant work. */ | |
| 4876 int set_regs_matched_done = 0; | |
| 4877 | |
| 4878 /* Used when we pop values we don't care about. */ | |
| 4879 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ | |
| 446 | 4880 re_char **reg_dummy; |
| 428 | 4881 register_info_type *reg_info_dummy; |
| 4882 #endif | |
| 4883 | |
| 4884 #ifdef DEBUG | |
| 4885 /* Counts the total number of registers pushed. */ | |
| 647 | 4886 int num_regs_pushed = 0; |
| 428 | 4887 #endif |
| 4888 | |
| 4889 /* 1 if this match ends in the same string (string1 or string2) | |
| 4890 as the best previous match. */ | |
| 460 | 4891 re_bool same_str_p; |
| 428 | 4892 |
| 4893 /* 1 if this match is the best seen so far. */ | |
| 460 | 4894 re_bool best_match_p; |
| 428 | 4895 |
| 826 | 4896 #ifdef emacs |
| 4897 Internal_Format fmt = buffer_or_other_internal_format (lispobj); | |
| 1346 | 4898 #ifdef REL_ALLOC |
| 4899 Ibyte *orig_buftext = | |
| 4900 BUFFERP (lispobj) ? | |
| 4901 BYTE_BUF_BYTE_ADDRESS (XBUFFER (lispobj), | |
| 4902 BYTE_BUF_BEGV (XBUFFER (lispobj))) : | |
| 4903 0; | |
| 4904 #endif | |
| 4905 | |
| 1333 | 4906 #ifdef ERROR_CHECK_MALLOC |
| 4907 int depth = bind_regex_malloc_disallowed (1); | |
| 4908 #endif | |
| 826 | 4909 #endif /* emacs */ |
| 771 | 4910 |
| 428 | 4911 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n"); |
| 4912 | |
| 1333 | 4913 BEGIN_REGEX_MALLOC_OK (); |
| 428 | 4914 INIT_FAIL_STACK (); |
| 1333 | 4915 END_REGEX_MALLOC_OK (); |
| 428 | 4916 |
| 4917 #ifdef MATCH_MAY_ALLOCATE | |
| 4918 /* Do not bother to initialize all the register variables if there are | |
| 4919 no groups in the pattern, as it takes a fair amount of time. If | |
| 4920 there are groups, we include space for register 0 (the whole | |
| 4921 pattern), even though we never use it, since it simplifies the | |
| 4922 array indexing. We should fix this. */ | |
| 502 | 4923 if (bufp->re_ngroups) |
| 428 | 4924 { |
| 1333 | 4925 BEGIN_REGEX_MALLOC_OK (); |
| 446 | 4926 regstart = REGEX_TALLOC (num_regs, re_char *); |
| 4927 regend = REGEX_TALLOC (num_regs, re_char *); | |
| 4928 old_regstart = REGEX_TALLOC (num_regs, re_char *); | |
| 4929 old_regend = REGEX_TALLOC (num_regs, re_char *); | |
| 4930 best_regstart = REGEX_TALLOC (num_regs, re_char *); | |
| 4931 best_regend = REGEX_TALLOC (num_regs, re_char *); | |
| 428 | 4932 reg_info = REGEX_TALLOC (num_regs, register_info_type); |
| 446 | 4933 reg_dummy = REGEX_TALLOC (num_regs, re_char *); |
| 428 | 4934 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); |
| 1333 | 4935 END_REGEX_MALLOC_OK (); |
| 428 | 4936 |
| 4937 if (!(regstart && regend && old_regstart && old_regend && reg_info | |
| 4938 && best_regstart && best_regend && reg_dummy && reg_info_dummy)) | |
| 4939 { | |
| 4940 FREE_VARIABLES (); | |
| 4941 return -2; | |
| 4942 } | |
| 4943 } | |
| 4944 else | |
| 4945 { | |
| 4946 /* We must initialize all our variables to NULL, so that | |
| 4947 `FREE_VARIABLES' doesn't try to free them. */ | |
| 4948 regstart = regend = old_regstart = old_regend = best_regstart | |
| 4949 = best_regend = reg_dummy = NULL; | |
| 4950 reg_info = reg_info_dummy = (register_info_type *) NULL; | |
| 4951 } | |
| 4952 #endif /* MATCH_MAY_ALLOCATE */ | |
| 4953 | |
| 1333 | 4954 #if defined (emacs) && defined (REL_ALLOC) |
| 4955 { | |
| 4956 /* If the allocations above (or the call to setup_syntax_cache() in | |
| 4957 re_match_2) caused a rel-alloc relocation, then fix up the data | |
| 4958 pointers */ | |
| 1346 | 4959 Bytecount offset = offset_post_relocation (lispobj, orig_buftext); |
| 1333 | 4960 if (offset) |
| 4961 { | |
| 4962 string1 += offset; | |
| 4963 string2 += offset; | |
| 4964 } | |
| 4965 } | |
| 4966 #endif /* defined (emacs) && defined (REL_ALLOC) */ | |
| 4967 | |
| 428 | 4968 /* The starting position is bogus. */ |
| 4969 if (pos < 0 || pos > size1 + size2) | |
| 4970 { | |
| 4971 FREE_VARIABLES (); | |
| 4972 return -1; | |
| 4973 } | |
| 4974 | |
| 4975 /* Initialize subexpression text positions to -1 to mark ones that no | |
| 4976 start_memory/stop_memory has been seen for. Also initialize the | |
| 4977 register information struct. */ | |
| 4978 for (mcnt = 1; mcnt < num_regs; mcnt++) | |
| 4979 { | |
| 4980 regstart[mcnt] = regend[mcnt] | |
| 4981 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; | |
| 4982 | |
| 4983 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; | |
| 4984 IS_ACTIVE (reg_info[mcnt]) = 0; | |
| 4985 MATCHED_SOMETHING (reg_info[mcnt]) = 0; | |
| 4986 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; | |
| 4987 } | |
| 4988 /* We move `string1' into `string2' if the latter's empty -- but not if | |
| 4989 `string1' is null. */ | |
| 4990 if (size2 == 0 && string1 != NULL) | |
| 4991 { | |
| 4992 string2 = string1; | |
| 4993 size2 = size1; | |
| 4994 string1 = 0; | |
| 4995 size1 = 0; | |
| 4996 } | |
| 4997 end1 = string1 + size1; | |
| 4998 end2 = string2 + size2; | |
| 4999 | |
| 5000 /* Compute where to stop matching, within the two strings. */ | |
| 5001 if (stop <= size1) | |
| 5002 { | |
| 5003 end_match_1 = string1 + stop; | |
| 5004 end_match_2 = string2; | |
| 5005 } | |
| 5006 else | |
| 5007 { | |
| 5008 end_match_1 = end1; | |
| 5009 end_match_2 = string2 + stop - size1; | |
| 5010 } | |
| 5011 | |
| 5012 /* `p' scans through the pattern as `d' scans through the data. | |
| 5013 `dend' is the end of the input string that `d' points within. `d' | |
| 5014 is advanced into the following input string whenever necessary, but | |
| 5015 this happens before fetching; therefore, at the beginning of the | |
| 5016 loop, `d' can be pointing at the end of a string, but it cannot | |
| 5017 equal `string2'. */ | |
| 5018 if (size1 > 0 && pos <= size1) | |
| 5019 { | |
| 5020 d = string1 + pos; | |
| 5021 dend = end_match_1; | |
| 5022 } | |
| 5023 else | |
| 5024 { | |
| 5025 d = string2 + pos - size1; | |
| 5026 dend = end_match_2; | |
| 5027 } | |
| 5028 | |
| 446 | 5029 DEBUG_PRINT1 ("The compiled pattern is: \n"); |
| 428 | 5030 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); |
| 5031 DEBUG_PRINT1 ("The string to match is: `"); | |
| 5032 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); | |
| 5033 DEBUG_PRINT1 ("'\n"); | |
| 5034 | |
| 5035 /* This loops over pattern commands. It exits by returning from the | |
| 5036 function if the match is complete, or it drops through if the match | |
| 5037 fails at this starting point in the input data. */ | |
| 5038 for (;;) | |
| 5039 { | |
| 5040 DEBUG_PRINT2 ("\n0x%lx: ", (long) p); | |
| 5041 #ifdef emacs /* XEmacs added, w/removal of immediate_quit */ | |
| 5042 if (!no_quit_in_re_search) | |
| 1333 | 5043 { |
| 5044 BEGIN_REGEX_MALLOC_OK (); | |
| 5045 QUIT; | |
| 5046 END_REGEX_MALLOC_OK (); | |
| 1346 | 5047 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); |
| 1333 | 5048 } |
| 428 | 5049 #endif |
| 5050 | |
| 5051 if (p == pend) | |
| 5052 { /* End of pattern means we might have succeeded. */ | |
| 5053 DEBUG_PRINT1 ("end of pattern ... "); | |
| 5054 | |
| 5055 /* If we haven't matched the entire string, and we want the | |
| 5056 longest match, try backtracking. */ | |
| 5057 if (d != end_match_2) | |
| 5058 { | |
| 5059 same_str_p = (FIRST_STRING_P (match_end) | |
| 5060 == MATCHING_IN_FIRST_STRING); | |
| 5061 | |
| 5062 /* AIX compiler got confused when this was combined | |
| 5063 with the previous declaration. */ | |
| 5064 if (same_str_p) | |
| 5065 best_match_p = d > match_end; | |
| 5066 else | |
| 5067 best_match_p = !MATCHING_IN_FIRST_STRING; | |
| 5068 | |
| 5069 DEBUG_PRINT1 ("backtracking.\n"); | |
| 5070 | |
| 5071 if (!FAIL_STACK_EMPTY ()) | |
| 5072 { /* More failure points to try. */ | |
| 5073 | |
| 5074 /* If exceeds best match so far, save it. */ | |
| 5075 if (!best_regs_set || best_match_p) | |
| 5076 { | |
| 5077 best_regs_set = true; | |
| 5078 match_end = d; | |
| 5079 | |
| 5080 DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); | |
| 5081 | |
| 5082 for (mcnt = 1; mcnt < num_regs; mcnt++) | |
| 5083 { | |
| 5084 best_regstart[mcnt] = regstart[mcnt]; | |
| 5085 best_regend[mcnt] = regend[mcnt]; | |
| 5086 } | |
| 5087 } | |
| 5088 goto fail; | |
| 5089 } | |
| 5090 | |
| 5091 /* If no failure points, don't restore garbage. And if | |
| 5092 last match is real best match, don't restore second | |
| 5093 best one. */ | |
| 5094 else if (best_regs_set && !best_match_p) | |
| 5095 { | |
| 5096 restore_best_regs: | |
| 5097 /* Restore best match. It may happen that `dend == | |
| 5098 end_match_1' while the restored d is in string2. | |
| 5099 For example, the pattern `x.*y.*z' against the | |
| 5100 strings `x-' and `y-z-', if the two strings are | |
| 5101 not consecutive in memory. */ | |
| 5102 DEBUG_PRINT1 ("Restoring best registers.\n"); | |
| 5103 | |
| 5104 d = match_end; | |
| 5105 dend = ((d >= string1 && d <= end1) | |
| 5106 ? end_match_1 : end_match_2); | |
| 5107 | |
| 5108 for (mcnt = 1; mcnt < num_regs; mcnt++) | |
| 5109 { | |
| 5110 regstart[mcnt] = best_regstart[mcnt]; | |
| 5111 regend[mcnt] = best_regend[mcnt]; | |
| 5112 } | |
| 5113 } | |
| 5114 } /* d != end_match_2 */ | |
| 5115 | |
| 5116 succeed_label: | |
| 5117 DEBUG_PRINT1 ("Accepting match.\n"); | |
| 5118 | |
| 5119 /* If caller wants register contents data back, do it. */ | |
| 1028 | 5120 { |
| 5121 int num_nonshy_regs = bufp->re_nsub + 1; | |
| 5122 if (regs && !bufp->no_sub) | |
| 5123 { | |
| 5124 /* Have the register data arrays been allocated? */ | |
| 5125 if (bufp->regs_allocated == REGS_UNALLOCATED) | |
| 5126 { /* No. So allocate them with malloc. We need one | |
| 5127 extra element beyond `num_regs' for the `-1' marker | |
| 5128 GNU code uses. */ | |
| 5129 regs->num_regs = MAX (RE_NREGS, num_nonshy_regs + 1); | |
| 1333 | 5130 BEGIN_REGEX_MALLOC_OK (); |
| 1028 | 5131 regs->start = TALLOC (regs->num_regs, regoff_t); |
| 5132 regs->end = TALLOC (regs->num_regs, regoff_t); | |
| 1333 | 5133 END_REGEX_MALLOC_OK (); |
| 5134 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 1028 | 5135 if (regs->start == NULL || regs->end == NULL) |
| 5136 { | |
| 5137 FREE_VARIABLES (); | |
| 5138 return -2; | |
| 5139 } | |
| 5140 bufp->regs_allocated = REGS_REALLOCATE; | |
| 5141 } | |
| 5142 else if (bufp->regs_allocated == REGS_REALLOCATE) | |
| 5143 { /* Yes. If we need more elements than were already | |
| 5144 allocated, reallocate them. If we need fewer, just | |
| 5145 leave it alone. */ | |
| 5146 if (regs->num_regs < num_nonshy_regs + 1) | |
| 5147 { | |
| 5148 regs->num_regs = num_nonshy_regs + 1; | |
| 1333 | 5149 BEGIN_REGEX_MALLOC_OK (); |
| 1028 | 5150 RETALLOC (regs->start, regs->num_regs, regoff_t); |
| 5151 RETALLOC (regs->end, regs->num_regs, regoff_t); | |
| 1333 | 5152 END_REGEX_MALLOC_OK (); |
| 5153 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 1028 | 5154 if (regs->start == NULL || regs->end == NULL) |
| 5155 { | |
| 5156 FREE_VARIABLES (); | |
| 5157 return -2; | |
| 5158 } | |
| 5159 } | |
| 5160 } | |
| 5161 else | |
| 5162 { | |
| 5163 /* The braces fend off a "empty body in an else-statement" | |
| 5164 warning under GCC when assert expands to nothing. */ | |
| 5165 assert (bufp->regs_allocated == REGS_FIXED); | |
| 5166 } | |
| 5167 | |
| 5168 /* Convert the pointer data in `regstart' and `regend' to | |
| 5169 indices. Register zero has to be set differently, | |
| 5170 since we haven't kept track of any info for it. */ | |
| 5171 if (regs->num_regs > 0) | |
| 5172 { | |
| 5173 regs->start[0] = pos; | |
| 5174 regs->end[0] = (MATCHING_IN_FIRST_STRING | |
| 5175 ? ((regoff_t) (d - string1)) | |
| 5176 : ((regoff_t) (d - string2 + size1))); | |
| 5177 } | |
| 5178 | |
| 2639 | 5179 /* Map over the NUM_NONSHY_REGS non-shy internal registers. |
| 5180 Copy each into the corresponding external register. | |
| 5181 MCNT indexes external registers. */ | |
| 1028 | 5182 for (mcnt = 1; mcnt < MIN (num_nonshy_regs, regs->num_regs); |
| 5183 mcnt++) | |
| 5184 { | |
| 5185 int internal_reg = bufp->external_to_internal_register[mcnt]; | |
| 5186 if (REG_UNSET (regstart[internal_reg]) || | |
| 5187 REG_UNSET (regend[internal_reg])) | |
| 5188 regs->start[mcnt] = regs->end[mcnt] = -1; | |
| 5189 else | |
| 5190 { | |
| 5191 regs->start[mcnt] = | |
| 5192 (regoff_t) POINTER_TO_OFFSET (regstart[internal_reg]); | |
| 5193 regs->end[mcnt] = | |
| 5194 (regoff_t) POINTER_TO_OFFSET (regend[internal_reg]); | |
| 5195 } | |
| 5196 } | |
| 5197 } /* regs && !bufp->no_sub */ | |
| 5198 | |
| 5199 /* If we have regs and the regs structure has more elements than | |
| 2639 | 5200 were in the pattern, set the extra elements starting with |
| 5201 NUM_NONSHY_REGS to -1. If we (re)allocated the registers, | |
| 5202 this is the case, because we always allocate enough to have | |
| 5203 at least one -1 at the end. | |
| 1028 | 5204 |
| 5205 We do this even when no_sub is set because some applications | |
| 5206 (XEmacs) reuse register structures which may contain stale | |
| 5207 information, and permit attempts to access those registers. | |
| 5208 | |
| 5209 It would be possible to require the caller to do this, but we'd | |
| 5210 have to change the API for this function to reflect that, and | |
| 1425 | 5211 audit all callers. Note: as of 2003-04-17 callers in XEmacs |
| 5212 do clear the registers, but it's safer to leave this code in | |
| 5213 because of reallocation. | |
| 5214 */ | |
| 1028 | 5215 if (regs && regs->num_regs > 0) |
| 5216 for (mcnt = num_nonshy_regs; mcnt < regs->num_regs; mcnt++) | |
| 5217 regs->start[mcnt] = regs->end[mcnt] = -1; | |
| 5218 } | |
| 428 | 5219 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n", |
| 5220 nfailure_points_pushed, nfailure_points_popped, | |
| 5221 nfailure_points_pushed - nfailure_points_popped); | |
| 5222 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); | |
| 5223 | |
| 5224 mcnt = d - pos - (MATCHING_IN_FIRST_STRING | |
| 5225 ? string1 | |
| 5226 : string2 - size1); | |
| 5227 | |
| 5228 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); | |
| 5229 | |
| 5230 FREE_VARIABLES (); | |
| 5231 return mcnt; | |
| 5232 } | |
| 5233 | |
| 5234 /* Otherwise match next pattern command. */ | |
| 5235 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++)) | |
| 5236 { | |
| 5237 /* Ignore these. Used to ignore the n of succeed_n's which | |
| 5238 currently have n == 0. */ | |
| 5239 case no_op: | |
| 5240 DEBUG_PRINT1 ("EXECUTING no_op.\n"); | |
| 5241 break; | |
| 5242 | |
| 5243 case succeed: | |
| 5244 DEBUG_PRINT1 ("EXECUTING succeed.\n"); | |
| 5245 goto succeed_label; | |
| 5246 | |
| 826 | 5247 /* Match exactly a string of length n in the pattern. The |
| 5248 following byte in the pattern defines n, and the n bytes after | |
| 5249 that make up the string to match. (Under Mule, this will be in | |
| 5250 the default internal format.) */ | |
| 428 | 5251 case exactn: |
| 5252 mcnt = *p++; | |
| 5253 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt); | |
| 5254 | |
| 5255 /* This is written out as an if-else so we don't waste time | |
| 5256 testing `translate' inside the loop. */ | |
| 446 | 5257 if (TRANSLATE_P (translate)) |
| 428 | 5258 { |
| 5259 do | |
| 5260 { | |
| 446 | 5261 #ifdef MULE |
| 5262 Bytecount pat_len; | |
| 5263 | |
| 450 | 5264 REGEX_PREFETCH (); |
| 867 | 5265 if (RE_TRANSLATE_1 (itext_ichar_fmt (d, fmt, lispobj)) |
| 5266 != itext_ichar (p)) | |
| 428 | 5267 goto fail; |
| 446 | 5268 |
| 867 | 5269 pat_len = itext_ichar_len (p); |
| 446 | 5270 p += pat_len; |
| 867 | 5271 INC_IBYTEPTR_FMT (d, fmt); |
| 446 | 5272 |
| 5273 mcnt -= pat_len; | |
| 5274 #else /* not MULE */ | |
| 450 | 5275 REGEX_PREFETCH (); |
| 826 | 5276 if ((unsigned char) RE_TRANSLATE_1 (*d++) != *p++) |
| 446 | 5277 goto fail; |
| 5278 mcnt--; | |
| 5279 #endif | |
| 428 | 5280 } |
| 446 | 5281 while (mcnt > 0); |
| 428 | 5282 } |
| 5283 else | |
| 5284 { | |
| 826 | 5285 #ifdef MULE |
| 5286 /* If buffer format is default, then we can shortcut and just | |
| 5287 compare the text directly, byte by byte. Otherwise, we | |
| 5288 need to go character by character. */ | |
| 5289 if (fmt != FORMAT_DEFAULT) | |
| 428 | 5290 { |
| 826 | 5291 do |
| 5292 { | |
| 5293 Bytecount pat_len; | |
| 5294 | |
| 5295 REGEX_PREFETCH (); | |
| 867 | 5296 if (itext_ichar_fmt (d, fmt, lispobj) != |
| 5297 itext_ichar (p)) | |
| 826 | 5298 goto fail; |
| 5299 | |
| 867 | 5300 pat_len = itext_ichar_len (p); |
| 826 | 5301 p += pat_len; |
| 867 | 5302 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 5303 |
| 5304 mcnt -= pat_len; | |
| 5305 } | |
| 5306 while (mcnt > 0); | |
| 428 | 5307 } |
| 826 | 5308 else |
| 5309 #endif | |
| 5310 { | |
| 5311 do | |
| 5312 { | |
| 5313 REGEX_PREFETCH (); | |
| 5314 if (*d++ != *p++) goto fail; | |
| 5315 mcnt--; | |
| 5316 } | |
| 5317 while (mcnt > 0); | |
| 5318 } | |
| 428 | 5319 } |
| 5320 SET_REGS_MATCHED (); | |
| 5321 break; | |
| 5322 | |
| 5323 | |
| 5324 /* Match any character except possibly a newline or a null. */ | |
| 5325 case anychar: | |
| 5326 DEBUG_PRINT1 ("EXECUTING anychar.\n"); | |
| 5327 | |
| 450 | 5328 REGEX_PREFETCH (); |
| 428 | 5329 |
| 826 | 5330 if ((!(bufp->syntax & RE_DOT_NEWLINE) && |
| 867 | 5331 RE_TRANSLATE (itext_ichar_fmt (d, fmt, lispobj)) == '\n') |
| 826 | 5332 || (bufp->syntax & RE_DOT_NOT_NULL && |
| 867 | 5333 RE_TRANSLATE (itext_ichar_fmt (d, fmt, lispobj)) == |
| 826 | 5334 '\000')) |
| 428 | 5335 goto fail; |
| 5336 | |
| 5337 SET_REGS_MATCHED (); | |
| 5338 DEBUG_PRINT2 (" Matched `%d'.\n", *d); | |
| 867 | 5339 INC_IBYTEPTR_FMT (d, fmt); /* XEmacs change */ |
| 428 | 5340 break; |
| 5341 | |
| 5342 | |
| 5343 case charset: | |
| 5344 case charset_not: | |
| 5345 { | |
| 1414 | 5346 REGISTER Ichar c; |
| 460 | 5347 re_bool not_p = (re_opcode_t) *(p - 1) == charset_not; |
| 458 | 5348 |
| 5349 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not_p ? "_not" : ""); | |
| 428 | 5350 |
| 450 | 5351 REGEX_PREFETCH (); |
| 867 | 5352 c = itext_ichar_fmt (d, fmt, lispobj); |
| 826 | 5353 c = RE_TRANSLATE (c); /* The character to match. */ |
| 428 | 5354 |
| 647 | 5355 /* Cast to `unsigned int' instead of `unsigned char' in case the |
| 428 | 5356 bit list is a full 32 bytes long. */ |
| 1414 | 5357 if ((unsigned int)c < (unsigned int) (*p * BYTEWIDTH) |
| 428 | 5358 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
| 458 | 5359 not_p = !not_p; |
| 428 | 5360 |
| 5361 p += 1 + *p; | |
| 5362 | |
| 458 | 5363 if (!not_p) goto fail; |
| 428 | 5364 |
| 5365 SET_REGS_MATCHED (); | |
| 867 | 5366 INC_IBYTEPTR_FMT (d, fmt); /* XEmacs change */ |
| 428 | 5367 break; |
| 5368 } | |
| 5369 | |
| 5370 #ifdef MULE | |
| 5371 case charset_mule: | |
| 5372 case charset_mule_not: | |
| 5373 { | |
| 867 | 5374 REGISTER Ichar c; |
| 460 | 5375 re_bool not_p = (re_opcode_t) *(p - 1) == charset_mule_not; |
| 458 | 5376 |
| 5377 DEBUG_PRINT2 ("EXECUTING charset_mule%s.\n", not_p ? "_not" : ""); | |
| 428 | 5378 |
| 450 | 5379 REGEX_PREFETCH (); |
| 867 | 5380 c = itext_ichar_fmt (d, fmt, lispobj); |
| 826 | 5381 c = RE_TRANSLATE (c); /* The character to match. */ |
| 428 | 5382 |
| 5383 if (EQ (Qt, unified_range_table_lookup (p, c, Qnil))) | |
| 458 | 5384 not_p = !not_p; |
| 428 | 5385 |
| 5386 p += unified_range_table_bytes_used (p); | |
| 5387 | |
| 458 | 5388 if (!not_p) goto fail; |
| 428 | 5389 |
| 5390 SET_REGS_MATCHED (); | |
| 867 | 5391 INC_IBYTEPTR_FMT (d, fmt); |
| 428 | 5392 break; |
| 5393 } | |
| 5394 #endif /* MULE */ | |
| 5395 | |
| 5396 | |
| 5397 /* The beginning of a group is represented by start_memory. | |
| 5398 The arguments are the register number in the next byte, and the | |
| 5399 number of groups inner to this one in the next. The text | |
| 5400 matched within the group is recorded (in the internal | |
| 5401 registers data structure) under the register number. */ | |
| 5402 case start_memory: | |
| 5403 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); | |
| 5404 | |
| 5405 /* Find out if this group can match the empty string. */ | |
| 5406 p1 = p; /* To send to group_match_null_string_p. */ | |
| 5407 | |
| 5408 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) | |
| 2639 | 5409 REG_MATCH_NULL_STRING_P (reg_info[*p]) |
| 5410 = group_match_null_string_p (&p1, pend, reg_info); | |
| 5411 | |
| 5412 DEBUG_PRINT2 (" group CAN%s match null string\n", | |
| 5413 REG_MATCH_NULL_STRING_P (reg_info[*p]) ? "NOT" : ""); | |
| 428 | 5414 |
| 5415 /* Save the position in the string where we were the last time | |
| 5416 we were at this open-group operator in case the group is | |
| 5417 operated upon by a repetition operator, e.g., with `(a*)*b' | |
| 5418 against `ab'; then we want to ignore where we are now in | |
| 5419 the string in case this attempt to match fails. */ | |
| 5420 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) | |
| 5421 ? REG_UNSET (regstart[*p]) ? d : regstart[*p] | |
| 5422 : regstart[*p]; | |
| 5423 DEBUG_PRINT2 (" old_regstart: %d\n", | |
| 5424 POINTER_TO_OFFSET (old_regstart[*p])); | |
| 5425 | |
| 5426 regstart[*p] = d; | |
| 5427 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); | |
| 5428 | |
| 5429 IS_ACTIVE (reg_info[*p]) = 1; | |
| 5430 MATCHED_SOMETHING (reg_info[*p]) = 0; | |
| 5431 | |
| 5432 /* Clear this whenever we change the register activity status. */ | |
| 5433 set_regs_matched_done = 0; | |
| 5434 | |
| 5435 /* This is the new highest active register. */ | |
| 5436 highest_active_reg = *p; | |
| 5437 | |
| 5438 /* If nothing was active before, this is the new lowest active | |
| 5439 register. */ | |
| 5440 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) | |
| 5441 lowest_active_reg = *p; | |
| 5442 | |
| 5443 /* Move past the register number and inner group count. */ | |
| 5444 p += 2; | |
| 5445 just_past_start_mem = p; | |
| 5446 | |
| 5447 break; | |
| 5448 | |
| 5449 | |
| 5450 /* The stop_memory opcode represents the end of a group. Its | |
| 5451 arguments are the same as start_memory's: the register | |
| 5452 number, and the number of inner groups. */ | |
| 5453 case stop_memory: | |
| 5454 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); | |
| 5455 | |
| 5456 /* We need to save the string position the last time we were at | |
| 5457 this close-group operator in case the group is operated | |
| 5458 upon by a repetition operator, e.g., with `((a*)*(b*)*)*' | |
| 5459 against `aba'; then we want to ignore where we are now in | |
| 5460 the string in case this attempt to match fails. */ | |
| 5461 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) | |
| 5462 ? REG_UNSET (regend[*p]) ? d : regend[*p] | |
| 5463 : regend[*p]; | |
| 5464 DEBUG_PRINT2 (" old_regend: %d\n", | |
| 5465 POINTER_TO_OFFSET (old_regend[*p])); | |
| 5466 | |
| 5467 regend[*p] = d; | |
| 5468 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); | |
| 5469 | |
| 5470 /* This register isn't active anymore. */ | |
| 5471 IS_ACTIVE (reg_info[*p]) = 0; | |
| 5472 | |
| 5473 /* Clear this whenever we change the register activity status. */ | |
| 5474 set_regs_matched_done = 0; | |
| 5475 | |
| 5476 /* If this was the only register active, nothing is active | |
| 5477 anymore. */ | |
| 5478 if (lowest_active_reg == highest_active_reg) | |
| 5479 { | |
| 5480 lowest_active_reg = NO_LOWEST_ACTIVE_REG; | |
| 5481 highest_active_reg = NO_HIGHEST_ACTIVE_REG; | |
| 5482 } | |
| 5483 else | |
| 5484 { /* We must scan for the new highest active register, since | |
| 5485 it isn't necessarily one less than now: consider | |
| 5486 (a(b)c(d(e)f)g). When group 3 ends, after the f), the | |
| 5487 new highest active register is 1. */ | |
| 5488 unsigned char r = *p - 1; | |
| 5489 while (r > 0 && !IS_ACTIVE (reg_info[r])) | |
| 5490 r--; | |
| 5491 | |
| 5492 /* If we end up at register zero, that means that we saved | |
| 5493 the registers as the result of an `on_failure_jump', not | |
| 5494 a `start_memory', and we jumped to past the innermost | |
| 5495 `stop_memory'. For example, in ((.)*) we save | |
| 5496 registers 1 and 2 as a result of the *, but when we pop | |
| 5497 back to the second ), we are at the stop_memory 1. | |
| 5498 Thus, nothing is active. */ | |
| 5499 if (r == 0) | |
| 5500 { | |
| 5501 lowest_active_reg = NO_LOWEST_ACTIVE_REG; | |
| 5502 highest_active_reg = NO_HIGHEST_ACTIVE_REG; | |
| 5503 } | |
| 5504 else | |
| 5505 { | |
| 5506 highest_active_reg = r; | |
| 5507 | |
| 5508 /* 98/9/21 jhod: We've also gotta set lowest_active_reg, don't we? */ | |
| 5509 r = 1; | |
| 5510 while (r < highest_active_reg && !IS_ACTIVE(reg_info[r])) | |
| 5511 r++; | |
| 5512 lowest_active_reg = r; | |
| 5513 } | |
| 5514 } | |
| 5515 | |
| 5516 /* If just failed to match something this time around with a | |
| 5517 group that's operated on by a repetition operator, try to | |
| 5518 force exit from the ``loop'', and restore the register | |
| 5519 information for this group that we had before trying this | |
| 5520 last match. */ | |
| 5521 if ((!MATCHED_SOMETHING (reg_info[*p]) | |
| 5522 || just_past_start_mem == p - 1) | |
| 5523 && (p + 2) < pend) | |
| 5524 { | |
| 460 | 5525 re_bool is_a_jump_n = false; |
| 428 | 5526 |
| 5527 p1 = p + 2; | |
| 5528 mcnt = 0; | |
| 5529 switch ((re_opcode_t) *p1++) | |
| 5530 { | |
| 5531 case jump_n: | |
| 5532 is_a_jump_n = true; | |
| 5533 case pop_failure_jump: | |
| 5534 case maybe_pop_jump: | |
| 5535 case jump: | |
| 5536 case dummy_failure_jump: | |
| 5537 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 5538 if (is_a_jump_n) | |
| 5539 p1 += 2; | |
| 5540 break; | |
| 5541 | |
| 5542 default: | |
| 5543 /* do nothing */ ; | |
| 5544 } | |
| 5545 p1 += mcnt; | |
| 5546 | |
| 5547 /* If the next operation is a jump backwards in the pattern | |
| 5548 to an on_failure_jump right before the start_memory | |
| 5549 corresponding to this stop_memory, exit from the loop | |
| 5550 by forcing a failure after pushing on the stack the | |
| 5551 on_failure_jump's jump in the pattern, and d. */ | |
| 5552 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump | |
| 5553 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p) | |
| 5554 { | |
| 5555 /* If this group ever matched anything, then restore | |
| 5556 what its registers were before trying this last | |
| 5557 failed match, e.g., with `(a*)*b' against `ab' for | |
| 5558 regstart[1], and, e.g., with `((a*)*(b*)*)*' | |
| 5559 against `aba' for regend[3]. | |
| 5560 | |
| 5561 Also restore the registers for inner groups for, | |
| 5562 e.g., `((a*)(b*))*' against `aba' (register 3 would | |
| 5563 otherwise get trashed). */ | |
| 5564 | |
| 5565 if (EVER_MATCHED_SOMETHING (reg_info[*p])) | |
| 5566 { | |
| 647 | 5567 int r; |
| 428 | 5568 |
| 5569 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; | |
| 5570 | |
| 5571 /* Restore this and inner groups' (if any) registers. */ | |
| 5572 for (r = *p; r < *p + *(p + 1); r++) | |
| 5573 { | |
| 5574 regstart[r] = old_regstart[r]; | |
| 5575 | |
| 5576 /* xx why this test? */ | |
| 5577 if (old_regend[r] >= regstart[r]) | |
| 5578 regend[r] = old_regend[r]; | |
| 5579 } | |
| 5580 } | |
| 5581 p1++; | |
| 5582 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 5583 PUSH_FAILURE_POINT (p1 + mcnt, d, -2); | |
| 5584 | |
| 5585 goto fail; | |
| 5586 } | |
| 5587 } | |
| 5588 | |
| 5589 /* Move past the register number and the inner group count. */ | |
| 5590 p += 2; | |
| 5591 break; | |
| 5592 | |
| 5593 | |
| 5594 /* \<digit> has been turned into a `duplicate' command which is | |
| 502 | 5595 followed by the numeric value of <digit> as the register number. |
| 5596 (Already passed through external-to-internal-register mapping, | |
| 5597 so it refers to the actual group number, not the non-shy-only | |
| 5598 numbering used in the external world.) */ | |
| 428 | 5599 case duplicate: |
| 5600 { | |
| 446 | 5601 REGISTER re_char *d2, *dend2; |
| 502 | 5602 /* Get which register to match against. */ |
| 5603 int regno = *p++; | |
| 428 | 5604 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno); |
| 5605 | |
| 5606 /* Can't back reference a group which we've never matched. */ | |
| 5607 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) | |
| 5608 goto fail; | |
| 5609 | |
| 5610 /* Where in input to try to start matching. */ | |
| 5611 d2 = regstart[regno]; | |
| 5612 | |
| 5613 /* Where to stop matching; if both the place to start and | |
| 5614 the place to stop matching are in the same string, then | |
| 5615 set to the place to stop, otherwise, for now have to use | |
| 5616 the end of the first string. */ | |
| 5617 | |
| 5618 dend2 = ((FIRST_STRING_P (regstart[regno]) | |
| 5619 == FIRST_STRING_P (regend[regno])) | |
| 5620 ? regend[regno] : end_match_1); | |
| 5621 for (;;) | |
| 5622 { | |
| 5623 /* If necessary, advance to next segment in register | |
| 5624 contents. */ | |
| 5625 while (d2 == dend2) | |
| 5626 { | |
| 5627 if (dend2 == end_match_2) break; | |
| 5628 if (dend2 == regend[regno]) break; | |
| 5629 | |
| 5630 /* End of string1 => advance to string2. */ | |
| 5631 d2 = string2; | |
| 5632 dend2 = regend[regno]; | |
| 5633 } | |
| 5634 /* At end of register contents => success */ | |
| 5635 if (d2 == dend2) break; | |
| 5636 | |
| 5637 /* If necessary, advance to next segment in data. */ | |
| 450 | 5638 REGEX_PREFETCH (); |
| 428 | 5639 |
| 5640 /* How many characters left in this segment to match. */ | |
| 5641 mcnt = dend - d; | |
| 5642 | |
| 5643 /* Want how many consecutive characters we can match in | |
| 5644 one shot, so, if necessary, adjust the count. */ | |
| 5645 if (mcnt > dend2 - d2) | |
| 5646 mcnt = dend2 - d2; | |
| 5647 | |
| 5648 /* Compare that many; failure if mismatch, else move | |
| 5649 past them. */ | |
| 446 | 5650 if (TRANSLATE_P (translate) |
| 826 | 5651 ? bcmp_translate (d, d2, mcnt, translate |
| 5652 #ifdef emacs | |
| 5653 , fmt, lispobj | |
| 5654 #endif | |
| 5655 ) | |
| 428 | 5656 : memcmp (d, d2, mcnt)) |
| 5657 goto fail; | |
| 5658 d += mcnt, d2 += mcnt; | |
| 5659 | |
| 5660 /* Do this because we've match some characters. */ | |
| 5661 SET_REGS_MATCHED (); | |
| 5662 } | |
| 5663 } | |
| 5664 break; | |
| 5665 | |
| 5666 | |
| 5667 /* begline matches the empty string at the beginning of the string | |
| 5668 (unless `not_bol' is set in `bufp'), and, if | |
| 5669 `newline_anchor' is set, after newlines. */ | |
| 5670 case begline: | |
| 5671 DEBUG_PRINT1 ("EXECUTING begline.\n"); | |
| 5672 | |
| 5673 if (AT_STRINGS_BEG (d)) | |
| 5674 { | |
| 5675 if (!bufp->not_bol) break; | |
| 5676 } | |
| 826 | 5677 else |
| 5678 { | |
| 5679 re_char *d2 = d; | |
| 867 | 5680 DEC_IBYTEPTR (d2); |
| 5681 if (itext_ichar_ascii_fmt (d2, fmt, lispobj) == '\n' && | |
| 826 | 5682 bufp->newline_anchor) |
| 5683 break; | |
| 5684 } | |
| 428 | 5685 /* In all other cases, we fail. */ |
| 5686 goto fail; | |
| 5687 | |
| 5688 | |
| 5689 /* endline is the dual of begline. */ | |
| 5690 case endline: | |
| 5691 DEBUG_PRINT1 ("EXECUTING endline.\n"); | |
| 5692 | |
| 5693 if (AT_STRINGS_END (d)) | |
| 5694 { | |
| 5695 if (!bufp->not_eol) break; | |
| 5696 } | |
| 5697 | |
| 5698 /* We have to ``prefetch'' the next character. */ | |
| 826 | 5699 else if ((d == end1 ? |
| 867 | 5700 itext_ichar_ascii_fmt (string2, fmt, lispobj) : |
| 5701 itext_ichar_ascii_fmt (d, fmt, lispobj)) == '\n' | |
| 428 | 5702 && bufp->newline_anchor) |
| 5703 { | |
| 5704 break; | |
| 5705 } | |
| 5706 goto fail; | |
| 5707 | |
| 5708 | |
| 5709 /* Match at the very beginning of the data. */ | |
| 5710 case begbuf: | |
| 5711 DEBUG_PRINT1 ("EXECUTING begbuf.\n"); | |
| 5712 if (AT_STRINGS_BEG (d)) | |
| 5713 break; | |
| 5714 goto fail; | |
| 5715 | |
| 5716 | |
| 5717 /* Match at the very end of the data. */ | |
| 5718 case endbuf: | |
| 5719 DEBUG_PRINT1 ("EXECUTING endbuf.\n"); | |
| 5720 if (AT_STRINGS_END (d)) | |
| 5721 break; | |
| 5722 goto fail; | |
| 5723 | |
| 5724 | |
| 5725 /* on_failure_keep_string_jump is used to optimize `.*\n'. It | |
| 5726 pushes NULL as the value for the string on the stack. Then | |
| 5727 `pop_failure_point' will keep the current value for the | |
| 5728 string, instead of restoring it. To see why, consider | |
| 5729 matching `foo\nbar' against `.*\n'. The .* matches the foo; | |
| 5730 then the . fails against the \n. But the next thing we want | |
| 5731 to do is match the \n against the \n; if we restored the | |
| 5732 string value, we would be back at the foo. | |
| 5733 | |
| 5734 Because this is used only in specific cases, we don't need to | |
| 5735 check all the things that `on_failure_jump' does, to make | |
| 5736 sure the right things get saved on the stack. Hence we don't | |
| 5737 share its code. The only reason to push anything on the | |
| 5738 stack at all is that otherwise we would have to change | |
| 5739 `anychar's code to do something besides goto fail in this | |
| 5740 case; that seems worse than this. */ | |
| 5741 case on_failure_keep_string_jump: | |
| 5742 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); | |
| 5743 | |
| 5744 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 5745 DEBUG_PRINT3 (" %d (to 0x%lx):\n", mcnt, (long) (p + mcnt)); | |
| 5746 | |
| 446 | 5747 PUSH_FAILURE_POINT (p + mcnt, (unsigned char *) 0, -2); |
| 428 | 5748 break; |
| 5749 | |
| 5750 | |
| 5751 /* Uses of on_failure_jump: | |
| 5752 | |
| 5753 Each alternative starts with an on_failure_jump that points | |
| 5754 to the beginning of the next alternative. Each alternative | |
| 5755 except the last ends with a jump that in effect jumps past | |
| 5756 the rest of the alternatives. (They really jump to the | |
| 5757 ending jump of the following alternative, because tensioning | |
| 5758 these jumps is a hassle.) | |
| 5759 | |
| 5760 Repeats start with an on_failure_jump that points past both | |
| 5761 the repetition text and either the following jump or | |
| 5762 pop_failure_jump back to this on_failure_jump. */ | |
| 5763 case on_failure_jump: | |
| 5764 on_failure: | |
| 5765 DEBUG_PRINT1 ("EXECUTING on_failure_jump"); | |
| 5766 | |
| 5767 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 5768 DEBUG_PRINT3 (" %d (to 0x%lx)", mcnt, (long) (p + mcnt)); | |
| 5769 | |
| 5770 /* If this on_failure_jump comes right before a group (i.e., | |
| 5771 the original * applied to a group), save the information | |
| 5772 for that group and all inner ones, so that if we fail back | |
| 5773 to this point, the group's information will be correct. | |
| 5774 For example, in \(a*\)*\1, we need the preceding group, | |
| 5775 and in \(\(a*\)b*\)\2, we need the inner group. */ | |
| 5776 | |
| 5777 /* We can't use `p' to check ahead because we push | |
| 5778 a failure point to `p + mcnt' after we do this. */ | |
| 5779 p1 = p; | |
| 5780 | |
| 5781 /* We need to skip no_op's before we look for the | |
| 5782 start_memory in case this on_failure_jump is happening as | |
| 5783 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 | |
| 5784 against aba. */ | |
| 5785 while (p1 < pend && (re_opcode_t) *p1 == no_op) | |
| 5786 p1++; | |
| 5787 | |
| 5788 if (p1 < pend && (re_opcode_t) *p1 == start_memory) | |
| 5789 { | |
| 5790 /* We have a new highest active register now. This will | |
| 5791 get reset at the start_memory we are about to get to, | |
| 5792 but we will have saved all the registers relevant to | |
| 5793 this repetition op, as described above. */ | |
| 5794 highest_active_reg = *(p1 + 1) + *(p1 + 2); | |
| 5795 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) | |
| 5796 lowest_active_reg = *(p1 + 1); | |
| 5797 } | |
| 5798 | |
| 5799 DEBUG_PRINT1 (":\n"); | |
| 5800 PUSH_FAILURE_POINT (p + mcnt, d, -2); | |
| 5801 break; | |
| 5802 | |
| 5803 | |
| 5804 /* A smart repeat ends with `maybe_pop_jump'. | |
| 5805 We change it to either `pop_failure_jump' or `jump'. */ | |
| 5806 case maybe_pop_jump: | |
| 5807 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 5808 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); | |
| 5809 { | |
| 5810 REGISTER unsigned char *p2 = p; | |
| 5811 | |
| 5812 /* Compare the beginning of the repeat with what in the | |
| 5813 pattern follows its end. If we can establish that there | |
| 5814 is nothing that they would both match, i.e., that we | |
| 5815 would have to backtrack because of (as in, e.g., `a*a') | |
| 5816 then we can change to pop_failure_jump, because we'll | |
| 5817 never have to backtrack. | |
| 5818 | |
| 5819 This is not true in the case of alternatives: in | |
| 5820 `(a|ab)*' we do need to backtrack to the `ab' alternative | |
| 5821 (e.g., if the string was `ab'). But instead of trying to | |
| 5822 detect that here, the alternative has put on a dummy | |
| 5823 failure point which is what we will end up popping. */ | |
| 5824 | |
| 5825 /* Skip over open/close-group commands. | |
| 5826 If what follows this loop is a ...+ construct, | |
| 5827 look at what begins its body, since we will have to | |
| 5828 match at least one of that. */ | |
| 5829 while (1) | |
| 5830 { | |
| 5831 if (p2 + 2 < pend | |
| 5832 && ((re_opcode_t) *p2 == stop_memory | |
| 5833 || (re_opcode_t) *p2 == start_memory)) | |
| 5834 p2 += 3; | |
| 5835 else if (p2 + 6 < pend | |
| 5836 && (re_opcode_t) *p2 == dummy_failure_jump) | |
| 5837 p2 += 6; | |
| 5838 else | |
| 5839 break; | |
| 5840 } | |
| 5841 | |
| 5842 p1 = p + mcnt; | |
| 5843 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding | |
| 5844 to the `maybe_finalize_jump' of this case. Examine what | |
| 5845 follows. */ | |
| 5846 | |
| 5847 /* If we're at the end of the pattern, we can change. */ | |
| 5848 if (p2 == pend) | |
| 5849 { | |
| 5850 /* Consider what happens when matching ":\(.*\)" | |
| 5851 against ":/". I don't really understand this code | |
| 5852 yet. */ | |
| 5853 p[-3] = (unsigned char) pop_failure_jump; | |
| 5854 DEBUG_PRINT1 | |
| 5855 (" End of pattern: change to `pop_failure_jump'.\n"); | |
| 5856 } | |
| 5857 | |
| 5858 else if ((re_opcode_t) *p2 == exactn | |
| 5859 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) | |
| 5860 { | |
| 5861 REGISTER unsigned char c | |
| 5862 = *p2 == (unsigned char) endline ? '\n' : p2[2]; | |
| 5863 | |
| 5864 if ((re_opcode_t) p1[3] == exactn && p1[5] != c) | |
| 5865 { | |
| 5866 p[-3] = (unsigned char) pop_failure_jump; | |
| 5867 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", | |
| 5868 c, p1[5]); | |
| 5869 } | |
| 5870 | |
| 5871 else if ((re_opcode_t) p1[3] == charset | |
| 5872 || (re_opcode_t) p1[3] == charset_not) | |
| 5873 { | |
| 458 | 5874 int not_p = (re_opcode_t) p1[3] == charset_not; |
| 428 | 5875 |
| 5876 if (c < (unsigned char) (p1[4] * BYTEWIDTH) | |
| 5877 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) | |
| 458 | 5878 not_p = !not_p; |
| 5879 | |
| 5880 /* `not_p' is equal to 1 if c would match, which means | |
| 428 | 5881 that we can't change to pop_failure_jump. */ |
| 458 | 5882 if (!not_p) |
| 428 | 5883 { |
| 5884 p[-3] = (unsigned char) pop_failure_jump; | |
| 5885 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); | |
| 5886 } | |
| 5887 } | |
| 5888 } | |
| 5889 else if ((re_opcode_t) *p2 == charset) | |
| 5890 { | |
| 5891 #ifdef DEBUG | |
| 5892 REGISTER unsigned char c | |
| 5893 = *p2 == (unsigned char) endline ? '\n' : p2[2]; | |
| 5894 #endif | |
| 5895 | |
| 5896 if ((re_opcode_t) p1[3] == exactn | |
| 5897 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5] | |
| 5898 && (p2[2 + p1[5] / BYTEWIDTH] | |
| 5899 & (1 << (p1[5] % BYTEWIDTH))))) | |
| 5900 { | |
| 5901 p[-3] = (unsigned char) pop_failure_jump; | |
| 5902 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", | |
| 5903 c, p1[5]); | |
| 5904 } | |
| 5905 | |
| 5906 else if ((re_opcode_t) p1[3] == charset_not) | |
| 5907 { | |
| 5908 int idx; | |
| 5909 /* We win if the charset_not inside the loop | |
| 5910 lists every character listed in the charset after. */ | |
| 5911 for (idx = 0; idx < (int) p2[1]; idx++) | |
| 5912 if (! (p2[2 + idx] == 0 | |
| 5913 || (idx < (int) p1[4] | |
| 5914 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0)))) | |
| 5915 break; | |
| 5916 | |
| 5917 if (idx == p2[1]) | |
| 5918 { | |
| 5919 p[-3] = (unsigned char) pop_failure_jump; | |
| 5920 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); | |
| 5921 } | |
| 5922 } | |
| 5923 else if ((re_opcode_t) p1[3] == charset) | |
| 5924 { | |
| 5925 int idx; | |
| 5926 /* We win if the charset inside the loop | |
| 5927 has no overlap with the one after the loop. */ | |
| 5928 for (idx = 0; | |
| 5929 idx < (int) p2[1] && idx < (int) p1[4]; | |
| 5930 idx++) | |
| 5931 if ((p2[2 + idx] & p1[5 + idx]) != 0) | |
| 5932 break; | |
| 5933 | |
| 5934 if (idx == p2[1] || idx == p1[4]) | |
| 5935 { | |
| 5936 p[-3] = (unsigned char) pop_failure_jump; | |
| 5937 DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); | |
| 5938 } | |
| 5939 } | |
| 5940 } | |
| 5941 } | |
| 5942 p -= 2; /* Point at relative address again. */ | |
| 5943 if ((re_opcode_t) p[-1] != pop_failure_jump) | |
| 5944 { | |
| 5945 p[-1] = (unsigned char) jump; | |
| 5946 DEBUG_PRINT1 (" Match => jump.\n"); | |
| 5947 goto unconditional_jump; | |
| 5948 } | |
| 5949 /* Note fall through. */ | |
| 5950 | |
| 5951 | |
| 5952 /* The end of a simple repeat has a pop_failure_jump back to | |
| 5953 its matching on_failure_jump, where the latter will push a | |
| 5954 failure point. The pop_failure_jump takes off failure | |
| 5955 points put on by this pop_failure_jump's matching | |
| 5956 on_failure_jump; we got through the pattern to here from the | |
| 5957 matching on_failure_jump, so didn't fail. */ | |
| 5958 case pop_failure_jump: | |
| 5959 { | |
| 5960 /* We need to pass separate storage for the lowest and | |
| 5961 highest registers, even though we don't care about the | |
| 5962 actual values. Otherwise, we will restore only one | |
| 5963 register from the stack, since lowest will == highest in | |
| 5964 `pop_failure_point'. */ | |
| 647 | 5965 int dummy_low_reg, dummy_high_reg; |
| 428 | 5966 unsigned char *pdummy; |
| 446 | 5967 re_char *sdummy = NULL; |
| 428 | 5968 |
| 5969 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n"); | |
| 5970 POP_FAILURE_POINT (sdummy, pdummy, | |
| 5971 dummy_low_reg, dummy_high_reg, | |
| 5972 reg_dummy, reg_dummy, reg_info_dummy); | |
| 5973 } | |
| 5974 /* Note fall through. */ | |
| 5975 | |
| 5976 | |
| 5977 /* Unconditionally jump (without popping any failure points). */ | |
| 5978 case jump: | |
| 5979 unconditional_jump: | |
| 5980 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ | |
| 5981 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt); | |
| 5982 p += mcnt; /* Do the jump. */ | |
| 5983 DEBUG_PRINT2 ("(to 0x%lx).\n", (long) p); | |
| 5984 break; | |
| 5985 | |
| 5986 | |
| 5987 /* We need this opcode so we can detect where alternatives end | |
| 5988 in `group_match_null_string_p' et al. */ | |
| 5989 case jump_past_alt: | |
| 5990 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n"); | |
| 5991 goto unconditional_jump; | |
| 5992 | |
| 5993 | |
| 5994 /* Normally, the on_failure_jump pushes a failure point, which | |
| 5995 then gets popped at pop_failure_jump. We will end up at | |
| 5996 pop_failure_jump, also, and with a pattern of, say, `a+', we | |
| 5997 are skipping over the on_failure_jump, so we have to push | |
| 5998 something meaningless for pop_failure_jump to pop. */ | |
| 5999 case dummy_failure_jump: | |
| 6000 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n"); | |
| 6001 /* It doesn't matter what we push for the string here. What | |
| 6002 the code at `fail' tests is the value for the pattern. */ | |
| 446 | 6003 PUSH_FAILURE_POINT ((unsigned char *) 0, (unsigned char *) 0, -2); |
| 428 | 6004 goto unconditional_jump; |
| 6005 | |
| 6006 | |
| 6007 /* At the end of an alternative, we need to push a dummy failure | |
| 6008 point in case we are followed by a `pop_failure_jump', because | |
| 6009 we don't want the failure point for the alternative to be | |
| 6010 popped. For example, matching `(a|ab)*' against `aab' | |
| 6011 requires that we match the `ab' alternative. */ | |
| 6012 case push_dummy_failure: | |
| 6013 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n"); | |
| 6014 /* See comments just above at `dummy_failure_jump' about the | |
| 6015 two zeroes. */ | |
| 446 | 6016 PUSH_FAILURE_POINT ((unsigned char *) 0, (unsigned char *) 0, -2); |
| 428 | 6017 break; |
| 6018 | |
| 6019 /* Have to succeed matching what follows at least n times. | |
| 6020 After that, handle like `on_failure_jump'. */ | |
| 6021 case succeed_n: | |
| 6022 EXTRACT_NUMBER (mcnt, p + 2); | |
| 6023 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); | |
| 6024 | |
| 6025 assert (mcnt >= 0); | |
| 6026 /* Originally, this is how many times we HAVE to succeed. */ | |
| 6027 if (mcnt > 0) | |
| 6028 { | |
| 6029 mcnt--; | |
| 6030 p += 2; | |
| 6031 STORE_NUMBER_AND_INCR (p, mcnt); | |
| 6032 DEBUG_PRINT3 (" Setting 0x%lx to %d.\n", (long) p, mcnt); | |
| 6033 } | |
| 6034 else if (mcnt == 0) | |
| 6035 { | |
| 6036 DEBUG_PRINT2 (" Setting two bytes from 0x%lx to no_op.\n", | |
| 6037 (long) (p+2)); | |
| 6038 p[2] = (unsigned char) no_op; | |
| 6039 p[3] = (unsigned char) no_op; | |
| 6040 goto on_failure; | |
| 6041 } | |
| 6042 break; | |
| 6043 | |
| 6044 case jump_n: | |
| 6045 EXTRACT_NUMBER (mcnt, p + 2); | |
| 6046 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); | |
| 6047 | |
| 6048 /* Originally, this is how many times we CAN jump. */ | |
| 6049 if (mcnt) | |
| 6050 { | |
| 6051 mcnt--; | |
| 6052 STORE_NUMBER (p + 2, mcnt); | |
| 6053 goto unconditional_jump; | |
| 6054 } | |
| 6055 /* If don't have to jump any more, skip over the rest of command. */ | |
| 6056 else | |
| 6057 p += 4; | |
| 6058 break; | |
| 6059 | |
| 6060 case set_number_at: | |
| 6061 { | |
| 6062 DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); | |
| 6063 | |
| 6064 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 6065 p1 = p + mcnt; | |
| 6066 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 6067 DEBUG_PRINT3 (" Setting 0x%lx to %d.\n", (long) p1, mcnt); | |
| 6068 STORE_NUMBER (p1, mcnt); | |
| 6069 break; | |
| 6070 } | |
| 6071 | |
| 6072 case wordbound: | |
| 6073 DEBUG_PRINT1 ("EXECUTING wordbound.\n"); | |
| 6074 should_succeed = 1; | |
| 6075 matchwordbound: | |
| 6076 { | |
| 6077 /* XEmacs change */ | |
| 1377 | 6078 /* Straightforward and (I hope) correct implementation. |
| 6079 Probably should be optimized by arranging to compute | |
| 1497 | 6080 charpos only once. */ |
| 1377 | 6081 /* emch1 is the character before d, syn1 is the syntax of |
| 6082 emch1, emch2 is the character at d, and syn2 is the | |
| 6083 syntax of emch2. */ | |
| 6084 Ichar emch1, emch2; | |
| 1468 | 6085 int syn1 = 0, |
| 6086 syn2 = 0; | |
| 1377 | 6087 re_char *d_before, *d_after; |
| 6088 int result, | |
| 6089 at_beg = AT_STRINGS_BEG (d), | |
| 6090 at_end = AT_STRINGS_END (d); | |
| 6091 #ifdef emacs | |
| 1497 | 6092 Charxpos charpos; |
| 1377 | 6093 #endif |
| 6094 | |
| 6095 if (at_beg && at_end) | |
| 6096 { | |
| 6097 result = 0; | |
| 6098 } | |
| 428 | 6099 else |
| 6100 { | |
| 1377 | 6101 if (!at_beg) |
| 6102 { | |
| 6103 d_before = POS_BEFORE_GAP_UNSAFE (d); | |
| 6104 DEC_IBYTEPTR_FMT (d_before, fmt); | |
| 6105 emch1 = itext_ichar_fmt (d_before, fmt, lispobj); | |
| 460 | 6106 #ifdef emacs |
| 1497 | 6107 charpos = offset_to_charxpos (lispobj, |
| 6108 PTR_TO_OFFSET (d)) - 1; | |
| 1377 | 6109 BEGIN_REGEX_MALLOC_OK (); |
| 1497 | 6110 UPDATE_SYNTAX_CACHE (scache, charpos); |
| 460 | 6111 #endif |
| 1377 | 6112 syn1 = SYNTAX_FROM_CACHE (scache, emch1); |
| 6113 END_REGEX_MALLOC_OK (); | |
| 6114 } | |
| 6115 if (!at_end) | |
| 6116 { | |
| 6117 d_after = POS_AFTER_GAP_UNSAFE (d); | |
| 6118 emch2 = itext_ichar_fmt (d_after, fmt, lispobj); | |
| 460 | 6119 #ifdef emacs |
| 1497 | 6120 charpos = offset_to_charxpos (lispobj, PTR_TO_OFFSET (d)); |
| 1377 | 6121 BEGIN_REGEX_MALLOC_OK (); |
| 1497 | 6122 UPDATE_SYNTAX_CACHE_FORWARD (scache, charpos); |
| 460 | 6123 #endif |
| 1377 | 6124 syn2 = SYNTAX_FROM_CACHE (scache, emch2); |
| 6125 END_REGEX_MALLOC_OK (); | |
| 6126 } | |
| 1333 | 6127 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); |
| 1377 | 6128 |
| 6129 if (at_beg) | |
| 6130 result = (syn2 == Sword); | |
| 6131 else if (at_end) | |
| 6132 result = (syn1 == Sword); | |
| 6133 else | |
| 6134 result = ((syn1 == Sword) != (syn2 == Sword)); | |
| 428 | 6135 } |
| 1377 | 6136 |
| 428 | 6137 if (result == should_succeed) |
| 6138 break; | |
| 6139 goto fail; | |
| 6140 } | |
| 6141 | |
| 6142 case notwordbound: | |
| 6143 DEBUG_PRINT1 ("EXECUTING notwordbound.\n"); | |
| 6144 should_succeed = 0; | |
| 6145 goto matchwordbound; | |
| 6146 | |
| 6147 case wordbeg: | |
| 6148 DEBUG_PRINT1 ("EXECUTING wordbeg.\n"); | |
| 460 | 6149 if (AT_STRINGS_END (d)) |
| 6150 goto fail; | |
| 428 | 6151 { |
| 6152 /* XEmacs: this originally read: | |
| 6153 | |
| 6154 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1))) | |
| 6155 break; | |
| 6156 | |
| 6157 */ | |
| 460 | 6158 re_char *dtmp = POS_AFTER_GAP_UNSAFE (d); |
| 867 | 6159 Ichar emch = itext_ichar_fmt (dtmp, fmt, lispobj); |
| 1333 | 6160 int tempres; |
| 1347 | 6161 #ifdef emacs |
| 6162 Charxpos charpos = offset_to_charxpos (lispobj, PTR_TO_OFFSET (d)); | |
| 6163 #endif | |
| 1333 | 6164 BEGIN_REGEX_MALLOC_OK (); |
| 460 | 6165 #ifdef emacs |
| 826 | 6166 UPDATE_SYNTAX_CACHE (scache, charpos); |
| 460 | 6167 #endif |
| 1333 | 6168 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); |
| 6169 END_REGEX_MALLOC_OK (); | |
| 6170 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6171 if (tempres) | |
| 428 | 6172 goto fail; |
| 6173 if (AT_STRINGS_BEG (d)) | |
| 6174 break; | |
| 460 | 6175 dtmp = POS_BEFORE_GAP_UNSAFE (d); |
| 867 | 6176 DEC_IBYTEPTR_FMT (dtmp, fmt); |
| 6177 emch = itext_ichar_fmt (dtmp, fmt, lispobj); | |
| 1333 | 6178 BEGIN_REGEX_MALLOC_OK (); |
| 460 | 6179 #ifdef emacs |
| 826 | 6180 UPDATE_SYNTAX_CACHE_BACKWARD (scache, charpos - 1); |
| 460 | 6181 #endif |
| 1333 | 6182 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); |
| 6183 END_REGEX_MALLOC_OK (); | |
| 6184 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6185 if (tempres) | |
| 428 | 6186 break; |
| 6187 goto fail; | |
| 6188 } | |
| 6189 | |
| 6190 case wordend: | |
| 6191 DEBUG_PRINT1 ("EXECUTING wordend.\n"); | |
| 460 | 6192 if (AT_STRINGS_BEG (d)) |
| 6193 goto fail; | |
| 428 | 6194 { |
| 6195 /* XEmacs: this originally read: | |
| 6196 | |
| 6197 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1) | |
| 6198 && (!WORDCHAR_P (d) || AT_STRINGS_END (d))) | |
| 6199 break; | |
| 6200 | |
| 6201 The or condition is incorrect (reversed). | |
| 6202 */ | |
| 460 | 6203 re_char *dtmp; |
| 867 | 6204 Ichar emch; |
| 1333 | 6205 int tempres; |
| 460 | 6206 #ifdef emacs |
| 826 | 6207 Charxpos charpos = offset_to_charxpos (lispobj, PTR_TO_OFFSET (d)); |
| 1347 | 6208 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 6209 UPDATE_SYNTAX_CACHE (scache, charpos); |
| 1333 | 6210 END_REGEX_MALLOC_OK (); |
| 6211 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 1347 | 6212 #endif |
| 460 | 6213 dtmp = POS_BEFORE_GAP_UNSAFE (d); |
| 867 | 6214 DEC_IBYTEPTR_FMT (dtmp, fmt); |
| 6215 emch = itext_ichar_fmt (dtmp, fmt, lispobj); | |
| 1333 | 6216 BEGIN_REGEX_MALLOC_OK (); |
| 6217 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); | |
| 6218 END_REGEX_MALLOC_OK (); | |
| 6219 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6220 if (tempres) | |
| 428 | 6221 goto fail; |
| 6222 if (AT_STRINGS_END (d)) | |
| 6223 break; | |
| 460 | 6224 dtmp = POS_AFTER_GAP_UNSAFE (d); |
| 867 | 6225 emch = itext_ichar_fmt (dtmp, fmt, lispobj); |
| 1333 | 6226 BEGIN_REGEX_MALLOC_OK (); |
| 460 | 6227 #ifdef emacs |
| 826 | 6228 UPDATE_SYNTAX_CACHE_FORWARD (scache, charpos + 1); |
| 460 | 6229 #endif |
| 1333 | 6230 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); |
| 6231 END_REGEX_MALLOC_OK (); | |
| 6232 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6233 if (tempres) | |
| 428 | 6234 break; |
| 6235 goto fail; | |
| 6236 } | |
| 6237 | |
| 6238 #ifdef emacs | |
| 6239 case before_dot: | |
| 6240 DEBUG_PRINT1 ("EXECUTING before_dot.\n"); | |
| 826 | 6241 if (!BUFFERP (lispobj) |
| 6242 || (BUF_PTR_BYTE_POS (XBUFFER (lispobj), (unsigned char *) d) | |
| 6243 >= BUF_PT (XBUFFER (lispobj)))) | |
| 428 | 6244 goto fail; |
| 6245 break; | |
| 6246 | |
| 6247 case at_dot: | |
| 6248 DEBUG_PRINT1 ("EXECUTING at_dot.\n"); | |
| 826 | 6249 if (!BUFFERP (lispobj) |
| 6250 || (BUF_PTR_BYTE_POS (XBUFFER (lispobj), (unsigned char *) d) | |
| 6251 != BUF_PT (XBUFFER (lispobj)))) | |
| 428 | 6252 goto fail; |
| 6253 break; | |
| 6254 | |
| 6255 case after_dot: | |
| 6256 DEBUG_PRINT1 ("EXECUTING after_dot.\n"); | |
| 826 | 6257 if (!BUFFERP (lispobj) |
| 6258 || (BUF_PTR_BYTE_POS (XBUFFER (lispobj), (unsigned char *) d) | |
| 6259 <= BUF_PT (XBUFFER (lispobj)))) | |
| 428 | 6260 goto fail; |
| 6261 break; | |
| 6262 | |
| 6263 case syntaxspec: | |
| 6264 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); | |
| 6265 mcnt = *p++; | |
| 6266 goto matchsyntax; | |
| 6267 | |
| 6268 case wordchar: | |
| 6269 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n"); | |
| 6270 mcnt = (int) Sword; | |
| 6271 matchsyntax: | |
| 6272 should_succeed = 1; | |
| 6273 matchornotsyntax: | |
| 6274 { | |
| 6275 int matches; | |
| 867 | 6276 Ichar emch; |
| 428 | 6277 |
| 450 | 6278 REGEX_PREFETCH (); |
| 1333 | 6279 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 6280 UPDATE_SYNTAX_CACHE |
| 6281 (scache, offset_to_charxpos (lispobj, PTR_TO_OFFSET (d))); | |
| 1333 | 6282 END_REGEX_MALLOC_OK (); |
| 6283 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 826 | 6284 |
| 867 | 6285 emch = itext_ichar_fmt (d, fmt, lispobj); |
| 1333 | 6286 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 6287 matches = (SYNTAX_FROM_CACHE (scache, emch) == |
| 6288 (enum syntaxcode) mcnt); | |
| 1333 | 6289 END_REGEX_MALLOC_OK (); |
| 6290 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 867 | 6291 INC_IBYTEPTR_FMT (d, fmt); |
| 428 | 6292 if (matches != should_succeed) |
| 6293 goto fail; | |
| 6294 SET_REGS_MATCHED (); | |
| 6295 } | |
| 6296 break; | |
| 6297 | |
| 6298 case notsyntaxspec: | |
| 6299 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); | |
| 6300 mcnt = *p++; | |
| 6301 goto matchnotsyntax; | |
| 6302 | |
| 6303 case notwordchar: | |
| 6304 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n"); | |
| 6305 mcnt = (int) Sword; | |
| 6306 matchnotsyntax: | |
| 6307 should_succeed = 0; | |
| 6308 goto matchornotsyntax; | |
| 6309 | |
| 6310 #ifdef MULE | |
| 6311 /* 97/2/17 jhod Mule category code patch */ | |
| 6312 case categoryspec: | |
| 6313 should_succeed = 1; | |
| 6314 matchornotcategory: | |
| 6315 { | |
| 867 | 6316 Ichar emch; |
| 428 | 6317 |
| 6318 mcnt = *p++; | |
| 450 | 6319 REGEX_PREFETCH (); |
| 867 | 6320 emch = itext_ichar_fmt (d, fmt, lispobj); |
| 6321 INC_IBYTEPTR_FMT (d, fmt); | |
| 826 | 6322 if (check_category_char (emch, BUFFER_CATEGORY_TABLE (lispbuf), |
| 6323 mcnt, should_succeed)) | |
| 428 | 6324 goto fail; |
| 6325 SET_REGS_MATCHED (); | |
| 6326 } | |
| 6327 break; | |
| 6328 | |
| 6329 case notcategoryspec: | |
| 6330 should_succeed = 0; | |
| 6331 goto matchornotcategory; | |
| 6332 /* end of category patch */ | |
| 6333 #endif /* MULE */ | |
| 6334 #else /* not emacs */ | |
| 6335 case wordchar: | |
| 6336 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n"); | |
| 450 | 6337 REGEX_PREFETCH (); |
| 826 | 6338 if (!WORDCHAR_P ((int) (*d))) |
| 428 | 6339 goto fail; |
| 6340 SET_REGS_MATCHED (); | |
| 6341 d++; | |
| 6342 break; | |
| 6343 | |
| 6344 case notwordchar: | |
| 6345 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); | |
| 450 | 6346 REGEX_PREFETCH (); |
| 826 | 6347 if (!WORDCHAR_P ((int) (*d))) |
| 428 | 6348 goto fail; |
| 6349 SET_REGS_MATCHED (); | |
| 6350 d++; | |
| 6351 break; | |
| 446 | 6352 #endif /* emacs */ |
| 428 | 6353 |
| 6354 default: | |
| 2500 | 6355 ABORT (); |
| 428 | 6356 } |
| 6357 continue; /* Successfully executed one pattern command; keep going. */ | |
| 6358 | |
| 6359 | |
| 6360 /* We goto here if a matching operation fails. */ | |
| 6361 fail: | |
| 6362 if (!FAIL_STACK_EMPTY ()) | |
| 6363 { /* A restart point is known. Restore to that state. */ | |
| 6364 DEBUG_PRINT1 ("\nFAIL:\n"); | |
| 6365 POP_FAILURE_POINT (d, p, | |
| 6366 lowest_active_reg, highest_active_reg, | |
| 6367 regstart, regend, reg_info); | |
| 6368 | |
| 6369 /* If this failure point is a dummy, try the next one. */ | |
| 6370 if (!p) | |
| 6371 goto fail; | |
| 6372 | |
| 6373 /* If we failed to the end of the pattern, don't examine *p. */ | |
| 6374 assert (p <= pend); | |
| 6375 if (p < pend) | |
| 6376 { | |
| 460 | 6377 re_bool is_a_jump_n = false; |
| 428 | 6378 |
| 6379 /* If failed to a backwards jump that's part of a repetition | |
| 6380 loop, need to pop this failure point and use the next one. */ | |
| 6381 switch ((re_opcode_t) *p) | |
| 6382 { | |
| 6383 case jump_n: | |
| 6384 is_a_jump_n = true; | |
| 6385 case maybe_pop_jump: | |
| 6386 case pop_failure_jump: | |
| 6387 case jump: | |
| 6388 p1 = p + 1; | |
| 6389 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6390 p1 += mcnt; | |
| 6391 | |
| 6392 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) | |
| 6393 || (!is_a_jump_n | |
| 6394 && (re_opcode_t) *p1 == on_failure_jump)) | |
| 6395 goto fail; | |
| 6396 break; | |
| 6397 default: | |
| 6398 /* do nothing */ ; | |
| 6399 } | |
| 6400 } | |
| 6401 | |
| 6402 if (d >= string1 && d <= end1) | |
| 6403 dend = end_match_1; | |
| 6404 } | |
| 6405 else | |
| 6406 break; /* Matching at this starting point really fails. */ | |
| 6407 } /* for (;;) */ | |
| 6408 | |
| 6409 if (best_regs_set) | |
| 6410 goto restore_best_regs; | |
| 6411 | |
| 6412 FREE_VARIABLES (); | |
| 6413 | |
| 6414 return -1; /* Failure to match. */ | |
| 1333 | 6415 } /* re_match_2_internal */ |
| 428 | 6416 |
| 6417 /* Subroutine definitions for re_match_2. */ | |
| 6418 | |
| 6419 | |
| 6420 /* We are passed P pointing to a register number after a start_memory. | |
| 6421 | |
| 6422 Return true if the pattern up to the corresponding stop_memory can | |
| 6423 match the empty string, and false otherwise. | |
| 6424 | |
| 6425 If we find the matching stop_memory, sets P to point to one past its number. | |
| 6426 Otherwise, sets P to an undefined byte less than or equal to END. | |
| 6427 | |
| 6428 We don't handle duplicates properly (yet). */ | |
| 6429 | |
| 460 | 6430 static re_bool |
| 428 | 6431 group_match_null_string_p (unsigned char **p, unsigned char *end, |
| 6432 register_info_type *reg_info) | |
| 6433 { | |
| 6434 int mcnt; | |
| 6435 /* Point to after the args to the start_memory. */ | |
| 6436 unsigned char *p1 = *p + 2; | |
| 6437 | |
| 6438 while (p1 < end) | |
| 6439 { | |
| 6440 /* Skip over opcodes that can match nothing, and return true or | |
| 6441 false, as appropriate, when we get to one that can't, or to the | |
| 6442 matching stop_memory. */ | |
| 6443 | |
| 6444 switch ((re_opcode_t) *p1) | |
| 6445 { | |
| 6446 /* Could be either a loop or a series of alternatives. */ | |
| 6447 case on_failure_jump: | |
| 6448 p1++; | |
| 6449 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6450 | |
| 6451 /* If the next operation is not a jump backwards in the | |
| 6452 pattern. */ | |
| 6453 | |
| 6454 if (mcnt >= 0) | |
| 6455 { | |
| 6456 /* Go through the on_failure_jumps of the alternatives, | |
| 6457 seeing if any of the alternatives cannot match nothing. | |
| 6458 The last alternative starts with only a jump, | |
| 6459 whereas the rest start with on_failure_jump and end | |
| 6460 with a jump, e.g., here is the pattern for `a|b|c': | |
| 6461 | |
| 6462 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 | |
| 6463 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 | |
| 6464 /exactn/1/c | |
| 6465 | |
| 6466 So, we have to first go through the first (n-1) | |
| 6467 alternatives and then deal with the last one separately. */ | |
| 6468 | |
| 6469 | |
| 6470 /* Deal with the first (n-1) alternatives, which start | |
| 6471 with an on_failure_jump (see above) that jumps to right | |
| 6472 past a jump_past_alt. */ | |
| 6473 | |
| 6474 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt) | |
| 6475 { | |
| 6476 /* `mcnt' holds how many bytes long the alternative | |
| 6477 is, including the ending `jump_past_alt' and | |
| 6478 its number. */ | |
| 6479 | |
| 6480 if (!alt_match_null_string_p (p1, p1 + mcnt - 3, | |
| 6481 reg_info)) | |
| 6482 return false; | |
| 6483 | |
| 6484 /* Move to right after this alternative, including the | |
| 6485 jump_past_alt. */ | |
| 6486 p1 += mcnt; | |
| 6487 | |
| 6488 /* Break if it's the beginning of an n-th alternative | |
| 6489 that doesn't begin with an on_failure_jump. */ | |
| 6490 if ((re_opcode_t) *p1 != on_failure_jump) | |
| 6491 break; | |
| 6492 | |
| 6493 /* Still have to check that it's not an n-th | |
| 6494 alternative that starts with an on_failure_jump. */ | |
| 6495 p1++; | |
| 6496 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6497 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt) | |
| 6498 { | |
| 6499 /* Get to the beginning of the n-th alternative. */ | |
| 6500 p1 -= 3; | |
| 6501 break; | |
| 6502 } | |
| 6503 } | |
| 6504 | |
| 6505 /* Deal with the last alternative: go back and get number | |
| 6506 of the `jump_past_alt' just before it. `mcnt' contains | |
| 6507 the length of the alternative. */ | |
| 6508 EXTRACT_NUMBER (mcnt, p1 - 2); | |
| 6509 | |
| 6510 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info)) | |
| 6511 return false; | |
| 6512 | |
| 6513 p1 += mcnt; /* Get past the n-th alternative. */ | |
| 6514 } /* if mcnt > 0 */ | |
| 6515 break; | |
| 6516 | |
| 6517 | |
| 6518 case stop_memory: | |
| 6519 assert (p1[1] == **p); | |
| 6520 *p = p1 + 2; | |
| 6521 return true; | |
| 6522 | |
| 6523 | |
| 6524 default: | |
| 6525 if (!common_op_match_null_string_p (&p1, end, reg_info)) | |
| 6526 return false; | |
| 6527 } | |
| 6528 } /* while p1 < end */ | |
| 6529 | |
| 6530 return false; | |
| 6531 } /* group_match_null_string_p */ | |
| 6532 | |
| 6533 | |
| 6534 /* Similar to group_match_null_string_p, but doesn't deal with alternatives: | |
| 6535 It expects P to be the first byte of a single alternative and END one | |
| 6536 byte past the last. The alternative can contain groups. */ | |
| 6537 | |
| 460 | 6538 static re_bool |
| 428 | 6539 alt_match_null_string_p (unsigned char *p, unsigned char *end, |
| 6540 register_info_type *reg_info) | |
| 6541 { | |
| 6542 int mcnt; | |
| 6543 unsigned char *p1 = p; | |
| 6544 | |
| 6545 while (p1 < end) | |
| 6546 { | |
| 6547 /* Skip over opcodes that can match nothing, and break when we get | |
| 6548 to one that can't. */ | |
| 6549 | |
| 6550 switch ((re_opcode_t) *p1) | |
| 6551 { | |
| 6552 /* It's a loop. */ | |
| 6553 case on_failure_jump: | |
| 6554 p1++; | |
| 6555 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6556 p1 += mcnt; | |
| 6557 break; | |
| 6558 | |
| 6559 default: | |
| 6560 if (!common_op_match_null_string_p (&p1, end, reg_info)) | |
| 6561 return false; | |
| 6562 } | |
| 6563 } /* while p1 < end */ | |
| 6564 | |
| 6565 return true; | |
| 6566 } /* alt_match_null_string_p */ | |
| 6567 | |
| 6568 | |
| 6569 /* Deals with the ops common to group_match_null_string_p and | |
| 6570 alt_match_null_string_p. | |
| 6571 | |
| 6572 Sets P to one after the op and its arguments, if any. */ | |
| 6573 | |
| 460 | 6574 static re_bool |
| 428 | 6575 common_op_match_null_string_p (unsigned char **p, unsigned char *end, |
| 6576 register_info_type *reg_info) | |
| 6577 { | |
| 6578 int mcnt; | |
| 460 | 6579 re_bool ret; |
| 428 | 6580 int reg_no; |
| 6581 unsigned char *p1 = *p; | |
| 6582 | |
| 6583 switch ((re_opcode_t) *p1++) | |
| 6584 { | |
| 6585 case no_op: | |
| 6586 case begline: | |
| 6587 case endline: | |
| 6588 case begbuf: | |
| 6589 case endbuf: | |
| 6590 case wordbeg: | |
| 6591 case wordend: | |
| 6592 case wordbound: | |
| 6593 case notwordbound: | |
| 6594 #ifdef emacs | |
| 6595 case before_dot: | |
| 6596 case at_dot: | |
| 6597 case after_dot: | |
| 6598 #endif | |
| 6599 break; | |
| 6600 | |
| 6601 case start_memory: | |
| 6602 reg_no = *p1; | |
| 6603 assert (reg_no > 0 && reg_no <= MAX_REGNUM); | |
| 6604 ret = group_match_null_string_p (&p1, end, reg_info); | |
| 6605 | |
| 6606 /* Have to set this here in case we're checking a group which | |
| 6607 contains a group and a back reference to it. */ | |
| 6608 | |
| 6609 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE) | |
| 6610 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret; | |
| 6611 | |
| 6612 if (!ret) | |
| 6613 return false; | |
| 6614 break; | |
| 6615 | |
| 6616 /* If this is an optimized succeed_n for zero times, make the jump. */ | |
| 6617 case jump: | |
| 6618 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6619 if (mcnt >= 0) | |
| 6620 p1 += mcnt; | |
| 6621 else | |
| 6622 return false; | |
| 6623 break; | |
| 6624 | |
| 6625 case succeed_n: | |
| 6626 /* Get to the number of times to succeed. */ | |
| 6627 p1 += 2; | |
| 6628 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6629 | |
| 6630 if (mcnt == 0) | |
| 6631 { | |
| 6632 p1 -= 4; | |
| 6633 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6634 p1 += mcnt; | |
| 6635 } | |
| 6636 else | |
| 6637 return false; | |
| 6638 break; | |
| 6639 | |
| 6640 case duplicate: | |
| 6641 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) | |
| 6642 return false; | |
| 6643 break; | |
| 6644 | |
| 6645 case set_number_at: | |
| 6646 p1 += 4; | |
| 6647 | |
| 6648 default: | |
| 6649 /* All other opcodes mean we cannot match the empty string. */ | |
| 6650 return false; | |
| 6651 } | |
| 6652 | |
| 6653 *p = p1; | |
| 6654 return true; | |
| 6655 } /* common_op_match_null_string_p */ | |
| 6656 | |
| 6657 | |
| 6658 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN | |
| 6659 bytes; nonzero otherwise. */ | |
| 6660 | |
| 6661 static int | |
| 446 | 6662 bcmp_translate (re_char *s1, re_char *s2, |
| 826 | 6663 REGISTER int len, RE_TRANSLATE_TYPE translate |
| 6664 #ifdef emacs | |
| 2333 | 6665 , Internal_Format USED_IF_MULE (fmt), |
| 6666 Lisp_Object USED_IF_MULE (lispobj) | |
| 826 | 6667 #endif |
| 6668 ) | |
| 428 | 6669 { |
| 826 | 6670 REGISTER re_char *p1 = s1, *p2 = s2; |
| 446 | 6671 #ifdef MULE |
| 826 | 6672 re_char *p1_end = s1 + len; |
| 6673 re_char *p2_end = s2 + len; | |
| 446 | 6674 |
| 6675 while (p1 != p1_end && p2 != p2_end) | |
| 6676 { | |
| 867 | 6677 Ichar p1_ch, p2_ch; |
| 6678 | |
| 6679 p1_ch = itext_ichar_fmt (p1, fmt, lispobj); | |
| 6680 p2_ch = itext_ichar_fmt (p2, fmt, lispobj); | |
| 826 | 6681 |
| 6682 if (RE_TRANSLATE_1 (p1_ch) | |
| 6683 != RE_TRANSLATE_1 (p2_ch)) | |
| 446 | 6684 return 1; |
| 867 | 6685 INC_IBYTEPTR_FMT (p1, fmt); |
| 6686 INC_IBYTEPTR_FMT (p2, fmt); | |
| 446 | 6687 } |
| 6688 #else /* not MULE */ | |
| 428 | 6689 while (len) |
| 6690 { | |
| 826 | 6691 if (RE_TRANSLATE_1 (*p1++) != RE_TRANSLATE_1 (*p2++)) return 1; |
| 428 | 6692 len--; |
| 6693 } | |
| 446 | 6694 #endif /* MULE */ |
| 428 | 6695 return 0; |
| 6696 } | |
| 6697 | |
| 6698 /* Entry points for GNU code. */ | |
| 6699 | |
| 6700 /* re_compile_pattern is the GNU regular expression compiler: it | |
| 6701 compiles PATTERN (of length SIZE) and puts the result in BUFP. | |
| 6702 Returns 0 if the pattern was valid, otherwise an error string. | |
| 6703 | |
| 6704 Assumes the `allocated' (and perhaps `buffer') and `translate' fields | |
| 6705 are set in BUFP on entry. | |
| 6706 | |
| 6707 We call regex_compile to do the actual compilation. */ | |
| 6708 | |
| 442 | 6709 const char * |
| 6710 re_compile_pattern (const char *pattern, int length, | |
| 428 | 6711 struct re_pattern_buffer *bufp) |
| 6712 { | |
| 6713 reg_errcode_t ret; | |
| 6714 | |
| 6715 /* GNU code is written to assume at least RE_NREGS registers will be set | |
| 6716 (and at least one extra will be -1). */ | |
| 6717 bufp->regs_allocated = REGS_UNALLOCATED; | |
| 6718 | |
| 6719 /* And GNU code determines whether or not to get register information | |
| 6720 by passing null for the REGS argument to re_match, etc., not by | |
| 6721 setting no_sub. */ | |
| 6722 bufp->no_sub = 0; | |
| 6723 | |
| 6724 /* Match anchors at newline. */ | |
| 6725 bufp->newline_anchor = 1; | |
| 6726 | |
| 826 | 6727 ret = regex_compile ((unsigned char *) pattern, length, re_syntax_options, |
| 6728 bufp); | |
| 428 | 6729 |
| 6730 if (!ret) | |
| 6731 return NULL; | |
| 6732 return gettext (re_error_msgid[(int) ret]); | |
| 6733 } | |
| 6734 | |
| 6735 /* Entry points compatible with 4.2 BSD regex library. We don't define | |
| 6736 them unless specifically requested. */ | |
| 6737 | |
| 6738 #ifdef _REGEX_RE_COMP | |
| 6739 | |
| 6740 /* BSD has one and only one pattern buffer. */ | |
| 6741 static struct re_pattern_buffer re_comp_buf; | |
| 6742 | |
| 6743 char * | |
| 442 | 6744 re_comp (const char *s) |
| 428 | 6745 { |
| 6746 reg_errcode_t ret; | |
| 6747 | |
| 6748 if (!s) | |
| 6749 { | |
| 6750 if (!re_comp_buf.buffer) | |
| 6751 return gettext ("No previous regular expression"); | |
| 6752 return 0; | |
| 6753 } | |
| 6754 | |
| 6755 if (!re_comp_buf.buffer) | |
| 6756 { | |
| 1333 | 6757 re_comp_buf.buffer = (unsigned char *) xmalloc (200); |
| 428 | 6758 if (re_comp_buf.buffer == NULL) |
| 6759 return gettext (re_error_msgid[(int) REG_ESPACE]); | |
| 6760 re_comp_buf.allocated = 200; | |
| 6761 | |
| 1333 | 6762 re_comp_buf.fastmap = (char *) xmalloc (1 << BYTEWIDTH); |
| 428 | 6763 if (re_comp_buf.fastmap == NULL) |
| 6764 return gettext (re_error_msgid[(int) REG_ESPACE]); | |
| 6765 } | |
| 6766 | |
| 6767 /* Since `re_exec' always passes NULL for the `regs' argument, we | |
| 6768 don't need to initialize the pattern buffer fields which affect it. */ | |
| 6769 | |
| 6770 /* Match anchors at newlines. */ | |
| 6771 re_comp_buf.newline_anchor = 1; | |
| 6772 | |
| 826 | 6773 ret = regex_compile ((unsigned char *)s, strlen (s), re_syntax_options, |
| 6774 &re_comp_buf); | |
| 428 | 6775 |
| 6776 if (!ret) | |
| 6777 return NULL; | |
| 6778 | |
| 442 | 6779 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */ |
| 428 | 6780 return (char *) gettext (re_error_msgid[(int) ret]); |
| 6781 } | |
| 6782 | |
| 6783 | |
| 6784 int | |
| 442 | 6785 re_exec (const char *s) |
| 428 | 6786 { |
| 442 | 6787 const int len = strlen (s); |
| 428 | 6788 return |
| 6789 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0); | |
| 6790 } | |
| 6791 #endif /* _REGEX_RE_COMP */ | |
| 6792 | |
| 6793 /* POSIX.2 functions. Don't define these for Emacs. */ | |
| 6794 | |
| 6795 #ifndef emacs | |
| 6796 | |
| 6797 /* regcomp takes a regular expression as a string and compiles it. | |
| 6798 | |
| 6799 PREG is a regex_t *. We do not expect any fields to be initialized, | |
| 6800 since POSIX says we shouldn't. Thus, we set | |
| 6801 | |
| 6802 `buffer' to the compiled pattern; | |
| 6803 `used' to the length of the compiled pattern; | |
| 6804 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the | |
| 6805 REG_EXTENDED bit in CFLAGS is set; otherwise, to | |
| 6806 RE_SYNTAX_POSIX_BASIC; | |
| 6807 `newline_anchor' to REG_NEWLINE being set in CFLAGS; | |
| 6808 `fastmap' and `fastmap_accurate' to zero; | |
| 6809 `re_nsub' to the number of subexpressions in PATTERN. | |
| 502 | 6810 (non-shy of course. POSIX probably doesn't know about |
| 6811 shy ones, and in any case they should be invisible.) | |
| 428 | 6812 |
| 6813 PATTERN is the address of the pattern string. | |
| 6814 | |
| 6815 CFLAGS is a series of bits which affect compilation. | |
| 6816 | |
| 6817 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we | |
| 6818 use POSIX basic syntax. | |
| 6819 | |
| 6820 If REG_NEWLINE is set, then . and [^...] don't match newline. | |
| 6821 Also, regexec will try a match beginning after every newline. | |
| 6822 | |
| 6823 If REG_ICASE is set, then we considers upper- and lowercase | |
| 6824 versions of letters to be equivalent when matching. | |
| 6825 | |
| 6826 If REG_NOSUB is set, then when PREG is passed to regexec, that | |
| 6827 routine will report only success or failure, and nothing about the | |
| 6828 registers. | |
| 6829 | |
| 6830 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for | |
| 6831 the return codes and their meanings.) */ | |
| 6832 | |
| 6833 int | |
| 442 | 6834 regcomp (regex_t *preg, const char *pattern, int cflags) |
| 428 | 6835 { |
| 6836 reg_errcode_t ret; | |
| 647 | 6837 unsigned int syntax |
| 428 | 6838 = (cflags & REG_EXTENDED) ? |
| 6839 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC; | |
| 6840 | |
| 6841 /* regex_compile will allocate the space for the compiled pattern. */ | |
| 6842 preg->buffer = 0; | |
| 6843 preg->allocated = 0; | |
| 6844 preg->used = 0; | |
| 6845 | |
| 6846 /* Don't bother to use a fastmap when searching. This simplifies the | |
| 6847 REG_NEWLINE case: if we used a fastmap, we'd have to put all the | |
| 6848 characters after newlines into the fastmap. This way, we just try | |
| 6849 every character. */ | |
| 6850 preg->fastmap = 0; | |
| 6851 | |
| 6852 if (cflags & REG_ICASE) | |
| 6853 { | |
| 647 | 6854 int i; |
| 428 | 6855 |
| 1333 | 6856 preg->translate = (char *) xmalloc (CHAR_SET_SIZE); |
| 428 | 6857 if (preg->translate == NULL) |
| 6858 return (int) REG_ESPACE; | |
| 6859 | |
| 6860 /* Map uppercase characters to corresponding lowercase ones. */ | |
| 6861 for (i = 0; i < CHAR_SET_SIZE; i++) | |
| 6862 preg->translate[i] = ISUPPER (i) ? tolower (i) : i; | |
| 6863 } | |
| 6864 else | |
| 6865 preg->translate = NULL; | |
| 6866 | |
| 6867 /* If REG_NEWLINE is set, newlines are treated differently. */ | |
| 6868 if (cflags & REG_NEWLINE) | |
| 6869 { /* REG_NEWLINE implies neither . nor [^...] match newline. */ | |
| 6870 syntax &= ~RE_DOT_NEWLINE; | |
| 6871 syntax |= RE_HAT_LISTS_NOT_NEWLINE; | |
| 6872 /* It also changes the matching behavior. */ | |
| 6873 preg->newline_anchor = 1; | |
| 6874 } | |
| 6875 else | |
| 6876 preg->newline_anchor = 0; | |
| 6877 | |
| 6878 preg->no_sub = !!(cflags & REG_NOSUB); | |
| 6879 | |
| 6880 /* POSIX says a null character in the pattern terminates it, so we | |
| 6881 can use strlen here in compiling the pattern. */ | |
| 446 | 6882 ret = regex_compile ((unsigned char *) pattern, strlen (pattern), syntax, preg); |
| 428 | 6883 |
| 6884 /* POSIX doesn't distinguish between an unmatched open-group and an | |
| 6885 unmatched close-group: both are REG_EPAREN. */ | |
| 6886 if (ret == REG_ERPAREN) ret = REG_EPAREN; | |
| 6887 | |
| 6888 return (int) ret; | |
| 6889 } | |
| 6890 | |
| 6891 | |
| 6892 /* regexec searches for a given pattern, specified by PREG, in the | |
| 6893 string STRING. | |
| 6894 | |
| 6895 If NMATCH is zero or REG_NOSUB was set in the cflags argument to | |
| 6896 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at | |
| 6897 least NMATCH elements, and we set them to the offsets of the | |
| 6898 corresponding matched substrings. | |
| 6899 | |
| 6900 EFLAGS specifies `execution flags' which affect matching: if | |
| 6901 REG_NOTBOL is set, then ^ does not match at the beginning of the | |
| 6902 string; if REG_NOTEOL is set, then $ does not match at the end. | |
| 6903 | |
| 6904 We return 0 if we find a match and REG_NOMATCH if not. */ | |
| 6905 | |
| 6906 int | |
| 442 | 6907 regexec (const regex_t *preg, const char *string, size_t nmatch, |
| 428 | 6908 regmatch_t pmatch[], int eflags) |
| 6909 { | |
| 6910 int ret; | |
| 6911 struct re_registers regs; | |
| 6912 regex_t private_preg; | |
| 6913 int len = strlen (string); | |
| 460 | 6914 re_bool want_reg_info = !preg->no_sub && nmatch > 0; |
| 428 | 6915 |
| 6916 private_preg = *preg; | |
| 6917 | |
| 6918 private_preg.not_bol = !!(eflags & REG_NOTBOL); | |
| 6919 private_preg.not_eol = !!(eflags & REG_NOTEOL); | |
| 6920 | |
| 6921 /* The user has told us exactly how many registers to return | |
| 6922 information about, via `nmatch'. We have to pass that on to the | |
| 6923 matching routines. */ | |
| 6924 private_preg.regs_allocated = REGS_FIXED; | |
| 6925 | |
| 6926 if (want_reg_info) | |
| 6927 { | |
| 647 | 6928 regs.num_regs = (int) nmatch; |
| 6929 regs.start = TALLOC ((int) nmatch, regoff_t); | |
| 6930 regs.end = TALLOC ((int) nmatch, regoff_t); | |
| 428 | 6931 if (regs.start == NULL || regs.end == NULL) |
| 6932 return (int) REG_NOMATCH; | |
| 6933 } | |
| 6934 | |
| 6935 /* Perform the searching operation. */ | |
| 6936 ret = re_search (&private_preg, string, len, | |
| 6937 /* start: */ 0, /* range: */ len, | |
| 6938 want_reg_info ? ®s : (struct re_registers *) 0); | |
| 6939 | |
| 6940 /* Copy the register information to the POSIX structure. */ | |
| 6941 if (want_reg_info) | |
| 6942 { | |
| 6943 if (ret >= 0) | |
| 6944 { | |
| 647 | 6945 int r; |
| 6946 | |
| 6947 for (r = 0; r < (int) nmatch; r++) | |
| 428 | 6948 { |
| 6949 pmatch[r].rm_so = regs.start[r]; | |
| 6950 pmatch[r].rm_eo = regs.end[r]; | |
| 6951 } | |
| 6952 } | |
| 6953 | |
| 6954 /* If we needed the temporary register info, free the space now. */ | |
| 1726 | 6955 xfree (regs.start, regoff_t *); |
| 6956 xfree (regs.end, regoff_t *); | |
| 428 | 6957 } |
| 6958 | |
| 6959 /* We want zero return to mean success, unlike `re_search'. */ | |
| 6960 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH; | |
| 6961 } | |
| 6962 | |
| 6963 | |
| 6964 /* Returns a message corresponding to an error code, ERRCODE, returned | |
| 6965 from either regcomp or regexec. We don't use PREG here. */ | |
| 6966 | |
| 6967 size_t | |
| 2286 | 6968 regerror (int errcode, const regex_t *UNUSED (preg), char *errbuf, |
| 647 | 6969 size_t errbuf_size) |
| 428 | 6970 { |
| 442 | 6971 const char *msg; |
| 665 | 6972 Bytecount msg_size; |
| 428 | 6973 |
| 6974 if (errcode < 0 | |
| 647 | 6975 || errcode >= (int) (sizeof (re_error_msgid) / |
| 6976 sizeof (re_error_msgid[0]))) | |
| 428 | 6977 /* Only error codes returned by the rest of the code should be passed |
| 6978 to this routine. If we are given anything else, or if other regex | |
| 6979 code generates an invalid error code, then the program has a bug. | |
| 6980 Dump core so we can fix it. */ | |
| 2500 | 6981 ABORT (); |
| 428 | 6982 |
| 6983 msg = gettext (re_error_msgid[errcode]); | |
| 6984 | |
| 6985 msg_size = strlen (msg) + 1; /* Includes the null. */ | |
| 6986 | |
| 6987 if (errbuf_size != 0) | |
| 6988 { | |
| 665 | 6989 if (msg_size > (Bytecount) errbuf_size) |
| 428 | 6990 { |
| 6991 strncpy (errbuf, msg, errbuf_size - 1); | |
| 6992 errbuf[errbuf_size - 1] = 0; | |
| 6993 } | |
| 6994 else | |
| 6995 strcpy (errbuf, msg); | |
| 6996 } | |
| 6997 | |
| 647 | 6998 return (size_t) msg_size; |
| 428 | 6999 } |
| 7000 | |
| 7001 | |
| 7002 /* Free dynamically allocated space used by PREG. */ | |
| 7003 | |
| 7004 void | |
| 7005 regfree (regex_t *preg) | |
| 7006 { | |
| 7007 if (preg->buffer != NULL) | |
| 1726 | 7008 xfree (preg->buffer, unsigned char *); |
| 428 | 7009 preg->buffer = NULL; |
| 7010 | |
| 7011 preg->allocated = 0; | |
| 7012 preg->used = 0; | |
| 7013 | |
| 7014 if (preg->fastmap != NULL) | |
| 1726 | 7015 xfree (preg->fastmap, char *); |
| 428 | 7016 preg->fastmap = NULL; |
| 7017 preg->fastmap_accurate = 0; | |
| 7018 | |
| 7019 if (preg->translate != NULL) | |
| 1726 | 7020 xfree (preg->translate, RE_TRANSLATE_TYPE); |
| 428 | 7021 preg->translate = NULL; |
| 7022 } | |
| 7023 | |
| 7024 #endif /* not emacs */ | |
| 7025 |