Mercurial > hg > xemacs-beta
annotate src/regex.c @ 5359:f5a5501814f5
Document the CL set functions and #'eql in the Lispref, not just cl.texi
man/ChangeLog addition:
2011-02-19 Aidan Kehoe <kehoea@parhasard.net>
* lispref/lists.texi (Sets And Lists):
Document #'member*, #'remove*, #'delete* in this file. Document
#'memq, #'member, #'remq, #'remove, #'delq, #'delete in terms of
the former functions.
Document #'subsetp, #'union, #'intersection, #'set-difference,
#'set-exclusive-or and their destructive analogues in this file.
* lispref/lists.texi (Association Lists):
Document #'assoc*, #'rassoc* in this file. Document #'assq,
#'assoc, #'rassq, #'rassoc in terms of the first two functions.
* lispref/objects.texi (Equality Predicates):
Document #'eql here, don't leave it to cl.texi.
src/ChangeLog addition:
2011-02-19 Aidan Kehoe <kehoea@parhasard.net>
* fns.c (Fset_exclusive_or):
This function accepts the :stable keyword too, document this in
its arglist.
| author | Aidan Kehoe <kehoea@parhasard.net> |
|---|---|
| date | Sat, 19 Feb 2011 11:03:46 +0000 |
| parents | efaa6cd845e5 |
| children | 308d34e9f07d |
| 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. | |
| 5041 | 8 Copyright (C) 1995, 2001, 2002, 2003, 2010 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 |
| 176 /* Get the interface, including the syntax bits. */ | |
| 177 #include "regex.h" | |
| 178 | |
| 179 /* isalpha etc. are used for the character classes. */ | |
| 180 #include <ctype.h> | |
| 181 | |
| 182 /* Jim Meyering writes: | |
| 183 | |
| 184 "... Some ctype macros are valid only for character codes that | |
| 185 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when | |
| 186 using /bin/cc or gcc but without giving an ansi option). So, all | |
| 187 ctype uses should be through macros like ISPRINT... If | |
| 188 STDC_HEADERS is defined, then autoconf has verified that the ctype | |
| 189 macros don't need to be guarded with references to isascii. ... | |
| 190 Defining isascii to 1 should let any compiler worth its salt | |
| 191 eliminate the && through constant folding." */ | |
| 192 | |
| 193 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII)) | |
| 194 #define ISASCII_1(c) 1 | |
| 195 #else | |
| 196 #define ISASCII_1(c) isascii(c) | |
| 197 #endif | |
| 198 | |
| 199 #ifdef MULE | |
| 200 /* The IS*() macros can be passed any character, including an extended | |
| 201 one. We need to make sure there are no crashes, which would occur | |
| 202 otherwise due to out-of-bounds array references. */ | |
| 203 #define ISASCII(c) (((EMACS_UINT) (c)) < 0x100 && ISASCII_1 (c)) | |
| 204 #else | |
| 205 #define ISASCII(c) ISASCII_1 (c) | |
| 206 #endif /* MULE */ | |
| 207 | |
| 208 #ifdef isblank | |
| 209 #define ISBLANK(c) (ISASCII (c) && isblank (c)) | |
| 210 #else | |
| 211 #define ISBLANK(c) ((c) == ' ' || (c) == '\t') | |
| 212 #endif | |
| 213 #ifdef isgraph | |
| 214 #define ISGRAPH(c) (ISASCII (c) && isgraph (c)) | |
| 215 #else | |
| 216 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c)) | |
| 217 #endif | |
| 218 | |
| 219 #define ISPRINT(c) (ISASCII (c) && isprint (c)) | |
| 220 #define ISDIGIT(c) (ISASCII (c) && isdigit (c)) | |
| 221 #define ISALNUM(c) (ISASCII (c) && isalnum (c)) | |
| 222 #define ISALPHA(c) (ISASCII (c) && isalpha (c)) | |
| 223 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c)) | |
| 224 #define ISLOWER(c) (ISASCII (c) && islower (c)) | |
| 225 #define ISPUNCT(c) (ISASCII (c) && ispunct (c)) | |
| 226 #define ISSPACE(c) (ISASCII (c) && isspace (c)) | |
| 227 #define ISUPPER(c) (ISASCII (c) && isupper (c)) | |
| 228 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c)) | |
| 229 | |
| 230 #ifndef NULL | |
| 231 #define NULL (void *)0 | |
| 232 #endif | |
| 233 | |
| 234 /* We remove any previous definition of `SIGN_EXTEND_CHAR', | |
| 235 since ours (we hope) works properly with all combinations of | |
| 236 machines, compilers, `char' and `unsigned char' argument types. | |
| 237 (Per Bothner suggested the basic approach.) */ | |
| 238 #undef SIGN_EXTEND_CHAR | |
| 239 #if __STDC__ | |
| 240 #define SIGN_EXTEND_CHAR(c) ((signed char) (c)) | |
| 241 #else /* not __STDC__ */ | |
| 242 /* As in Harbison and Steele. */ | |
| 243 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128) | |
| 244 #endif | |
| 245 | |
| 246 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we | |
| 247 use `alloca' instead of `malloc'. This is because using malloc in | |
| 248 re_search* or re_match* could cause memory leaks when C-g is used in | |
| 249 Emacs; also, malloc is slower and causes storage fragmentation. On | |
| 250 the other hand, malloc is more portable, and easier to debug. | |
| 251 | |
| 252 Because we sometimes use alloca, some routines have to be macros, | |
| 253 not functions -- `alloca'-allocated space disappears at the end of the | |
| 254 function it is called in. */ | |
| 255 | |
| 1333 | 256 #ifndef emacs |
| 257 #define ALLOCA alloca | |
| 258 #define xmalloc malloc | |
| 259 #define xrealloc realloc | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
260 #define xfree free |
| 1333 | 261 #endif |
| 262 | |
| 263 #ifdef emacs | |
| 264 #define ALLOCA_GARBAGE_COLLECT() \ | |
| 265 do \ | |
| 266 { \ | |
| 267 if (need_to_check_c_alloca) \ | |
| 268 xemacs_c_alloca (0); \ | |
| 269 } while (0) | |
| 270 #elif defined (C_ALLOCA) | |
| 271 #define ALLOCA_GARBAGE_COLLECT() alloca (0) | |
| 272 #else | |
| 273 #define ALLOCA_GARBAGE_COLLECT() | |
| 274 #endif | |
| 275 | |
| 276 #ifndef emacs | |
| 277 /* So we can use just it to conditionalize on */ | |
| 278 #undef ERROR_CHECK_MALLOC | |
| 279 #endif | |
| 280 | |
| 281 #ifdef ERROR_CHECK_MALLOC | |
| 282 /* When REL_ALLOC, malloc() is problematic because it could potentially | |
| 283 cause all rel-alloc()ed data -- including buffer text -- to be relocated. | |
| 284 We deal with this by checking for such relocation whenever we have | |
| 285 executed a statement that may call malloc() -- or alloca(), which may | |
| 286 end up calling malloc() in some circumstances -- and recomputing all | |
| 287 of our string pointers in re_match_2_internal() and re_search_2(). | |
| 288 However, if malloc() or alloca() happens and we don't know about it, | |
| 289 we could still be screwed. So we set up a system where we indicate all | |
| 290 places where we are prepared for malloc() or alloca(), and in any | |
| 291 other circumstances, calls to those functions (from anywhere inside of | |
| 2500 | 292 XEmacs!) will ABORT(). We do this even when REL_ALLOC is not defined |
| 1333 | 293 so that we catch these problems sooner, since many developers and beta |
| 294 testers will not be running with REL_ALLOC. */ | |
| 295 int regex_malloc_disallowed; | |
| 296 #define BEGIN_REGEX_MALLOC_OK() regex_malloc_disallowed = 0 | |
| 297 #define END_REGEX_MALLOC_OK() regex_malloc_disallowed = 1 | |
| 298 #define UNBIND_REGEX_MALLOC_CHECK() unbind_to (depth) | |
| 299 #else | |
| 300 #define BEGIN_REGEX_MALLOC_OK() | |
| 301 #define END_REGEX_MALLOC_OK() | |
| 302 #define UNBIND_REGEX_MALLOC_CHECK() | |
| 303 #endif | |
| 304 | |
| 305 | |
| 428 | 306 #ifdef REGEX_MALLOC |
| 307 | |
| 1333 | 308 #define REGEX_ALLOCATE xmalloc |
| 309 #define REGEX_REALLOCATE(source, osize, nsize) xrealloc (source, nsize) | |
| 310 #define REGEX_FREE xfree | |
| 428 | 311 |
| 312 #else /* not REGEX_MALLOC */ | |
| 313 | |
| 314 /* Emacs already defines alloca, sometimes. */ | |
| 315 #ifndef alloca | |
| 316 | |
| 317 /* Make alloca work the best possible way. */ | |
| 318 #ifdef __GNUC__ | |
| 319 #define alloca __builtin_alloca | |
| 771 | 320 #elif defined (__DECC) /* XEmacs: added next 3 lines, similar to config.h.in */ |
| 321 #include <alloca.h> | |
| 322 #pragma intrinsic(alloca) | |
| 428 | 323 #else /* not __GNUC__ */ |
| 324 #if HAVE_ALLOCA_H | |
| 325 #include <alloca.h> | |
| 326 #else /* not __GNUC__ or HAVE_ALLOCA_H */ | |
| 327 #ifndef _AIX /* Already did AIX, up at the top. */ | |
| 444 | 328 void *alloca (); |
| 428 | 329 #endif /* not _AIX */ |
| 446 | 330 #endif /* HAVE_ALLOCA_H */ |
| 331 #endif /* __GNUC__ */ | |
| 428 | 332 |
| 333 #endif /* not alloca */ | |
| 334 | |
| 1333 | 335 #define REGEX_ALLOCATE ALLOCA |
| 428 | 336 |
| 2367 | 337 /* !!#### Needs review */ |
| 428 | 338 /* Assumes a `char *destination' variable. */ |
| 339 #define REGEX_REALLOCATE(source, osize, nsize) \ | |
| 1333 | 340 (destination = (char *) ALLOCA (nsize), \ |
| 428 | 341 memmove (destination, source, osize), \ |
| 342 destination) | |
| 343 | |
| 1726 | 344 /* No need to do anything to free, after alloca. |
| 345 Do nothing! But inhibit gcc warning. */ | |
| 346 #define REGEX_FREE(arg,type) ((void)0) | |
| 428 | 347 |
| 446 | 348 #endif /* REGEX_MALLOC */ |
| 428 | 349 |
| 350 /* Define how to allocate the failure stack. */ | |
| 351 | |
| 771 | 352 #ifdef REGEX_REL_ALLOC |
| 428 | 353 #define REGEX_ALLOCATE_STACK(size) \ |
| 1346 | 354 r_alloc ((unsigned char **) &failure_stack_ptr, (size)) |
| 428 | 355 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \ |
| 1346 | 356 r_re_alloc ((unsigned char **) &failure_stack_ptr, (nsize)) |
| 428 | 357 #define REGEX_FREE_STACK(ptr) \ |
| 1346 | 358 r_alloc_free ((unsigned char **) &failure_stack_ptr) |
| 428 | 359 |
| 771 | 360 #else /* not REGEX_REL_ALLOC */ |
| 428 | 361 |
| 362 #ifdef REGEX_MALLOC | |
| 363 | |
| 1333 | 364 #define REGEX_ALLOCATE_STACK xmalloc |
| 365 #define REGEX_REALLOCATE_STACK(source, osize, nsize) xrealloc (source, nsize) | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
366 #define REGEX_FREE_STACK(arg) xfree (arg) |
| 428 | 367 |
| 368 #else /* not REGEX_MALLOC */ | |
| 369 | |
| 1333 | 370 #define REGEX_ALLOCATE_STACK ALLOCA |
| 428 | 371 |
| 372 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \ | |
| 373 REGEX_REALLOCATE (source, osize, nsize) | |
| 374 /* No need to explicitly free anything. */ | |
| 375 #define REGEX_FREE_STACK(arg) | |
| 376 | |
| 446 | 377 #endif /* REGEX_MALLOC */ |
| 771 | 378 #endif /* REGEX_REL_ALLOC */ |
| 428 | 379 |
| 380 | |
| 381 /* True if `size1' is non-NULL and PTR is pointing anywhere inside | |
| 382 `string1' or just past its end. This works if PTR is NULL, which is | |
| 383 a good thing. */ | |
| 384 #define FIRST_STRING_P(ptr) \ | |
| 385 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1) | |
| 386 | |
| 387 /* (Re)Allocate N items of type T using malloc, or fail. */ | |
| 1333 | 388 #define TALLOC(n, t) ((t *) xmalloc ((n) * sizeof (t))) |
| 389 #define RETALLOC(addr, n, t) ((addr) = (t *) xrealloc (addr, (n) * sizeof (t))) | |
| 428 | 390 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) |
| 391 | |
| 392 #define BYTEWIDTH 8 /* In bits. */ | |
| 393 | |
| 434 | 394 #define STREQ(s1, s2) (strcmp (s1, s2) == 0) |
| 428 | 395 |
| 396 #undef MAX | |
| 397 #undef MIN | |
| 398 #define MAX(a, b) ((a) > (b) ? (a) : (b)) | |
| 399 #define MIN(a, b) ((a) < (b) ? (a) : (b)) | |
| 400 | |
| 446 | 401 /* Type of source-pattern and string chars. */ |
| 402 typedef const unsigned char re_char; | |
| 403 | |
| 460 | 404 typedef char re_bool; |
| 428 | 405 #define false 0 |
| 406 #define true 1 | |
| 407 | |
| 408 | |
| 1346 | 409 #ifdef emacs |
| 410 | |
| 411 #ifdef MULE | |
| 412 | |
| 413 Lisp_Object Vthe_lisp_rangetab; | |
| 414 | |
| 415 void | |
| 416 vars_of_regex (void) | |
| 417 { | |
| 2421 | 418 Vthe_lisp_rangetab = Fmake_range_table (Qstart_closed_end_closed); |
| 1346 | 419 staticpro (&Vthe_lisp_rangetab); |
| 420 } | |
| 421 | |
| 422 #else /* not MULE */ | |
| 423 | |
| 424 void | |
| 425 vars_of_regex (void) | |
| 426 { | |
| 427 } | |
| 428 | |
| 429 #endif /* MULE */ | |
| 430 | |
| 431 /* Convert an offset from the start of the logical text string formed by | |
| 432 concatenating the two strings together into a character position in the | |
| 433 Lisp buffer or string that the text represents. Knows that | |
| 434 when handling buffer text, the "string" we're passed in is always | |
| 435 BEGV - ZV. */ | |
| 436 | |
| 437 static Charxpos | |
| 438 offset_to_charxpos (Lisp_Object lispobj, int off) | |
| 439 { | |
| 440 if (STRINGP (lispobj)) | |
| 441 return string_index_byte_to_char (lispobj, off); | |
| 442 else if (BUFFERP (lispobj)) | |
| 443 return bytebpos_to_charbpos (XBUFFER (lispobj), | |
| 444 off + BYTE_BUF_BEGV (XBUFFER (lispobj))); | |
| 445 else | |
| 446 return 0; | |
| 447 } | |
| 448 | |
| 449 #ifdef REL_ALLOC | |
| 450 | |
| 451 /* STRING1 is the value of STRING1 given to re_match_2(). LISPOBJ is | |
| 452 the Lisp object (if any) from which the string is taken. If LISPOBJ | |
| 453 is a buffer, return a relocation offset to be added to all pointers to | |
| 454 string data so that they will be accurate again, after an allocation or | |
| 455 reallocation that potentially relocated the buffer data. | |
| 456 */ | |
| 457 static Bytecount | |
| 458 offset_post_relocation (Lisp_Object lispobj, Ibyte *orig_buftext) | |
| 459 { | |
| 460 if (!BUFFERP (lispobj)) | |
| 461 return 0; | |
| 462 return (BYTE_BUF_BYTE_ADDRESS (XBUFFER (lispobj), | |
| 463 BYTE_BUF_BEGV (XBUFFER (lispobj))) - | |
| 464 orig_buftext); | |
| 465 } | |
| 466 | |
| 467 #endif /* REL_ALLOC */ | |
| 468 | |
| 469 #ifdef ERROR_CHECK_MALLOC | |
| 470 | |
| 471 /* NOTE that this can run malloc() so you need to adjust afterwards. */ | |
| 472 | |
| 473 static int | |
| 474 bind_regex_malloc_disallowed (int value) | |
| 475 { | |
| 476 /* Tricky, because the act of binding can run malloc(). */ | |
| 477 int old_regex_malloc_disallowed = regex_malloc_disallowed; | |
| 478 int depth; | |
| 479 regex_malloc_disallowed = 0; | |
| 480 depth = record_unwind_protect_restoring_int (®ex_malloc_disallowed, | |
| 481 old_regex_malloc_disallowed); | |
| 482 regex_malloc_disallowed = value; | |
| 483 return depth; | |
| 484 } | |
| 485 | |
| 486 #endif /* ERROR_CHECK_MALLOC */ | |
| 487 | |
| 488 #endif /* emacs */ | |
| 489 | |
| 490 | |
| 428 | 491 /* These are the command codes that appear in compiled regular |
| 492 expressions. Some opcodes are followed by argument bytes. A | |
| 493 command code can specify any interpretation whatsoever for its | |
| 494 arguments. Zero bytes may appear in the compiled regular expression. */ | |
| 495 | |
| 496 typedef enum | |
| 497 { | |
| 498 no_op = 0, | |
| 499 | |
| 500 /* Succeed right away--no more backtracking. */ | |
| 501 succeed, | |
| 502 | |
| 503 /* Followed by one byte giving n, then by n literal bytes. */ | |
| 504 exactn, | |
| 505 | |
| 506 /* Matches any (more or less) character. */ | |
| 507 anychar, | |
| 508 | |
| 509 /* Matches any one char belonging to specified set. First | |
| 510 following byte is number of bitmap bytes. Then come bytes | |
| 511 for a bitmap saying which chars are in. Bits in each byte | |
| 512 are ordered low-bit-first. A character is in the set if its | |
| 513 bit is 1. A character too large to have a bit in the map is | |
| 514 automatically not in the set. */ | |
| 515 charset, | |
| 516 | |
| 517 /* Same parameters as charset, but match any character that is | |
| 518 not one of those specified. */ | |
| 519 charset_not, | |
| 520 | |
| 521 /* Start remembering the text that is matched, for storing in a | |
| 522 register. Followed by one byte with the register number, in | |
| 502 | 523 the range 1 to the pattern buffer's re_ngroups |
| 428 | 524 field. Then followed by one byte with the number of groups |
| 525 inner to this one. (This last has to be part of the | |
| 526 start_memory only because we need it in the on_failure_jump | |
| 527 of re_match_2.) */ | |
| 528 start_memory, | |
| 529 | |
| 530 /* Stop remembering the text that is matched and store it in a | |
| 531 memory register. Followed by one byte with the register | |
| 502 | 532 number, in the range 1 to `re_ngroups' in the |
| 428 | 533 pattern buffer, and one byte with the number of inner groups, |
| 534 just like `start_memory'. (We need the number of inner | |
| 535 groups here because we don't have any easy way of finding the | |
| 536 corresponding start_memory when we're at a stop_memory.) */ | |
| 537 stop_memory, | |
| 538 | |
| 539 /* Match a duplicate of something remembered. Followed by one | |
| 540 byte containing the register number. */ | |
| 541 duplicate, | |
| 542 | |
| 543 /* Fail unless at beginning of line. */ | |
| 544 begline, | |
| 545 | |
| 546 /* Fail unless at end of line. */ | |
| 547 endline, | |
| 548 | |
| 549 /* Succeeds if at beginning of buffer (if emacs) or at beginning | |
| 550 of string to be matched (if not). */ | |
| 551 begbuf, | |
| 552 | |
| 553 /* Analogously, for end of buffer/string. */ | |
| 554 endbuf, | |
| 555 | |
| 556 /* Followed by two byte relative address to which to jump. */ | |
| 557 jump, | |
| 558 | |
| 559 /* Same as jump, but marks the end of an alternative. */ | |
| 560 jump_past_alt, | |
| 561 | |
| 562 /* Followed by two-byte relative address of place to resume at | |
| 563 in case of failure. */ | |
| 564 on_failure_jump, | |
| 565 | |
| 566 /* Like on_failure_jump, but pushes a placeholder instead of the | |
| 567 current string position when executed. */ | |
| 568 on_failure_keep_string_jump, | |
| 569 | |
| 570 /* Throw away latest failure point and then jump to following | |
| 571 two-byte relative address. */ | |
| 572 pop_failure_jump, | |
| 573 | |
| 574 /* Change to pop_failure_jump if know won't have to backtrack to | |
| 575 match; otherwise change to jump. This is used to jump | |
| 576 back to the beginning of a repeat. If what follows this jump | |
| 577 clearly won't match what the repeat does, such that we can be | |
| 578 sure that there is no use backtracking out of repetitions | |
| 579 already matched, then we change it to a pop_failure_jump. | |
| 580 Followed by two-byte address. */ | |
| 581 maybe_pop_jump, | |
| 582 | |
| 583 /* Jump to following two-byte address, and push a dummy failure | |
| 584 point. This failure point will be thrown away if an attempt | |
| 585 is made to use it for a failure. A `+' construct makes this | |
| 586 before the first repeat. Also used as an intermediary kind | |
| 587 of jump when compiling an alternative. */ | |
| 588 dummy_failure_jump, | |
| 589 | |
| 590 /* Push a dummy failure point and continue. Used at the end of | |
| 591 alternatives. */ | |
| 592 push_dummy_failure, | |
| 593 | |
| 594 /* Followed by two-byte relative address and two-byte number n. | |
| 595 After matching N times, jump to the address upon failure. */ | |
| 596 succeed_n, | |
| 597 | |
| 598 /* Followed by two-byte relative address, and two-byte number n. | |
| 599 Jump to the address N times, then fail. */ | |
| 600 jump_n, | |
| 601 | |
| 602 /* Set the following two-byte relative address to the | |
| 603 subsequent two-byte number. The address *includes* the two | |
| 604 bytes of number. */ | |
| 605 set_number_at, | |
| 606 | |
| 607 wordchar, /* Matches any word-constituent character. */ | |
| 608 notwordchar, /* Matches any char that is not a word-constituent. */ | |
| 609 | |
| 610 wordbeg, /* Succeeds if at word beginning. */ | |
| 611 wordend, /* Succeeds if at word end. */ | |
| 612 | |
| 613 wordbound, /* Succeeds if at a word boundary. */ | |
| 614 notwordbound /* Succeeds if not at a word boundary. */ | |
| 615 | |
| 616 #ifdef emacs | |
| 617 ,before_dot, /* Succeeds if before point. */ | |
| 618 at_dot, /* Succeeds if at point. */ | |
| 619 after_dot, /* Succeeds if after point. */ | |
| 620 | |
| 621 /* Matches any character whose syntax is specified. Followed by | |
| 622 a byte which contains a syntax code, e.g., Sword. */ | |
| 623 syntaxspec, | |
| 624 | |
| 625 /* Matches any character whose syntax is not that specified. */ | |
| 626 notsyntaxspec | |
| 627 | |
| 628 #endif /* emacs */ | |
| 629 | |
| 630 #ifdef MULE | |
| 631 /* need extra stuff to be able to properly work with XEmacs/Mule | |
| 632 characters (which may take up more than one byte) */ | |
| 633 | |
| 634 ,charset_mule, /* Matches any character belonging to specified set. | |
| 635 The set is stored in "unified range-table | |
| 636 format"; see rangetab.c. Unlike the `charset' | |
| 637 opcode, this can handle arbitrary characters. */ | |
| 638 | |
| 639 charset_mule_not /* Same parameters as charset_mule, but match any | |
| 640 character that is not one of those specified. */ | |
| 641 | |
| 642 /* 97/2/17 jhod: The following two were merged back in from the Mule | |
| 643 2.3 code to enable some language specific processing */ | |
| 644 ,categoryspec, /* Matches entries in the character category tables */ | |
| 645 notcategoryspec /* The opposite of the above */ | |
| 646 #endif /* MULE */ | |
| 647 | |
| 648 } re_opcode_t; | |
| 649 | |
| 650 /* Common operations on the compiled pattern. */ | |
| 651 | |
| 652 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */ | |
| 653 | |
| 654 #define STORE_NUMBER(destination, number) \ | |
| 655 do { \ | |
| 656 (destination)[0] = (number) & 0377; \ | |
| 657 (destination)[1] = (number) >> 8; \ | |
| 658 } while (0) | |
| 659 | |
| 660 /* Same as STORE_NUMBER, except increment DESTINATION to | |
| 661 the byte after where the number is stored. Therefore, DESTINATION | |
| 662 must be an lvalue. */ | |
| 663 | |
| 664 #define STORE_NUMBER_AND_INCR(destination, number) \ | |
| 665 do { \ | |
| 666 STORE_NUMBER (destination, number); \ | |
| 667 (destination) += 2; \ | |
| 668 } while (0) | |
| 669 | |
| 670 /* Put into DESTINATION a number stored in two contiguous bytes starting | |
| 671 at SOURCE. */ | |
| 672 | |
| 673 #define EXTRACT_NUMBER(destination, source) \ | |
| 674 do { \ | |
| 675 (destination) = *(source) & 0377; \ | |
| 676 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \ | |
| 677 } while (0) | |
| 678 | |
| 679 #ifdef DEBUG | |
| 680 static void | |
| 446 | 681 extract_number (int *dest, re_char *source) |
| 428 | 682 { |
| 683 int temp = SIGN_EXTEND_CHAR (*(source + 1)); | |
| 684 *dest = *source & 0377; | |
| 685 *dest += temp << 8; | |
| 686 } | |
| 687 | |
| 688 #ifndef EXTRACT_MACROS /* To debug the macros. */ | |
| 689 #undef EXTRACT_NUMBER | |
| 690 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) | |
| 691 #endif /* not EXTRACT_MACROS */ | |
| 692 | |
| 693 #endif /* DEBUG */ | |
| 694 | |
| 695 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number. | |
| 696 SOURCE must be an lvalue. */ | |
| 697 | |
| 698 #define EXTRACT_NUMBER_AND_INCR(destination, source) \ | |
| 699 do { \ | |
| 700 EXTRACT_NUMBER (destination, source); \ | |
| 701 (source) += 2; \ | |
| 702 } while (0) | |
| 703 | |
| 704 #ifdef DEBUG | |
| 705 static void | |
| 706 extract_number_and_incr (int *destination, unsigned char **source) | |
| 707 { | |
| 708 extract_number (destination, *source); | |
| 709 *source += 2; | |
| 710 } | |
| 711 | |
| 712 #ifndef EXTRACT_MACROS | |
| 713 #undef EXTRACT_NUMBER_AND_INCR | |
| 714 #define EXTRACT_NUMBER_AND_INCR(dest, src) \ | |
| 715 extract_number_and_incr (&dest, &src) | |
| 716 #endif /* not EXTRACT_MACROS */ | |
| 717 | |
| 718 #endif /* DEBUG */ | |
| 719 | |
| 720 /* If DEBUG is defined, Regex prints many voluminous messages about what | |
| 721 it is doing (if the variable `debug' is nonzero). If linked with the | |
| 722 main program in `iregex.c', you can enter patterns and strings | |
| 723 interactively. And if linked with the main program in `main.c' and | |
| 724 the other test files, you can run the already-written tests. */ | |
| 725 | |
| 726 #if defined (DEBUG) | |
| 727 | |
| 728 /* We use standard I/O for debugging. */ | |
| 729 #include <stdio.h> | |
| 730 | |
| 731 #ifndef emacs | |
| 732 /* XEmacs provides its own version of assert() */ | |
| 733 /* It is useful to test things that ``must'' be true when debugging. */ | |
| 734 #include <assert.h> | |
| 735 #endif | |
| 736 | |
| 5041 | 737 extern int debug_regexps; |
| 428 | 738 |
| 739 #define DEBUG_STATEMENT(e) e | |
| 5041 | 740 |
| 741 #define DEBUG_PRINT1(x) if (debug_regexps) printf (x) | |
| 742 #define DEBUG_PRINT2(x1, x2) if (debug_regexps) printf (x1, x2) | |
| 743 #define DEBUG_PRINT3(x1, x2, x3) if (debug_regexps) printf (x1, x2, x3) | |
| 744 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug_regexps) printf (x1, x2, x3, x4) | |
| 428 | 745 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ |
| 5041 | 746 if (debug_regexps) print_partial_compiled_pattern (s, e) |
| 428 | 747 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ |
| 5041 | 748 if (debug_regexps) print_double_string (w, s1, sz1, s2, sz2) |
| 749 | |
| 750 #define DEBUG_FAIL_PRINT1(x) \ | |
| 751 if (debug_regexps & RE_DEBUG_FAILURE_POINT) printf (x) | |
| 752 #define DEBUG_FAIL_PRINT2(x1, x2) \ | |
| 753 if (debug_regexps & RE_DEBUG_FAILURE_POINT) printf (x1, x2) | |
| 754 #define DEBUG_FAIL_PRINT3(x1, x2, x3) \ | |
| 755 if (debug_regexps & RE_DEBUG_FAILURE_POINT) printf (x1, x2, x3) | |
| 756 #define DEBUG_FAIL_PRINT4(x1, x2, x3, x4) \ | |
| 757 if (debug_regexps & RE_DEBUG_FAILURE_POINT) printf (x1, x2, x3, x4) | |
| 758 #define DEBUG_FAIL_PRINT_COMPILED_PATTERN(p, s, e) \ | |
| 759 if (debug_regexps & RE_DEBUG_FAILURE_POINT) \ | |
| 760 print_partial_compiled_pattern (s, e) | |
| 761 #define DEBUG_FAIL_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ | |
| 762 if (debug_regexps & RE_DEBUG_FAILURE_POINT) \ | |
| 763 print_double_string (w, s1, sz1, s2, sz2) | |
| 764 | |
| 765 #define DEBUG_MATCH_PRINT1(x) \ | |
| 766 if (debug_regexps & RE_DEBUG_MATCHING) printf (x) | |
| 767 #define DEBUG_MATCH_PRINT2(x1, x2) \ | |
| 768 if (debug_regexps & RE_DEBUG_MATCHING) printf (x1, x2) | |
| 769 #define DEBUG_MATCH_PRINT3(x1, x2, x3) \ | |
| 770 if (debug_regexps & RE_DEBUG_MATCHING) printf (x1, x2, x3) | |
| 771 #define DEBUG_MATCH_PRINT4(x1, x2, x3, x4) \ | |
| 772 if (debug_regexps & RE_DEBUG_MATCHING) printf (x1, x2, x3, x4) | |
| 773 #define DEBUG_MATCH_PRINT_COMPILED_PATTERN(p, s, e) \ | |
| 774 if (debug_regexps & RE_DEBUG_MATCHING) \ | |
| 775 print_partial_compiled_pattern (s, e) | |
| 776 #define DEBUG_MATCH_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ | |
| 777 if (debug_regexps & RE_DEBUG_MATCHING) \ | |
| 778 print_double_string (w, s1, sz1, s2, sz2) | |
| 428 | 779 |
| 780 | |
| 781 /* Print the fastmap in human-readable form. */ | |
| 782 | |
| 783 static void | |
| 784 print_fastmap (char *fastmap) | |
| 785 { | |
| 647 | 786 int was_a_range = 0; |
| 787 int i = 0; | |
| 428 | 788 |
| 789 while (i < (1 << BYTEWIDTH)) | |
| 790 { | |
| 791 if (fastmap[i++]) | |
| 792 { | |
| 793 was_a_range = 0; | |
| 794 putchar (i - 1); | |
| 795 while (i < (1 << BYTEWIDTH) && fastmap[i]) | |
| 796 { | |
| 797 was_a_range = 1; | |
| 798 i++; | |
| 799 } | |
| 800 if (was_a_range) | |
| 801 { | |
| 802 putchar ('-'); | |
| 803 putchar (i - 1); | |
| 804 } | |
| 805 } | |
| 806 } | |
| 807 putchar ('\n'); | |
| 808 } | |
| 809 | |
| 810 | |
| 811 /* Print a compiled pattern string in human-readable form, starting at | |
| 812 the START pointer into it and ending just before the pointer END. */ | |
| 813 | |
| 814 static void | |
| 446 | 815 print_partial_compiled_pattern (re_char *start, re_char *end) |
| 428 | 816 { |
| 817 int mcnt, mcnt2; | |
| 446 | 818 unsigned char *p = (unsigned char *) start; |
| 819 re_char *pend = end; | |
| 428 | 820 |
| 821 if (start == NULL) | |
| 822 { | |
| 823 puts ("(null)"); | |
| 824 return; | |
| 825 } | |
| 826 | |
| 827 /* Loop over pattern commands. */ | |
| 828 while (p < pend) | |
| 829 { | |
| 830 printf ("%ld:\t", (long)(p - start)); | |
| 831 | |
| 832 switch ((re_opcode_t) *p++) | |
| 833 { | |
| 834 case no_op: | |
| 835 printf ("/no_op"); | |
| 836 break; | |
| 837 | |
| 838 case exactn: | |
| 839 mcnt = *p++; | |
| 840 printf ("/exactn/%d", mcnt); | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
841 while (mcnt--) |
| 428 | 842 { |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
843 putchar ('/'); |
| 428 | 844 putchar (*p++); |
| 845 } | |
| 846 break; | |
| 847 | |
| 848 case start_memory: | |
| 849 mcnt = *p++; | |
| 850 printf ("/start_memory/%d/%d", mcnt, *p++); | |
| 851 break; | |
| 852 | |
| 853 case stop_memory: | |
| 854 mcnt = *p++; | |
| 855 printf ("/stop_memory/%d/%d", mcnt, *p++); | |
| 856 break; | |
| 857 | |
| 858 case duplicate: | |
| 859 printf ("/duplicate/%d", *p++); | |
| 860 break; | |
| 861 | |
| 862 case anychar: | |
| 863 printf ("/anychar"); | |
| 864 break; | |
| 865 | |
| 866 case charset: | |
| 867 case charset_not: | |
| 868 { | |
| 869 REGISTER int c, last = -100; | |
| 870 REGISTER int in_range = 0; | |
| 871 | |
| 872 printf ("/charset [%s", | |
| 873 (re_opcode_t) *(p - 1) == charset_not ? "^" : ""); | |
| 874 | |
| 875 assert (p + *p < pend); | |
| 876 | |
| 877 for (c = 0; c < 256; c++) | |
| 878 if (((unsigned char) (c / 8) < *p) | |
| 879 && (p[1 + (c/8)] & (1 << (c % 8)))) | |
| 880 { | |
| 881 /* Are we starting a range? */ | |
| 882 if (last + 1 == c && ! in_range) | |
| 883 { | |
| 884 putchar ('-'); | |
| 885 in_range = 1; | |
| 886 } | |
| 887 /* Have we broken a range? */ | |
| 888 else if (last + 1 != c && in_range) | |
| 889 { | |
| 890 putchar (last); | |
| 891 in_range = 0; | |
| 892 } | |
| 893 | |
| 894 if (! in_range) | |
| 895 putchar (c); | |
| 896 | |
| 897 last = c; | |
| 898 } | |
| 899 | |
| 900 if (in_range) | |
| 901 putchar (last); | |
| 902 | |
| 903 putchar (']'); | |
| 904 | |
| 905 p += 1 + *p; | |
| 906 } | |
| 907 break; | |
| 908 | |
| 909 #ifdef MULE | |
| 910 case charset_mule: | |
| 911 case charset_mule_not: | |
| 912 { | |
| 913 int nentries, i; | |
| 914 | |
| 915 printf ("/charset_mule [%s", | |
| 916 (re_opcode_t) *(p - 1) == charset_mule_not ? "^" : ""); | |
| 917 nentries = unified_range_table_nentries (p); | |
| 918 for (i = 0; i < nentries; i++) | |
| 919 { | |
| 920 EMACS_INT first, last; | |
| 921 Lisp_Object dummy_val; | |
| 922 | |
| 923 unified_range_table_get_range (p, i, &first, &last, | |
| 924 &dummy_val); | |
| 925 if (first < 0x100) | |
| 926 putchar (first); | |
| 927 else | |
| 928 printf ("(0x%lx)", (long)first); | |
| 929 if (first != last) | |
| 930 { | |
| 931 putchar ('-'); | |
| 932 if (last < 0x100) | |
| 933 putchar (last); | |
| 934 else | |
| 935 printf ("(0x%lx)", (long)last); | |
| 936 } | |
| 937 } | |
| 938 putchar (']'); | |
| 939 p += unified_range_table_bytes_used (p); | |
| 940 } | |
| 941 break; | |
| 942 #endif | |
| 943 | |
| 944 case begline: | |
| 945 printf ("/begline"); | |
| 946 break; | |
| 947 | |
| 948 case endline: | |
| 949 printf ("/endline"); | |
| 950 break; | |
| 951 | |
| 952 case on_failure_jump: | |
| 953 extract_number_and_incr (&mcnt, &p); | |
| 954 printf ("/on_failure_jump to %ld", (long)(p + mcnt - start)); | |
| 955 break; | |
| 956 | |
| 957 case on_failure_keep_string_jump: | |
| 958 extract_number_and_incr (&mcnt, &p); | |
| 959 printf ("/on_failure_keep_string_jump to %ld", (long)(p + mcnt - start)); | |
| 960 break; | |
| 961 | |
| 962 case dummy_failure_jump: | |
| 963 extract_number_and_incr (&mcnt, &p); | |
| 964 printf ("/dummy_failure_jump to %ld", (long)(p + mcnt - start)); | |
| 965 break; | |
| 966 | |
| 967 case push_dummy_failure: | |
| 968 printf ("/push_dummy_failure"); | |
| 969 break; | |
| 970 | |
| 971 case maybe_pop_jump: | |
| 972 extract_number_and_incr (&mcnt, &p); | |
| 973 printf ("/maybe_pop_jump to %ld", (long)(p + mcnt - start)); | |
| 974 break; | |
| 975 | |
| 976 case pop_failure_jump: | |
| 977 extract_number_and_incr (&mcnt, &p); | |
| 978 printf ("/pop_failure_jump to %ld", (long)(p + mcnt - start)); | |
| 979 break; | |
| 980 | |
| 981 case jump_past_alt: | |
| 982 extract_number_and_incr (&mcnt, &p); | |
| 983 printf ("/jump_past_alt to %ld", (long)(p + mcnt - start)); | |
| 984 break; | |
| 985 | |
| 986 case jump: | |
| 987 extract_number_and_incr (&mcnt, &p); | |
| 988 printf ("/jump to %ld", (long)(p + mcnt - start)); | |
| 989 break; | |
| 990 | |
| 991 case succeed_n: | |
| 992 extract_number_and_incr (&mcnt, &p); | |
| 993 extract_number_and_incr (&mcnt2, &p); | |
| 994 printf ("/succeed_n to %ld, %d times", (long)(p + mcnt - start), mcnt2); | |
| 995 break; | |
| 996 | |
| 997 case jump_n: | |
| 998 extract_number_and_incr (&mcnt, &p); | |
| 999 extract_number_and_incr (&mcnt2, &p); | |
| 1000 printf ("/jump_n to %ld, %d times", (long)(p + mcnt - start), mcnt2); | |
| 1001 break; | |
| 1002 | |
| 1003 case set_number_at: | |
| 1004 extract_number_and_incr (&mcnt, &p); | |
| 1005 extract_number_and_incr (&mcnt2, &p); | |
| 1006 printf ("/set_number_at location %ld to %d", (long)(p + mcnt - start), mcnt2); | |
| 1007 break; | |
| 1008 | |
| 1009 case wordbound: | |
| 1010 printf ("/wordbound"); | |
| 1011 break; | |
| 1012 | |
| 1013 case notwordbound: | |
| 1014 printf ("/notwordbound"); | |
| 1015 break; | |
| 1016 | |
| 1017 case wordbeg: | |
| 1018 printf ("/wordbeg"); | |
| 1019 break; | |
| 1020 | |
| 1021 case wordend: | |
| 1022 printf ("/wordend"); | |
| 1023 | |
| 1024 #ifdef emacs | |
| 1025 case before_dot: | |
| 1026 printf ("/before_dot"); | |
| 1027 break; | |
| 1028 | |
| 1029 case at_dot: | |
| 1030 printf ("/at_dot"); | |
| 1031 break; | |
| 1032 | |
| 1033 case after_dot: | |
| 1034 printf ("/after_dot"); | |
| 1035 break; | |
| 1036 | |
| 1037 case syntaxspec: | |
| 1038 printf ("/syntaxspec"); | |
| 1039 mcnt = *p++; | |
| 1040 printf ("/%d", mcnt); | |
| 1041 break; | |
| 1042 | |
| 1043 case notsyntaxspec: | |
| 1044 printf ("/notsyntaxspec"); | |
| 1045 mcnt = *p++; | |
| 1046 printf ("/%d", mcnt); | |
| 1047 break; | |
| 1048 | |
| 1049 #ifdef MULE | |
| 1050 /* 97/2/17 jhod Mule category patch */ | |
| 1051 case categoryspec: | |
| 1052 printf ("/categoryspec"); | |
| 1053 mcnt = *p++; | |
| 1054 printf ("/%d", mcnt); | |
| 1055 break; | |
| 1056 | |
| 1057 case notcategoryspec: | |
| 1058 printf ("/notcategoryspec"); | |
| 1059 mcnt = *p++; | |
| 1060 printf ("/%d", mcnt); | |
| 1061 break; | |
| 1062 /* end of category patch */ | |
| 1063 #endif /* MULE */ | |
| 1064 #endif /* emacs */ | |
| 1065 | |
| 1066 case wordchar: | |
| 1067 printf ("/wordchar"); | |
| 1068 break; | |
| 1069 | |
| 1070 case notwordchar: | |
| 1071 printf ("/notwordchar"); | |
| 1072 break; | |
| 1073 | |
| 1074 case begbuf: | |
| 1075 printf ("/begbuf"); | |
| 1076 break; | |
| 1077 | |
| 1078 case endbuf: | |
| 1079 printf ("/endbuf"); | |
| 1080 break; | |
| 1081 | |
| 1082 default: | |
| 1083 printf ("?%d", *(p-1)); | |
| 1084 } | |
| 1085 | |
| 1086 putchar ('\n'); | |
| 1087 } | |
| 1088 | |
| 1089 printf ("%ld:\tend of pattern.\n", (long)(p - start)); | |
| 1090 } | |
| 1091 | |
| 1092 | |
| 1093 static void | |
| 1094 print_compiled_pattern (struct re_pattern_buffer *bufp) | |
| 1095 { | |
| 446 | 1096 re_char *buffer = bufp->buffer; |
| 428 | 1097 |
| 1098 print_partial_compiled_pattern (buffer, buffer + bufp->used); | |
| 1099 printf ("%ld bytes used/%ld bytes allocated.\n", bufp->used, | |
| 1100 bufp->allocated); | |
| 1101 | |
| 1102 if (bufp->fastmap_accurate && bufp->fastmap) | |
| 1103 { | |
| 1104 printf ("fastmap: "); | |
| 1105 print_fastmap (bufp->fastmap); | |
| 1106 } | |
| 1107 | |
| 1108 printf ("re_nsub: %ld\t", (long)bufp->re_nsub); | |
| 502 | 1109 printf ("re_ngroups: %ld\t", (long)bufp->re_ngroups); |
| 428 | 1110 printf ("regs_alloc: %d\t", bufp->regs_allocated); |
| 1111 printf ("can_be_null: %d\t", bufp->can_be_null); | |
| 1112 printf ("newline_anchor: %d\n", bufp->newline_anchor); | |
| 1113 printf ("no_sub: %d\t", bufp->no_sub); | |
| 1114 printf ("not_bol: %d\t", bufp->not_bol); | |
| 1115 printf ("not_eol: %d\t", bufp->not_eol); | |
| 1116 printf ("syntax: %d\n", bufp->syntax); | |
| 1117 /* Perhaps we should print the translate table? */ | |
| 1118 /* and maybe the category table? */ | |
| 502 | 1119 |
| 1120 if (bufp->external_to_internal_register) | |
| 1121 { | |
| 1122 int i; | |
| 1123 | |
| 1124 printf ("external_to_internal_register:\n"); | |
| 1125 for (i = 0; i <= bufp->re_nsub; i++) | |
| 1126 { | |
| 1127 if (i > 0) | |
| 1128 printf (", "); | |
| 1129 printf ("%d -> %d", i, bufp->external_to_internal_register[i]); | |
| 1130 } | |
| 1131 printf ("\n"); | |
| 1132 } | |
| 428 | 1133 } |
| 1134 | |
| 1135 | |
| 1136 static void | |
| 446 | 1137 print_double_string (re_char *where, re_char *string1, int size1, |
| 1138 re_char *string2, int size2) | |
| 428 | 1139 { |
| 1140 if (where == NULL) | |
| 1141 printf ("(null)"); | |
| 1142 else | |
| 1143 { | |
| 647 | 1144 int this_char; |
| 428 | 1145 |
| 1146 if (FIRST_STRING_P (where)) | |
| 1147 { | |
| 1148 for (this_char = where - string1; this_char < size1; this_char++) | |
| 1149 putchar (string1[this_char]); | |
| 1150 | |
| 1151 where = string2; | |
| 1152 } | |
| 1153 | |
| 1154 for (this_char = where - string2; this_char < size2; this_char++) | |
| 1155 putchar (string2[this_char]); | |
| 1156 } | |
| 1157 } | |
| 1158 | |
| 1159 #else /* not DEBUG */ | |
| 1160 | |
| 771 | 1161 #ifndef emacs |
| 428 | 1162 #undef assert |
| 771 | 1163 #define assert(e) ((void) (1)) |
| 1164 #endif | |
| 428 | 1165 |
| 1166 #define DEBUG_STATEMENT(e) | |
| 5041 | 1167 |
| 428 | 1168 #define DEBUG_PRINT1(x) |
| 1169 #define DEBUG_PRINT2(x1, x2) | |
| 1170 #define DEBUG_PRINT3(x1, x2, x3) | |
| 1171 #define DEBUG_PRINT4(x1, x2, x3, x4) | |
| 1172 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) | |
| 1173 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) | |
| 1174 | |
| 5041 | 1175 #define DEBUG_FAIL_PRINT1(x) |
| 1176 #define DEBUG_FAIL_PRINT2(x1, x2) | |
| 1177 #define DEBUG_FAIL_PRINT3(x1, x2, x3) | |
| 1178 #define DEBUG_FAIL_PRINT4(x1, x2, x3, x4) | |
| 1179 #define DEBUG_FAIL_PRINT_COMPILED_PATTERN(p, s, e) | |
| 1180 #define DEBUG_FAIL_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) | |
| 1181 | |
| 1182 #define DEBUG_MATCH_PRINT1(x) | |
| 1183 #define DEBUG_MATCH_PRINT2(x1, x2) | |
| 1184 #define DEBUG_MATCH_PRINT3(x1, x2, x3) | |
| 1185 #define DEBUG_MATCH_PRINT4(x1, x2, x3, x4) | |
| 1186 #define DEBUG_MATCH_PRINT_COMPILED_PATTERN(p, s, e) | |
| 1187 #define DEBUG_MATCH_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) | |
| 1188 | |
| 446 | 1189 #endif /* DEBUG */ |
| 428 | 1190 |
| 1191 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can | |
| 1192 also be assigned to arbitrarily: each pattern buffer stores its own | |
| 1193 syntax, so it can be changed between regex compilations. */ | |
| 1194 /* This has no initializer because initialized variables in Emacs | |
| 1195 become read-only after dumping. */ | |
| 1196 reg_syntax_t re_syntax_options; | |
| 1197 | |
| 1198 | |
| 1199 /* Specify the precise syntax of regexps for compilation. This provides | |
| 1200 for compatibility for various utilities which historically have | |
| 1201 different, incompatible syntaxes. | |
| 1202 | |
| 1203 The argument SYNTAX is a bit mask comprised of the various bits | |
| 1204 defined in regex.h. We return the old syntax. */ | |
| 1205 | |
| 1206 reg_syntax_t | |
| 1207 re_set_syntax (reg_syntax_t syntax) | |
| 1208 { | |
| 1209 reg_syntax_t ret = re_syntax_options; | |
| 1210 | |
| 1211 re_syntax_options = syntax; | |
| 1212 return ret; | |
| 1213 } | |
| 1214 | |
| 1215 /* This table gives an error message for each of the error codes listed | |
| 1216 in regex.h. Obviously the order here has to be same as there. | |
| 1217 POSIX doesn't require that we do anything for REG_NOERROR, | |
| 1218 but why not be nice? */ | |
| 1219 | |
| 442 | 1220 static const char *re_error_msgid[] = |
| 428 | 1221 { |
| 1222 "Success", /* REG_NOERROR */ | |
| 1223 "No match", /* REG_NOMATCH */ | |
| 1224 "Invalid regular expression", /* REG_BADPAT */ | |
| 1225 "Invalid collation character", /* REG_ECOLLATE */ | |
| 1226 "Invalid character class name", /* REG_ECTYPE */ | |
| 1227 "Trailing backslash", /* REG_EESCAPE */ | |
| 1228 "Invalid back reference", /* REG_ESUBREG */ | |
| 1229 "Unmatched [ or [^", /* REG_EBRACK */ | |
| 1230 "Unmatched ( or \\(", /* REG_EPAREN */ | |
| 1231 "Unmatched \\{", /* REG_EBRACE */ | |
| 1232 "Invalid content of \\{\\}", /* REG_BADBR */ | |
| 1233 "Invalid range end", /* REG_ERANGE */ | |
| 1234 "Memory exhausted", /* REG_ESPACE */ | |
| 1235 "Invalid preceding regular expression", /* REG_BADRPT */ | |
| 1236 "Premature end of regular expression", /* REG_EEND */ | |
| 1237 "Regular expression too big", /* REG_ESIZE */ | |
| 1238 "Unmatched ) or \\)", /* REG_ERPAREN */ | |
| 1239 #ifdef emacs | |
| 1240 "Invalid syntax designator", /* REG_ESYNTAX */ | |
| 1241 #endif | |
| 1242 #ifdef MULE | |
| 1243 "Ranges may not span charsets", /* REG_ERANGESPAN */ | |
| 1244 "Invalid category designator", /* REG_ECATEGORY */ | |
| 1245 #endif | |
| 1246 }; | |
| 1247 | |
| 1248 /* Avoiding alloca during matching, to placate r_alloc. */ | |
| 1249 | |
| 1333 | 1250 /* About these various flags: |
| 1251 | |
| 1252 MATCH_MAY_ALLOCATE indicates that it's OK to do allocation in the | |
| 1253 searching and matching functions. In this case, we use local variables | |
| 1254 to hold the values allocated. If not, we use *global* variables, which | |
| 1255 are pre-allocated. NOTE: XEmacs ***MUST*** run with MATCH_MAY_ALLOCATE, | |
| 1256 because the regexp routines may get called reentrantly as a result of | |
| 1257 QUIT processing (e.g. under Windows: re_match -> QUIT -> quit_p -> drain | |
| 1258 events -> process WM_INITMENU -> call filter -> re_match; see stack | |
| 1259 trace in signal.c), so we cannot have any global variables (unless we do | |
| 1260 lots of trickiness including some unwind-protects, which isn't worth it | |
| 1261 at this point). | |
| 1262 | |
| 1263 REL_ALLOC means that the relocating allocator is in use, for buffers | |
| 1264 and such. REGEX_REL_ALLOC means that we use rel-alloc to manage the | |
| 1265 fail stack, which may grow quite large. REGEX_MALLOC means we use | |
| 1266 malloc() in place of alloca() to allocate the fail stack -- only | |
| 1267 applicable if REGEX_REL_ALLOC is not defined. | |
| 1268 */ | |
| 1269 | |
| 428 | 1270 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the |
| 1271 searching and matching functions should not call alloca. On some | |
| 1272 systems, alloca is implemented in terms of malloc, and if we're | |
| 1273 using the relocating allocator routines, then malloc could cause a | |
| 1274 relocation, which might (if the strings being searched are in the | |
| 1275 ralloc heap) shift the data out from underneath the regexp | |
| 771 | 1276 routines. [To clarify: The purpose of rel-alloc is to allow data to |
| 1277 be moved in memory from one place to another so that all data | |
| 1278 blocks can be consolidated together and excess memory released back | |
| 1279 to the operating system. This requires that all the blocks that | |
| 1280 are managed by rel-alloc go at the very end of the program's heap, | |
| 1281 after all regularly malloc()ed data. malloc(), however, is used to | |
| 1282 owning the end of the heap, so that when more memory is needed, it | |
| 1283 just expands the heap using sbrk(). This is reconciled by using a | |
| 1284 malloc() (such as malloc.c, gmalloc.c, or recent versions of | |
| 1285 malloc() in libc) where the sbrk() call can be replaced with a | |
| 1286 user-specified call -- in this case, to rel-alloc's r_alloc_sbrk() | |
| 1287 routine. This routine calls the real sbrk(), but then shifts all | |
| 1288 the rel-alloc-managed blocks forward to the end of the heap again, | |
| 1289 so that malloc() gets the memory it needs in the location it needs | |
| 1290 it at. The regex routines may well have pointers to buffer data as | |
| 1291 their arguments, and buffers are managed by rel-alloc if rel-alloc | |
| 1292 has been enabled, so calling malloc() may potentially screw things | |
| 1293 up badly if it runs out of space and asks for more from the OS.] | |
| 1294 | |
| 1295 [[Here's another reason to avoid allocation: Emacs processes input | |
| 1296 from X in a signal handler; processing X input may call malloc; if | |
| 1297 input arrives while a matching routine is calling malloc, then | |
| 1298 we're scrod. But Emacs can't just block input while calling | |
| 1299 matching routines; then we don't notice interrupts when they come | |
| 1300 in. So, Emacs blocks input around all regexp calls except the | |
| 1301 matching calls, which it leaves unprotected, in the faith that they | |
| 1333 | 1302 will not malloc.]] This previous paragraph is irrelevant under XEmacs, |
| 1303 as we *do not* do anything so stupid as process input from within a | |
| 1304 signal handler. | |
| 1305 | |
| 1306 However, the regexp routines may get called reentrantly as a result of | |
| 1307 QUIT processing (e.g. under Windows: re_match -> QUIT -> quit_p -> drain | |
| 1308 events -> process WM_INITMENU -> call filter -> re_match; see stack | |
| 1309 trace in signal.c), so we cannot have any global variables (unless we do | |
| 1310 lots of trickiness including some unwind-protects, which isn't worth it | |
| 1311 at this point). Hence we MUST have MATCH_MAY_ALLOCATE defined. | |
| 1312 | |
| 1313 Also, the first paragraph does not make complete sense to me -- what | |
| 1314 about the use of rel-alloc to handle the fail stacks? Shouldn't these | |
| 1315 reallocations potentially cause buffer data to be relocated as well? I | |
| 826 | 1316 must be missing something, though -- perhaps the writer above is |
| 1317 assuming that the failure stack(s) will always be allocated after the | |
| 1318 buffer data, and thus reallocating them with rel-alloc won't move buffer | |
| 1333 | 1319 data. (In fact, a cursory glance at the code in ralloc.c seems to |
| 1320 confirm this.) --ben */ | |
| 428 | 1321 |
| 1322 /* Normally, this is fine. */ | |
| 1323 #define MATCH_MAY_ALLOCATE | |
| 1324 | |
| 1325 /* When using GNU C, we are not REALLY using the C alloca, no matter | |
| 1326 what config.h may say. So don't take precautions for it. */ | |
| 1327 #ifdef __GNUC__ | |
| 1328 #undef C_ALLOCA | |
| 1329 #endif | |
| 1330 | |
| 1331 /* The match routines may not allocate if (1) they would do it with malloc | |
| 1332 and (2) it's not safe for them to use malloc. | |
| 1333 Note that if REL_ALLOC is defined, matching would not use malloc for the | |
| 1334 failure stack, but we would still use it for the register vectors; | |
| 1335 so REL_ALLOC should not affect this. */ | |
| 771 | 1336 |
| 1333 | 1337 /* XEmacs can handle REL_ALLOC and malloc() OK */ |
| 1338 #if !defined (emacs) && (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (REL_ALLOC) | |
| 428 | 1339 #undef MATCH_MAY_ALLOCATE |
| 1340 #endif | |
| 1341 | |
| 1333 | 1342 #if !defined (MATCH_MAY_ALLOCATE) && defined (emacs) |
| 771 | 1343 #error regex must be handle reentrancy; MATCH_MAY_ALLOCATE must be defined |
| 1344 #endif | |
| 1345 | |
| 428 | 1346 |
| 1347 /* Failure stack declarations and macros; both re_compile_fastmap and | |
| 1348 re_match_2 use a failure stack. These have to be macros because of | |
| 1349 REGEX_ALLOCATE_STACK. */ | |
| 1350 | |
| 1351 | |
| 1352 /* Number of failure points for which to initially allocate space | |
| 1353 when matching. If this number is exceeded, we allocate more | |
| 1354 space, so it is not a hard limit. */ | |
| 1355 #ifndef INIT_FAILURE_ALLOC | |
| 3300 | 1356 #define INIT_FAILURE_ALLOC 20 |
| 428 | 1357 #endif |
| 1358 | |
| 1359 /* Roughly the maximum number of failure points on the stack. Would be | |
| 1360 exactly that if always used MAX_FAILURE_SPACE each time we failed. | |
| 1361 This is a variable only so users of regex can assign to it; we never | |
| 1362 change it ourselves. */ | |
| 1363 #if defined (MATCH_MAY_ALLOCATE) | |
| 1364 /* 4400 was enough to cause a crash on Alpha OSF/1, | |
| 1365 whose default stack limit is 2mb. */ | |
| 3300 | 1366 int re_max_failures = 40000; |
| 428 | 1367 #else |
| 3300 | 1368 int re_max_failures = 4000; |
| 428 | 1369 #endif |
| 1370 | |
| 1371 union fail_stack_elt | |
| 1372 { | |
| 446 | 1373 re_char *pointer; |
| 428 | 1374 int integer; |
| 1375 }; | |
| 1376 | |
| 1377 typedef union fail_stack_elt fail_stack_elt_t; | |
| 1378 | |
| 1379 typedef struct | |
| 1380 { | |
| 1381 fail_stack_elt_t *stack; | |
| 665 | 1382 Elemcount size; |
| 1383 Elemcount avail; /* Offset of next open position. */ | |
| 428 | 1384 } fail_stack_type; |
| 1385 | |
| 1386 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0) | |
| 1387 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0) | |
| 1388 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size) | |
| 1389 | |
| 1390 | |
| 1391 /* Define macros to initialize and free the failure stack. | |
| 1392 Do `return -2' if the alloc fails. */ | |
| 1393 | |
| 1394 #ifdef MATCH_MAY_ALLOCATE | |
| 1333 | 1395 #define INIT_FAIL_STACK() \ |
| 1396 do { \ | |
| 1397 fail_stack.stack = (fail_stack_elt_t *) \ | |
| 1398 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * \ | |
| 1399 sizeof (fail_stack_elt_t)); \ | |
| 1400 \ | |
| 1401 if (fail_stack.stack == NULL) \ | |
| 1402 { \ | |
| 1403 UNBIND_REGEX_MALLOC_CHECK (); \ | |
| 1404 return -2; \ | |
| 1405 } \ | |
| 1406 \ | |
| 1407 fail_stack.size = INIT_FAILURE_ALLOC; \ | |
| 1408 fail_stack.avail = 0; \ | |
| 428 | 1409 } while (0) |
| 1410 | |
| 1411 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack) | |
| 1412 #else | |
| 1413 #define INIT_FAIL_STACK() \ | |
| 1414 do { \ | |
| 1415 fail_stack.avail = 0; \ | |
| 1416 } while (0) | |
| 1417 | |
| 1418 #define RESET_FAIL_STACK() | |
| 1419 #endif | |
| 1420 | |
| 1421 | |
| 1422 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. | |
| 1423 | |
| 1424 Return 1 if succeeds, and 0 if either ran out of memory | |
| 1425 allocating space for it or it was already too large. | |
| 1426 | |
| 1427 REGEX_REALLOCATE_STACK requires `destination' be declared. */ | |
| 1428 | |
| 1429 #define DOUBLE_FAIL_STACK(fail_stack) \ | |
| 1430 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \ | |
| 1431 ? 0 \ | |
| 1432 : ((fail_stack).stack = (fail_stack_elt_t *) \ | |
| 1433 REGEX_REALLOCATE_STACK ((fail_stack).stack, \ | |
| 1434 (fail_stack).size * sizeof (fail_stack_elt_t), \ | |
| 1435 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ | |
| 1436 \ | |
| 1437 (fail_stack).stack == NULL \ | |
| 1438 ? 0 \ | |
| 1439 : ((fail_stack).size <<= 1, \ | |
| 1440 1))) | |
| 1441 | |
| 1333 | 1442 #if !defined (emacs) || !defined (REL_ALLOC) |
| 1443 #define RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS() | |
| 1444 #else | |
| 1445 /* Don't change NULL pointers */ | |
| 1446 #define ADD_IF_NZ(val) if (val) val += rmdp_offset | |
| 1346 | 1447 #define RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS() \ |
| 1448 do \ | |
| 1449 { \ | |
| 1450 Bytecount rmdp_offset = offset_post_relocation (lispobj, orig_buftext); \ | |
| 1451 \ | |
| 1452 if (rmdp_offset) \ | |
| 1453 { \ | |
| 1454 int i; \ | |
| 1455 \ | |
| 1456 ADD_IF_NZ (string1); \ | |
| 1457 ADD_IF_NZ (string2); \ | |
| 1458 ADD_IF_NZ (d); \ | |
| 1459 ADD_IF_NZ (dend); \ | |
| 1460 ADD_IF_NZ (end1); \ | |
| 1461 ADD_IF_NZ (end2); \ | |
| 1462 ADD_IF_NZ (end_match_1); \ | |
| 1463 ADD_IF_NZ (end_match_2); \ | |
| 1464 \ | |
| 1465 if (bufp->re_ngroups) \ | |
| 1466 { \ | |
| 1467 for (i = 0; i < num_regs; i++) \ | |
| 1468 { \ | |
| 1469 ADD_IF_NZ (regstart[i]); \ | |
| 1470 ADD_IF_NZ (regend[i]); \ | |
| 1471 ADD_IF_NZ (old_regstart[i]); \ | |
| 1472 ADD_IF_NZ (old_regend[i]); \ | |
| 1473 ADD_IF_NZ (best_regstart[i]); \ | |
| 1474 ADD_IF_NZ (best_regend[i]); \ | |
| 1475 ADD_IF_NZ (reg_dummy[i]); \ | |
| 1476 } \ | |
| 1477 } \ | |
| 1478 \ | |
| 1479 ADD_IF_NZ (match_end); \ | |
| 1480 } \ | |
| 1333 | 1481 } while (0) |
| 1482 #endif /* !defined (emacs) || !defined (REL_ALLOC) */ | |
| 1483 | |
| 1484 #if !defined (emacs) || !defined (REL_ALLOC) | |
| 1485 #define RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS() | |
| 1486 #else | |
| 1346 | 1487 #define RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS() \ |
| 1488 do \ | |
| 1489 { \ | |
| 1490 Bytecount rmdp_offset = offset_post_relocation (lispobj, orig_buftext); \ | |
| 1491 \ | |
| 1492 if (rmdp_offset) \ | |
| 1493 { \ | |
| 1494 ADD_IF_NZ (str1); \ | |
| 1495 ADD_IF_NZ (str2); \ | |
| 1496 ADD_IF_NZ (string1); \ | |
| 1497 ADD_IF_NZ (string2); \ | |
| 1498 ADD_IF_NZ (d); \ | |
| 1499 } \ | |
| 1333 | 1500 } while (0) |
| 1501 | |
| 1502 #endif /* emacs */ | |
| 428 | 1503 |
| 1504 /* Push pointer POINTER on FAIL_STACK. | |
| 1505 Return 1 if was able to do so and 0 if ran out of memory allocating | |
| 1506 space to do so. */ | |
| 1507 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \ | |
| 1508 ((FAIL_STACK_FULL () \ | |
| 1509 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \ | |
| 1510 ? 0 \ | |
| 1511 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \ | |
| 1512 1)) | |
| 1513 | |
| 1514 /* Push a pointer value onto the failure stack. | |
| 1515 Assumes the variable `fail_stack'. Probably should only | |
| 1516 be called from within `PUSH_FAILURE_POINT'. */ | |
| 1517 #define PUSH_FAILURE_POINTER(item) \ | |
| 1518 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item) | |
| 1519 | |
| 1520 /* This pushes an integer-valued item onto the failure stack. | |
| 1521 Assumes the variable `fail_stack'. Probably should only | |
| 1522 be called from within `PUSH_FAILURE_POINT'. */ | |
| 1523 #define PUSH_FAILURE_INT(item) \ | |
| 1524 fail_stack.stack[fail_stack.avail++].integer = (item) | |
| 1525 | |
| 1526 /* Push a fail_stack_elt_t value onto the failure stack. | |
| 1527 Assumes the variable `fail_stack'. Probably should only | |
| 1528 be called from within `PUSH_FAILURE_POINT'. */ | |
| 1529 #define PUSH_FAILURE_ELT(item) \ | |
| 1530 fail_stack.stack[fail_stack.avail++] = (item) | |
| 1531 | |
| 1532 /* These three POP... operations complement the three PUSH... operations. | |
| 1533 All assume that `fail_stack' is nonempty. */ | |
| 1534 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer | |
| 1535 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer | |
| 1536 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail] | |
| 1537 | |
| 1538 /* Used to omit pushing failure point id's when we're not debugging. */ | |
| 1539 #ifdef DEBUG | |
| 1540 #define DEBUG_PUSH PUSH_FAILURE_INT | |
| 1541 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT () | |
| 1542 #else | |
| 1543 #define DEBUG_PUSH(item) | |
| 1544 #define DEBUG_POP(item_addr) | |
| 1545 #endif | |
| 1546 | |
| 1547 | |
| 1548 /* Push the information about the state we will need | |
| 1549 if we ever fail back to it. | |
| 1550 | |
| 1551 Requires variables fail_stack, regstart, regend, reg_info, and | |
| 1552 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be | |
| 1553 declared. | |
| 1554 | |
| 1555 Does `return FAILURE_CODE' if runs out of memory. */ | |
| 1556 | |
| 771 | 1557 #if !defined (REGEX_MALLOC) && !defined (REGEX_REL_ALLOC) |
| 456 | 1558 #define DECLARE_DESTINATION char *destination |
| 428 | 1559 #else |
| 456 | 1560 #define DECLARE_DESTINATION DECLARE_NOTHING |
| 428 | 1561 #endif |
| 1562 | |
| 1563 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ | |
| 456 | 1564 do { \ |
| 1565 DECLARE_DESTINATION; \ | |
| 1566 /* Must be int, so when we don't save any registers, the arithmetic \ | |
| 1567 of 0 + -1 isn't done as unsigned. */ \ | |
| 1568 int this_reg; \ | |
| 428 | 1569 \ |
| 456 | 1570 DEBUG_STATEMENT (failure_id++); \ |
| 1571 DEBUG_STATEMENT (nfailure_points_pushed++); \ | |
| 5041 | 1572 DEBUG_FAIL_PRINT2 ("\nPUSH_FAILURE_POINT #%d:\n", failure_id); \ |
| 1573 DEBUG_FAIL_PRINT2 (" Before push, next avail: %ld\n", \ | |
| 647 | 1574 (long) (fail_stack).avail); \ |
| 5041 | 1575 DEBUG_FAIL_PRINT2 (" size: %ld\n", \ |
| 647 | 1576 (long) (fail_stack).size); \ |
| 456 | 1577 \ |
| 5041 | 1578 DEBUG_FAIL_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \ |
| 1579 DEBUG_FAIL_PRINT2 (" available: %ld\n", \ | |
| 456 | 1580 (long) REMAINING_AVAIL_SLOTS); \ |
| 428 | 1581 \ |
| 456 | 1582 /* Ensure we have enough space allocated for what we will push. */ \ |
| 1583 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ | |
| 1584 { \ | |
| 1333 | 1585 BEGIN_REGEX_MALLOC_OK (); \ |
| 456 | 1586 if (!DOUBLE_FAIL_STACK (fail_stack)) \ |
| 1333 | 1587 { \ |
| 1588 END_REGEX_MALLOC_OK (); \ | |
| 1589 UNBIND_REGEX_MALLOC_CHECK (); \ | |
| 1590 return failure_code; \ | |
| 1591 } \ | |
| 1592 END_REGEX_MALLOC_OK (); \ | |
| 5041 | 1593 DEBUG_FAIL_PRINT2 ("\n Doubled stack; size now: %ld\n", \ |
| 647 | 1594 (long) (fail_stack).size); \ |
| 5041 | 1595 DEBUG_FAIL_PRINT2 (" slots available: %ld\n", \ |
| 456 | 1596 (long) REMAINING_AVAIL_SLOTS); \ |
| 1333 | 1597 \ |
| 1598 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); \ | |
| 456 | 1599 } \ |
| 428 | 1600 \ |
| 456 | 1601 /* Push the info, starting with the registers. */ \ |
| 5041 | 1602 DEBUG_FAIL_PRINT1 ("\n"); \ |
| 428 | 1603 \ |
| 456 | 1604 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \ |
| 1605 this_reg++) \ | |
| 1606 { \ | |
| 5041 | 1607 DEBUG_FAIL_PRINT2 (" Pushing reg: %d\n", this_reg); \ |
| 456 | 1608 DEBUG_STATEMENT (num_regs_pushed++); \ |
| 428 | 1609 \ |
| 5041 | 1610 DEBUG_FAIL_PRINT2 (" start: 0x%lx\n", (long) regstart[this_reg]); \ |
| 456 | 1611 PUSH_FAILURE_POINTER (regstart[this_reg]); \ |
| 1612 \ | |
| 5041 | 1613 DEBUG_FAIL_PRINT2 (" end: 0x%lx\n", (long) regend[this_reg]); \ |
| 456 | 1614 PUSH_FAILURE_POINTER (regend[this_reg]); \ |
| 428 | 1615 \ |
| 5041 | 1616 DEBUG_FAIL_PRINT2 (" info: 0x%lx\n ", \ |
| 456 | 1617 * (long *) (®_info[this_reg])); \ |
| 5041 | 1618 DEBUG_FAIL_PRINT2 (" match_null=%d", \ |
| 456 | 1619 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \ |
| 5041 | 1620 DEBUG_FAIL_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \ |
| 1621 DEBUG_FAIL_PRINT2 (" matched_something=%d", \ | |
| 456 | 1622 MATCHED_SOMETHING (reg_info[this_reg])); \ |
| 5041 | 1623 DEBUG_FAIL_PRINT2 (" ever_matched_something=%d", \ |
| 456 | 1624 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \ |
| 5041 | 1625 DEBUG_FAIL_PRINT1 ("\n"); \ |
| 456 | 1626 PUSH_FAILURE_ELT (reg_info[this_reg].word); \ |
| 1627 } \ | |
| 428 | 1628 \ |
| 5041 | 1629 DEBUG_FAIL_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg); \ |
| 456 | 1630 PUSH_FAILURE_INT (lowest_active_reg); \ |
| 428 | 1631 \ |
| 5041 | 1632 DEBUG_FAIL_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg); \ |
| 456 | 1633 PUSH_FAILURE_INT (highest_active_reg); \ |
| 428 | 1634 \ |
| 5041 | 1635 DEBUG_FAIL_PRINT2 (" Pushing pattern 0x%lx: \n", (long) pattern_place); \ |
| 1636 DEBUG_FAIL_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ | |
| 456 | 1637 PUSH_FAILURE_POINTER (pattern_place); \ |
| 428 | 1638 \ |
| 5041 | 1639 DEBUG_FAIL_PRINT2 (" Pushing string 0x%lx: `", (long) string_place); \ |
| 1640 DEBUG_FAIL_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ | |
| 456 | 1641 size2); \ |
| 5041 | 1642 DEBUG_FAIL_PRINT1 ("'\n"); \ |
| 456 | 1643 PUSH_FAILURE_POINTER (string_place); \ |
| 428 | 1644 \ |
| 5041 | 1645 DEBUG_FAIL_PRINT2 (" Pushing failure id: %u\n", failure_id); \ |
| 456 | 1646 DEBUG_PUSH (failure_id); \ |
| 1647 } while (0) | |
| 428 | 1648 |
| 1649 /* This is the number of items that are pushed and popped on the stack | |
| 1650 for each register. */ | |
| 1651 #define NUM_REG_ITEMS 3 | |
| 1652 | |
| 1653 /* Individual items aside from the registers. */ | |
| 1654 #ifdef DEBUG | |
| 1655 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */ | |
| 1656 #else | |
| 1657 #define NUM_NONREG_ITEMS 4 | |
| 1658 #endif | |
| 1659 | |
| 1660 /* We push at most this many items on the stack. */ | |
| 1661 /* We used to use (num_regs - 1), which is the number of registers | |
| 1662 this regexp will save; but that was changed to 5 | |
| 1663 to avoid stack overflow for a regexp with lots of parens. */ | |
| 1664 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS) | |
| 1665 | |
| 1666 /* We actually push this many items. */ | |
| 1667 #define NUM_FAILURE_ITEMS \ | |
| 1668 ((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \ | |
| 1669 + NUM_NONREG_ITEMS) | |
| 1670 | |
| 1671 /* How many items can still be added to the stack without overflowing it. */ | |
| 1672 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail) | |
| 1673 | |
| 1674 | |
| 1675 /* Pops what PUSH_FAIL_STACK pushes. | |
| 1676 | |
| 1677 We restore into the parameters, all of which should be lvalues: | |
| 1678 STR -- the saved data position. | |
| 1679 PAT -- the saved pattern position. | |
| 1680 LOW_REG, HIGH_REG -- the highest and lowest active registers. | |
| 1681 REGSTART, REGEND -- arrays of string positions. | |
| 1682 REG_INFO -- array of information about each subexpression. | |
| 1683 | |
| 1684 Also assumes the variables `fail_stack' and (if debugging), `bufp', | |
| 1685 `pend', `string1', `size1', `string2', and `size2'. */ | |
| 1686 | |
| 456 | 1687 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, \ |
| 1688 regstart, regend, reg_info) \ | |
| 1689 do { \ | |
| 428 | 1690 DEBUG_STATEMENT (fail_stack_elt_t ffailure_id;) \ |
| 1691 int this_reg; \ | |
| 442 | 1692 const unsigned char *string_temp; \ |
| 428 | 1693 \ |
| 1694 assert (!FAIL_STACK_EMPTY ()); \ | |
| 1695 \ | |
| 1696 /* Remove failure points and point to how many regs pushed. */ \ | |
| 5041 | 1697 DEBUG_FAIL_PRINT1 ("POP_FAILURE_POINT:\n"); \ |
| 1698 DEBUG_FAIL_PRINT2 (" Before pop, next avail: %ld\n", \ | |
| 647 | 1699 (long) fail_stack.avail); \ |
| 5041 | 1700 DEBUG_FAIL_PRINT2 (" size: %ld\n", \ |
| 647 | 1701 (long) fail_stack.size); \ |
| 428 | 1702 \ |
| 1703 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ | |
| 1704 \ | |
| 1705 DEBUG_POP (&ffailure_id.integer); \ | |
| 5041 | 1706 DEBUG_FAIL_PRINT2 (" Popping failure id: %d\n", \ |
| 647 | 1707 * (int *) &ffailure_id); \ |
| 428 | 1708 \ |
| 1709 /* If the saved string location is NULL, it came from an \ | |
| 1710 on_failure_keep_string_jump opcode, and we want to throw away the \ | |
| 1711 saved NULL, thus retaining our current position in the string. */ \ | |
| 1712 string_temp = POP_FAILURE_POINTER (); \ | |
| 1713 if (string_temp != NULL) \ | |
| 446 | 1714 str = string_temp; \ |
| 428 | 1715 \ |
| 5041 | 1716 DEBUG_FAIL_PRINT2 (" Popping string 0x%lx: `", (long) str); \ |
| 1717 DEBUG_FAIL_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \ | |
| 1718 DEBUG_FAIL_PRINT1 ("'\n"); \ | |
| 428 | 1719 \ |
| 1720 pat = (unsigned char *) POP_FAILURE_POINTER (); \ | |
| 5041 | 1721 DEBUG_FAIL_PRINT2 (" Popping pattern 0x%lx: ", (long) pat); \ |
| 1722 DEBUG_FAIL_PRINT_COMPILED_PATTERN (bufp, pat, pend); \ | |
| 428 | 1723 \ |
| 1724 /* Restore register info. */ \ | |
| 647 | 1725 high_reg = POP_FAILURE_INT (); \ |
| 5041 | 1726 DEBUG_FAIL_PRINT2 (" Popping high active reg: %d\n", high_reg); \ |
| 428 | 1727 \ |
| 647 | 1728 low_reg = POP_FAILURE_INT (); \ |
| 5041 | 1729 DEBUG_FAIL_PRINT2 (" Popping low active reg: %d\n", low_reg); \ |
| 428 | 1730 \ |
| 1731 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ | |
| 1732 { \ | |
| 5041 | 1733 DEBUG_FAIL_PRINT2 (" Popping reg: %d\n", this_reg); \ |
| 428 | 1734 \ |
| 1735 reg_info[this_reg].word = POP_FAILURE_ELT (); \ | |
| 5041 | 1736 DEBUG_FAIL_PRINT2 (" info: 0x%lx\n", \ |
| 428 | 1737 * (long *) ®_info[this_reg]); \ |
| 1738 \ | |
| 446 | 1739 regend[this_reg] = POP_FAILURE_POINTER (); \ |
| 5041 | 1740 DEBUG_FAIL_PRINT2 (" end: 0x%lx\n", (long) regend[this_reg]); \ |
| 428 | 1741 \ |
| 446 | 1742 regstart[this_reg] = POP_FAILURE_POINTER (); \ |
| 5041 | 1743 DEBUG_FAIL_PRINT2 (" start: 0x%lx\n", (long) regstart[this_reg]); \ |
| 428 | 1744 } \ |
| 1745 \ | |
| 1746 set_regs_matched_done = 0; \ | |
| 1747 DEBUG_STATEMENT (nfailure_points_popped++); \ | |
| 456 | 1748 } while (0) /* POP_FAILURE_POINT */ |
| 428 | 1749 |
| 1750 | |
| 1751 | |
| 1752 /* Structure for per-register (a.k.a. per-group) information. | |
| 1753 Other register information, such as the | |
| 1754 starting and ending positions (which are addresses), and the list of | |
| 1755 inner groups (which is a bits list) are maintained in separate | |
| 1756 variables. | |
| 1757 | |
| 1758 We are making a (strictly speaking) nonportable assumption here: that | |
| 1759 the compiler will pack our bit fields into something that fits into | |
| 1760 the type of `word', i.e., is something that fits into one item on the | |
| 1761 failure stack. */ | |
| 1762 | |
| 1763 typedef union | |
| 1764 { | |
| 1765 fail_stack_elt_t word; | |
| 1766 struct | |
| 1767 { | |
| 1768 /* This field is one if this group can match the empty string, | |
| 1769 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ | |
| 1770 #define MATCH_NULL_UNSET_VALUE 3 | |
| 647 | 1771 unsigned int match_null_string_p : 2; |
| 1772 unsigned int is_active : 1; | |
| 1773 unsigned int matched_something : 1; | |
| 1774 unsigned int ever_matched_something : 1; | |
| 428 | 1775 } bits; |
| 1776 } register_info_type; | |
| 1777 | |
| 1778 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p) | |
| 1779 #define IS_ACTIVE(R) ((R).bits.is_active) | |
| 1780 #define MATCHED_SOMETHING(R) ((R).bits.matched_something) | |
| 1781 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something) | |
| 1782 | |
| 1783 | |
| 1784 /* Call this when have matched a real character; it sets `matched' flags | |
| 1785 for the subexpressions which we are currently inside. Also records | |
| 1786 that those subexprs have matched. */ | |
| 1787 #define SET_REGS_MATCHED() \ | |
| 1788 do \ | |
| 1789 { \ | |
| 1790 if (!set_regs_matched_done) \ | |
| 1791 { \ | |
| 647 | 1792 int r; \ |
| 428 | 1793 set_regs_matched_done = 1; \ |
| 1794 for (r = lowest_active_reg; r <= highest_active_reg; r++) \ | |
| 1795 { \ | |
| 1796 MATCHED_SOMETHING (reg_info[r]) \ | |
| 1797 = EVER_MATCHED_SOMETHING (reg_info[r]) \ | |
| 1798 = 1; \ | |
| 1799 } \ | |
| 1800 } \ | |
| 1801 } \ | |
| 1802 while (0) | |
| 1803 | |
| 1804 /* Registers are set to a sentinel when they haven't yet matched. */ | |
| 446 | 1805 static unsigned char reg_unset_dummy; |
| 428 | 1806 #define REG_UNSET_VALUE (®_unset_dummy) |
| 1807 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE) | |
| 1808 | |
| 1809 /* Subroutine declarations and macros for regex_compile. */ | |
| 1810 | |
| 1811 /* Fetch the next character in the uncompiled pattern---translating it | |
| 826 | 1812 if necessary. */ |
| 428 | 1813 #define PATFETCH(c) \ |
| 446 | 1814 do { \ |
| 1815 PATFETCH_RAW (c); \ | |
| 826 | 1816 c = RE_TRANSLATE (c); \ |
| 428 | 1817 } while (0) |
| 1818 | |
| 1819 /* Fetch the next character in the uncompiled pattern, with no | |
| 1820 translation. */ | |
| 1821 #define PATFETCH_RAW(c) \ | |
| 1822 do {if (p == pend) return REG_EEND; \ | |
| 1823 assert (p < pend); \ | |
| 867 | 1824 c = itext_ichar (p); \ |
| 1825 INC_IBYTEPTR (p); \ | |
| 428 | 1826 } while (0) |
| 1827 | |
| 1828 /* Go backwards one character in the pattern. */ | |
| 867 | 1829 #define PATUNFETCH DEC_IBYTEPTR (p) |
| 428 | 1830 |
| 1831 /* If `translate' is non-null, return translate[D], else just D. We | |
| 1832 cast the subscript to translate because some data is declared as | |
| 1833 `char *', to avoid warnings when a string constant is passed. But | |
| 1834 when we use a character as a subscript we must make it unsigned. */ | |
| 826 | 1835 #define RE_TRANSLATE(d) \ |
| 1836 (TRANSLATE_P (translate) ? RE_TRANSLATE_1 (d) : (d)) | |
| 428 | 1837 |
| 1838 /* Macros for outputting the compiled pattern into `buffer'. */ | |
| 1839 | |
| 1840 /* If the buffer isn't allocated when it comes in, use this. */ | |
| 1841 #define INIT_BUF_SIZE 32 | |
| 1842 | |
| 1843 /* Make sure we have at least N more bytes of space in buffer. */ | |
| 1844 #define GET_BUFFER_SPACE(n) \ | |
| 647 | 1845 while (buf_end - bufp->buffer + (n) > (ptrdiff_t) bufp->allocated) \ |
| 428 | 1846 EXTEND_BUFFER () |
| 1847 | |
| 1848 /* Make sure we have one more byte of buffer space and then add C to it. */ | |
| 1849 #define BUF_PUSH(c) \ | |
| 1850 do { \ | |
| 1851 GET_BUFFER_SPACE (1); \ | |
| 446 | 1852 *buf_end++ = (unsigned char) (c); \ |
| 428 | 1853 } while (0) |
| 1854 | |
| 1855 | |
| 1856 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */ | |
| 1857 #define BUF_PUSH_2(c1, c2) \ | |
| 1858 do { \ | |
| 1859 GET_BUFFER_SPACE (2); \ | |
| 446 | 1860 *buf_end++ = (unsigned char) (c1); \ |
| 1861 *buf_end++ = (unsigned char) (c2); \ | |
| 428 | 1862 } while (0) |
| 1863 | |
| 1864 | |
| 1865 /* As with BUF_PUSH_2, except for three bytes. */ | |
| 1866 #define BUF_PUSH_3(c1, c2, c3) \ | |
| 1867 do { \ | |
| 1868 GET_BUFFER_SPACE (3); \ | |
| 446 | 1869 *buf_end++ = (unsigned char) (c1); \ |
| 1870 *buf_end++ = (unsigned char) (c2); \ | |
| 1871 *buf_end++ = (unsigned char) (c3); \ | |
| 428 | 1872 } while (0) |
| 1873 | |
| 1874 | |
| 1875 /* Store a jump with opcode OP at LOC to location TO. We store a | |
| 1876 relative address offset by the three bytes the jump itself occupies. */ | |
| 1877 #define STORE_JUMP(op, loc, to) \ | |
| 1878 store_op1 (op, loc, (to) - (loc) - 3) | |
| 1879 | |
| 1880 /* Likewise, for a two-argument jump. */ | |
| 1881 #define STORE_JUMP2(op, loc, to, arg) \ | |
| 1882 store_op2 (op, loc, (to) - (loc) - 3, arg) | |
| 1883 | |
| 446 | 1884 /* Like `STORE_JUMP', but for inserting. Assume `buf_end' is the |
| 1885 buffer end. */ | |
| 428 | 1886 #define INSERT_JUMP(op, loc, to) \ |
| 446 | 1887 insert_op1 (op, loc, (to) - (loc) - 3, buf_end) |
| 1888 | |
| 1889 /* Like `STORE_JUMP2', but for inserting. Assume `buf_end' is the | |
| 1890 buffer end. */ | |
| 428 | 1891 #define INSERT_JUMP2(op, loc, to, arg) \ |
| 446 | 1892 insert_op2 (op, loc, (to) - (loc) - 3, arg, buf_end) |
| 428 | 1893 |
| 1894 | |
| 1895 /* This is not an arbitrary limit: the arguments which represent offsets | |
| 1896 into the pattern are two bytes long. So if 2^16 bytes turns out to | |
| 1897 be too small, many things would have to change. */ | |
| 1898 #define MAX_BUF_SIZE (1L << 16) | |
| 1899 | |
| 1900 | |
| 1901 /* Extend the buffer by twice its current size via realloc and | |
| 1902 reset the pointers that pointed into the old block to point to the | |
| 1903 correct places in the new one. If extending the buffer results in it | |
| 1904 being larger than MAX_BUF_SIZE, then flag memory exhausted. */ | |
| 1333 | 1905 #define EXTEND_BUFFER() \ |
| 1906 do { \ | |
| 1907 re_char *old_buffer = bufp->buffer; \ | |
| 1908 if (bufp->allocated == MAX_BUF_SIZE) \ | |
| 1909 return REG_ESIZE; \ | |
| 1910 bufp->allocated <<= 1; \ | |
| 1911 if (bufp->allocated > MAX_BUF_SIZE) \ | |
| 1912 bufp->allocated = MAX_BUF_SIZE; \ | |
| 1913 bufp->buffer = \ | |
| 1914 (unsigned char *) xrealloc (bufp->buffer, bufp->allocated); \ | |
| 1915 if (bufp->buffer == NULL) \ | |
| 1916 return REG_ESPACE; \ | |
| 1917 /* If the buffer moved, move all the pointers into it. */ \ | |
| 1918 if (old_buffer != bufp->buffer) \ | |
| 1919 { \ | |
| 1920 buf_end = (buf_end - old_buffer) + bufp->buffer; \ | |
| 1921 begalt = (begalt - old_buffer) + bufp->buffer; \ | |
| 1922 if (fixup_alt_jump) \ | |
| 1923 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer; \ | |
| 1924 if (laststart) \ | |
| 1925 laststart = (laststart - old_buffer) + bufp->buffer; \ | |
| 1926 if (pending_exact) \ | |
| 1927 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ | |
| 1928 } \ | |
| 428 | 1929 } while (0) |
| 1930 | |
| 1931 | |
| 1932 /* Since we have one byte reserved for the register number argument to | |
| 1933 {start,stop}_memory, the maximum number of groups we can report | |
| 1934 things about is what fits in that byte. */ | |
| 1935 #define MAX_REGNUM 255 | |
| 1936 | |
| 1937 /* But patterns can have more than `MAX_REGNUM' registers. We just | |
| 502 | 1938 ignore the excess. |
| 1939 #### not true! groups past this will fail in lots of ways, if we | |
| 1940 ever have to backtrack. | |
| 1941 */ | |
| 647 | 1942 typedef int regnum_t; |
| 428 | 1943 |
| 502 | 1944 #define INIT_REG_TRANSLATE_SIZE 5 |
| 428 | 1945 |
| 1946 /* Macros for the compile stack. */ | |
| 1947 | |
| 1948 /* Since offsets can go either forwards or backwards, this type needs to | |
| 1949 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ | |
| 1950 typedef int pattern_offset_t; | |
| 1951 | |
| 1952 typedef struct | |
| 1953 { | |
| 1954 pattern_offset_t begalt_offset; | |
| 1955 pattern_offset_t fixup_alt_jump; | |
| 1956 pattern_offset_t inner_group_offset; | |
| 1957 pattern_offset_t laststart_offset; | |
| 1958 regnum_t regnum; | |
| 1959 } compile_stack_elt_t; | |
| 1960 | |
| 1961 | |
| 1962 typedef struct | |
| 1963 { | |
| 1964 compile_stack_elt_t *stack; | |
| 647 | 1965 int size; |
| 1966 int avail; /* Offset of next open position. */ | |
| 428 | 1967 } compile_stack_type; |
| 1968 | |
| 1969 | |
| 1970 #define INIT_COMPILE_STACK_SIZE 32 | |
| 1971 | |
| 1972 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0) | |
| 1973 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) | |
| 1974 | |
| 1975 /* The next available element. */ | |
| 1976 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) | |
| 1977 | |
| 1978 | |
| 1979 /* Set the bit for character C in a bit vector. */ | |
| 1980 #define SET_LIST_BIT(c) \ | |
| 446 | 1981 (buf_end[((unsigned char) (c)) / BYTEWIDTH] \ |
| 428 | 1982 |= 1 << (((unsigned char) c) % BYTEWIDTH)) |
| 1983 | |
| 1984 #ifdef MULE | |
| 1985 | |
| 1986 /* Set the "bit" for character C in a range table. */ | |
| 1987 #define SET_RANGETAB_BIT(c) put_range_table (rtab, c, c, Qt) | |
| 1988 | |
| 1989 /* Set the "bit" for character c in the appropriate table. */ | |
| 1990 #define SET_EITHER_BIT(c) \ | |
| 1991 do { \ | |
| 1992 if (has_extended_chars) \ | |
| 1993 SET_RANGETAB_BIT (c); \ | |
| 1994 else \ | |
| 1995 SET_LIST_BIT (c); \ | |
| 1996 } while (0) | |
| 1997 | |
| 1998 #else /* not MULE */ | |
| 1999 | |
| 2000 #define SET_EITHER_BIT(c) SET_LIST_BIT (c) | |
| 2001 | |
| 2002 #endif | |
| 2003 | |
| 2004 | |
| 2005 /* Get the next unsigned number in the uncompiled pattern. */ | |
| 2006 #define GET_UNSIGNED_NUMBER(num) \ | |
| 2007 { if (p != pend) \ | |
| 2008 { \ | |
| 2009 PATFETCH (c); \ | |
| 2010 while (ISDIGIT (c)) \ | |
| 2011 { \ | |
| 2012 if (num < 0) \ | |
| 2013 num = 0; \ | |
| 2014 num = num * 10 + c - '0'; \ | |
| 2015 if (p == pend) \ | |
| 2016 break; \ | |
| 2017 PATFETCH (c); \ | |
| 2018 } \ | |
| 2019 } \ | |
| 2020 } | |
| 2021 | |
| 2022 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ | |
| 2023 | |
| 2024 #define IS_CHAR_CLASS(string) \ | |
| 2025 (STREQ (string, "alpha") || STREQ (string, "upper") \ | |
| 2026 || STREQ (string, "lower") || STREQ (string, "digit") \ | |
| 2027 || STREQ (string, "alnum") || STREQ (string, "xdigit") \ | |
| 2028 || STREQ (string, "space") || STREQ (string, "print") \ | |
| 2029 || STREQ (string, "punct") || STREQ (string, "graph") \ | |
| 2030 || STREQ (string, "cntrl") || STREQ (string, "blank")) | |
| 2031 | |
| 2032 static void store_op1 (re_opcode_t op, unsigned char *loc, int arg); | |
| 2033 static void store_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2); | |
| 2034 static void insert_op1 (re_opcode_t op, unsigned char *loc, int arg, | |
| 2035 unsigned char *end); | |
| 2036 static void insert_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2, | |
| 2037 unsigned char *end); | |
| 460 | 2038 static re_bool at_begline_loc_p (re_char *pattern, re_char *p, |
| 428 | 2039 reg_syntax_t syntax); |
| 460 | 2040 static re_bool at_endline_loc_p (re_char *p, re_char *pend, int syntax); |
| 2041 static re_bool group_in_compile_stack (compile_stack_type compile_stack, | |
| 428 | 2042 regnum_t regnum); |
| 446 | 2043 static reg_errcode_t compile_range (re_char **p_ptr, re_char *pend, |
| 2044 RE_TRANSLATE_TYPE translate, | |
| 2045 reg_syntax_t syntax, | |
| 428 | 2046 unsigned char *b); |
| 2047 #ifdef MULE | |
| 446 | 2048 static reg_errcode_t compile_extended_range (re_char **p_ptr, |
| 2049 re_char *pend, | |
| 2050 RE_TRANSLATE_TYPE translate, | |
| 428 | 2051 reg_syntax_t syntax, |
| 2052 Lisp_Object rtab); | |
| 2053 #endif /* MULE */ | |
| 460 | 2054 static re_bool group_match_null_string_p (unsigned char **p, |
| 428 | 2055 unsigned char *end, |
| 2056 register_info_type *reg_info); | |
| 460 | 2057 static re_bool alt_match_null_string_p (unsigned char *p, unsigned char *end, |
| 428 | 2058 register_info_type *reg_info); |
| 460 | 2059 static re_bool common_op_match_null_string_p (unsigned char **p, |
| 428 | 2060 unsigned char *end, |
| 2061 register_info_type *reg_info); | |
| 826 | 2062 static int bcmp_translate (re_char *s1, re_char *s2, |
| 2063 REGISTER int len, RE_TRANSLATE_TYPE translate | |
| 2064 #ifdef emacs | |
| 2065 , Internal_Format fmt, Lisp_Object lispobj | |
| 2066 #endif | |
| 2067 ); | |
| 428 | 2068 static int re_match_2_internal (struct re_pattern_buffer *bufp, |
| 446 | 2069 re_char *string1, int size1, |
| 2070 re_char *string2, int size2, int pos, | |
| 826 | 2071 struct re_registers *regs, int stop |
| 2072 RE_LISP_CONTEXT_ARGS_DECL); | |
| 428 | 2073 |
| 2074 #ifndef MATCH_MAY_ALLOCATE | |
| 2075 | |
| 2076 /* If we cannot allocate large objects within re_match_2_internal, | |
| 2077 we make the fail stack and register vectors global. | |
| 2078 The fail stack, we grow to the maximum size when a regexp | |
| 2079 is compiled. | |
| 2080 The register vectors, we adjust in size each time we | |
| 2081 compile a regexp, according to the number of registers it needs. */ | |
| 2082 | |
| 2083 static fail_stack_type fail_stack; | |
| 2084 | |
| 2085 /* Size with which the following vectors are currently allocated. | |
| 2086 That is so we can make them bigger as needed, | |
| 2087 but never make them smaller. */ | |
| 2088 static int regs_allocated_size; | |
| 2089 | |
| 446 | 2090 static re_char ** regstart, ** regend; |
| 2091 static re_char ** old_regstart, ** old_regend; | |
| 2092 static re_char **best_regstart, **best_regend; | |
| 428 | 2093 static register_info_type *reg_info; |
| 446 | 2094 static re_char **reg_dummy; |
| 428 | 2095 static register_info_type *reg_info_dummy; |
| 2096 | |
| 2097 /* Make the register vectors big enough for NUM_REGS registers, | |
| 2098 but don't make them smaller. */ | |
| 2099 | |
| 2100 static | |
| 2101 regex_grow_registers (int num_regs) | |
| 2102 { | |
| 2103 if (num_regs > regs_allocated_size) | |
| 2104 { | |
| 551 | 2105 RETALLOC (regstart, num_regs, re_char *); |
| 2106 RETALLOC (regend, num_regs, re_char *); | |
| 2107 RETALLOC (old_regstart, num_regs, re_char *); | |
| 2108 RETALLOC (old_regend, num_regs, re_char *); | |
| 2109 RETALLOC (best_regstart, num_regs, re_char *); | |
| 2110 RETALLOC (best_regend, num_regs, re_char *); | |
| 2111 RETALLOC (reg_info, num_regs, register_info_type); | |
| 2112 RETALLOC (reg_dummy, num_regs, re_char *); | |
| 2113 RETALLOC (reg_info_dummy, num_regs, register_info_type); | |
| 428 | 2114 |
| 2115 regs_allocated_size = num_regs; | |
| 2116 } | |
| 2117 } | |
| 2118 | |
| 2119 #endif /* not MATCH_MAY_ALLOCATE */ | |
| 2120 | |
| 2121 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX. | |
| 2122 Returns one of error codes defined in `regex.h', or zero for success. | |
| 2123 | |
| 2124 Assumes the `allocated' (and perhaps `buffer') and `translate' | |
| 2125 fields are set in BUFP on entry. | |
| 2126 | |
| 2127 If it succeeds, results are put in BUFP (if it returns an error, the | |
| 2128 contents of BUFP are undefined): | |
| 2129 `buffer' is the compiled pattern; | |
| 2130 `syntax' is set to SYNTAX; | |
| 2131 `used' is set to the length of the compiled pattern; | |
| 2132 `fastmap_accurate' is zero; | |
| 502 | 2133 `re_ngroups' is the number of groups/subexpressions (including shy |
| 2134 groups) in PATTERN; | |
| 2135 `re_nsub' is the number of non-shy groups in PATTERN; | |
| 428 | 2136 `not_bol' and `not_eol' are zero; |
| 2137 | |
| 2138 The `fastmap' and `newline_anchor' fields are neither | |
| 2139 examined nor set. */ | |
| 2140 | |
| 2141 /* Return, freeing storage we allocated. */ | |
| 1726 | 2142 #define FREE_STACK_RETURN(value) \ |
| 2143 do \ | |
| 2144 { \ | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
2145 xfree (compile_stack.stack); \ |
| 1726 | 2146 return value; \ |
| 1333 | 2147 } while (0) |
| 428 | 2148 |
| 2149 static reg_errcode_t | |
| 446 | 2150 regex_compile (re_char *pattern, int size, reg_syntax_t syntax, |
| 428 | 2151 struct re_pattern_buffer *bufp) |
| 2152 { | |
| 2153 /* We fetch characters from PATTERN here. We declare these as int | |
| 2154 (or possibly long) so that chars above 127 can be used as | |
| 2155 array indices. The macros that fetch a character from the pattern | |
| 2156 make sure to coerce to unsigned char before assigning, so we won't | |
| 2157 get bitten by negative numbers here. */ | |
| 2158 /* XEmacs change: used to be unsigned char. */ | |
| 2159 REGISTER EMACS_INT c, c1; | |
| 2160 | |
| 2161 /* A random temporary spot in PATTERN. */ | |
| 446 | 2162 re_char *p1; |
| 428 | 2163 |
| 2164 /* Points to the end of the buffer, where we should append. */ | |
| 446 | 2165 REGISTER unsigned char *buf_end; |
| 428 | 2166 |
| 2167 /* Keeps track of unclosed groups. */ | |
| 2168 compile_stack_type compile_stack; | |
| 2169 | |
| 2170 /* Points to the current (ending) position in the pattern. */ | |
| 446 | 2171 re_char *p = pattern; |
| 2172 re_char *pend = pattern + size; | |
| 428 | 2173 |
| 2174 /* How to translate the characters in the pattern. */ | |
| 446 | 2175 RE_TRANSLATE_TYPE translate = bufp->translate; |
| 428 | 2176 |
| 2177 /* Address of the count-byte of the most recently inserted `exactn' | |
| 2178 command. This makes it possible to tell if a new exact-match | |
| 2179 character can be added to that command or if the character requires | |
| 2180 a new `exactn' command. */ | |
| 2181 unsigned char *pending_exact = 0; | |
| 2182 | |
| 2183 /* Address of start of the most recently finished expression. | |
| 2184 This tells, e.g., postfix * where to find the start of its | |
| 2185 operand. Reset at the beginning of groups and alternatives. */ | |
| 2186 unsigned char *laststart = 0; | |
| 2187 | |
| 2188 /* Address of beginning of regexp, or inside of last group. */ | |
| 2189 unsigned char *begalt; | |
| 2190 | |
| 2191 /* Place in the uncompiled pattern (i.e., the {) to | |
| 2192 which to go back if the interval is invalid. */ | |
| 446 | 2193 re_char *beg_interval; |
| 428 | 2194 |
| 2195 /* Address of the place where a forward jump should go to the end of | |
| 2196 the containing expression. Each alternative of an `or' -- except the | |
| 2197 last -- ends with a forward jump of this sort. */ | |
| 2198 unsigned char *fixup_alt_jump = 0; | |
| 2199 | |
| 2200 /* Counts open-groups as they are encountered. Remembered for the | |
| 2201 matching close-group on the compile stack, so the same register | |
| 2202 number is put in the stop_memory as the start_memory. */ | |
| 2203 regnum_t regnum = 0; | |
| 2204 | |
| 2205 #ifdef DEBUG | |
| 5041 | 2206 if (debug_regexps & RE_DEBUG_COMPILATION) |
| 428 | 2207 { |
| 647 | 2208 int debug_count; |
| 428 | 2209 |
| 5041 | 2210 DEBUG_PRINT1 ("\nCompiling pattern: "); |
| 428 | 2211 for (debug_count = 0; debug_count < size; debug_count++) |
| 2212 putchar (pattern[debug_count]); | |
| 2213 putchar ('\n'); | |
| 2214 } | |
| 2215 #endif /* DEBUG */ | |
| 2216 | |
| 2217 /* Initialize the compile stack. */ | |
| 2218 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t); | |
| 2219 if (compile_stack.stack == NULL) | |
| 2220 return REG_ESPACE; | |
| 2221 | |
| 2222 compile_stack.size = INIT_COMPILE_STACK_SIZE; | |
| 2223 compile_stack.avail = 0; | |
| 2224 | |
| 2225 /* Initialize the pattern buffer. */ | |
| 2226 bufp->syntax = syntax; | |
| 2227 bufp->fastmap_accurate = 0; | |
| 2228 bufp->not_bol = bufp->not_eol = 0; | |
| 2229 | |
| 2230 /* Set `used' to zero, so that if we return an error, the pattern | |
| 2231 printer (for debugging) will think there's no pattern. We reset it | |
| 2232 at the end. */ | |
| 2233 bufp->used = 0; | |
| 2234 | |
| 2235 /* Always count groups, whether or not bufp->no_sub is set. */ | |
| 2236 bufp->re_nsub = 0; | |
| 502 | 2237 bufp->re_ngroups = 0; |
| 2238 | |
| 2239 bufp->warned_about_incompatible_back_references = 0; | |
| 2240 | |
| 2241 if (bufp->external_to_internal_register == 0) | |
| 2242 { | |
| 2243 bufp->external_to_internal_register_size = INIT_REG_TRANSLATE_SIZE; | |
| 2244 RETALLOC (bufp->external_to_internal_register, | |
| 2245 bufp->external_to_internal_register_size, | |
| 2246 int); | |
| 2247 } | |
| 2248 | |
| 2249 { | |
| 2250 int i; | |
| 2251 | |
| 2252 bufp->external_to_internal_register[0] = 0; | |
| 2253 for (i = 1; i < bufp->external_to_internal_register_size; i++) | |
| 2254 bufp->external_to_internal_register[i] = (int) 0xDEADBEEF; | |
| 2255 } | |
| 428 | 2256 |
| 2257 #if !defined (emacs) && !defined (SYNTAX_TABLE) | |
| 2258 /* Initialize the syntax table. */ | |
| 2259 init_syntax_once (); | |
| 2260 #endif | |
| 2261 | |
| 2262 if (bufp->allocated == 0) | |
| 2263 { | |
| 2264 if (bufp->buffer) | |
| 2265 { /* If zero allocated, but buffer is non-null, try to realloc | |
| 2266 enough space. This loses if buffer's address is bogus, but | |
| 2267 that is the user's responsibility. */ | |
| 2268 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); | |
| 2269 } | |
| 2270 else | |
| 2271 { /* Caller did not allocate a buffer. Do it for them. */ | |
| 2272 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); | |
| 2273 } | |
| 2274 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE); | |
| 2275 | |
| 2276 bufp->allocated = INIT_BUF_SIZE; | |
| 2277 } | |
| 2278 | |
| 446 | 2279 begalt = buf_end = bufp->buffer; |
| 428 | 2280 |
| 2281 /* Loop through the uncompiled pattern until we're at the end. */ | |
| 2282 while (p != pend) | |
| 2283 { | |
| 2284 PATFETCH (c); | |
| 2285 | |
| 2286 switch (c) | |
| 2287 { | |
| 2288 case '^': | |
| 2289 { | |
| 2290 if ( /* If at start of pattern, it's an operator. */ | |
| 2291 p == pattern + 1 | |
| 2292 /* If context independent, it's an operator. */ | |
| 2293 || syntax & RE_CONTEXT_INDEP_ANCHORS | |
| 2294 /* Otherwise, depends on what's come before. */ | |
| 2295 || at_begline_loc_p (pattern, p, syntax)) | |
| 2296 BUF_PUSH (begline); | |
| 2297 else | |
| 2298 goto normal_char; | |
| 2299 } | |
| 2300 break; | |
| 2301 | |
| 2302 | |
| 2303 case '$': | |
| 2304 { | |
| 2305 if ( /* If at end of pattern, it's an operator. */ | |
| 2306 p == pend | |
| 2307 /* If context independent, it's an operator. */ | |
| 2308 || syntax & RE_CONTEXT_INDEP_ANCHORS | |
| 2309 /* Otherwise, depends on what's next. */ | |
| 2310 || at_endline_loc_p (p, pend, syntax)) | |
| 2311 BUF_PUSH (endline); | |
| 2312 else | |
| 2313 goto normal_char; | |
| 2314 } | |
| 2315 break; | |
| 2316 | |
| 2317 | |
| 2318 case '+': | |
| 2319 case '?': | |
| 2320 if ((syntax & RE_BK_PLUS_QM) | |
| 2321 || (syntax & RE_LIMITED_OPS)) | |
| 2322 goto normal_char; | |
| 2323 handle_plus: | |
| 2324 case '*': | |
| 2325 /* If there is no previous pattern... */ | |
| 2326 if (!laststart) | |
| 2327 { | |
| 2328 if (syntax & RE_CONTEXT_INVALID_OPS) | |
| 2329 FREE_STACK_RETURN (REG_BADRPT); | |
| 2330 else if (!(syntax & RE_CONTEXT_INDEP_OPS)) | |
| 2331 goto normal_char; | |
| 2332 } | |
| 2333 | |
| 2334 { | |
| 2335 /* true means zero/many matches are allowed. */ | |
| 460 | 2336 re_bool zero_times_ok = c != '+'; |
| 2337 re_bool many_times_ok = c != '?'; | |
| 428 | 2338 |
| 2339 /* true means match shortest string possible. */ | |
| 460 | 2340 re_bool minimal = false; |
| 428 | 2341 |
| 2342 /* If there is a sequence of repetition chars, collapse it | |
| 2343 down to just one (the right one). We can't combine | |
| 2344 interval operators with these because of, e.g., `a{2}*', | |
| 2345 which should only match an even number of `a's. */ | |
| 2346 while (p != pend) | |
| 2347 { | |
| 2348 PATFETCH (c); | |
| 2349 | |
| 2350 if (c == '*' || (!(syntax & RE_BK_PLUS_QM) | |
| 2351 && (c == '+' || c == '?'))) | |
| 2352 ; | |
| 2353 | |
| 2354 else if (syntax & RE_BK_PLUS_QM && c == '\\') | |
| 2355 { | |
| 2356 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | |
| 2357 | |
| 2358 PATFETCH (c1); | |
| 2359 if (!(c1 == '+' || c1 == '?')) | |
| 2360 { | |
| 2361 PATUNFETCH; | |
| 2362 PATUNFETCH; | |
| 2363 break; | |
| 2364 } | |
| 2365 | |
| 2366 c = c1; | |
| 2367 } | |
| 2368 else | |
| 2369 { | |
| 2370 PATUNFETCH; | |
| 2371 break; | |
| 2372 } | |
| 2373 | |
| 2374 /* If we get here, we found another repeat character. */ | |
| 2375 if (!(syntax & RE_NO_MINIMAL_MATCHING)) | |
| 2376 { | |
| 440 | 2377 /* "*?" and "+?" and "??" are okay (and mean match |
| 2378 minimally), but other sequences (such as "*??" and | |
| 2379 "+++") are rejected (reserved for future use). */ | |
| 428 | 2380 if (minimal || c != '?') |
| 2381 FREE_STACK_RETURN (REG_BADRPT); | |
| 2382 minimal = true; | |
| 2383 } | |
| 2384 else | |
| 2385 { | |
| 2386 zero_times_ok |= c != '+'; | |
| 2387 many_times_ok |= c != '?'; | |
| 2388 } | |
| 2389 } | |
| 2390 | |
| 2391 /* Star, etc. applied to an empty pattern is equivalent | |
| 2392 to an empty pattern. */ | |
| 2393 if (!laststart) | |
| 2394 break; | |
| 2395 | |
| 2396 /* Now we know whether zero matches is allowed | |
| 2397 and whether two or more matches is allowed | |
| 2398 and whether we want minimal or maximal matching. */ | |
| 2399 if (minimal) | |
| 2400 { | |
| 2401 if (!many_times_ok) | |
| 2402 { | |
| 2403 /* "a??" becomes: | |
| 2404 0: /on_failure_jump to 6 | |
| 2405 3: /jump to 9 | |
| 2406 6: /exactn/1/A | |
| 2407 9: end of pattern. | |
| 2408 */ | |
| 2409 GET_BUFFER_SPACE (6); | |
| 446 | 2410 INSERT_JUMP (jump, laststart, buf_end + 3); |
| 2411 buf_end += 3; | |
| 428 | 2412 INSERT_JUMP (on_failure_jump, laststart, laststart + 6); |
| 446 | 2413 buf_end += 3; |
| 428 | 2414 } |
| 2415 else if (zero_times_ok) | |
| 2416 { | |
| 2417 /* "a*?" becomes: | |
| 2418 0: /jump to 6 | |
| 2419 3: /exactn/1/A | |
| 2420 6: /on_failure_jump to 3 | |
| 2421 9: end of pattern. | |
| 2422 */ | |
| 2423 GET_BUFFER_SPACE (6); | |
| 446 | 2424 INSERT_JUMP (jump, laststart, buf_end + 3); |
| 2425 buf_end += 3; | |
| 2426 STORE_JUMP (on_failure_jump, buf_end, laststart + 3); | |
| 2427 buf_end += 3; | |
| 428 | 2428 } |
| 2429 else | |
| 2430 { | |
| 2431 /* "a+?" becomes: | |
| 2432 0: /exactn/1/A | |
| 2433 3: /on_failure_jump to 0 | |
| 2434 6: end of pattern. | |
| 2435 */ | |
| 2436 GET_BUFFER_SPACE (3); | |
| 446 | 2437 STORE_JUMP (on_failure_jump, buf_end, laststart); |
| 2438 buf_end += 3; | |
| 428 | 2439 } |
| 2440 } | |
| 2441 else | |
| 2442 { | |
| 2443 /* Are we optimizing this jump? */ | |
| 460 | 2444 re_bool keep_string_p = false; |
| 428 | 2445 |
| 2446 if (many_times_ok) | |
| 446 | 2447 { /* More than one repetition is allowed, so put in |
| 2448 at the end a backward relative jump from | |
| 2449 `buf_end' to before the next jump we're going | |
| 2450 to put in below (which jumps from laststart to | |
| 2451 after this jump). | |
| 428 | 2452 |
| 2453 But if we are at the `*' in the exact sequence `.*\n', | |
| 2454 insert an unconditional jump backwards to the ., | |
| 2455 instead of the beginning of the loop. This way we only | |
| 2456 push a failure point once, instead of every time | |
| 2457 through the loop. */ | |
| 2458 assert (p - 1 > pattern); | |
| 2459 | |
| 2460 /* Allocate the space for the jump. */ | |
| 2461 GET_BUFFER_SPACE (3); | |
| 2462 | |
| 2463 /* We know we are not at the first character of the | |
| 2464 pattern, because laststart was nonzero. And we've | |
| 2465 already incremented `p', by the way, to be the | |
| 2466 character after the `*'. Do we have to do something | |
| 2467 analogous here for null bytes, because of | |
| 2468 RE_DOT_NOT_NULL? */ | |
| 446 | 2469 if (*(p - 2) == '.' |
| 428 | 2470 && zero_times_ok |
| 446 | 2471 && p < pend && *p == '\n' |
| 428 | 2472 && !(syntax & RE_DOT_NEWLINE)) |
| 2473 { /* We have .*\n. */ | |
| 446 | 2474 STORE_JUMP (jump, buf_end, laststart); |
| 428 | 2475 keep_string_p = true; |
| 2476 } | |
| 2477 else | |
| 2478 /* Anything else. */ | |
| 446 | 2479 STORE_JUMP (maybe_pop_jump, buf_end, laststart - 3); |
| 428 | 2480 |
| 2481 /* We've added more stuff to the buffer. */ | |
| 446 | 2482 buf_end += 3; |
| 428 | 2483 } |
| 2484 | |
| 446 | 2485 /* On failure, jump from laststart to buf_end + 3, |
| 2486 which will be the end of the buffer after this jump | |
| 2487 is inserted. */ | |
| 428 | 2488 GET_BUFFER_SPACE (3); |
| 2489 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump | |
| 2490 : on_failure_jump, | |
| 446 | 2491 laststart, buf_end + 3); |
| 2492 buf_end += 3; | |
| 428 | 2493 |
| 2494 if (!zero_times_ok) | |
| 2495 { | |
| 2496 /* At least one repetition is required, so insert a | |
| 2497 `dummy_failure_jump' before the initial | |
| 2498 `on_failure_jump' instruction of the loop. This | |
| 2499 effects a skip over that instruction the first time | |
| 2500 we hit that loop. */ | |
| 2501 GET_BUFFER_SPACE (3); | |
| 2502 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); | |
| 446 | 2503 buf_end += 3; |
| 428 | 2504 } |
| 2505 } | |
| 2506 pending_exact = 0; | |
| 2507 } | |
| 2508 break; | |
| 2509 | |
| 2510 | |
| 2511 case '.': | |
| 446 | 2512 laststart = buf_end; |
| 428 | 2513 BUF_PUSH (anychar); |
| 2514 break; | |
| 2515 | |
| 2516 | |
| 2517 case '[': | |
| 2518 { | |
| 2519 /* XEmacs change: this whole section */ | |
| 460 | 2520 re_bool had_char_class = false; |
| 428 | 2521 #ifdef MULE |
| 460 | 2522 re_bool has_extended_chars = false; |
| 428 | 2523 REGISTER Lisp_Object rtab = Qnil; |
| 2524 #endif | |
| 2525 | |
| 2526 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2527 | |
| 2528 /* Ensure that we have enough space to push a charset: the | |
| 2529 opcode, the length count, and the bitset; 34 bytes in all. */ | |
| 2530 GET_BUFFER_SPACE (34); | |
| 2531 | |
| 446 | 2532 laststart = buf_end; |
| 428 | 2533 |
| 2534 /* We test `*p == '^' twice, instead of using an if | |
| 2535 statement, so we only need one BUF_PUSH. */ | |
| 2536 BUF_PUSH (*p == '^' ? charset_not : charset); | |
| 2537 if (*p == '^') | |
| 2538 p++; | |
| 2539 | |
| 2540 /* Remember the first position in the bracket expression. */ | |
| 2541 p1 = p; | |
| 2542 | |
| 2543 /* Push the number of bytes in the bitmap. */ | |
| 2544 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); | |
| 2545 | |
| 2546 /* Clear the whole map. */ | |
| 446 | 2547 memset (buf_end, 0, (1 << BYTEWIDTH) / BYTEWIDTH); |
| 428 | 2548 |
| 2549 /* charset_not matches newline according to a syntax bit. */ | |
| 446 | 2550 if ((re_opcode_t) buf_end[-2] == charset_not |
| 428 | 2551 && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) |
| 2552 SET_LIST_BIT ('\n'); | |
| 2553 | |
| 2554 #ifdef MULE | |
| 2555 start_over_with_extended: | |
| 2556 if (has_extended_chars) | |
| 2557 { | |
| 2558 /* There are extended chars here, which means we need to start | |
| 2559 over and shift to unified range-table format. */ | |
| 446 | 2560 if (buf_end[-2] == charset) |
| 2561 buf_end[-2] = charset_mule; | |
| 428 | 2562 else |
| 446 | 2563 buf_end[-2] = charset_mule_not; |
| 2564 buf_end--; | |
| 428 | 2565 p = p1; /* go back to the beginning of the charset, after |
| 2566 a possible ^. */ | |
| 2567 rtab = Vthe_lisp_rangetab; | |
| 2568 Fclear_range_table (rtab); | |
| 2569 | |
| 2570 /* charset_not matches newline according to a syntax bit. */ | |
| 446 | 2571 if ((re_opcode_t) buf_end[-1] == charset_mule_not |
| 428 | 2572 && (syntax & RE_HAT_LISTS_NOT_NEWLINE)) |
| 2573 SET_EITHER_BIT ('\n'); | |
| 2574 } | |
| 2575 #endif /* MULE */ | |
| 2576 | |
| 2577 /* Read in characters and ranges, setting map bits. */ | |
| 2578 for (;;) | |
| 2579 { | |
| 2580 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2581 | |
| 446 | 2582 PATFETCH (c); |
| 428 | 2583 |
| 2584 #ifdef MULE | |
| 2585 if (c >= 0x80 && !has_extended_chars) | |
| 2586 { | |
| 2587 has_extended_chars = 1; | |
| 2588 /* Frumble-bumble, we've found some extended chars. | |
| 2589 Need to start over, process everything using | |
| 2590 the general extended-char mechanism, and need | |
| 2591 to use charset_mule and charset_mule_not instead | |
| 2592 of charset and charset_not. */ | |
| 2593 goto start_over_with_extended; | |
| 2594 } | |
| 2595 #endif /* MULE */ | |
| 2596 /* \ might escape characters inside [...] and [^...]. */ | |
| 2597 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') | |
| 2598 { | |
| 2599 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | |
| 2600 | |
| 446 | 2601 PATFETCH (c1); |
| 428 | 2602 #ifdef MULE |
| 2603 if (c1 >= 0x80 && !has_extended_chars) | |
| 2604 { | |
| 2605 has_extended_chars = 1; | |
| 2606 goto start_over_with_extended; | |
| 2607 } | |
| 2608 #endif /* MULE */ | |
| 2609 SET_EITHER_BIT (c1); | |
| 2610 continue; | |
| 2611 } | |
| 2612 | |
| 2613 /* Could be the end of the bracket expression. If it's | |
| 2614 not (i.e., when the bracket expression is `[]' so | |
| 2615 far), the ']' character bit gets set way below. */ | |
| 2616 if (c == ']' && p != p1 + 1) | |
| 2617 break; | |
| 2618 | |
| 2619 /* Look ahead to see if it's a range when the last thing | |
| 2620 was a character class. */ | |
| 2621 if (had_char_class && c == '-' && *p != ']') | |
| 2622 FREE_STACK_RETURN (REG_ERANGE); | |
| 2623 | |
| 2624 /* Look ahead to see if it's a range when the last thing | |
| 2625 was a character: if this is a hyphen not at the | |
| 2626 beginning or the end of a list, then it's the range | |
| 2627 operator. */ | |
| 2628 if (c == '-' | |
| 2629 && !(p - 2 >= pattern && p[-2] == '[') | |
| 446 | 2630 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') |
| 428 | 2631 && *p != ']') |
| 2632 { | |
| 2633 reg_errcode_t ret; | |
| 2634 | |
| 2635 #ifdef MULE | |
| 2636 if (* (unsigned char *) p >= 0x80 && !has_extended_chars) | |
| 2637 { | |
| 2638 has_extended_chars = 1; | |
| 2639 goto start_over_with_extended; | |
| 2640 } | |
| 2641 if (has_extended_chars) | |
| 2642 ret = compile_extended_range (&p, pend, translate, | |
| 2643 syntax, rtab); | |
| 2644 else | |
| 2645 #endif /* MULE */ | |
| 446 | 2646 ret = compile_range (&p, pend, translate, syntax, buf_end); |
| 428 | 2647 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); |
| 2648 } | |
| 2649 | |
| 2650 else if (p[0] == '-' && p[1] != ']') | |
| 2651 { /* This handles ranges made up of characters only. */ | |
| 2652 reg_errcode_t ret; | |
| 2653 | |
| 2654 /* Move past the `-'. */ | |
| 2655 PATFETCH (c1); | |
| 2656 | |
| 2657 #ifdef MULE | |
| 2658 if (* (unsigned char *) p >= 0x80 && !has_extended_chars) | |
| 2659 { | |
| 2660 has_extended_chars = 1; | |
| 2661 goto start_over_with_extended; | |
| 2662 } | |
| 2663 if (has_extended_chars) | |
| 2664 ret = compile_extended_range (&p, pend, translate, | |
| 2665 syntax, rtab); | |
| 2666 else | |
| 2667 #endif /* MULE */ | |
| 446 | 2668 ret = compile_range (&p, pend, translate, syntax, buf_end); |
| 428 | 2669 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret); |
| 2670 } | |
| 2671 | |
| 2672 /* See if we're at the beginning of a possible character | |
| 2673 class. */ | |
| 2674 | |
| 2675 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') | |
| 2676 { /* Leave room for the null. */ | |
| 2677 char str[CHAR_CLASS_MAX_LENGTH + 1]; | |
| 2678 | |
| 2679 PATFETCH (c); | |
| 2680 c1 = 0; | |
| 2681 | |
| 2682 /* If pattern is `[[:'. */ | |
| 2683 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2684 | |
| 2685 for (;;) | |
| 2686 { | |
| 446 | 2687 /* #### This code is unused. |
| 2688 Correctness is not checked after TRT | |
| 2689 table change. */ | |
| 428 | 2690 PATFETCH (c); |
| 2691 if (c == ':' || c == ']' || p == pend | |
| 2692 || c1 == CHAR_CLASS_MAX_LENGTH) | |
| 2693 break; | |
| 442 | 2694 str[c1++] = (char) c; |
| 428 | 2695 } |
| 2696 str[c1] = '\0'; | |
| 2697 | |
| 446 | 2698 /* If isn't a word bracketed by `[:' and `:]': |
| 428 | 2699 undo the ending character, the letters, and leave |
| 2700 the leading `:' and `[' (but set bits for them). */ | |
| 2701 if (c == ':' && *p == ']') | |
| 2702 { | |
| 2703 int ch; | |
| 460 | 2704 re_bool is_alnum = STREQ (str, "alnum"); |
| 2705 re_bool is_alpha = STREQ (str, "alpha"); | |
| 2706 re_bool is_blank = STREQ (str, "blank"); | |
| 2707 re_bool is_cntrl = STREQ (str, "cntrl"); | |
| 2708 re_bool is_digit = STREQ (str, "digit"); | |
| 2709 re_bool is_graph = STREQ (str, "graph"); | |
| 2710 re_bool is_lower = STREQ (str, "lower"); | |
| 2711 re_bool is_print = STREQ (str, "print"); | |
| 2712 re_bool is_punct = STREQ (str, "punct"); | |
| 2713 re_bool is_space = STREQ (str, "space"); | |
| 2714 re_bool is_upper = STREQ (str, "upper"); | |
| 2715 re_bool is_xdigit = STREQ (str, "xdigit"); | |
| 428 | 2716 |
| 2717 if (!IS_CHAR_CLASS (str)) | |
| 2718 FREE_STACK_RETURN (REG_ECTYPE); | |
| 2719 | |
| 2720 /* Throw away the ] at the end of the character | |
| 2721 class. */ | |
| 2722 PATFETCH (c); | |
| 2723 | |
| 2724 if (p == pend) FREE_STACK_RETURN (REG_EBRACK); | |
| 2725 | |
| 2726 for (ch = 0; ch < 1 << BYTEWIDTH; ch++) | |
| 2727 { | |
| 2728 /* This was split into 3 if's to | |
| 2729 avoid an arbitrary limit in some compiler. */ | |
| 2730 if ( (is_alnum && ISALNUM (ch)) | |
| 2731 || (is_alpha && ISALPHA (ch)) | |
| 2732 || (is_blank && ISBLANK (ch)) | |
| 2733 || (is_cntrl && ISCNTRL (ch))) | |
| 2734 SET_EITHER_BIT (ch); | |
| 2735 if ( (is_digit && ISDIGIT (ch)) | |
| 2736 || (is_graph && ISGRAPH (ch)) | |
| 2737 || (is_lower && ISLOWER (ch)) | |
| 2738 || (is_print && ISPRINT (ch))) | |
| 2739 SET_EITHER_BIT (ch); | |
| 2740 if ( (is_punct && ISPUNCT (ch)) | |
| 2741 || (is_space && ISSPACE (ch)) | |
| 2742 || (is_upper && ISUPPER (ch)) | |
| 2743 || (is_xdigit && ISXDIGIT (ch))) | |
| 2744 SET_EITHER_BIT (ch); | |
| 2745 } | |
| 2746 had_char_class = true; | |
| 2747 } | |
| 2748 else | |
| 2749 { | |
| 2750 c1++; | |
| 2751 while (c1--) | |
| 2752 PATUNFETCH; | |
| 2753 SET_EITHER_BIT ('['); | |
| 2754 SET_EITHER_BIT (':'); | |
| 2755 had_char_class = false; | |
| 2756 } | |
| 2757 } | |
| 2758 else | |
| 2759 { | |
| 2760 had_char_class = false; | |
| 2761 SET_EITHER_BIT (c); | |
| 2762 } | |
| 2763 } | |
| 2764 | |
| 2765 #ifdef MULE | |
| 2766 if (has_extended_chars) | |
| 2767 { | |
| 2768 /* We have a range table, not a bit vector. */ | |
| 2769 int bytes_needed = | |
| 2770 unified_range_table_bytes_needed (rtab); | |
| 2771 GET_BUFFER_SPACE (bytes_needed); | |
| 446 | 2772 unified_range_table_copy_data (rtab, buf_end); |
| 2773 buf_end += unified_range_table_bytes_used (buf_end); | |
| 428 | 2774 break; |
| 2775 } | |
| 2776 #endif /* MULE */ | |
| 2777 /* Discard any (non)matching list bytes that are all 0 at the | |
| 2778 end of the map. Decrease the map-length byte too. */ | |
| 446 | 2779 while ((int) buf_end[-1] > 0 && buf_end[buf_end[-1] - 1] == 0) |
| 2780 buf_end[-1]--; | |
| 2781 buf_end += buf_end[-1]; | |
| 428 | 2782 } |
| 2783 break; | |
| 2784 | |
| 2785 | |
| 2786 case '(': | |
| 2787 if (syntax & RE_NO_BK_PARENS) | |
| 2788 goto handle_open; | |
| 2789 else | |
| 2790 goto normal_char; | |
| 2791 | |
| 2792 | |
| 2793 case ')': | |
| 2794 if (syntax & RE_NO_BK_PARENS) | |
| 2795 goto handle_close; | |
| 2796 else | |
| 2797 goto normal_char; | |
| 2798 | |
| 2799 | |
| 2800 case '\n': | |
| 2801 if (syntax & RE_NEWLINE_ALT) | |
| 2802 goto handle_alt; | |
| 2803 else | |
| 2804 goto normal_char; | |
| 2805 | |
| 2806 | |
| 2807 case '|': | |
| 2808 if (syntax & RE_NO_BK_VBAR) | |
| 2809 goto handle_alt; | |
| 2810 else | |
| 2811 goto normal_char; | |
| 2812 | |
| 2813 | |
| 2814 case '{': | |
| 2815 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) | |
| 2816 goto handle_interval; | |
| 2817 else | |
| 2818 goto normal_char; | |
| 2819 | |
| 2820 | |
| 2821 case '\\': | |
| 2822 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE); | |
| 2823 | |
| 2824 /* Do not translate the character after the \, so that we can | |
| 2825 distinguish, e.g., \B from \b, even if we normally would | |
| 2826 translate, e.g., B to b. */ | |
| 2827 PATFETCH_RAW (c); | |
| 2828 | |
| 2829 switch (c) | |
| 2830 { | |
| 2831 case '(': | |
| 2832 if (syntax & RE_NO_BK_PARENS) | |
| 2833 goto normal_backslash; | |
| 2834 | |
| 2835 handle_open: | |
| 2836 { | |
| 2837 regnum_t r; | |
| 502 | 2838 int shy = 0; |
| 428 | 2839 |
| 2840 if (!(syntax & RE_NO_SHY_GROUPS) | |
| 2841 && p != pend | |
| 446 | 2842 && *p == '?') |
| 428 | 2843 { |
| 2844 p++; | |
| 446 | 2845 PATFETCH (c); |
| 428 | 2846 switch (c) |
| 2847 { | |
| 2848 case ':': /* shy groups */ | |
| 502 | 2849 shy = 1; |
| 428 | 2850 break; |
| 2851 | |
| 2852 /* All others are reserved for future constructs. */ | |
| 2853 default: | |
| 2854 FREE_STACK_RETURN (REG_BADPAT); | |
| 2855 } | |
| 2856 } | |
| 502 | 2857 |
| 2858 r = ++regnum; | |
| 2859 bufp->re_ngroups++; | |
| 2860 if (!shy) | |
| 2861 { | |
| 2862 bufp->re_nsub++; | |
| 2863 while (bufp->external_to_internal_register_size <= | |
| 2864 bufp->re_nsub) | |
| 2865 { | |
| 2866 int i; | |
| 2867 int old_size = | |
| 2868 bufp->external_to_internal_register_size; | |
| 2869 bufp->external_to_internal_register_size += 5; | |
| 2870 RETALLOC (bufp->external_to_internal_register, | |
| 2871 bufp->external_to_internal_register_size, | |
| 2872 int); | |
| 2873 /* debugging */ | |
| 2874 for (i = old_size; | |
| 2875 i < bufp->external_to_internal_register_size; i++) | |
| 2876 bufp->external_to_internal_register[i] = | |
| 2877 (int) 0xDEADBEEF; | |
| 2878 } | |
| 2879 | |
| 2880 bufp->external_to_internal_register[bufp->re_nsub] = | |
| 2881 bufp->re_ngroups; | |
| 2882 } | |
| 428 | 2883 |
| 2884 if (COMPILE_STACK_FULL) | |
| 2885 { | |
| 2886 RETALLOC (compile_stack.stack, compile_stack.size << 1, | |
| 2887 compile_stack_elt_t); | |
| 2888 if (compile_stack.stack == NULL) return REG_ESPACE; | |
| 2889 | |
| 2890 compile_stack.size <<= 1; | |
| 2891 } | |
| 2892 | |
| 2893 /* These are the values to restore when we hit end of this | |
| 2894 group. They are all relative offsets, so that if the | |
| 2895 whole pattern moves because of realloc, they will still | |
| 2896 be valid. */ | |
| 2897 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; | |
| 2898 COMPILE_STACK_TOP.fixup_alt_jump | |
| 2899 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; | |
| 446 | 2900 COMPILE_STACK_TOP.laststart_offset = buf_end - bufp->buffer; |
| 428 | 2901 COMPILE_STACK_TOP.regnum = r; |
| 2902 | |
| 2903 /* We will eventually replace the 0 with the number of | |
| 2904 groups inner to this one. But do not push a | |
| 2905 start_memory for groups beyond the last one we can | |
| 502 | 2906 represent in the compiled pattern. |
| 2907 #### bad bad bad. this will fail in lots of ways, if we | |
| 2908 ever have to backtrack for these groups. | |
| 2909 */ | |
| 428 | 2910 if (r <= MAX_REGNUM) |
| 2911 { | |
| 2912 COMPILE_STACK_TOP.inner_group_offset | |
| 446 | 2913 = buf_end - bufp->buffer + 2; |
| 428 | 2914 BUF_PUSH_3 (start_memory, r, 0); |
| 2915 } | |
| 2916 | |
| 2917 compile_stack.avail++; | |
| 2918 | |
| 2919 fixup_alt_jump = 0; | |
| 2920 laststart = 0; | |
| 446 | 2921 begalt = buf_end; |
| 428 | 2922 /* If we've reached MAX_REGNUM groups, then this open |
| 2923 won't actually generate any code, so we'll have to | |
| 2924 clear pending_exact explicitly. */ | |
| 2925 pending_exact = 0; | |
| 2926 } | |
| 2927 break; | |
| 2928 | |
| 2929 | |
| 2930 case ')': | |
| 2931 if (syntax & RE_NO_BK_PARENS) goto normal_backslash; | |
| 2932 | |
| 2933 if (COMPILE_STACK_EMPTY) { | |
| 2934 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) | |
| 2935 goto normal_backslash; | |
| 2936 else | |
| 2937 FREE_STACK_RETURN (REG_ERPAREN); | |
| 2938 } | |
| 2939 | |
| 2940 handle_close: | |
| 2941 if (fixup_alt_jump) | |
| 2942 { /* Push a dummy failure point at the end of the | |
| 2943 alternative for a possible future | |
| 2944 `pop_failure_jump' to pop. See comments at | |
| 2945 `push_dummy_failure' in `re_match_2'. */ | |
| 2946 BUF_PUSH (push_dummy_failure); | |
| 2947 | |
| 2948 /* We allocated space for this jump when we assigned | |
| 2949 to `fixup_alt_jump', in the `handle_alt' case below. */ | |
| 446 | 2950 STORE_JUMP (jump_past_alt, fixup_alt_jump, buf_end - 1); |
| 428 | 2951 } |
| 2952 | |
| 2953 /* See similar code for backslashed left paren above. */ | |
| 2954 if (COMPILE_STACK_EMPTY) { | |
| 2955 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) | |
| 2956 goto normal_char; | |
| 2957 else | |
| 2958 FREE_STACK_RETURN (REG_ERPAREN); | |
| 2959 } | |
| 2960 | |
| 2961 /* Since we just checked for an empty stack above, this | |
| 2962 ``can't happen''. */ | |
| 2963 assert (compile_stack.avail != 0); | |
| 2964 { | |
| 2965 /* We don't just want to restore into `regnum', because | |
| 2966 later groups should continue to be numbered higher, | |
| 2967 as in `(ab)c(de)' -- the second group is #2. */ | |
| 2968 regnum_t this_group_regnum; | |
| 2969 | |
| 2970 compile_stack.avail--; | |
| 2971 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; | |
| 2972 fixup_alt_jump | |
| 2973 = COMPILE_STACK_TOP.fixup_alt_jump | |
| 2974 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 | |
| 2975 : 0; | |
| 2976 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; | |
| 2977 this_group_regnum = COMPILE_STACK_TOP.regnum; | |
| 2978 /* If we've reached MAX_REGNUM groups, then this open | |
| 2979 won't actually generate any code, so we'll have to | |
| 2980 clear pending_exact explicitly. */ | |
| 2981 pending_exact = 0; | |
| 2982 | |
| 2983 /* We're at the end of the group, so now we know how many | |
| 2984 groups were inside this one. */ | |
| 2985 if (this_group_regnum <= MAX_REGNUM) | |
| 2986 { | |
| 2987 unsigned char *inner_group_loc | |
| 2988 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; | |
| 2989 | |
| 2990 *inner_group_loc = regnum - this_group_regnum; | |
| 2991 BUF_PUSH_3 (stop_memory, this_group_regnum, | |
| 2992 regnum - this_group_regnum); | |
| 2993 } | |
| 2994 } | |
| 2995 break; | |
| 2996 | |
| 2997 | |
| 2998 case '|': /* `\|'. */ | |
| 2999 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) | |
| 3000 goto normal_backslash; | |
| 3001 handle_alt: | |
| 3002 if (syntax & RE_LIMITED_OPS) | |
| 3003 goto normal_char; | |
| 3004 | |
| 3005 /* Insert before the previous alternative a jump which | |
| 3006 jumps to this alternative if the former fails. */ | |
| 3007 GET_BUFFER_SPACE (3); | |
| 446 | 3008 INSERT_JUMP (on_failure_jump, begalt, buf_end + 6); |
| 428 | 3009 pending_exact = 0; |
| 446 | 3010 buf_end += 3; |
| 428 | 3011 |
| 3012 /* The alternative before this one has a jump after it | |
| 3013 which gets executed if it gets matched. Adjust that | |
| 3014 jump so it will jump to this alternative's analogous | |
| 3015 jump (put in below, which in turn will jump to the next | |
| 3016 (if any) alternative's such jump, etc.). The last such | |
| 3017 jump jumps to the correct final destination. A picture: | |
| 3018 _____ _____ | |
| 3019 | | | | | |
| 3020 | v | v | |
| 3021 a | b | c | |
| 3022 | |
| 3023 If we are at `b', then fixup_alt_jump right now points to a | |
| 3024 three-byte space after `a'. We'll put in the jump, set | |
| 3025 fixup_alt_jump to right after `b', and leave behind three | |
| 3026 bytes which we'll fill in when we get to after `c'. */ | |
| 3027 | |
| 3028 if (fixup_alt_jump) | |
| 446 | 3029 STORE_JUMP (jump_past_alt, fixup_alt_jump, buf_end); |
| 428 | 3030 |
| 3031 /* Mark and leave space for a jump after this alternative, | |
| 3032 to be filled in later either by next alternative or | |
| 3033 when know we're at the end of a series of alternatives. */ | |
| 446 | 3034 fixup_alt_jump = buf_end; |
| 428 | 3035 GET_BUFFER_SPACE (3); |
| 446 | 3036 buf_end += 3; |
| 428 | 3037 |
| 3038 laststart = 0; | |
| 446 | 3039 begalt = buf_end; |
| 428 | 3040 break; |
| 3041 | |
| 3042 | |
| 3043 case '{': | |
| 3044 /* If \{ is a literal. */ | |
| 3045 if (!(syntax & RE_INTERVALS) | |
| 3046 /* If we're at `\{' and it's not the open-interval | |
| 3047 operator. */ | |
| 3048 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) | |
| 3049 || (p - 2 == pattern && p == pend)) | |
| 3050 goto normal_backslash; | |
| 3051 | |
| 3052 handle_interval: | |
| 3053 { | |
| 3054 /* If got here, then the syntax allows intervals. */ | |
| 3055 | |
| 3056 /* At least (most) this many matches must be made. */ | |
| 3057 int lower_bound = -1, upper_bound = -1; | |
| 3058 | |
| 3059 beg_interval = p - 1; | |
| 3060 | |
| 3061 if (p == pend) | |
| 3062 { | |
| 3063 if (syntax & RE_NO_BK_BRACES) | |
| 3064 goto unfetch_interval; | |
| 3065 else | |
| 3066 FREE_STACK_RETURN (REG_EBRACE); | |
| 3067 } | |
| 3068 | |
| 3069 GET_UNSIGNED_NUMBER (lower_bound); | |
| 3070 | |
| 3071 if (c == ',') | |
| 3072 { | |
| 3073 GET_UNSIGNED_NUMBER (upper_bound); | |
| 3074 if (upper_bound < 0) upper_bound = RE_DUP_MAX; | |
| 3075 } | |
| 3076 else | |
| 3077 /* Interval such as `{1}' => match exactly once. */ | |
| 3078 upper_bound = lower_bound; | |
| 3079 | |
| 3080 if (lower_bound < 0 || upper_bound > RE_DUP_MAX | |
| 3081 || lower_bound > upper_bound) | |
| 3082 { | |
| 3083 if (syntax & RE_NO_BK_BRACES) | |
| 3084 goto unfetch_interval; | |
| 3085 else | |
| 3086 FREE_STACK_RETURN (REG_BADBR); | |
| 3087 } | |
| 3088 | |
| 3089 if (!(syntax & RE_NO_BK_BRACES)) | |
| 3090 { | |
| 3091 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE); | |
| 3092 | |
| 3093 PATFETCH (c); | |
| 3094 } | |
| 3095 | |
| 3096 if (c != '}') | |
| 3097 { | |
| 3098 if (syntax & RE_NO_BK_BRACES) | |
| 3099 goto unfetch_interval; | |
| 3100 else | |
| 3101 FREE_STACK_RETURN (REG_BADBR); | |
| 3102 } | |
| 3103 | |
| 3104 /* We just parsed a valid interval. */ | |
| 3105 | |
| 3106 /* If it's invalid to have no preceding re. */ | |
| 3107 if (!laststart) | |
| 3108 { | |
| 3109 if (syntax & RE_CONTEXT_INVALID_OPS) | |
| 3110 FREE_STACK_RETURN (REG_BADRPT); | |
| 3111 else if (syntax & RE_CONTEXT_INDEP_OPS) | |
| 446 | 3112 laststart = buf_end; |
| 428 | 3113 else |
| 3114 goto unfetch_interval; | |
| 3115 } | |
| 3116 | |
| 3117 /* If the upper bound is zero, don't want to succeed at | |
| 3118 all; jump from `laststart' to `b + 3', which will be | |
| 3119 the end of the buffer after we insert the jump. */ | |
| 3120 if (upper_bound == 0) | |
| 3121 { | |
| 3122 GET_BUFFER_SPACE (3); | |
| 446 | 3123 INSERT_JUMP (jump, laststart, buf_end + 3); |
| 3124 buf_end += 3; | |
| 428 | 3125 } |
| 3126 | |
| 3127 /* Otherwise, we have a nontrivial interval. When | |
| 3128 we're all done, the pattern will look like: | |
| 3129 set_number_at <jump count> <upper bound> | |
| 3130 set_number_at <succeed_n count> <lower bound> | |
| 3131 succeed_n <after jump addr> <succeed_n count> | |
| 3132 <body of loop> | |
| 3133 jump_n <succeed_n addr> <jump count> | |
| 3134 (The upper bound and `jump_n' are omitted if | |
| 3135 `upper_bound' is 1, though.) */ | |
| 3136 else | |
| 3137 { /* If the upper bound is > 1, we need to insert | |
| 3138 more at the end of the loop. */ | |
| 647 | 3139 int nbytes = 10 + (upper_bound > 1) * 10; |
| 428 | 3140 |
| 3141 GET_BUFFER_SPACE (nbytes); | |
| 3142 | |
| 3143 /* Initialize lower bound of the `succeed_n', even | |
| 3144 though it will be set during matching by its | |
| 3145 attendant `set_number_at' (inserted next), | |
| 3146 because `re_compile_fastmap' needs to know. | |
| 3147 Jump to the `jump_n' we might insert below. */ | |
| 3148 INSERT_JUMP2 (succeed_n, laststart, | |
| 446 | 3149 buf_end + 5 + (upper_bound > 1) * 5, |
| 428 | 3150 lower_bound); |
| 446 | 3151 buf_end += 5; |
| 428 | 3152 |
| 3153 /* Code to initialize the lower bound. Insert | |
| 3154 before the `succeed_n'. The `5' is the last two | |
| 3155 bytes of this `set_number_at', plus 3 bytes of | |
| 3156 the following `succeed_n'. */ | |
| 446 | 3157 insert_op2 (set_number_at, laststart, 5, lower_bound, buf_end); |
| 3158 buf_end += 5; | |
| 428 | 3159 |
| 3160 if (upper_bound > 1) | |
| 3161 { /* More than one repetition is allowed, so | |
| 3162 append a backward jump to the `succeed_n' | |
| 3163 that starts this interval. | |
| 3164 | |
| 3165 When we've reached this during matching, | |
| 3166 we'll have matched the interval once, so | |
| 3167 jump back only `upper_bound - 1' times. */ | |
| 446 | 3168 STORE_JUMP2 (jump_n, buf_end, laststart + 5, |
| 428 | 3169 upper_bound - 1); |
| 446 | 3170 buf_end += 5; |
| 428 | 3171 |
| 3172 /* The location we want to set is the second | |
| 3173 parameter of the `jump_n'; that is `b-2' as | |
| 3174 an absolute address. `laststart' will be | |
| 3175 the `set_number_at' we're about to insert; | |
| 3176 `laststart+3' the number to set, the source | |
| 3177 for the relative address. But we are | |
| 3178 inserting into the middle of the pattern -- | |
| 3179 so everything is getting moved up by 5. | |
| 3180 Conclusion: (b - 2) - (laststart + 3) + 5, | |
| 3181 i.e., b - laststart. | |
| 3182 | |
| 3183 We insert this at the beginning of the loop | |
| 3184 so that if we fail during matching, we'll | |
| 3185 reinitialize the bounds. */ | |
| 446 | 3186 insert_op2 (set_number_at, laststart, |
| 3187 buf_end - laststart, | |
| 3188 upper_bound - 1, buf_end); | |
| 3189 buf_end += 5; | |
| 428 | 3190 } |
| 3191 } | |
| 3192 pending_exact = 0; | |
| 3193 beg_interval = NULL; | |
| 3194 } | |
| 3195 break; | |
| 3196 | |
| 3197 unfetch_interval: | |
| 3198 /* If an invalid interval, match the characters as literals. */ | |
| 3199 assert (beg_interval); | |
| 3200 p = beg_interval; | |
| 3201 beg_interval = NULL; | |
| 3202 | |
| 3203 /* normal_char and normal_backslash need `c'. */ | |
| 3204 PATFETCH (c); | |
| 3205 | |
| 3206 if (!(syntax & RE_NO_BK_BRACES)) | |
| 3207 { | |
| 3208 if (p > pattern && p[-1] == '\\') | |
| 3209 goto normal_backslash; | |
| 3210 } | |
| 3211 goto normal_char; | |
| 3212 | |
| 3213 #ifdef emacs | |
| 3214 /* There is no way to specify the before_dot and after_dot | |
| 3215 operators. rms says this is ok. --karl */ | |
| 3216 case '=': | |
| 3217 BUF_PUSH (at_dot); | |
| 3218 break; | |
| 3219 | |
| 3220 case 's': | |
| 446 | 3221 laststart = buf_end; |
| 428 | 3222 PATFETCH (c); |
| 3223 /* XEmacs addition */ | |
| 3224 if (c >= 0x80 || syntax_spec_code[c] == 0377) | |
| 3225 FREE_STACK_RETURN (REG_ESYNTAX); | |
| 3226 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); | |
| 3227 break; | |
| 3228 | |
| 3229 case 'S': | |
| 446 | 3230 laststart = buf_end; |
| 428 | 3231 PATFETCH (c); |
| 3232 /* XEmacs addition */ | |
| 3233 if (c >= 0x80 || syntax_spec_code[c] == 0377) | |
| 3234 FREE_STACK_RETURN (REG_ESYNTAX); | |
| 3235 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); | |
| 3236 break; | |
| 3237 | |
| 3238 #ifdef MULE | |
| 3239 /* 97.2.17 jhod merged in to XEmacs from mule-2.3 */ | |
| 3240 case 'c': | |
| 446 | 3241 laststart = buf_end; |
| 428 | 3242 PATFETCH_RAW (c); |
| 3243 if (c < 32 || c > 127) | |
| 3244 FREE_STACK_RETURN (REG_ECATEGORY); | |
| 3245 BUF_PUSH_2 (categoryspec, c); | |
| 3246 break; | |
| 3247 | |
| 3248 case 'C': | |
| 446 | 3249 laststart = buf_end; |
| 428 | 3250 PATFETCH_RAW (c); |
| 3251 if (c < 32 || c > 127) | |
| 3252 FREE_STACK_RETURN (REG_ECATEGORY); | |
| 3253 BUF_PUSH_2 (notcategoryspec, c); | |
| 3254 break; | |
| 3255 /* end of category patch */ | |
| 3256 #endif /* MULE */ | |
| 3257 #endif /* emacs */ | |
| 3258 | |
| 3259 | |
| 3260 case 'w': | |
| 446 | 3261 laststart = buf_end; |
| 428 | 3262 BUF_PUSH (wordchar); |
| 3263 break; | |
| 3264 | |
| 3265 | |
| 3266 case 'W': | |
| 446 | 3267 laststart = buf_end; |
| 428 | 3268 BUF_PUSH (notwordchar); |
| 3269 break; | |
| 3270 | |
| 3271 | |
| 3272 case '<': | |
| 3273 BUF_PUSH (wordbeg); | |
| 3274 break; | |
| 3275 | |
| 3276 case '>': | |
| 3277 BUF_PUSH (wordend); | |
| 3278 break; | |
| 3279 | |
| 3280 case 'b': | |
| 3281 BUF_PUSH (wordbound); | |
| 3282 break; | |
| 3283 | |
| 3284 case 'B': | |
| 3285 BUF_PUSH (notwordbound); | |
| 3286 break; | |
| 3287 | |
| 3288 case '`': | |
| 3289 BUF_PUSH (begbuf); | |
| 3290 break; | |
| 3291 | |
| 3292 case '\'': | |
| 3293 BUF_PUSH (endbuf); | |
| 3294 break; | |
| 3295 | |
| 3296 case '1': case '2': case '3': case '4': case '5': | |
| 3297 case '6': case '7': case '8': case '9': | |
| 446 | 3298 { |
| 502 | 3299 regnum_t reg, regint; |
| 3300 int may_need_to_unfetch = 0; | |
| 446 | 3301 if (syntax & RE_NO_BK_REFS) |
| 3302 goto normal_char; | |
| 3303 | |
| 502 | 3304 /* This only goes up to 99. It could be extended to work |
| 3305 up to 255 (the maximum number of registers that can be | |
| 3306 handled by the current regexp engine, because it stores | |
| 3307 its register numbers in the compiled pattern as one byte, | |
| 3308 ugh). Doing that's a bit trickier, because you might | |
| 3309 have the case where \25 a back-ref but \255 is not, ... */ | |
| 446 | 3310 reg = c - '0'; |
| 502 | 3311 if (p < pend) |
| 3312 { | |
| 3313 PATFETCH (c); | |
| 3314 if (c >= '0' && c <= '9') | |
| 3315 { | |
| 3316 regnum_t new_reg = reg * 10 + c - '0'; | |
| 3317 if (new_reg <= bufp->re_nsub) | |
| 3318 { | |
| 3319 reg = new_reg; | |
| 3320 may_need_to_unfetch = 1; | |
| 3321 } | |
| 3322 else | |
| 3323 PATUNFETCH; | |
| 3324 } | |
| 523 | 3325 else |
| 3326 PATUNFETCH; | |
| 502 | 3327 } |
| 3328 | |
| 3329 if (reg > bufp->re_nsub) | |
| 446 | 3330 FREE_STACK_RETURN (REG_ESUBREG); |
| 3331 | |
| 502 | 3332 regint = bufp->external_to_internal_register[reg]; |
| 446 | 3333 /* Can't back reference to a subexpression if inside of it. */ |
| 502 | 3334 if (group_in_compile_stack (compile_stack, regint)) |
| 3335 { | |
| 3336 if (may_need_to_unfetch) | |
| 3337 PATUNFETCH; | |
| 3338 goto normal_char; | |
| 3339 } | |
| 3340 | |
| 3341 #ifdef emacs | |
| 3342 if (reg > 9 && | |
| 3343 bufp->warned_about_incompatible_back_references == 0) | |
| 3344 { | |
| 3345 bufp->warned_about_incompatible_back_references = 1; | |
| 3346 warn_when_safe (intern ("regex"), Qinfo, | |
| 3347 "Back reference \\%d now has new " | |
| 3348 "semantics in %s", reg, pattern); | |
| 3349 } | |
| 3350 #endif | |
| 446 | 3351 |
| 3352 laststart = buf_end; | |
| 502 | 3353 BUF_PUSH_2 (duplicate, regint); |
| 446 | 3354 } |
| 428 | 3355 break; |
| 3356 | |
| 3357 | |
| 3358 case '+': | |
| 3359 case '?': | |
| 3360 if (syntax & RE_BK_PLUS_QM) | |
| 3361 goto handle_plus; | |
| 3362 else | |
| 3363 goto normal_backslash; | |
| 3364 | |
| 3365 default: | |
| 3366 normal_backslash: | |
| 3367 /* You might think it would be useful for \ to mean | |
| 3368 not to translate; but if we don't translate it, | |
| 3369 it will never match anything. */ | |
| 826 | 3370 c = RE_TRANSLATE (c); |
| 428 | 3371 goto normal_char; |
| 3372 } | |
| 3373 break; | |
| 3374 | |
| 3375 | |
| 3376 default: | |
| 3377 /* Expects the character in `c'. */ | |
| 3378 /* `p' points to the location after where `c' came from. */ | |
| 3379 normal_char: | |
| 3380 { | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3381 /* The following conditional synced to GNU Emacs 22.1. */ |
| 428 | 3382 /* If no exactn currently being built. */ |
| 3383 if (!pending_exact | |
| 3384 | |
| 3385 /* If last exactn not at current position. */ | |
| 446 | 3386 || pending_exact + *pending_exact + 1 != buf_end |
| 428 | 3387 |
| 3388 /* 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
|
3389 || *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
|
3390 |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3391 /* If followed by a repetition operator. |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3392 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
|
3393 trailing backslash will get caught later. */ |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3394 || (p != pend && (*p == '*' || *p == '^')) |
| 428 | 3395 || ((syntax & RE_BK_PLUS_QM) |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3396 ? 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
|
3397 : p != pend && (*p == '+' || *p == '?')) |
| 428 | 3398 || ((syntax & RE_INTERVALS) |
| 3399 && ((syntax & RE_NO_BK_BRACES) | |
|
4750
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3400 ? p != pend && *p == '{' |
|
b5f21bb36684
Fix crash in regex.c (closes issue630).
Stephen J. Turnbull <stephen@xemacs.org>
parents:
4527
diff
changeset
|
3401 : p + 1 < pend && (p[0] == '\\' && p[1] == '{')))) |
| 428 | 3402 { |
| 3403 /* Start building a new exactn. */ | |
| 3404 | |
| 446 | 3405 laststart = buf_end; |
| 428 | 3406 |
| 3407 BUF_PUSH_2 (exactn, 0); | |
| 446 | 3408 pending_exact = buf_end - 1; |
| 428 | 3409 } |
| 3410 | |
| 446 | 3411 #ifndef MULE |
| 428 | 3412 BUF_PUSH (c); |
| 3413 (*pending_exact)++; | |
| 446 | 3414 #else |
| 3415 { | |
| 3416 Bytecount bt_count; | |
| 867 | 3417 Ibyte tmp_buf[MAX_ICHAR_LEN]; |
| 446 | 3418 int i; |
| 3419 | |
| 867 | 3420 bt_count = set_itext_ichar (tmp_buf, c); |
| 446 | 3421 |
| 3422 for (i = 0; i < bt_count; i++) | |
| 3423 { | |
| 3424 BUF_PUSH (tmp_buf[i]); | |
| 3425 (*pending_exact)++; | |
| 3426 } | |
| 3427 } | |
| 3428 #endif | |
| 428 | 3429 break; |
| 3430 } | |
| 3431 } /* switch (c) */ | |
| 3432 } /* while p != pend */ | |
| 3433 | |
| 3434 | |
| 3435 /* Through the pattern now. */ | |
| 3436 | |
| 3437 if (fixup_alt_jump) | |
| 446 | 3438 STORE_JUMP (jump_past_alt, fixup_alt_jump, buf_end); |
| 428 | 3439 |
| 3440 if (!COMPILE_STACK_EMPTY) | |
| 3441 FREE_STACK_RETURN (REG_EPAREN); | |
| 3442 | |
| 3443 /* If we don't want backtracking, force success | |
| 3444 the first time we reach the end of the compiled pattern. */ | |
| 3445 if (syntax & RE_NO_POSIX_BACKTRACKING) | |
| 3446 BUF_PUSH (succeed); | |
| 3447 | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
3448 xfree (compile_stack.stack); |
| 428 | 3449 |
| 3450 /* We have succeeded; set the length of the buffer. */ | |
| 446 | 3451 bufp->used = buf_end - bufp->buffer; |
| 428 | 3452 |
| 3453 #ifdef DEBUG | |
| 5041 | 3454 if (debug_regexps & RE_DEBUG_COMPILATION) |
| 428 | 3455 { |
| 3456 DEBUG_PRINT1 ("\nCompiled pattern: \n"); | |
| 3457 print_compiled_pattern (bufp); | |
| 3458 } | |
| 3459 #endif /* DEBUG */ | |
| 3460 | |
| 3461 #ifndef MATCH_MAY_ALLOCATE | |
| 3462 /* Initialize the failure stack to the largest possible stack. This | |
| 3463 isn't necessary unless we're trying to avoid calling alloca in | |
| 3464 the search and match routines. */ | |
| 3465 { | |
| 502 | 3466 int num_regs = bufp->re_ngroups + 1; |
| 428 | 3467 |
| 3468 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size | |
| 3469 is strictly greater than re_max_failures, the largest possible stack | |
| 3470 is 2 * re_max_failures failure points. */ | |
| 3471 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS)) | |
| 3472 { | |
| 3473 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS); | |
| 3474 | |
| 3475 if (! fail_stack.stack) | |
| 3476 fail_stack.stack | |
| 3477 = (fail_stack_elt_t *) xmalloc (fail_stack.size | |
| 3478 * sizeof (fail_stack_elt_t)); | |
| 3479 else | |
| 3480 fail_stack.stack | |
| 3481 = (fail_stack_elt_t *) xrealloc (fail_stack.stack, | |
| 3482 (fail_stack.size | |
| 3483 * sizeof (fail_stack_elt_t))); | |
| 3484 } | |
| 3485 | |
| 3486 regex_grow_registers (num_regs); | |
| 3487 } | |
| 3488 #endif /* not MATCH_MAY_ALLOCATE */ | |
| 3489 | |
| 3490 return REG_NOERROR; | |
| 3491 } /* regex_compile */ | |
| 3492 | |
| 3493 /* Subroutines for `regex_compile'. */ | |
| 3494 | |
| 3495 /* Store OP at LOC followed by two-byte integer parameter ARG. */ | |
| 3496 | |
| 3497 static void | |
| 3498 store_op1 (re_opcode_t op, unsigned char *loc, int arg) | |
| 3499 { | |
| 3500 *loc = (unsigned char) op; | |
| 3501 STORE_NUMBER (loc + 1, arg); | |
| 3502 } | |
| 3503 | |
| 3504 | |
| 3505 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */ | |
| 3506 | |
| 3507 static void | |
| 3508 store_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2) | |
| 3509 { | |
| 3510 *loc = (unsigned char) op; | |
| 3511 STORE_NUMBER (loc + 1, arg1); | |
| 3512 STORE_NUMBER (loc + 3, arg2); | |
| 3513 } | |
| 3514 | |
| 3515 | |
| 3516 /* Copy the bytes from LOC to END to open up three bytes of space at LOC | |
| 3517 for OP followed by two-byte integer parameter ARG. */ | |
| 3518 | |
| 3519 static void | |
| 3520 insert_op1 (re_opcode_t op, unsigned char *loc, int arg, unsigned char *end) | |
| 3521 { | |
| 3522 REGISTER unsigned char *pfrom = end; | |
| 3523 REGISTER unsigned char *pto = end + 3; | |
| 3524 | |
| 3525 while (pfrom != loc) | |
| 3526 *--pto = *--pfrom; | |
| 3527 | |
| 3528 store_op1 (op, loc, arg); | |
| 3529 } | |
| 3530 | |
| 3531 | |
| 3532 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */ | |
| 3533 | |
| 3534 static void | |
| 3535 insert_op2 (re_opcode_t op, unsigned char *loc, int arg1, int arg2, | |
| 3536 unsigned char *end) | |
| 3537 { | |
| 3538 REGISTER unsigned char *pfrom = end; | |
| 3539 REGISTER unsigned char *pto = end + 5; | |
| 3540 | |
| 3541 while (pfrom != loc) | |
| 3542 *--pto = *--pfrom; | |
| 3543 | |
| 3544 store_op2 (op, loc, arg1, arg2); | |
| 3545 } | |
| 3546 | |
| 3547 | |
| 3548 /* P points to just after a ^ in PATTERN. Return true if that ^ comes | |
| 3549 after an alternative or a begin-subexpression. We assume there is at | |
| 3550 least one character before the ^. */ | |
| 3551 | |
| 460 | 3552 static re_bool |
| 446 | 3553 at_begline_loc_p (re_char *pattern, re_char *p, reg_syntax_t syntax) |
| 428 | 3554 { |
| 446 | 3555 re_char *prev = p - 2; |
| 460 | 3556 re_bool prev_prev_backslash = prev > pattern && prev[-1] == '\\'; |
| 428 | 3557 |
| 3558 return | |
| 3559 /* After a subexpression? */ | |
| 3560 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) | |
| 3561 /* After an alternative? */ | |
| 3562 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); | |
| 3563 } | |
| 3564 | |
| 3565 | |
| 3566 /* The dual of at_begline_loc_p. This one is for $. We assume there is | |
| 3567 at least one character after the $, i.e., `P < PEND'. */ | |
| 3568 | |
| 460 | 3569 static re_bool |
| 446 | 3570 at_endline_loc_p (re_char *p, re_char *pend, int syntax) |
| 428 | 3571 { |
| 446 | 3572 re_char *next = p; |
| 460 | 3573 re_bool next_backslash = *next == '\\'; |
| 446 | 3574 re_char *next_next = p + 1 < pend ? p + 1 : 0; |
| 428 | 3575 |
| 3576 return | |
| 3577 /* Before a subexpression? */ | |
| 3578 (syntax & RE_NO_BK_PARENS ? *next == ')' | |
| 3579 : next_backslash && next_next && *next_next == ')') | |
| 3580 /* Before an alternative? */ | |
| 3581 || (syntax & RE_NO_BK_VBAR ? *next == '|' | |
| 3582 : next_backslash && next_next && *next_next == '|'); | |
| 3583 } | |
| 3584 | |
| 3585 | |
| 3586 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and | |
| 3587 false if it's not. */ | |
| 3588 | |
| 460 | 3589 static re_bool |
| 428 | 3590 group_in_compile_stack (compile_stack_type compile_stack, regnum_t regnum) |
| 3591 { | |
| 3592 int this_element; | |
| 3593 | |
| 3594 for (this_element = compile_stack.avail - 1; | |
| 3595 this_element >= 0; | |
| 3596 this_element--) | |
| 3597 if (compile_stack.stack[this_element].regnum == regnum) | |
| 3598 return true; | |
| 3599 | |
| 3600 return false; | |
| 3601 } | |
| 3602 | |
| 3603 | |
| 3604 /* Read the ending character of a range (in a bracket expression) from the | |
| 3605 uncompiled pattern *P_PTR (which ends at PEND). We assume the | |
| 3606 starting character is in `P[-2]'. (`P[-1]' is the character `-'.) | |
| 3607 Then we set the translation of all bits between the starting and | |
| 3608 ending characters (inclusive) in the compiled pattern B. | |
| 3609 | |
| 3610 Return an error code. | |
| 3611 | |
| 3612 We use these short variable names so we can use the same macros as | |
| 826 | 3613 `regex_compile' itself. |
| 3614 | |
| 3615 Under Mule, this is only called when both chars of the range are | |
| 3616 ASCII. */ | |
| 428 | 3617 |
| 3618 static reg_errcode_t | |
| 446 | 3619 compile_range (re_char **p_ptr, re_char *pend, RE_TRANSLATE_TYPE translate, |
| 3620 reg_syntax_t syntax, unsigned char *buf_end) | |
| 428 | 3621 { |
| 867 | 3622 Ichar this_char; |
| 428 | 3623 |
| 446 | 3624 re_char *p = *p_ptr; |
| 428 | 3625 int range_start, range_end; |
| 3626 | |
| 3627 if (p == pend) | |
| 3628 return REG_ERANGE; | |
| 3629 | |
| 3630 /* Even though the pattern is a signed `char *', we need to fetch | |
| 3631 with unsigned char *'s; if the high bit of the pattern character | |
| 3632 is set, the range endpoints will be negative if we fetch using a | |
| 3633 signed char *. | |
| 3634 | |
| 3635 We also want to fetch the endpoints without translating them; the | |
| 3636 appropriate translation is done in the bit-setting loop below. */ | |
| 442 | 3637 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */ |
| 3638 range_start = ((const unsigned char *) p)[-2]; | |
| 3639 range_end = ((const unsigned char *) p)[0]; | |
| 428 | 3640 |
| 3641 /* Have to increment the pointer into the pattern string, so the | |
| 3642 caller isn't still at the ending character. */ | |
| 3643 (*p_ptr)++; | |
| 3644 | |
| 3645 /* If the start is after the end, the range is empty. */ | |
| 3646 if (range_start > range_end) | |
| 3647 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; | |
| 3648 | |
| 3649 /* Here we see why `this_char' has to be larger than an `unsigned | |
| 3650 char' -- the range is inclusive, so if `range_end' == 0xff | |
| 3651 (assuming 8-bit characters), we would otherwise go into an infinite | |
| 3652 loop, since all characters <= 0xff. */ | |
| 3653 for (this_char = range_start; this_char <= range_end; this_char++) | |
| 3654 { | |
| 826 | 3655 SET_LIST_BIT (RE_TRANSLATE (this_char)); |
| 428 | 3656 } |
| 3657 | |
| 3658 return REG_NOERROR; | |
| 3659 } | |
| 3660 | |
| 3661 #ifdef MULE | |
| 3662 | |
| 3663 static reg_errcode_t | |
| 446 | 3664 compile_extended_range (re_char **p_ptr, re_char *pend, |
| 3665 RE_TRANSLATE_TYPE translate, | |
| 428 | 3666 reg_syntax_t syntax, Lisp_Object rtab) |
| 3667 { | |
| 867 | 3668 Ichar this_char, range_start, range_end; |
| 3669 const Ibyte *p; | |
| 428 | 3670 |
| 3671 if (*p_ptr == pend) | |
| 3672 return REG_ERANGE; | |
| 3673 | |
| 867 | 3674 p = (const Ibyte *) *p_ptr; |
| 3675 range_end = itext_ichar (p); | |
| 428 | 3676 p--; /* back to '-' */ |
| 867 | 3677 DEC_IBYTEPTR (p); /* back to start of range */ |
| 428 | 3678 /* We also want to fetch the endpoints without translating them; the |
| 3679 appropriate translation is done in the bit-setting loop below. */ | |
| 867 | 3680 range_start = itext_ichar (p); |
| 3681 INC_IBYTEPTR (*p_ptr); | |
| 428 | 3682 |
| 3683 /* If the start is after the end, the range is empty. */ | |
| 3684 if (range_start > range_end) | |
| 3685 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; | |
| 3686 | |
| 3687 /* Can't have ranges spanning different charsets, except maybe for | |
| 3688 ranges entirely within the first 256 chars. */ | |
| 3689 | |
| 3690 if ((range_start >= 0x100 || range_end >= 0x100) | |
| 867 | 3691 && ichar_leading_byte (range_start) != |
| 3692 ichar_leading_byte (range_end)) | |
| 428 | 3693 return REG_ERANGESPAN; |
| 3694 | |
| 826 | 3695 /* #### This might be way inefficient if the range encompasses 10,000 |
| 3696 chars or something. To be efficient, you'd have to do something like | |
| 3697 this: | |
| 428 | 3698 |
| 3699 range_table a; | |
| 3700 range_table b; | |
| 3701 map over translation table in [range_start, range_end] of | |
| 3702 (put the mapped range in a; | |
| 3703 put the translation in b) | |
| 3704 invert the range in a and truncate to [range_start, range_end] | |
| 3705 compute the union of a, b | |
| 3706 union the result into rtab | |
| 3707 */ | |
| 826 | 3708 for (this_char = range_start; this_char <= range_end; this_char++) |
| 428 | 3709 { |
| 826 | 3710 SET_RANGETAB_BIT (RE_TRANSLATE (this_char)); |
| 428 | 3711 } |
| 3712 | |
| 3713 if (this_char <= range_end) | |
| 3714 put_range_table (rtab, this_char, range_end, Qt); | |
| 3715 | |
| 3716 return REG_NOERROR; | |
| 3717 } | |
| 3718 | |
| 3719 #endif /* MULE */ | |
| 3720 | |
| 3721 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in | |
| 3722 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible | |
| 3723 characters can start a string that matches the pattern. This fastmap | |
| 3724 is used by re_search to skip quickly over impossible starting points. | |
| 3725 | |
| 3726 The caller must supply the address of a (1 << BYTEWIDTH)-byte data | |
| 3727 area as BUFP->fastmap. | |
| 3728 | |
| 3729 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in | |
| 3730 the pattern buffer. | |
| 3731 | |
| 3732 Returns 0 if we succeed, -2 if an internal error. */ | |
| 3733 | |
| 3734 int | |
| 826 | 3735 re_compile_fastmap (struct re_pattern_buffer *bufp |
| 3736 RE_LISP_SHORT_CONTEXT_ARGS_DECL) | |
| 428 | 3737 { |
| 3738 int j, k; | |
| 3739 #ifdef MATCH_MAY_ALLOCATE | |
| 3740 fail_stack_type fail_stack; | |
| 3741 #endif | |
| 456 | 3742 DECLARE_DESTINATION; |
| 428 | 3743 /* We don't push any register information onto the failure stack. */ |
| 3744 | |
| 826 | 3745 /* &&#### this should be changed for 8-bit-fixed, for efficiency. see |
| 3746 comment marked with &&#### in re_search_2. */ | |
| 3747 | |
| 428 | 3748 REGISTER char *fastmap = bufp->fastmap; |
| 3749 unsigned char *pattern = bufp->buffer; | |
| 647 | 3750 long size = bufp->used; |
| 428 | 3751 unsigned char *p = pattern; |
| 3752 REGISTER unsigned char *pend = pattern + size; | |
| 3753 | |
| 771 | 3754 #ifdef REGEX_REL_ALLOC |
| 428 | 3755 /* This holds the pointer to the failure stack, when |
| 3756 it is allocated relocatably. */ | |
| 3757 fail_stack_elt_t *failure_stack_ptr; | |
| 3758 #endif | |
| 3759 | |
| 3760 /* Assume that each path through the pattern can be null until | |
| 3761 proven otherwise. We set this false at the bottom of switch | |
| 3762 statement, to which we get only if a particular path doesn't | |
| 3763 match the empty string. */ | |
| 460 | 3764 re_bool path_can_be_null = true; |
| 428 | 3765 |
| 3766 /* We aren't doing a `succeed_n' to begin with. */ | |
| 460 | 3767 re_bool succeed_n_p = false; |
| 428 | 3768 |
| 1333 | 3769 #ifdef ERROR_CHECK_MALLOC |
| 3770 /* The pattern comes from string data, not buffer data. We don't access | |
| 3771 any buffer data, so we don't have to worry about malloc() (but the | |
| 3772 disallowed flag may have been set by a caller). */ | |
| 3773 int depth = bind_regex_malloc_disallowed (0); | |
| 3774 #endif | |
| 3775 | |
| 428 | 3776 assert (fastmap != NULL && p != NULL); |
| 3777 | |
| 3778 INIT_FAIL_STACK (); | |
| 3779 memset (fastmap, 0, 1 << BYTEWIDTH); /* Assume nothing's valid. */ | |
| 3780 bufp->fastmap_accurate = 1; /* It will be when we're done. */ | |
| 3781 bufp->can_be_null = 0; | |
| 3782 | |
| 3783 while (1) | |
| 3784 { | |
| 3785 if (p == pend || *p == succeed) | |
| 3786 { | |
| 3787 /* We have reached the (effective) end of pattern. */ | |
| 3788 if (!FAIL_STACK_EMPTY ()) | |
| 3789 { | |
| 3790 bufp->can_be_null |= path_can_be_null; | |
| 3791 | |
| 3792 /* Reset for next path. */ | |
| 3793 path_can_be_null = true; | |
| 3794 | |
| 446 | 3795 p = (unsigned char *) fail_stack.stack[--fail_stack.avail].pointer; |
| 428 | 3796 |
| 3797 continue; | |
| 3798 } | |
| 3799 else | |
| 3800 break; | |
| 3801 } | |
| 3802 | |
| 3803 /* We should never be about to go beyond the end of the pattern. */ | |
| 3804 assert (p < pend); | |
| 3805 | |
|
4759
aa5ed11f473b
Remove support for obsolete systems. See xemacs-patches message with ID
Jerry James <james@xemacs.org>
parents:
4750
diff
changeset
|
3806 switch ((re_opcode_t) *p++) |
| 428 | 3807 { |
| 3808 | |
| 3809 /* I guess the idea here is to simply not bother with a fastmap | |
| 3810 if a backreference is used, since it's too hard to figure out | |
| 3811 the fastmap for the corresponding group. Setting | |
| 3812 `can_be_null' stops `re_search_2' from using the fastmap, so | |
| 3813 that is all we do. */ | |
| 3814 case duplicate: | |
| 3815 bufp->can_be_null = 1; | |
| 3816 goto done; | |
| 3817 | |
| 3818 | |
| 3819 /* Following are the cases which match a character. These end | |
| 3820 with `break'. */ | |
| 3821 | |
| 3822 case exactn: | |
| 3823 fastmap[p[1]] = 1; | |
| 3824 break; | |
| 3825 | |
| 3826 | |
| 3827 case charset: | |
| 3828 /* XEmacs: Under Mule, these bit vectors will | |
| 3829 only contain values for characters below 0x80. */ | |
| 3830 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
| 3831 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) | |
| 3832 fastmap[j] = 1; | |
| 3833 break; | |
| 3834 | |
| 3835 | |
| 3836 case charset_not: | |
| 3837 /* Chars beyond end of map must be allowed. */ | |
| 3838 #ifdef MULE | |
| 3839 for (j = *p * BYTEWIDTH; j < 0x80; j++) | |
| 3840 fastmap[j] = 1; | |
| 3841 /* And all extended characters must be allowed, too. */ | |
| 3842 for (j = 0x80; j < 0xA0; j++) | |
| 3843 fastmap[j] = 1; | |
| 446 | 3844 #else /* not MULE */ |
| 428 | 3845 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) |
| 3846 fastmap[j] = 1; | |
| 446 | 3847 #endif /* MULE */ |
| 428 | 3848 |
| 3849 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) | |
| 3850 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) | |
| 3851 fastmap[j] = 1; | |
| 3852 break; | |
| 3853 | |
| 3854 #ifdef MULE | |
| 3855 case charset_mule: | |
| 3856 { | |
| 3857 int nentries; | |
| 3858 int i; | |
| 3859 | |
| 3860 nentries = unified_range_table_nentries (p); | |
| 3861 for (i = 0; i < nentries; i++) | |
| 3862 { | |
| 3863 EMACS_INT first, last; | |
| 3864 Lisp_Object dummy_val; | |
| 3865 int jj; | |
| 867 | 3866 Ibyte strr[MAX_ICHAR_LEN]; |
| 428 | 3867 |
| 3868 unified_range_table_get_range (p, i, &first, &last, | |
| 3869 &dummy_val); | |
| 3870 for (jj = first; jj <= last && jj < 0x80; jj++) | |
| 3871 fastmap[jj] = 1; | |
| 3872 /* Ranges below 0x100 can span charsets, but there | |
| 3873 are only two (Control-1 and Latin-1), and | |
| 3874 either first or last has to be in them. */ | |
| 867 | 3875 set_itext_ichar (strr, first); |
| 428 | 3876 fastmap[*strr] = 1; |
| 3877 if (last < 0x100) | |
| 3878 { | |
| 867 | 3879 set_itext_ichar (strr, last); |
| 428 | 3880 fastmap[*strr] = 1; |
| 3881 } | |
| 3882 } | |
| 3883 } | |
| 3884 break; | |
| 3885 | |
| 3886 case charset_mule_not: | |
| 3887 { | |
| 3888 int nentries; | |
| 3889 int i; | |
|
4832
07fa38c30fdf
fix messed-up fastmap calculation in charset_mule_not
Ben Wing <ben@xemacs.org>
parents:
4759
diff
changeset
|
3890 int smallest_prev = 0; |
| 428 | 3891 |
| 3892 nentries = unified_range_table_nentries (p); | |
| 3893 for (i = 0; i < nentries; i++) | |
| 3894 { | |
| 3895 EMACS_INT first, last; | |
| 3896 Lisp_Object dummy_val; | |
| 3897 int jj; | |
| 3898 | |
| 3899 unified_range_table_get_range (p, i, &first, &last, | |
| 3900 &dummy_val); | |
| 3901 for (jj = smallest_prev; jj < first && jj < 0x80; jj++) | |
| 3902 fastmap[jj] = 1; | |
| 3903 smallest_prev = last + 1; | |
| 3904 if (smallest_prev >= 0x80) | |
| 3905 break; | |
| 3906 } | |
|
4832
07fa38c30fdf
fix messed-up fastmap calculation in charset_mule_not
Ben Wing <ben@xemacs.org>
parents:
4759
diff
changeset
|
3907 |
|
07fa38c30fdf
fix messed-up fastmap calculation in charset_mule_not
Ben Wing <ben@xemacs.org>
parents:
4759
diff
changeset
|
3908 /* Also set lead bytes after the end */ |
|
07fa38c30fdf
fix messed-up fastmap calculation in charset_mule_not
Ben Wing <ben@xemacs.org>
parents:
4759
diff
changeset
|
3909 for (i = smallest_prev; i < 0x80; i++) |
|
07fa38c30fdf
fix messed-up fastmap calculation in charset_mule_not
Ben Wing <ben@xemacs.org>
parents:
4759
diff
changeset
|
3910 fastmap[i] = 1; |
|
07fa38c30fdf
fix messed-up fastmap calculation in charset_mule_not
Ben Wing <ben@xemacs.org>
parents:
4759
diff
changeset
|
3911 |
| 428 | 3912 /* Calculating which leading bytes are actually allowed |
| 3913 here is rather difficult, so we just punt and allow | |
| 3914 all of them. */ | |
| 3915 for (i = 0x80; i < 0xA0; i++) | |
| 3916 fastmap[i] = 1; | |
| 3917 } | |
| 3918 break; | |
| 3919 #endif /* MULE */ | |
| 3920 | |
| 3921 | |
| 3922 case anychar: | |
| 3923 { | |
| 3924 int fastmap_newline = fastmap['\n']; | |
| 3925 | |
| 3926 /* `.' matches anything ... */ | |
| 3927 #ifdef MULE | |
| 3928 /* "anything" only includes bytes that can be the | |
| 3929 first byte of a character. */ | |
| 3930 for (j = 0; j < 0xA0; j++) | |
| 3931 fastmap[j] = 1; | |
| 3932 #else | |
| 3933 for (j = 0; j < (1 << BYTEWIDTH); j++) | |
| 3934 fastmap[j] = 1; | |
| 3935 #endif | |
| 3936 | |
| 3937 /* ... except perhaps newline. */ | |
| 3938 if (!(bufp->syntax & RE_DOT_NEWLINE)) | |
| 3939 fastmap['\n'] = fastmap_newline; | |
| 3940 | |
| 3941 /* Return if we have already set `can_be_null'; if we have, | |
| 3942 then the fastmap is irrelevant. Something's wrong here. */ | |
| 3943 else if (bufp->can_be_null) | |
| 3944 goto done; | |
| 3945 | |
| 3946 /* Otherwise, have to check alternative paths. */ | |
| 3947 break; | |
| 3948 } | |
| 3949 | |
| 826 | 3950 #ifndef emacs |
| 3951 case wordchar: | |
| 3952 for (j = 0; j < (1 << BYTEWIDTH); j++) | |
| 3953 if (SYNTAX (ignored, j) == Sword) | |
| 3954 fastmap[j] = 1; | |
| 3955 break; | |
| 3956 | |
| 3957 case notwordchar: | |
| 3958 for (j = 0; j < (1 << BYTEWIDTH); j++) | |
| 3959 if (SYNTAX (ignored, j) != Sword) | |
| 3960 fastmap[j] = 1; | |
| 3961 break; | |
| 3962 #else /* emacs */ | |
| 3963 case wordchar: | |
| 3964 case notwordchar: | |
| 460 | 3965 case wordbound: |
| 3966 case notwordbound: | |
| 3967 case wordbeg: | |
| 3968 case wordend: | |
| 3969 case notsyntaxspec: | |
| 3970 case syntaxspec: | |
| 3971 /* This match depends on text properties. These end with | |
| 3972 aborting optimizations. */ | |
| 3973 bufp->can_be_null = 1; | |
| 3974 goto done; | |
| 826 | 3975 #if 0 /* all of the following code is unused now that the `syntax-table' |
| 3976 property exists -- it's trickier to do this than just look in | |
| 3977 the buffer. &&#### but we could just use the syntax-cache stuff | |
| 3978 instead; why don't we? --ben */ | |
| 3979 case wordchar: | |
| 3980 k = (int) Sword; | |
| 3981 goto matchsyntax; | |
| 3982 | |
| 3983 case notwordchar: | |
| 3984 k = (int) Sword; | |
| 3985 goto matchnotsyntax; | |
| 3986 | |
| 428 | 3987 case syntaxspec: |
| 3988 k = *p++; | |
| 826 | 3989 matchsyntax: |
| 428 | 3990 #ifdef MULE |
| 3991 for (j = 0; j < 0x80; j++) | |
| 826 | 3992 if (SYNTAX |
| 3993 (XCHAR_TABLE (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) == | |
| 428 | 3994 (enum syntaxcode) k) |
| 3995 fastmap[j] = 1; | |
| 3996 for (j = 0x80; j < 0xA0; j++) | |
| 3997 { | |
| 826 | 3998 if (leading_byte_prefix_p ((unsigned char) j)) |
| 428 | 3999 /* too complicated to calculate this right */ |
| 4000 fastmap[j] = 1; | |
| 4001 else | |
| 4002 { | |
| 4003 int multi_p; | |
| 4004 Lisp_Object cset; | |
| 4005 | |
| 826 | 4006 cset = charset_by_leading_byte (j); |
| 428 | 4007 if (CHARSETP (cset)) |
| 4008 { | |
| 826 | 4009 if (charset_syntax (lispbuf, cset, &multi_p) |
| 428 | 4010 == Sword || multi_p) |
| 4011 fastmap[j] = 1; | |
| 4012 } | |
| 4013 } | |
| 4014 } | |
| 446 | 4015 #else /* not MULE */ |
| 428 | 4016 for (j = 0; j < (1 << BYTEWIDTH); j++) |
| 826 | 4017 if (SYNTAX |
| 4018 (XCHAR_TABLE (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) == | |
| 428 | 4019 (enum syntaxcode) k) |
| 4020 fastmap[j] = 1; | |
| 446 | 4021 #endif /* MULE */ |
| 428 | 4022 break; |
| 4023 | |
| 4024 | |
| 4025 case notsyntaxspec: | |
| 4026 k = *p++; | |
| 826 | 4027 matchnotsyntax: |
| 428 | 4028 #ifdef MULE |
| 4029 for (j = 0; j < 0x80; j++) | |
| 826 | 4030 if (SYNTAX |
| 428 | 4031 (XCHAR_TABLE |
| 826 | 4032 (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) != |
| 428 | 4033 (enum syntaxcode) k) |
| 4034 fastmap[j] = 1; | |
| 4035 for (j = 0x80; j < 0xA0; j++) | |
| 4036 { | |
| 826 | 4037 if (leading_byte_prefix_p ((unsigned char) j)) |
| 428 | 4038 /* too complicated to calculate this right */ |
| 4039 fastmap[j] = 1; | |
| 4040 else | |
| 4041 { | |
| 4042 int multi_p; | |
| 4043 Lisp_Object cset; | |
| 4044 | |
| 826 | 4045 cset = charset_by_leading_byte (j); |
| 428 | 4046 if (CHARSETP (cset)) |
| 4047 { | |
| 826 | 4048 if (charset_syntax (lispbuf, cset, &multi_p) |
| 428 | 4049 != Sword || multi_p) |
| 4050 fastmap[j] = 1; | |
| 4051 } | |
| 4052 } | |
| 4053 } | |
| 446 | 4054 #else /* not MULE */ |
| 428 | 4055 for (j = 0; j < (1 << BYTEWIDTH); j++) |
| 826 | 4056 if (SYNTAX |
| 428 | 4057 (XCHAR_TABLE |
| 826 | 4058 (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf)), j) != |
| 428 | 4059 (enum syntaxcode) k) |
| 4060 fastmap[j] = 1; | |
| 446 | 4061 #endif /* MULE */ |
| 428 | 4062 break; |
| 826 | 4063 #endif /* 0 */ |
| 428 | 4064 |
| 4065 #ifdef MULE | |
| 4066 /* 97/2/17 jhod category patch */ | |
| 4067 case categoryspec: | |
| 4068 case notcategoryspec: | |
| 4069 bufp->can_be_null = 1; | |
| 1333 | 4070 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4071 return 0; |
| 4072 /* end if category patch */ | |
| 4073 #endif /* MULE */ | |
| 4074 | |
| 4075 /* All cases after this match the empty string. These end with | |
| 4076 `continue'. */ | |
| 4077 case before_dot: | |
| 4078 case at_dot: | |
| 4079 case after_dot: | |
| 4080 continue; | |
| 826 | 4081 #endif /* emacs */ |
| 428 | 4082 |
| 4083 | |
| 4084 case no_op: | |
| 4085 case begline: | |
| 4086 case endline: | |
| 4087 case begbuf: | |
| 4088 case endbuf: | |
| 460 | 4089 #ifndef emacs |
| 428 | 4090 case wordbound: |
| 4091 case notwordbound: | |
| 4092 case wordbeg: | |
| 4093 case wordend: | |
| 460 | 4094 #endif |
| 428 | 4095 case push_dummy_failure: |
| 4096 continue; | |
| 4097 | |
| 4098 | |
| 4099 case jump_n: | |
| 4100 case pop_failure_jump: | |
| 4101 case maybe_pop_jump: | |
| 4102 case jump: | |
| 4103 case jump_past_alt: | |
| 4104 case dummy_failure_jump: | |
| 4105 EXTRACT_NUMBER_AND_INCR (j, p); | |
| 4106 p += j; | |
| 4107 if (j > 0) | |
| 4108 continue; | |
| 4109 | |
| 4110 /* Jump backward implies we just went through the body of a | |
| 4111 loop and matched nothing. Opcode jumped to should be | |
| 4112 `on_failure_jump' or `succeed_n'. Just treat it like an | |
| 4113 ordinary jump. For a * loop, it has pushed its failure | |
| 4114 point already; if so, discard that as redundant. */ | |
| 4115 if ((re_opcode_t) *p != on_failure_jump | |
| 4116 && (re_opcode_t) *p != succeed_n) | |
| 4117 continue; | |
| 4118 | |
| 4119 p++; | |
| 4120 EXTRACT_NUMBER_AND_INCR (j, p); | |
| 4121 p += j; | |
| 4122 | |
| 4123 /* If what's on the stack is where we are now, pop it. */ | |
| 4124 if (!FAIL_STACK_EMPTY () | |
| 4125 && fail_stack.stack[fail_stack.avail - 1].pointer == p) | |
| 4126 fail_stack.avail--; | |
| 4127 | |
| 4128 continue; | |
| 4129 | |
| 4130 | |
| 4131 case on_failure_jump: | |
| 4132 case on_failure_keep_string_jump: | |
| 4133 handle_on_failure_jump: | |
| 4134 EXTRACT_NUMBER_AND_INCR (j, p); | |
| 4135 | |
| 4136 /* For some patterns, e.g., `(a?)?', `p+j' here points to the | |
| 4137 end of the pattern. We don't want to push such a point, | |
| 4138 since when we restore it above, entering the switch will | |
| 4139 increment `p' past the end of the pattern. We don't need | |
| 4140 to push such a point since we obviously won't find any more | |
| 4141 fastmap entries beyond `pend'. Such a pattern can match | |
| 4142 the null string, though. */ | |
| 4143 if (p + j < pend) | |
| 4144 { | |
| 4145 if (!PUSH_PATTERN_OP (p + j, fail_stack)) | |
| 4146 { | |
| 4147 RESET_FAIL_STACK (); | |
| 1333 | 4148 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4149 return -2; |
| 4150 } | |
| 4151 } | |
| 4152 else | |
| 4153 bufp->can_be_null = 1; | |
| 4154 | |
| 4155 if (succeed_n_p) | |
| 4156 { | |
| 4157 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ | |
| 4158 succeed_n_p = false; | |
| 4159 } | |
| 4160 | |
| 4161 continue; | |
| 4162 | |
| 4163 | |
| 4164 case succeed_n: | |
| 4165 /* Get to the number of times to succeed. */ | |
| 4166 p += 2; | |
| 4167 | |
| 4168 /* Increment p past the n for when k != 0. */ | |
| 4169 EXTRACT_NUMBER_AND_INCR (k, p); | |
| 4170 if (k == 0) | |
| 4171 { | |
| 4172 p -= 4; | |
| 4173 succeed_n_p = true; /* Spaghetti code alert. */ | |
| 4174 goto handle_on_failure_jump; | |
| 4175 } | |
| 4176 continue; | |
| 4177 | |
| 4178 | |
| 4179 case set_number_at: | |
| 4180 p += 4; | |
| 4181 continue; | |
| 4182 | |
| 4183 | |
| 4184 case start_memory: | |
| 4185 case stop_memory: | |
| 4186 p += 2; | |
| 4187 continue; | |
| 4188 | |
| 4189 | |
| 4190 default: | |
| 2500 | 4191 ABORT (); /* We have listed all the cases. */ |
| 428 | 4192 } /* switch *p++ */ |
| 4193 | |
| 4194 /* Getting here means we have found the possible starting | |
| 4195 characters for one path of the pattern -- and that the empty | |
| 4196 string does not match. We need not follow this path further. | |
| 4197 Instead, look at the next alternative (remembered on the | |
| 4198 stack), or quit if no more. The test at the top of the loop | |
| 4199 does these things. */ | |
| 4200 path_can_be_null = false; | |
| 4201 p = pend; | |
| 4202 } /* while p */ | |
| 4203 | |
| 4204 /* Set `can_be_null' for the last path (also the first path, if the | |
| 4205 pattern is empty). */ | |
| 4206 bufp->can_be_null |= path_can_be_null; | |
| 4207 | |
| 4208 done: | |
| 4209 RESET_FAIL_STACK (); | |
| 1333 | 4210 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4211 return 0; |
| 4212 } /* re_compile_fastmap */ | |
| 4213 | |
| 4214 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and | |
| 4215 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use | |
| 4216 this memory for recording register information. STARTS and ENDS | |
| 4217 must be allocated using the malloc library routine, and must each | |
| 4218 be at least NUM_REGS * sizeof (regoff_t) bytes long. | |
| 4219 | |
| 4220 If NUM_REGS == 0, then subsequent matches should allocate their own | |
| 4221 register data. | |
| 4222 | |
| 4223 Unless this function is called, the first search or match using | |
| 4224 PATTERN_BUFFER will allocate its own register data, without | |
| 4225 freeing the old data. */ | |
| 4226 | |
| 4227 void | |
| 4228 re_set_registers (struct re_pattern_buffer *bufp, struct re_registers *regs, | |
| 647 | 4229 int num_regs, regoff_t *starts, regoff_t *ends) |
| 428 | 4230 { |
| 4231 if (num_regs) | |
| 4232 { | |
| 4233 bufp->regs_allocated = REGS_REALLOCATE; | |
| 4234 regs->num_regs = num_regs; | |
| 4235 regs->start = starts; | |
| 4236 regs->end = ends; | |
| 4237 } | |
| 4238 else | |
| 4239 { | |
| 4240 bufp->regs_allocated = REGS_UNALLOCATED; | |
| 4241 regs->num_regs = 0; | |
| 4242 regs->start = regs->end = (regoff_t *) 0; | |
| 4243 } | |
| 4244 } | |
| 4245 | |
| 4246 /* Searching routines. */ | |
| 4247 | |
| 4248 /* Like re_search_2, below, but only one string is specified, and | |
| 4249 doesn't let you say where to stop matching. */ | |
| 4250 | |
| 4251 int | |
| 442 | 4252 re_search (struct re_pattern_buffer *bufp, const char *string, int size, |
| 826 | 4253 int startpos, int range, struct re_registers *regs |
| 4254 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4255 { |
| 4256 return re_search_2 (bufp, NULL, 0, string, size, startpos, range, | |
| 826 | 4257 regs, size RE_LISP_CONTEXT_ARGS); |
| 428 | 4258 } |
| 4259 | |
| 4260 /* Using the compiled pattern in BUFP->buffer, first tries to match the | |
| 4261 virtual concatenation of STRING1 and STRING2, starting first at index | |
| 4262 STARTPOS, then at STARTPOS + 1, and so on. | |
| 4263 | |
| 4264 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. | |
| 4265 | |
| 4266 RANGE is how far to scan while trying to match. RANGE = 0 means try | |
| 4267 only at STARTPOS; in general, the last start tried is STARTPOS + | |
| 4268 RANGE. | |
| 4269 | |
| 826 | 4270 All sizes and positions refer to bytes (not chars); under Mule, the code |
| 4271 knows about the format of the text and will only check at positions | |
| 4272 where a character starts. | |
| 4273 | |
| 428 | 4274 With MULE, RANGE is a byte position, not a char position. The last |
| 4275 start tried is the character starting <= STARTPOS + RANGE. | |
| 4276 | |
| 4277 In REGS, return the indices of the virtual concatenation of STRING1 | |
| 4278 and STRING2 that matched the entire BUFP->buffer and its contained | |
| 4279 subexpressions. | |
| 4280 | |
| 4281 Do not consider matching one past the index STOP in the virtual | |
| 4282 concatenation of STRING1 and STRING2. | |
| 4283 | |
| 4284 We return either the position in the strings at which the match was | |
| 4285 found, -1 if no match, or -2 if error (such as failure | |
| 4286 stack overflow). */ | |
| 4287 | |
| 4288 int | |
| 446 | 4289 re_search_2 (struct re_pattern_buffer *bufp, const char *str1, |
| 4290 int size1, const char *str2, int size2, int startpos, | |
| 826 | 4291 int range, struct re_registers *regs, int stop |
| 4292 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4293 { |
| 4294 int val; | |
| 446 | 4295 re_char *string1 = (re_char *) str1; |
| 4296 re_char *string2 = (re_char *) str2; | |
| 428 | 4297 REGISTER char *fastmap = bufp->fastmap; |
| 446 | 4298 REGISTER RE_TRANSLATE_TYPE translate = bufp->translate; |
| 428 | 4299 int total_size = size1 + size2; |
| 4300 int endpos = startpos + range; | |
| 4301 #ifdef REGEX_BEGLINE_CHECK | |
| 4302 int anchored_at_begline = 0; | |
| 4303 #endif | |
| 446 | 4304 re_char *d; |
| 826 | 4305 #ifdef emacs |
| 4306 Internal_Format fmt = buffer_or_other_internal_format (lispobj); | |
| 1346 | 4307 #ifdef REL_ALLOC |
| 4308 Ibyte *orig_buftext = | |
| 4309 BUFFERP (lispobj) ? | |
| 4310 BYTE_BUF_BYTE_ADDRESS (XBUFFER (lispobj), | |
| 4311 BYTE_BUF_BEGV (XBUFFER (lispobj))) : | |
| 4312 0; | |
| 4313 #endif | |
| 1333 | 4314 #ifdef ERROR_CHECK_MALLOC |
| 4315 int depth; | |
| 4316 #endif | |
| 826 | 4317 #endif /* emacs */ |
| 4318 #if 1 | |
| 4319 int forward_search_p; | |
| 4320 #endif | |
| 428 | 4321 |
| 4322 /* Check for out-of-range STARTPOS. */ | |
| 4323 if (startpos < 0 || startpos > total_size) | |
| 4324 return -1; | |
| 4325 | |
| 4326 /* Fix up RANGE if it might eventually take us outside | |
| 4327 the virtual concatenation of STRING1 and STRING2. */ | |
| 4328 if (endpos < 0) | |
| 4329 range = 0 - startpos; | |
| 4330 else if (endpos > total_size) | |
| 4331 range = total_size - startpos; | |
| 4332 | |
| 826 | 4333 #if 1 |
| 4334 forward_search_p = range > 0; | |
| 4335 #endif | |
| 4336 | |
| 428 | 4337 /* If the search isn't to be a backwards one, don't waste time in a |
| 4338 search for a pattern that must be anchored. */ | |
| 4339 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0) | |
| 4340 { | |
| 4341 if (startpos > 0) | |
| 4342 return -1; | |
| 4343 else | |
| 4344 { | |
| 442 | 4345 d = ((const unsigned char *) |
| 428 | 4346 (startpos >= size1 ? string2 - size1 : string1) + startpos); |
| 867 | 4347 range = itext_ichar_len_fmt (d, fmt); |
| 428 | 4348 } |
| 4349 } | |
| 4350 | |
| 460 | 4351 #ifdef emacs |
| 4352 /* In a forward search for something that starts with \=. | |
| 4353 don't keep searching past point. */ | |
| 4354 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0) | |
| 4355 { | |
| 826 | 4356 if (!BUFFERP (lispobj)) |
| 4357 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
|
4358 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
|
4359 - BYTE_BUF_BEGV (XBUFFER (lispobj)) - startpos); |
| 460 | 4360 if (range < 0) |
| 4361 return -1; | |
| 4362 } | |
| 4363 #endif /* emacs */ | |
| 4364 | |
| 1333 | 4365 #ifdef ERROR_CHECK_MALLOC |
| 4366 /* Do this after the above return()s. */ | |
| 4367 depth = bind_regex_malloc_disallowed (1); | |
| 4368 #endif | |
| 4369 | |
| 428 | 4370 /* Update the fastmap now if not correct already. */ |
| 1333 | 4371 BEGIN_REGEX_MALLOC_OK (); |
| 428 | 4372 if (fastmap && !bufp->fastmap_accurate) |
| 826 | 4373 if (re_compile_fastmap (bufp RE_LISP_SHORT_CONTEXT_ARGS) == -2) |
| 1333 | 4374 { |
| 4375 END_REGEX_MALLOC_OK (); | |
| 4376 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4377 return -2; | |
| 4378 } | |
| 4379 | |
| 4380 END_REGEX_MALLOC_OK (); | |
| 4381 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 428 | 4382 |
| 4383 #ifdef REGEX_BEGLINE_CHECK | |
| 4384 { | |
| 647 | 4385 long i = 0; |
| 428 | 4386 |
| 4387 while (i < bufp->used) | |
| 4388 { | |
| 4389 if (bufp->buffer[i] == start_memory || | |
| 4390 bufp->buffer[i] == stop_memory) | |
| 4391 i += 2; | |
| 4392 else | |
| 4393 break; | |
| 4394 } | |
| 4395 anchored_at_begline = i < bufp->used && bufp->buffer[i] == begline; | |
| 4396 } | |
| 4397 #endif | |
| 4398 | |
| 460 | 4399 #ifdef emacs |
| 1333 | 4400 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 4401 scache = setup_syntax_cache (scache, lispobj, lispbuf, |
| 4402 offset_to_charxpos (lispobj, startpos), | |
| 4403 1); | |
| 1333 | 4404 END_REGEX_MALLOC_OK (); |
| 4405 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 460 | 4406 #endif |
| 4407 | |
| 428 | 4408 /* Loop through the string, looking for a place to start matching. */ |
| 4409 for (;;) | |
| 4410 { | |
| 4411 #ifdef REGEX_BEGLINE_CHECK | |
| 826 | 4412 /* If the regex is anchored at the beginning of a line (i.e. with a |
| 4413 ^), then we can speed things up by skipping to the next | |
| 4414 beginning-of-line. However, to determine "beginning of line" we | |
| 4415 need to look at the previous char, so can't do this check if at | |
| 4416 beginning of either string. (Well, we could if at the beginning of | |
| 4417 the second string, but it would require additional code, and this | |
| 4418 is just an optimization.) */ | |
| 4419 if (anchored_at_begline && startpos > 0 && startpos != size1) | |
| 428 | 4420 { |
| 826 | 4421 if (range > 0) |
| 4422 { | |
| 4423 /* whose stupid idea was it anyway to make this | |
| 4424 function take two strings to match?? */ | |
| 4425 int lim = 0; | |
| 4426 re_char *orig_d; | |
| 4427 re_char *stop_d; | |
| 4428 | |
| 4429 /* Compute limit as below in fastmap code, so we are guaranteed | |
| 4430 to remain within a single string. */ | |
| 4431 if (startpos < size1 && startpos + range >= size1) | |
| 4432 lim = range - (size1 - startpos); | |
| 4433 | |
| 4434 d = ((const unsigned char *) | |
| 4435 (startpos >= size1 ? string2 - size1 : string1) + startpos); | |
| 4436 orig_d = d; | |
| 4437 stop_d = d + range - lim; | |
| 4438 | |
| 4439 /* We want to find the next location (including the current | |
| 4440 one) where the previous char is a newline, so back up one | |
| 4441 and search forward for a newline. */ | |
| 867 | 4442 DEC_IBYTEPTR_FMT (d, fmt); /* Ok, since startpos != size1. */ |
| 826 | 4443 |
| 4444 /* Written out as an if-else to avoid testing `translate' | |
| 4445 inside the loop. */ | |
| 4446 if (TRANSLATE_P (translate)) | |
| 4447 while (d < stop_d && | |
| 867 | 4448 RE_TRANSLATE_1 (itext_ichar_fmt (d, fmt, lispobj)) |
| 826 | 4449 != '\n') |
| 867 | 4450 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4451 else |
| 4452 while (d < stop_d && | |
| 867 | 4453 itext_ichar_ascii_fmt (d, fmt, lispobj) != '\n') |
| 4454 INC_IBYTEPTR_FMT (d, fmt); | |
| 826 | 4455 |
| 4456 /* If we were stopped by a newline, skip forward over it. | |
| 4457 Otherwise we will get in an infloop when our start position | |
| 4458 was at begline. */ | |
| 4459 if (d < stop_d) | |
| 867 | 4460 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4461 range -= d - orig_d; |
| 4462 startpos += d - orig_d; | |
| 4463 #if 1 | |
| 4464 assert (!forward_search_p || range >= 0); | |
| 4465 #endif | |
| 4466 } | |
| 4467 else if (range < 0) | |
| 4468 { | |
| 4469 /* We're lazy, like in the fastmap code below */ | |
| 867 | 4470 Ichar c; |
| 826 | 4471 |
| 4472 d = ((const unsigned char *) | |
| 4473 (startpos >= size1 ? string2 - size1 : string1) + startpos); | |
| 867 | 4474 DEC_IBYTEPTR_FMT (d, fmt); |
| 4475 c = itext_ichar_fmt (d, fmt, lispobj); | |
| 826 | 4476 c = RE_TRANSLATE (c); |
| 4477 if (c != '\n') | |
| 4478 goto advance; | |
| 4479 } | |
| 428 | 4480 } |
| 4481 #endif /* REGEX_BEGLINE_CHECK */ | |
| 4482 | |
| 4483 /* If a fastmap is supplied, skip quickly over characters that | |
| 4484 cannot be the start of a match. If the pattern can match the | |
| 4485 null string, however, we don't need to skip characters; we want | |
| 4486 the first null string. */ | |
| 4487 if (fastmap && startpos < total_size && !bufp->can_be_null) | |
| 4488 { | |
| 826 | 4489 /* For the moment, fastmap always works as if buffer |
| 4490 is in default format, so convert chars in the search strings | |
| 4491 into default format as we go along, if necessary. | |
| 4492 | |
| 4493 &&#### fastmap needs rethinking for 8-bit-fixed so | |
| 4494 it's faster. We need it to reflect the raw | |
| 4495 8-bit-fixed values. That isn't so hard if we assume | |
| 4496 that the top 96 bytes represent a single 1-byte | |
| 4497 charset. For 16-bit/32-bit stuff it's probably not | |
| 4498 worth it to make the fastmap represent the raw, due to | |
| 4499 its nature -- we'd have to use the LSB for the | |
| 4500 fastmap, and that causes lots of problems with Mule | |
| 4501 chars, where it essentially wipes out the usefulness | |
| 4502 of the fastmap entirely. */ | |
| 428 | 4503 if (range > 0) /* Searching forwards. */ |
| 4504 { | |
| 4505 int lim = 0; | |
| 4506 int irange = range; | |
| 4507 | |
| 4508 if (startpos < size1 && startpos + range >= size1) | |
| 4509 lim = range - (size1 - startpos); | |
| 4510 | |
| 442 | 4511 d = ((const unsigned char *) |
| 428 | 4512 (startpos >= size1 ? string2 - size1 : string1) + startpos); |
| 4513 | |
| 4514 /* Written out as an if-else to avoid testing `translate' | |
| 4515 inside the loop. */ | |
| 446 | 4516 if (TRANSLATE_P (translate)) |
| 826 | 4517 { |
| 4518 while (range > lim) | |
| 4519 { | |
| 4520 re_char *old_d = d; | |
| 428 | 4521 #ifdef MULE |
| 867 | 4522 Ibyte tempch[MAX_ICHAR_LEN]; |
| 4523 Ichar buf_ch = | |
| 4524 RE_TRANSLATE_1 (itext_ichar_fmt (d, fmt, lispobj)); | |
| 4525 set_itext_ichar (tempch, buf_ch); | |
| 826 | 4526 if (fastmap[*tempch]) |
| 4527 break; | |
| 446 | 4528 #else |
| 826 | 4529 if (fastmap[(unsigned char) RE_TRANSLATE_1 (*d)]) |
| 4530 break; | |
| 446 | 4531 #endif /* MULE */ |
| 867 | 4532 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4533 range -= (d - old_d); |
| 4534 #if 1 | |
| 1333 | 4535 assert (!forward_search_p || range >= 0); |
| 826 | 4536 #endif |
| 4537 } | |
| 4538 } | |
| 4539 #ifdef MULE | |
| 4540 else if (fmt != FORMAT_DEFAULT) | |
| 4541 { | |
| 4542 while (range > lim) | |
| 4543 { | |
| 4544 re_char *old_d = d; | |
| 867 | 4545 Ibyte tempch[MAX_ICHAR_LEN]; |
| 4546 Ichar buf_ch = itext_ichar_fmt (d, fmt, lispobj); | |
| 4547 set_itext_ichar (tempch, buf_ch); | |
| 826 | 4548 if (fastmap[*tempch]) |
| 4549 break; | |
| 867 | 4550 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 4551 range -= (d - old_d); |
| 4552 #if 1 | |
| 1333 | 4553 assert (!forward_search_p || range >= 0); |
| 826 | 4554 #endif |
| 4555 } | |
| 4556 } | |
| 4557 #endif /* MULE */ | |
| 428 | 4558 else |
| 826 | 4559 { |
| 4560 while (range > lim && !fastmap[*d]) | |
| 4561 { | |
| 4562 re_char *old_d = d; | |
| 867 | 4563 INC_IBYTEPTR (d); |
| 826 | 4564 range -= (d - old_d); |
| 4565 #if 1 | |
| 4566 assert (!forward_search_p || range >= 0); | |
| 4567 #endif | |
| 4568 } | |
| 4569 } | |
| 428 | 4570 |
| 4571 startpos += irange - range; | |
| 4572 } | |
| 4573 else /* Searching backwards. */ | |
| 4574 { | |
| 826 | 4575 /* #### It's not clear why we don't just write a loop, like |
| 4576 for the moving-forward case. Perhaps the writer got lazy, | |
| 4577 since backward searches aren't so common. */ | |
| 4578 d = ((const unsigned char *) | |
| 4579 (startpos >= size1 ? string2 - size1 : string1) + startpos); | |
| 428 | 4580 #ifdef MULE |
| 826 | 4581 { |
| 867 | 4582 Ibyte tempch[MAX_ICHAR_LEN]; |
| 4583 Ichar buf_ch = | |
| 4584 RE_TRANSLATE (itext_ichar_fmt (d, fmt, lispobj)); | |
| 4585 set_itext_ichar (tempch, buf_ch); | |
| 826 | 4586 if (!fastmap[*tempch]) |
| 4587 goto advance; | |
| 4588 } | |
| 428 | 4589 #else |
| 826 | 4590 if (!fastmap[(unsigned char) RE_TRANSLATE (*d)]) |
| 446 | 4591 goto advance; |
| 826 | 4592 #endif /* MULE */ |
| 428 | 4593 } |
| 4594 } | |
| 4595 | |
| 4596 /* If can't match the null string, and that's all we have left, fail. */ | |
| 4597 if (range >= 0 && startpos == total_size && fastmap | |
| 4598 && !bufp->can_be_null) | |
| 1333 | 4599 { |
| 4600 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4601 return -1; | |
| 4602 } | |
| 428 | 4603 |
| 4604 #ifdef emacs /* XEmacs added, w/removal of immediate_quit */ | |
| 4605 if (!no_quit_in_re_search) | |
| 1333 | 4606 { |
| 4607 BEGIN_REGEX_MALLOC_OK (); | |
| 4608 QUIT; | |
| 4609 END_REGEX_MALLOC_OK (); | |
| 4610 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 4611 } | |
| 4612 | |
| 428 | 4613 #endif |
| 1333 | 4614 BEGIN_REGEX_MALLOC_OK (); |
| 428 | 4615 val = re_match_2_internal (bufp, string1, size1, string2, size2, |
| 826 | 4616 startpos, regs, stop |
| 4617 RE_LISP_CONTEXT_ARGS); | |
| 428 | 4618 #ifndef REGEX_MALLOC |
| 1333 | 4619 ALLOCA_GARBAGE_COLLECT (); |
| 428 | 4620 #endif |
| 1333 | 4621 END_REGEX_MALLOC_OK (); |
| 4622 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 428 | 4623 |
| 4624 if (val >= 0) | |
| 1333 | 4625 { |
| 4626 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4627 return startpos; | |
| 4628 } | |
| 428 | 4629 |
| 4630 if (val == -2) | |
| 1333 | 4631 { |
| 4632 UNBIND_REGEX_MALLOC_CHECK (); | |
| 4633 return -2; | |
| 4634 } | |
| 4635 | |
| 4636 RE_SEARCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 428 | 4637 advance: |
| 4638 if (!range) | |
| 4639 break; | |
| 4640 else if (range > 0) | |
| 4641 { | |
| 826 | 4642 Bytecount d_size; |
| 442 | 4643 d = ((const unsigned char *) |
| 428 | 4644 (startpos >= size1 ? string2 - size1 : string1) + startpos); |
| 867 | 4645 d_size = itext_ichar_len_fmt (d, fmt); |
| 428 | 4646 range -= d_size; |
| 826 | 4647 #if 1 |
| 4648 assert (!forward_search_p || range >= 0); | |
| 4649 #endif | |
| 428 | 4650 startpos += d_size; |
| 4651 } | |
| 4652 else | |
| 4653 { | |
| 826 | 4654 Bytecount d_size; |
| 428 | 4655 /* Note startpos > size1 not >=. If we are on the |
| 4656 string1/string2 boundary, we want to backup into string1. */ | |
| 442 | 4657 d = ((const unsigned char *) |
| 428 | 4658 (startpos > size1 ? string2 - size1 : string1) + startpos); |
| 867 | 4659 DEC_IBYTEPTR_FMT (d, fmt); |
| 4660 d_size = itext_ichar_len_fmt (d, fmt); | |
| 428 | 4661 range += d_size; |
| 826 | 4662 #if 1 |
| 4663 assert (!forward_search_p || range >= 0); | |
| 4664 #endif | |
| 428 | 4665 startpos -= d_size; |
| 4666 } | |
| 4667 } | |
| 1333 | 4668 UNBIND_REGEX_MALLOC_CHECK (); |
| 428 | 4669 return -1; |
| 4670 } /* re_search_2 */ | |
| 826 | 4671 |
| 428 | 4672 |
| 4673 /* Declarations and macros for re_match_2. */ | |
| 4674 | |
| 4675 /* This converts PTR, a pointer into one of the search strings `string1' | |
| 4676 and `string2' into an offset from the beginning of that string. */ | |
| 4677 #define POINTER_TO_OFFSET(ptr) \ | |
| 4678 (FIRST_STRING_P (ptr) \ | |
| 4679 ? ((regoff_t) ((ptr) - string1)) \ | |
| 4680 : ((regoff_t) ((ptr) - string2 + size1))) | |
| 4681 | |
| 4682 /* Macros for dealing with the split strings in re_match_2. */ | |
| 4683 | |
| 4684 #define MATCHING_IN_FIRST_STRING (dend == end_match_1) | |
| 4685 | |
| 4686 /* Call before fetching a character with *d. This switches over to | |
| 4687 string2 if necessary. */ | |
| 826 | 4688 #define REGEX_PREFETCH() \ |
| 428 | 4689 while (d == dend) \ |
| 4690 { \ | |
| 4691 /* End of string2 => fail. */ \ | |
| 4692 if (dend == end_match_2) \ | |
| 4693 goto fail; \ | |
| 4694 /* End of string1 => advance to string2. */ \ | |
| 4695 d = string2; \ | |
| 4696 dend = end_match_2; \ | |
| 4697 } | |
| 4698 | |
| 4699 | |
| 4700 /* Test if at very beginning or at very end of the virtual concatenation | |
| 4701 of `string1' and `string2'. If only one string, it's `string2'. */ | |
| 4702 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) | |
| 4703 #define AT_STRINGS_END(d) ((d) == end2) | |
| 4704 | |
| 4705 /* XEmacs change: | |
| 4706 If the given position straddles the string gap, return the equivalent | |
| 4707 position that is before or after the gap, respectively; otherwise, | |
| 4708 return the same position. */ | |
| 4709 #define POS_BEFORE_GAP_UNSAFE(d) ((d) == string2 ? end1 : (d)) | |
| 4710 #define POS_AFTER_GAP_UNSAFE(d) ((d) == end1 ? string2 : (d)) | |
| 4711 | |
| 4712 /* Test if CH is a word-constituent character. (XEmacs change) */ | |
| 826 | 4713 #define WORDCHAR_P(ch) \ |
| 4714 (SYNTAX (BUFFER_MIRROR_SYNTAX_TABLE (lispbuf), ch) == Sword) | |
| 428 | 4715 |
| 4716 /* Free everything we malloc. */ | |
| 4717 #ifdef MATCH_MAY_ALLOCATE | |
| 1726 | 4718 #define FREE_VAR(var,type) if (var) REGEX_FREE (var, type); var = NULL |
| 428 | 4719 #define FREE_VARIABLES() \ |
| 4720 do { \ | |
| 1333 | 4721 UNBIND_REGEX_MALLOC_CHECK (); \ |
| 428 | 4722 REGEX_FREE_STACK (fail_stack.stack); \ |
| 1726 | 4723 FREE_VAR (regstart, re_char **); \ |
| 4724 FREE_VAR (regend, re_char **); \ | |
| 4725 FREE_VAR (old_regstart, re_char **); \ | |
| 4726 FREE_VAR (old_regend, re_char **); \ | |
| 4727 FREE_VAR (best_regstart, re_char **); \ | |
| 4728 FREE_VAR (best_regend, re_char **); \ | |
| 4729 FREE_VAR (reg_info, register_info_type *); \ | |
| 4730 FREE_VAR (reg_dummy, re_char **); \ | |
| 4731 FREE_VAR (reg_info_dummy, register_info_type *); \ | |
| 428 | 4732 } while (0) |
| 446 | 4733 #else /* not MATCH_MAY_ALLOCATE */ |
| 1333 | 4734 #define FREE_VARIABLES() \ |
| 4735 do { \ | |
| 4736 UNBIND_REGEX_MALLOC_CHECK (); \ | |
| 4737 } while (0) | |
| 446 | 4738 #endif /* MATCH_MAY_ALLOCATE */ |
| 428 | 4739 |
| 4740 /* These values must meet several constraints. They must not be valid | |
| 4741 register values; since we have a limit of 255 registers (because | |
| 4742 we use only one byte in the pattern for the register number), we can | |
| 4743 use numbers larger than 255. They must differ by 1, because of | |
| 4744 NUM_FAILURE_ITEMS above. And the value for the lowest register must | |
| 4745 be larger than the value for the highest register, so we do not try | |
| 4746 to actually save any registers when none are active. */ | |
| 4747 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) | |
| 4748 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) | |
| 4749 | |
| 4750 /* Matching routines. */ | |
| 4751 | |
| 826 | 4752 #ifndef emacs /* XEmacs never uses this. */ |
| 428 | 4753 /* re_match is like re_match_2 except it takes only a single string. */ |
| 4754 | |
| 4755 int | |
| 442 | 4756 re_match (struct re_pattern_buffer *bufp, const char *string, int size, |
| 826 | 4757 int pos, struct re_registers *regs |
| 4758 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4759 { |
| 446 | 4760 int result = re_match_2_internal (bufp, NULL, 0, (re_char *) string, size, |
| 826 | 4761 pos, regs, size |
| 4762 RE_LISP_CONTEXT_ARGS); | |
| 1333 | 4763 ALLOCA_GARBAGE_COLLECT (); |
| 428 | 4764 return result; |
| 4765 } | |
| 4766 #endif /* not emacs */ | |
| 4767 | |
| 4768 /* re_match_2 matches the compiled pattern in BUFP against the | |
| 4769 (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 and | |
| 4770 SIZE2, respectively). We start matching at POS, and stop matching | |
| 4771 at STOP. | |
| 4772 | |
| 4773 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we | |
| 4774 store offsets for the substring each group matched in REGS. See the | |
| 4775 documentation for exactly how many groups we fill. | |
| 4776 | |
| 4777 We return -1 if no match, -2 if an internal error (such as the | |
| 4778 failure stack overflowing). Otherwise, we return the length of the | |
| 4779 matched substring. */ | |
| 4780 | |
| 4781 int | |
| 442 | 4782 re_match_2 (struct re_pattern_buffer *bufp, const char *string1, |
| 4783 int size1, const char *string2, int size2, int pos, | |
| 826 | 4784 struct re_registers *regs, int stop |
| 4785 RE_LISP_CONTEXT_ARGS_DECL) | |
| 428 | 4786 { |
| 460 | 4787 int result; |
| 4788 | |
| 4789 #ifdef emacs | |
| 826 | 4790 scache = setup_syntax_cache (scache, lispobj, lispbuf, |
| 4791 offset_to_charxpos (lispobj, pos), | |
| 4792 1); | |
| 460 | 4793 #endif |
| 4794 | |
| 4795 result = re_match_2_internal (bufp, (re_char *) string1, size1, | |
| 4796 (re_char *) string2, size2, | |
| 826 | 4797 pos, regs, stop |
| 4798 RE_LISP_CONTEXT_ARGS); | |
| 460 | 4799 |
| 1333 | 4800 ALLOCA_GARBAGE_COLLECT (); |
| 428 | 4801 return result; |
| 4802 } | |
| 4803 | |
| 4804 /* This is a separate function so that we can force an alloca cleanup | |
| 4805 afterwards. */ | |
| 4806 static int | |
| 446 | 4807 re_match_2_internal (struct re_pattern_buffer *bufp, re_char *string1, |
| 4808 int size1, re_char *string2, int size2, int pos, | |
| 826 | 4809 struct re_registers *regs, int stop |
| 2333 | 4810 RE_LISP_CONTEXT_ARGS_MULE_DECL) |
| 428 | 4811 { |
| 4812 /* General temporaries. */ | |
| 4813 int mcnt; | |
| 4814 unsigned char *p1; | |
| 4815 int should_succeed; /* XEmacs change */ | |
| 4816 | |
| 4817 /* Just past the end of the corresponding string. */ | |
| 446 | 4818 re_char *end1, *end2; |
| 428 | 4819 |
| 4820 /* Pointers into string1 and string2, just past the last characters in | |
| 4821 each to consider matching. */ | |
| 446 | 4822 re_char *end_match_1, *end_match_2; |
| 428 | 4823 |
| 4824 /* Where we are in the data, and the end of the current string. */ | |
| 446 | 4825 re_char *d, *dend; |
| 428 | 4826 |
| 4827 /* Where we are in the pattern, and the end of the pattern. */ | |
| 4828 unsigned char *p = bufp->buffer; | |
| 4829 REGISTER unsigned char *pend = p + bufp->used; | |
| 4830 | |
| 4831 /* Mark the opcode just after a start_memory, so we can test for an | |
| 4832 empty subpattern when we get to the stop_memory. */ | |
| 446 | 4833 re_char *just_past_start_mem = 0; |
| 428 | 4834 |
| 4835 /* We use this to map every character in the string. */ | |
| 446 | 4836 RE_TRANSLATE_TYPE translate = bufp->translate; |
| 428 | 4837 |
| 4838 /* Failure point stack. Each place that can handle a failure further | |
| 4839 down the line pushes a failure point on this stack. It consists of | |
| 4840 restart, regend, and reg_info for all registers corresponding to | |
| 4841 the subexpressions we're currently inside, plus the number of such | |
| 4842 registers, and, finally, two char *'s. The first char * is where | |
| 4843 to resume scanning the pattern; the second one is where to resume | |
| 4844 scanning the strings. If the latter is zero, the failure point is | |
| 4845 a ``dummy''; if a failure happens and the failure point is a dummy, | |
| 4846 it gets discarded and the next one is tried. */ | |
| 4847 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ | |
| 4848 fail_stack_type fail_stack; | |
| 4849 #endif | |
| 4850 #ifdef DEBUG | |
| 647 | 4851 static int failure_id; |
| 4852 int nfailure_points_pushed = 0, nfailure_points_popped = 0; | |
| 428 | 4853 #endif |
| 4854 | |
| 771 | 4855 #ifdef REGEX_REL_ALLOC |
| 428 | 4856 /* This holds the pointer to the failure stack, when |
| 4857 it is allocated relocatably. */ | |
| 4858 fail_stack_elt_t *failure_stack_ptr; | |
| 4859 #endif | |
| 4860 | |
| 4861 /* We fill all the registers internally, independent of what we | |
| 4862 return, for use in backreferences. The number here includes | |
| 4863 an element for register zero. */ | |
| 647 | 4864 int num_regs = bufp->re_ngroups + 1; |
| 428 | 4865 |
| 4866 /* The currently active registers. */ | |
| 647 | 4867 int lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
| 4868 int highest_active_reg = NO_HIGHEST_ACTIVE_REG; | |
| 428 | 4869 |
| 4870 /* Information on the contents of registers. These are pointers into | |
| 4871 the input strings; they record just what was matched (on this | |
| 4872 attempt) by a subexpression part of the pattern, that is, the | |
| 4873 regnum-th regstart pointer points to where in the pattern we began | |
| 4874 matching and the regnum-th regend points to right after where we | |
| 4875 stopped matching the regnum-th subexpression. (The zeroth register | |
| 4876 keeps track of what the whole pattern matches.) */ | |
| 4877 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ | |
| 446 | 4878 re_char **regstart, **regend; |
| 428 | 4879 #endif |
| 4880 | |
| 4881 /* If a group that's operated upon by a repetition operator fails to | |
| 4882 match anything, then the register for its start will need to be | |
| 4883 restored because it will have been set to wherever in the string we | |
| 4884 are when we last see its open-group operator. Similarly for a | |
| 4885 register's end. */ | |
| 4886 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ | |
| 446 | 4887 re_char **old_regstart, **old_regend; |
| 428 | 4888 #endif |
| 4889 | |
| 4890 /* The is_active field of reg_info helps us keep track of which (possibly | |
| 4891 nested) subexpressions we are currently in. The matched_something | |
| 4892 field of reg_info[reg_num] helps us tell whether or not we have | |
| 4893 matched any of the pattern so far this time through the reg_num-th | |
| 4894 subexpression. These two fields get reset each time through any | |
| 4895 loop their register is in. */ | |
| 4896 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */ | |
| 4897 register_info_type *reg_info; | |
| 4898 #endif | |
| 4899 | |
| 4900 /* The following record the register info as found in the above | |
| 4901 variables when we find a match better than any we've seen before. | |
| 4902 This happens as we backtrack through the failure points, which in | |
| 4903 turn happens only if we have not yet matched the entire string. */ | |
| 647 | 4904 int best_regs_set = false; |
| 428 | 4905 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ |
| 446 | 4906 re_char **best_regstart, **best_regend; |
| 428 | 4907 #endif |
| 4908 | |
| 4909 /* Logically, this is `best_regend[0]'. But we don't want to have to | |
| 4910 allocate space for that if we're not allocating space for anything | |
| 4911 else (see below). Also, we never need info about register 0 for | |
| 4912 any of the other register vectors, and it seems rather a kludge to | |
| 4913 treat `best_regend' differently than the rest. So we keep track of | |
| 4914 the end of the best match so far in a separate variable. We | |
| 4915 initialize this to NULL so that when we backtrack the first time | |
| 4916 and need to test it, it's not garbage. */ | |
| 446 | 4917 re_char *match_end = NULL; |
| 428 | 4918 |
| 4919 /* This helps SET_REGS_MATCHED avoid doing redundant work. */ | |
| 4920 int set_regs_matched_done = 0; | |
| 4921 | |
| 4922 /* Used when we pop values we don't care about. */ | |
| 4923 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */ | |
| 446 | 4924 re_char **reg_dummy; |
| 428 | 4925 register_info_type *reg_info_dummy; |
| 4926 #endif | |
| 4927 | |
| 4928 #ifdef DEBUG | |
| 4929 /* Counts the total number of registers pushed. */ | |
| 647 | 4930 int num_regs_pushed = 0; |
| 428 | 4931 #endif |
| 4932 | |
| 4933 /* 1 if this match ends in the same string (string1 or string2) | |
| 4934 as the best previous match. */ | |
| 460 | 4935 re_bool same_str_p; |
| 428 | 4936 |
| 4937 /* 1 if this match is the best seen so far. */ | |
| 460 | 4938 re_bool best_match_p; |
| 428 | 4939 |
| 826 | 4940 #ifdef emacs |
| 4941 Internal_Format fmt = buffer_or_other_internal_format (lispobj); | |
| 1346 | 4942 #ifdef REL_ALLOC |
| 4943 Ibyte *orig_buftext = | |
| 4944 BUFFERP (lispobj) ? | |
| 4945 BYTE_BUF_BYTE_ADDRESS (XBUFFER (lispobj), | |
| 4946 BYTE_BUF_BEGV (XBUFFER (lispobj))) : | |
| 4947 0; | |
| 4948 #endif | |
| 4949 | |
| 1333 | 4950 #ifdef ERROR_CHECK_MALLOC |
| 4951 int depth = bind_regex_malloc_disallowed (1); | |
| 4952 #endif | |
| 826 | 4953 #endif /* emacs */ |
| 771 | 4954 |
| 5041 | 4955 DEBUG_MATCH_PRINT1 ("\n\nEntering re_match_2.\n"); |
| 428 | 4956 |
| 1333 | 4957 BEGIN_REGEX_MALLOC_OK (); |
| 428 | 4958 INIT_FAIL_STACK (); |
| 1333 | 4959 END_REGEX_MALLOC_OK (); |
| 428 | 4960 |
| 4961 #ifdef MATCH_MAY_ALLOCATE | |
| 4962 /* Do not bother to initialize all the register variables if there are | |
| 4963 no groups in the pattern, as it takes a fair amount of time. If | |
| 4964 there are groups, we include space for register 0 (the whole | |
| 4965 pattern), even though we never use it, since it simplifies the | |
| 4966 array indexing. We should fix this. */ | |
| 502 | 4967 if (bufp->re_ngroups) |
| 428 | 4968 { |
| 1333 | 4969 BEGIN_REGEX_MALLOC_OK (); |
| 446 | 4970 regstart = REGEX_TALLOC (num_regs, re_char *); |
| 4971 regend = REGEX_TALLOC (num_regs, re_char *); | |
| 4972 old_regstart = REGEX_TALLOC (num_regs, re_char *); | |
| 4973 old_regend = REGEX_TALLOC (num_regs, re_char *); | |
| 4974 best_regstart = REGEX_TALLOC (num_regs, re_char *); | |
| 4975 best_regend = REGEX_TALLOC (num_regs, re_char *); | |
| 428 | 4976 reg_info = REGEX_TALLOC (num_regs, register_info_type); |
| 446 | 4977 reg_dummy = REGEX_TALLOC (num_regs, re_char *); |
| 428 | 4978 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); |
| 1333 | 4979 END_REGEX_MALLOC_OK (); |
| 428 | 4980 |
| 4981 if (!(regstart && regend && old_regstart && old_regend && reg_info | |
| 4982 && best_regstart && best_regend && reg_dummy && reg_info_dummy)) | |
| 4983 { | |
| 4984 FREE_VARIABLES (); | |
| 4985 return -2; | |
| 4986 } | |
| 4987 } | |
| 4988 else | |
| 4989 { | |
| 4990 /* We must initialize all our variables to NULL, so that | |
| 4991 `FREE_VARIABLES' doesn't try to free them. */ | |
| 4992 regstart = regend = old_regstart = old_regend = best_regstart | |
| 4993 = best_regend = reg_dummy = NULL; | |
| 4994 reg_info = reg_info_dummy = (register_info_type *) NULL; | |
| 4995 } | |
| 4996 #endif /* MATCH_MAY_ALLOCATE */ | |
| 4997 | |
| 1333 | 4998 #if defined (emacs) && defined (REL_ALLOC) |
| 4999 { | |
| 5000 /* If the allocations above (or the call to setup_syntax_cache() in | |
| 5001 re_match_2) caused a rel-alloc relocation, then fix up the data | |
| 5002 pointers */ | |
| 1346 | 5003 Bytecount offset = offset_post_relocation (lispobj, orig_buftext); |
| 1333 | 5004 if (offset) |
| 5005 { | |
| 5006 string1 += offset; | |
| 5007 string2 += offset; | |
| 5008 } | |
| 5009 } | |
| 5010 #endif /* defined (emacs) && defined (REL_ALLOC) */ | |
| 5011 | |
| 428 | 5012 /* The starting position is bogus. */ |
| 5013 if (pos < 0 || pos > size1 + size2) | |
| 5014 { | |
| 5015 FREE_VARIABLES (); | |
| 5016 return -1; | |
| 5017 } | |
| 5018 | |
| 5019 /* Initialize subexpression text positions to -1 to mark ones that no | |
| 5020 start_memory/stop_memory has been seen for. Also initialize the | |
| 5021 register information struct. */ | |
| 5022 for (mcnt = 1; mcnt < num_regs; mcnt++) | |
| 5023 { | |
| 5024 regstart[mcnt] = regend[mcnt] | |
| 5025 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; | |
| 5026 | |
| 5027 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; | |
| 5028 IS_ACTIVE (reg_info[mcnt]) = 0; | |
| 5029 MATCHED_SOMETHING (reg_info[mcnt]) = 0; | |
| 5030 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; | |
| 5031 } | |
| 5032 /* We move `string1' into `string2' if the latter's empty -- but not if | |
| 5033 `string1' is null. */ | |
| 5034 if (size2 == 0 && string1 != NULL) | |
| 5035 { | |
| 5036 string2 = string1; | |
| 5037 size2 = size1; | |
| 5038 string1 = 0; | |
| 5039 size1 = 0; | |
| 5040 } | |
| 5041 end1 = string1 + size1; | |
| 5042 end2 = string2 + size2; | |
| 5043 | |
| 5044 /* Compute where to stop matching, within the two strings. */ | |
| 5045 if (stop <= size1) | |
| 5046 { | |
| 5047 end_match_1 = string1 + stop; | |
| 5048 end_match_2 = string2; | |
| 5049 } | |
| 5050 else | |
| 5051 { | |
| 5052 end_match_1 = end1; | |
| 5053 end_match_2 = string2 + stop - size1; | |
| 5054 } | |
| 5055 | |
| 5056 /* `p' scans through the pattern as `d' scans through the data. | |
| 5057 `dend' is the end of the input string that `d' points within. `d' | |
| 5058 is advanced into the following input string whenever necessary, but | |
| 5059 this happens before fetching; therefore, at the beginning of the | |
| 5060 loop, `d' can be pointing at the end of a string, but it cannot | |
| 5061 equal `string2'. */ | |
| 5062 if (size1 > 0 && pos <= size1) | |
| 5063 { | |
| 5064 d = string1 + pos; | |
| 5065 dend = end_match_1; | |
| 5066 } | |
| 5067 else | |
| 5068 { | |
| 5069 d = string2 + pos - size1; | |
| 5070 dend = end_match_2; | |
| 5071 } | |
| 5072 | |
| 5041 | 5073 DEBUG_MATCH_PRINT1 ("The compiled pattern is: \n"); |
| 5074 DEBUG_MATCH_PRINT_COMPILED_PATTERN (bufp, p, pend); | |
| 5075 DEBUG_MATCH_PRINT1 ("The string to match is: `"); | |
| 5076 DEBUG_MATCH_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); | |
| 5077 DEBUG_MATCH_PRINT1 ("'\n"); | |
| 428 | 5078 |
| 5079 /* This loops over pattern commands. It exits by returning from the | |
| 5080 function if the match is complete, or it drops through if the match | |
| 5081 fails at this starting point in the input data. */ | |
| 5082 for (;;) | |
| 5083 { | |
| 5041 | 5084 DEBUG_MATCH_PRINT2 ("\n0x%lx: ", (long) p); |
| 428 | 5085 #ifdef emacs /* XEmacs added, w/removal of immediate_quit */ |
| 5086 if (!no_quit_in_re_search) | |
| 1333 | 5087 { |
| 5088 BEGIN_REGEX_MALLOC_OK (); | |
| 5089 QUIT; | |
| 5090 END_REGEX_MALLOC_OK (); | |
| 1346 | 5091 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); |
| 1333 | 5092 } |
| 428 | 5093 #endif |
| 5094 | |
| 5095 if (p == pend) | |
| 5096 { /* End of pattern means we might have succeeded. */ | |
| 5041 | 5097 DEBUG_MATCH_PRINT1 ("end of pattern ... "); |
| 428 | 5098 |
| 5099 /* If we haven't matched the entire string, and we want the | |
| 5100 longest match, try backtracking. */ | |
| 5101 if (d != end_match_2) | |
| 5102 { | |
| 5103 same_str_p = (FIRST_STRING_P (match_end) | |
| 5104 == MATCHING_IN_FIRST_STRING); | |
| 5105 | |
| 5106 /* AIX compiler got confused when this was combined | |
| 5107 with the previous declaration. */ | |
| 5108 if (same_str_p) | |
| 5109 best_match_p = d > match_end; | |
| 5110 else | |
| 5111 best_match_p = !MATCHING_IN_FIRST_STRING; | |
| 5112 | |
| 5041 | 5113 DEBUG_MATCH_PRINT1 ("backtracking.\n"); |
| 428 | 5114 |
| 5115 if (!FAIL_STACK_EMPTY ()) | |
| 5116 { /* More failure points to try. */ | |
| 5117 | |
| 5118 /* If exceeds best match so far, save it. */ | |
| 5119 if (!best_regs_set || best_match_p) | |
| 5120 { | |
| 5121 best_regs_set = true; | |
| 5122 match_end = d; | |
| 5123 | |
| 5041 | 5124 DEBUG_MATCH_PRINT1 ("\nSAVING match as best so far.\n"); |
| 428 | 5125 |
| 5126 for (mcnt = 1; mcnt < num_regs; mcnt++) | |
| 5127 { | |
| 5128 best_regstart[mcnt] = regstart[mcnt]; | |
| 5129 best_regend[mcnt] = regend[mcnt]; | |
| 5130 } | |
| 5131 } | |
| 5132 goto fail; | |
| 5133 } | |
| 5134 | |
| 5135 /* If no failure points, don't restore garbage. And if | |
| 5136 last match is real best match, don't restore second | |
| 5137 best one. */ | |
| 5138 else if (best_regs_set && !best_match_p) | |
| 5139 { | |
| 5140 restore_best_regs: | |
| 5141 /* Restore best match. It may happen that `dend == | |
| 5142 end_match_1' while the restored d is in string2. | |
| 5143 For example, the pattern `x.*y.*z' against the | |
| 5144 strings `x-' and `y-z-', if the two strings are | |
| 5145 not consecutive in memory. */ | |
| 5041 | 5146 DEBUG_MATCH_PRINT1 ("Restoring best registers.\n"); |
| 428 | 5147 |
| 5148 d = match_end; | |
| 5149 dend = ((d >= string1 && d <= end1) | |
| 5150 ? end_match_1 : end_match_2); | |
| 5151 | |
| 5152 for (mcnt = 1; mcnt < num_regs; mcnt++) | |
| 5153 { | |
| 5154 regstart[mcnt] = best_regstart[mcnt]; | |
| 5155 regend[mcnt] = best_regend[mcnt]; | |
| 5156 } | |
| 5157 } | |
| 5158 } /* d != end_match_2 */ | |
| 5159 | |
| 5160 succeed_label: | |
| 5041 | 5161 DEBUG_MATCH_PRINT1 ("Accepting match.\n"); |
| 428 | 5162 |
| 5163 /* If caller wants register contents data back, do it. */ | |
| 1028 | 5164 { |
| 5165 int num_nonshy_regs = bufp->re_nsub + 1; | |
| 5166 if (regs && !bufp->no_sub) | |
| 5167 { | |
| 5168 /* Have the register data arrays been allocated? */ | |
| 5169 if (bufp->regs_allocated == REGS_UNALLOCATED) | |
| 5170 { /* No. So allocate them with malloc. We need one | |
| 5171 extra element beyond `num_regs' for the `-1' marker | |
| 5172 GNU code uses. */ | |
| 5173 regs->num_regs = MAX (RE_NREGS, num_nonshy_regs + 1); | |
| 1333 | 5174 BEGIN_REGEX_MALLOC_OK (); |
| 1028 | 5175 regs->start = TALLOC (regs->num_regs, regoff_t); |
| 5176 regs->end = TALLOC (regs->num_regs, regoff_t); | |
| 1333 | 5177 END_REGEX_MALLOC_OK (); |
| 5178 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 1028 | 5179 if (regs->start == NULL || regs->end == NULL) |
| 5180 { | |
| 5181 FREE_VARIABLES (); | |
| 5182 return -2; | |
| 5183 } | |
| 5184 bufp->regs_allocated = REGS_REALLOCATE; | |
| 5185 } | |
| 5186 else if (bufp->regs_allocated == REGS_REALLOCATE) | |
| 5187 { /* Yes. If we need more elements than were already | |
| 5188 allocated, reallocate them. If we need fewer, just | |
| 5189 leave it alone. */ | |
| 5190 if (regs->num_regs < num_nonshy_regs + 1) | |
| 5191 { | |
| 5192 regs->num_regs = num_nonshy_regs + 1; | |
| 1333 | 5193 BEGIN_REGEX_MALLOC_OK (); |
| 1028 | 5194 RETALLOC (regs->start, regs->num_regs, regoff_t); |
| 5195 RETALLOC (regs->end, regs->num_regs, regoff_t); | |
| 1333 | 5196 END_REGEX_MALLOC_OK (); |
| 5197 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 1028 | 5198 if (regs->start == NULL || regs->end == NULL) |
| 5199 { | |
| 5200 FREE_VARIABLES (); | |
| 5201 return -2; | |
| 5202 } | |
| 5203 } | |
| 5204 } | |
| 5205 else | |
| 5206 { | |
| 5207 /* The braces fend off a "empty body in an else-statement" | |
| 5208 warning under GCC when assert expands to nothing. */ | |
| 5209 assert (bufp->regs_allocated == REGS_FIXED); | |
| 5210 } | |
| 5211 | |
| 5212 /* Convert the pointer data in `regstart' and `regend' to | |
| 5213 indices. Register zero has to be set differently, | |
| 5214 since we haven't kept track of any info for it. */ | |
| 5215 if (regs->num_regs > 0) | |
| 5216 { | |
| 5217 regs->start[0] = pos; | |
| 5218 regs->end[0] = (MATCHING_IN_FIRST_STRING | |
| 5219 ? ((regoff_t) (d - string1)) | |
| 5220 : ((regoff_t) (d - string2 + size1))); | |
| 5221 } | |
| 5222 | |
| 2639 | 5223 /* Map over the NUM_NONSHY_REGS non-shy internal registers. |
| 5224 Copy each into the corresponding external register. | |
| 5225 MCNT indexes external registers. */ | |
| 1028 | 5226 for (mcnt = 1; mcnt < MIN (num_nonshy_regs, regs->num_regs); |
| 5227 mcnt++) | |
| 5228 { | |
| 5229 int internal_reg = bufp->external_to_internal_register[mcnt]; | |
| 5230 if (REG_UNSET (regstart[internal_reg]) || | |
| 5231 REG_UNSET (regend[internal_reg])) | |
| 5232 regs->start[mcnt] = regs->end[mcnt] = -1; | |
| 5233 else | |
| 5234 { | |
| 5235 regs->start[mcnt] = | |
| 5236 (regoff_t) POINTER_TO_OFFSET (regstart[internal_reg]); | |
| 5237 regs->end[mcnt] = | |
| 5238 (regoff_t) POINTER_TO_OFFSET (regend[internal_reg]); | |
| 5239 } | |
| 5240 } | |
| 5241 } /* regs && !bufp->no_sub */ | |
| 5242 | |
| 5243 /* If we have regs and the regs structure has more elements than | |
| 2639 | 5244 were in the pattern, set the extra elements starting with |
| 5245 NUM_NONSHY_REGS to -1. If we (re)allocated the registers, | |
| 5246 this is the case, because we always allocate enough to have | |
| 5247 at least one -1 at the end. | |
| 1028 | 5248 |
| 5249 We do this even when no_sub is set because some applications | |
| 5250 (XEmacs) reuse register structures which may contain stale | |
| 5251 information, and permit attempts to access those registers. | |
| 5252 | |
| 5253 It would be possible to require the caller to do this, but we'd | |
| 5254 have to change the API for this function to reflect that, and | |
| 1425 | 5255 audit all callers. Note: as of 2003-04-17 callers in XEmacs |
| 5256 do clear the registers, but it's safer to leave this code in | |
| 5257 because of reallocation. | |
| 5258 */ | |
| 1028 | 5259 if (regs && regs->num_regs > 0) |
| 5260 for (mcnt = num_nonshy_regs; mcnt < regs->num_regs; mcnt++) | |
| 5261 regs->start[mcnt] = regs->end[mcnt] = -1; | |
| 5262 } | |
| 5041 | 5263 DEBUG_MATCH_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n", |
| 428 | 5264 nfailure_points_pushed, nfailure_points_popped, |
| 5265 nfailure_points_pushed - nfailure_points_popped); | |
| 5041 | 5266 DEBUG_MATCH_PRINT2 ("%u registers pushed.\n", num_regs_pushed); |
| 428 | 5267 |
| 5268 mcnt = d - pos - (MATCHING_IN_FIRST_STRING | |
| 5269 ? string1 | |
| 5270 : string2 - size1); | |
| 5271 | |
| 5041 | 5272 DEBUG_MATCH_PRINT2 ("Returning %d from re_match_2.\n", mcnt); |
| 428 | 5273 |
| 5274 FREE_VARIABLES (); | |
| 5275 return mcnt; | |
| 5276 } | |
| 5277 | |
| 5278 /* Otherwise match next pattern command. */ | |
|
4759
aa5ed11f473b
Remove support for obsolete systems. See xemacs-patches message with ID
Jerry James <james@xemacs.org>
parents:
4750
diff
changeset
|
5279 switch ((re_opcode_t) *p++) |
| 428 | 5280 { |
| 5281 /* Ignore these. Used to ignore the n of succeed_n's which | |
| 5282 currently have n == 0. */ | |
| 5283 case no_op: | |
| 5041 | 5284 DEBUG_MATCH_PRINT1 ("EXECUTING no_op.\n"); |
| 428 | 5285 break; |
| 5286 | |
| 5287 case succeed: | |
| 5041 | 5288 DEBUG_MATCH_PRINT1 ("EXECUTING succeed.\n"); |
| 428 | 5289 goto succeed_label; |
| 5290 | |
| 826 | 5291 /* Match exactly a string of length n in the pattern. The |
| 5292 following byte in the pattern defines n, and the n bytes after | |
| 5293 that make up the string to match. (Under Mule, this will be in | |
| 5294 the default internal format.) */ | |
| 428 | 5295 case exactn: |
| 5296 mcnt = *p++; | |
| 5041 | 5297 DEBUG_MATCH_PRINT2 ("EXECUTING exactn %d.\n", mcnt); |
| 428 | 5298 |
| 5299 /* This is written out as an if-else so we don't waste time | |
| 5300 testing `translate' inside the loop. */ | |
| 446 | 5301 if (TRANSLATE_P (translate)) |
| 428 | 5302 { |
| 5303 do | |
| 5304 { | |
| 446 | 5305 #ifdef MULE |
| 5306 Bytecount pat_len; | |
| 5307 | |
| 450 | 5308 REGEX_PREFETCH (); |
| 867 | 5309 if (RE_TRANSLATE_1 (itext_ichar_fmt (d, fmt, lispobj)) |
| 5310 != itext_ichar (p)) | |
| 428 | 5311 goto fail; |
| 446 | 5312 |
| 867 | 5313 pat_len = itext_ichar_len (p); |
| 446 | 5314 p += pat_len; |
| 867 | 5315 INC_IBYTEPTR_FMT (d, fmt); |
| 446 | 5316 |
| 5317 mcnt -= pat_len; | |
| 5318 #else /* not MULE */ | |
| 450 | 5319 REGEX_PREFETCH (); |
| 826 | 5320 if ((unsigned char) RE_TRANSLATE_1 (*d++) != *p++) |
| 446 | 5321 goto fail; |
| 5322 mcnt--; | |
| 5323 #endif | |
| 428 | 5324 } |
| 446 | 5325 while (mcnt > 0); |
| 428 | 5326 } |
| 5327 else | |
| 5328 { | |
| 826 | 5329 #ifdef MULE |
| 5330 /* If buffer format is default, then we can shortcut and just | |
| 5331 compare the text directly, byte by byte. Otherwise, we | |
| 5332 need to go character by character. */ | |
| 5333 if (fmt != FORMAT_DEFAULT) | |
| 428 | 5334 { |
| 826 | 5335 do |
| 5336 { | |
| 5337 Bytecount pat_len; | |
| 5338 | |
| 5339 REGEX_PREFETCH (); | |
| 867 | 5340 if (itext_ichar_fmt (d, fmt, lispobj) != |
| 5341 itext_ichar (p)) | |
| 826 | 5342 goto fail; |
| 5343 | |
| 867 | 5344 pat_len = itext_ichar_len (p); |
| 826 | 5345 p += pat_len; |
| 867 | 5346 INC_IBYTEPTR_FMT (d, fmt); |
| 826 | 5347 |
| 5348 mcnt -= pat_len; | |
| 5349 } | |
| 5350 while (mcnt > 0); | |
| 428 | 5351 } |
| 826 | 5352 else |
| 5353 #endif | |
| 5354 { | |
| 5355 do | |
| 5356 { | |
| 5357 REGEX_PREFETCH (); | |
| 5358 if (*d++ != *p++) goto fail; | |
| 5359 mcnt--; | |
| 5360 } | |
| 5361 while (mcnt > 0); | |
| 5362 } | |
| 428 | 5363 } |
| 5364 SET_REGS_MATCHED (); | |
| 5365 break; | |
| 5366 | |
| 5367 | |
| 5368 /* Match any character except possibly a newline or a null. */ | |
| 5369 case anychar: | |
| 5041 | 5370 DEBUG_MATCH_PRINT1 ("EXECUTING anychar.\n"); |
| 428 | 5371 |
| 450 | 5372 REGEX_PREFETCH (); |
| 428 | 5373 |
| 826 | 5374 if ((!(bufp->syntax & RE_DOT_NEWLINE) && |
| 867 | 5375 RE_TRANSLATE (itext_ichar_fmt (d, fmt, lispobj)) == '\n') |
| 826 | 5376 || (bufp->syntax & RE_DOT_NOT_NULL && |
| 867 | 5377 RE_TRANSLATE (itext_ichar_fmt (d, fmt, lispobj)) == |
| 826 | 5378 '\000')) |
| 428 | 5379 goto fail; |
| 5380 | |
| 5381 SET_REGS_MATCHED (); | |
| 5041 | 5382 DEBUG_MATCH_PRINT2 (" Matched `%d'.\n", *d); |
| 867 | 5383 INC_IBYTEPTR_FMT (d, fmt); /* XEmacs change */ |
| 428 | 5384 break; |
| 5385 | |
| 5386 | |
| 5387 case charset: | |
| 5388 case charset_not: | |
| 5389 { | |
| 1414 | 5390 REGISTER Ichar c; |
| 460 | 5391 re_bool not_p = (re_opcode_t) *(p - 1) == charset_not; |
| 458 | 5392 |
| 5041 | 5393 DEBUG_MATCH_PRINT2 ("EXECUTING charset%s.\n", not_p ? "_not" : ""); |
| 428 | 5394 |
| 450 | 5395 REGEX_PREFETCH (); |
| 867 | 5396 c = itext_ichar_fmt (d, fmt, lispobj); |
| 826 | 5397 c = RE_TRANSLATE (c); /* The character to match. */ |
| 428 | 5398 |
| 647 | 5399 /* Cast to `unsigned int' instead of `unsigned char' in case the |
| 428 | 5400 bit list is a full 32 bytes long. */ |
| 1414 | 5401 if ((unsigned int)c < (unsigned int) (*p * BYTEWIDTH) |
| 428 | 5402 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
| 458 | 5403 not_p = !not_p; |
| 428 | 5404 |
| 5405 p += 1 + *p; | |
| 5406 | |
| 458 | 5407 if (!not_p) goto fail; |
| 428 | 5408 |
| 5409 SET_REGS_MATCHED (); | |
| 867 | 5410 INC_IBYTEPTR_FMT (d, fmt); /* XEmacs change */ |
| 428 | 5411 break; |
| 5412 } | |
| 5413 | |
| 5414 #ifdef MULE | |
| 5415 case charset_mule: | |
| 5416 case charset_mule_not: | |
| 5417 { | |
| 867 | 5418 REGISTER Ichar c; |
| 460 | 5419 re_bool not_p = (re_opcode_t) *(p - 1) == charset_mule_not; |
| 458 | 5420 |
| 5041 | 5421 DEBUG_MATCH_PRINT2 ("EXECUTING charset_mule%s.\n", not_p ? "_not" : ""); |
| 428 | 5422 |
| 450 | 5423 REGEX_PREFETCH (); |
| 867 | 5424 c = itext_ichar_fmt (d, fmt, lispobj); |
| 826 | 5425 c = RE_TRANSLATE (c); /* The character to match. */ |
| 428 | 5426 |
| 5427 if (EQ (Qt, unified_range_table_lookup (p, c, Qnil))) | |
| 458 | 5428 not_p = !not_p; |
| 428 | 5429 |
| 5430 p += unified_range_table_bytes_used (p); | |
| 5431 | |
| 458 | 5432 if (!not_p) goto fail; |
| 428 | 5433 |
| 5434 SET_REGS_MATCHED (); | |
| 867 | 5435 INC_IBYTEPTR_FMT (d, fmt); |
| 428 | 5436 break; |
| 5437 } | |
| 5438 #endif /* MULE */ | |
| 5439 | |
| 5440 | |
| 5441 /* The beginning of a group is represented by start_memory. | |
| 5442 The arguments are the register number in the next byte, and the | |
| 5443 number of groups inner to this one in the next. The text | |
| 5444 matched within the group is recorded (in the internal | |
| 5445 registers data structure) under the register number. */ | |
| 5446 case start_memory: | |
| 5041 | 5447 DEBUG_MATCH_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); |
| 428 | 5448 |
| 5449 /* Find out if this group can match the empty string. */ | |
| 5450 p1 = p; /* To send to group_match_null_string_p. */ | |
| 5451 | |
| 5452 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) | |
| 2639 | 5453 REG_MATCH_NULL_STRING_P (reg_info[*p]) |
| 5454 = group_match_null_string_p (&p1, pend, reg_info); | |
| 5455 | |
| 5041 | 5456 DEBUG_MATCH_PRINT2 (" group CAN%s match null string\n", |
| 2639 | 5457 REG_MATCH_NULL_STRING_P (reg_info[*p]) ? "NOT" : ""); |
| 428 | 5458 |
| 5459 /* Save the position in the string where we were the last time | |
| 5460 we were at this open-group operator in case the group is | |
| 5461 operated upon by a repetition operator, e.g., with `(a*)*b' | |
| 5462 against `ab'; then we want to ignore where we are now in | |
| 5463 the string in case this attempt to match fails. */ | |
| 5464 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) | |
| 5465 ? REG_UNSET (regstart[*p]) ? d : regstart[*p] | |
| 5466 : regstart[*p]; | |
| 5041 | 5467 DEBUG_MATCH_PRINT2 (" old_regstart: %d\n", |
| 428 | 5468 POINTER_TO_OFFSET (old_regstart[*p])); |
| 5469 | |
| 5470 regstart[*p] = d; | |
| 5041 | 5471 DEBUG_MATCH_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); |
| 428 | 5472 |
| 5473 IS_ACTIVE (reg_info[*p]) = 1; | |
| 5474 MATCHED_SOMETHING (reg_info[*p]) = 0; | |
| 5475 | |
| 5476 /* Clear this whenever we change the register activity status. */ | |
| 5477 set_regs_matched_done = 0; | |
| 5478 | |
| 5479 /* This is the new highest active register. */ | |
| 5480 highest_active_reg = *p; | |
| 5481 | |
| 5482 /* If nothing was active before, this is the new lowest active | |
| 5483 register. */ | |
| 5484 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) | |
| 5485 lowest_active_reg = *p; | |
| 5486 | |
| 5487 /* Move past the register number and inner group count. */ | |
| 5488 p += 2; | |
| 5489 just_past_start_mem = p; | |
| 5490 | |
| 5491 break; | |
| 5492 | |
| 5493 | |
| 5494 /* The stop_memory opcode represents the end of a group. Its | |
| 5495 arguments are the same as start_memory's: the register | |
| 5496 number, and the number of inner groups. */ | |
| 5497 case stop_memory: | |
| 5041 | 5498 DEBUG_MATCH_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); |
| 428 | 5499 |
| 5500 /* We need to save the string position the last time we were at | |
| 5501 this close-group operator in case the group is operated | |
| 5502 upon by a repetition operator, e.g., with `((a*)*(b*)*)*' | |
| 5503 against `aba'; then we want to ignore where we are now in | |
| 5504 the string in case this attempt to match fails. */ | |
| 5505 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) | |
| 5506 ? REG_UNSET (regend[*p]) ? d : regend[*p] | |
| 5507 : regend[*p]; | |
| 5041 | 5508 DEBUG_MATCH_PRINT2 (" old_regend: %d\n", |
| 428 | 5509 POINTER_TO_OFFSET (old_regend[*p])); |
| 5510 | |
| 5511 regend[*p] = d; | |
| 5041 | 5512 DEBUG_MATCH_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); |
| 428 | 5513 |
| 5514 /* This register isn't active anymore. */ | |
| 5515 IS_ACTIVE (reg_info[*p]) = 0; | |
| 5516 | |
| 5517 /* Clear this whenever we change the register activity status. */ | |
| 5518 set_regs_matched_done = 0; | |
| 5519 | |
| 5520 /* If this was the only register active, nothing is active | |
| 5521 anymore. */ | |
| 5522 if (lowest_active_reg == highest_active_reg) | |
| 5523 { | |
| 5524 lowest_active_reg = NO_LOWEST_ACTIVE_REG; | |
| 5525 highest_active_reg = NO_HIGHEST_ACTIVE_REG; | |
| 5526 } | |
| 5527 else | |
| 5528 { /* We must scan for the new highest active register, since | |
| 5529 it isn't necessarily one less than now: consider | |
| 5530 (a(b)c(d(e)f)g). When group 3 ends, after the f), the | |
| 5531 new highest active register is 1. */ | |
| 5532 unsigned char r = *p - 1; | |
| 5533 while (r > 0 && !IS_ACTIVE (reg_info[r])) | |
| 5534 r--; | |
| 5535 | |
| 5536 /* If we end up at register zero, that means that we saved | |
| 5537 the registers as the result of an `on_failure_jump', not | |
| 5538 a `start_memory', and we jumped to past the innermost | |
| 5539 `stop_memory'. For example, in ((.)*) we save | |
| 5540 registers 1 and 2 as a result of the *, but when we pop | |
| 5541 back to the second ), we are at the stop_memory 1. | |
| 5542 Thus, nothing is active. */ | |
| 5543 if (r == 0) | |
| 5544 { | |
| 5545 lowest_active_reg = NO_LOWEST_ACTIVE_REG; | |
| 5546 highest_active_reg = NO_HIGHEST_ACTIVE_REG; | |
| 5547 } | |
| 5548 else | |
| 5549 { | |
| 5550 highest_active_reg = r; | |
| 5551 | |
| 5552 /* 98/9/21 jhod: We've also gotta set lowest_active_reg, don't we? */ | |
| 5553 r = 1; | |
| 5554 while (r < highest_active_reg && !IS_ACTIVE(reg_info[r])) | |
| 5555 r++; | |
| 5556 lowest_active_reg = r; | |
| 5557 } | |
| 5558 } | |
| 5559 | |
| 5560 /* If just failed to match something this time around with a | |
| 5561 group that's operated on by a repetition operator, try to | |
| 5562 force exit from the ``loop'', and restore the register | |
| 5563 information for this group that we had before trying this | |
| 5564 last match. */ | |
| 5565 if ((!MATCHED_SOMETHING (reg_info[*p]) | |
| 5566 || just_past_start_mem == p - 1) | |
| 5567 && (p + 2) < pend) | |
| 5568 { | |
| 460 | 5569 re_bool is_a_jump_n = false; |
| 428 | 5570 |
| 5571 p1 = p + 2; | |
| 5572 mcnt = 0; | |
| 5573 switch ((re_opcode_t) *p1++) | |
| 5574 { | |
| 5575 case jump_n: | |
| 5576 is_a_jump_n = true; | |
| 5577 case pop_failure_jump: | |
| 5578 case maybe_pop_jump: | |
| 5579 case jump: | |
| 5580 case dummy_failure_jump: | |
| 5581 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 5582 if (is_a_jump_n) | |
| 5583 p1 += 2; | |
| 5584 break; | |
| 5585 | |
| 5586 default: | |
| 5587 /* do nothing */ ; | |
| 5588 } | |
| 5589 p1 += mcnt; | |
| 5590 | |
| 5591 /* If the next operation is a jump backwards in the pattern | |
| 5592 to an on_failure_jump right before the start_memory | |
| 5593 corresponding to this stop_memory, exit from the loop | |
| 5594 by forcing a failure after pushing on the stack the | |
| 5595 on_failure_jump's jump in the pattern, and d. */ | |
| 5596 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump | |
| 5597 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p) | |
| 5598 { | |
| 5599 /* If this group ever matched anything, then restore | |
| 5600 what its registers were before trying this last | |
| 5601 failed match, e.g., with `(a*)*b' against `ab' for | |
| 5602 regstart[1], and, e.g., with `((a*)*(b*)*)*' | |
| 5603 against `aba' for regend[3]. | |
| 5604 | |
| 5605 Also restore the registers for inner groups for, | |
| 5606 e.g., `((a*)(b*))*' against `aba' (register 3 would | |
| 5607 otherwise get trashed). */ | |
| 5608 | |
| 5609 if (EVER_MATCHED_SOMETHING (reg_info[*p])) | |
| 5610 { | |
| 647 | 5611 int r; |
| 428 | 5612 |
| 5613 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; | |
| 5614 | |
| 5615 /* Restore this and inner groups' (if any) registers. */ | |
| 5616 for (r = *p; r < *p + *(p + 1); r++) | |
| 5617 { | |
| 5618 regstart[r] = old_regstart[r]; | |
| 5619 | |
| 5620 /* xx why this test? */ | |
| 5621 if (old_regend[r] >= regstart[r]) | |
| 5622 regend[r] = old_regend[r]; | |
| 5623 } | |
| 5624 } | |
| 5625 p1++; | |
| 5626 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 5627 PUSH_FAILURE_POINT (p1 + mcnt, d, -2); | |
| 5628 | |
| 5629 goto fail; | |
| 5630 } | |
| 5631 } | |
| 5632 | |
| 5633 /* Move past the register number and the inner group count. */ | |
| 5634 p += 2; | |
| 5635 break; | |
| 5636 | |
| 5637 | |
| 5638 /* \<digit> has been turned into a `duplicate' command which is | |
| 502 | 5639 followed by the numeric value of <digit> as the register number. |
| 5640 (Already passed through external-to-internal-register mapping, | |
| 5641 so it refers to the actual group number, not the non-shy-only | |
| 5642 numbering used in the external world.) */ | |
| 428 | 5643 case duplicate: |
| 5644 { | |
| 446 | 5645 REGISTER re_char *d2, *dend2; |
| 502 | 5646 /* Get which register to match against. */ |
| 5647 int regno = *p++; | |
| 5041 | 5648 DEBUG_MATCH_PRINT2 ("EXECUTING duplicate %d.\n", regno); |
| 428 | 5649 |
| 5650 /* Can't back reference a group which we've never matched. */ | |
| 5651 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) | |
| 5652 goto fail; | |
| 5653 | |
| 5654 /* Where in input to try to start matching. */ | |
| 5655 d2 = regstart[regno]; | |
| 5656 | |
| 5657 /* Where to stop matching; if both the place to start and | |
| 5658 the place to stop matching are in the same string, then | |
| 5659 set to the place to stop, otherwise, for now have to use | |
| 5660 the end of the first string. */ | |
| 5661 | |
| 5662 dend2 = ((FIRST_STRING_P (regstart[regno]) | |
| 5663 == FIRST_STRING_P (regend[regno])) | |
| 5664 ? regend[regno] : end_match_1); | |
| 5665 for (;;) | |
| 5666 { | |
| 5667 /* If necessary, advance to next segment in register | |
| 5668 contents. */ | |
| 5669 while (d2 == dend2) | |
| 5670 { | |
| 5671 if (dend2 == end_match_2) break; | |
| 5672 if (dend2 == regend[regno]) break; | |
| 5673 | |
| 5674 /* End of string1 => advance to string2. */ | |
| 5675 d2 = string2; | |
| 5676 dend2 = regend[regno]; | |
| 5677 } | |
| 5678 /* At end of register contents => success */ | |
| 5679 if (d2 == dend2) break; | |
| 5680 | |
| 5681 /* If necessary, advance to next segment in data. */ | |
| 450 | 5682 REGEX_PREFETCH (); |
| 428 | 5683 |
| 5684 /* How many characters left in this segment to match. */ | |
| 5685 mcnt = dend - d; | |
| 5686 | |
| 5687 /* Want how many consecutive characters we can match in | |
| 5688 one shot, so, if necessary, adjust the count. */ | |
| 5689 if (mcnt > dend2 - d2) | |
| 5690 mcnt = dend2 - d2; | |
| 5691 | |
| 5692 /* Compare that many; failure if mismatch, else move | |
| 5693 past them. */ | |
| 446 | 5694 if (TRANSLATE_P (translate) |
| 826 | 5695 ? bcmp_translate (d, d2, mcnt, translate |
| 5696 #ifdef emacs | |
| 5697 , fmt, lispobj | |
| 5698 #endif | |
| 5699 ) | |
| 428 | 5700 : memcmp (d, d2, mcnt)) |
| 5701 goto fail; | |
| 5702 d += mcnt, d2 += mcnt; | |
| 5703 | |
| 5704 /* Do this because we've match some characters. */ | |
| 5705 SET_REGS_MATCHED (); | |
| 5706 } | |
| 5707 } | |
| 5708 break; | |
| 5709 | |
| 5710 | |
| 5711 /* begline matches the empty string at the beginning of the string | |
| 5712 (unless `not_bol' is set in `bufp'), and, if | |
| 5713 `newline_anchor' is set, after newlines. */ | |
| 5714 case begline: | |
| 5041 | 5715 DEBUG_MATCH_PRINT1 ("EXECUTING begline.\n"); |
| 428 | 5716 |
| 5717 if (AT_STRINGS_BEG (d)) | |
| 5718 { | |
| 5719 if (!bufp->not_bol) break; | |
| 5720 } | |
| 826 | 5721 else |
| 5722 { | |
| 5723 re_char *d2 = d; | |
| 867 | 5724 DEC_IBYTEPTR (d2); |
| 5725 if (itext_ichar_ascii_fmt (d2, fmt, lispobj) == '\n' && | |
| 826 | 5726 bufp->newline_anchor) |
| 5727 break; | |
| 5728 } | |
| 428 | 5729 /* In all other cases, we fail. */ |
| 5730 goto fail; | |
| 5731 | |
| 5732 | |
| 5733 /* endline is the dual of begline. */ | |
| 5734 case endline: | |
| 5041 | 5735 DEBUG_MATCH_PRINT1 ("EXECUTING endline.\n"); |
| 428 | 5736 |
| 5737 if (AT_STRINGS_END (d)) | |
| 5738 { | |
| 5739 if (!bufp->not_eol) break; | |
| 5740 } | |
| 5741 | |
| 5742 /* We have to ``prefetch'' the next character. */ | |
| 826 | 5743 else if ((d == end1 ? |
| 867 | 5744 itext_ichar_ascii_fmt (string2, fmt, lispobj) : |
| 5745 itext_ichar_ascii_fmt (d, fmt, lispobj)) == '\n' | |
| 428 | 5746 && bufp->newline_anchor) |
| 5747 { | |
| 5748 break; | |
| 5749 } | |
| 5750 goto fail; | |
| 5751 | |
| 5752 | |
| 5753 /* Match at the very beginning of the data. */ | |
| 5754 case begbuf: | |
| 5041 | 5755 DEBUG_MATCH_PRINT1 ("EXECUTING begbuf.\n"); |
| 428 | 5756 if (AT_STRINGS_BEG (d)) |
| 5757 break; | |
| 5758 goto fail; | |
| 5759 | |
| 5760 | |
| 5761 /* Match at the very end of the data. */ | |
| 5762 case endbuf: | |
| 5041 | 5763 DEBUG_MATCH_PRINT1 ("EXECUTING endbuf.\n"); |
| 428 | 5764 if (AT_STRINGS_END (d)) |
| 5765 break; | |
| 5766 goto fail; | |
| 5767 | |
| 5768 | |
| 5769 /* on_failure_keep_string_jump is used to optimize `.*\n'. It | |
| 5770 pushes NULL as the value for the string on the stack. Then | |
| 5771 `pop_failure_point' will keep the current value for the | |
| 5772 string, instead of restoring it. To see why, consider | |
| 5773 matching `foo\nbar' against `.*\n'. The .* matches the foo; | |
| 5774 then the . fails against the \n. But the next thing we want | |
| 5775 to do is match the \n against the \n; if we restored the | |
| 5776 string value, we would be back at the foo. | |
| 5777 | |
| 5778 Because this is used only in specific cases, we don't need to | |
| 5779 check all the things that `on_failure_jump' does, to make | |
| 5780 sure the right things get saved on the stack. Hence we don't | |
| 5781 share its code. The only reason to push anything on the | |
| 5782 stack at all is that otherwise we would have to change | |
| 5783 `anychar's code to do something besides goto fail in this | |
| 5784 case; that seems worse than this. */ | |
| 5785 case on_failure_keep_string_jump: | |
| 5041 | 5786 DEBUG_MATCH_PRINT1 ("EXECUTING on_failure_keep_string_jump"); |
| 428 | 5787 |
| 5788 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 5041 | 5789 DEBUG_MATCH_PRINT3 (" %d (to 0x%lx):\n", mcnt, (long) (p + mcnt)); |
| 428 | 5790 |
| 446 | 5791 PUSH_FAILURE_POINT (p + mcnt, (unsigned char *) 0, -2); |
| 428 | 5792 break; |
| 5793 | |
| 5794 | |
| 5795 /* Uses of on_failure_jump: | |
| 5796 | |
| 5797 Each alternative starts with an on_failure_jump that points | |
| 5798 to the beginning of the next alternative. Each alternative | |
| 5799 except the last ends with a jump that in effect jumps past | |
| 5800 the rest of the alternatives. (They really jump to the | |
| 5801 ending jump of the following alternative, because tensioning | |
| 5802 these jumps is a hassle.) | |
| 5803 | |
| 5804 Repeats start with an on_failure_jump that points past both | |
| 5805 the repetition text and either the following jump or | |
| 5806 pop_failure_jump back to this on_failure_jump. */ | |
| 5807 case on_failure_jump: | |
| 5808 on_failure: | |
| 5041 | 5809 DEBUG_MATCH_PRINT1 ("EXECUTING on_failure_jump"); |
| 428 | 5810 |
| 5811 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 5041 | 5812 DEBUG_MATCH_PRINT3 (" %d (to 0x%lx)", mcnt, (long) (p + mcnt)); |
| 428 | 5813 |
| 5814 /* If this on_failure_jump comes right before a group (i.e., | |
| 5815 the original * applied to a group), save the information | |
| 5816 for that group and all inner ones, so that if we fail back | |
| 5817 to this point, the group's information will be correct. | |
| 5818 For example, in \(a*\)*\1, we need the preceding group, | |
| 5819 and in \(\(a*\)b*\)\2, we need the inner group. */ | |
| 5820 | |
| 5821 /* We can't use `p' to check ahead because we push | |
| 5822 a failure point to `p + mcnt' after we do this. */ | |
| 5823 p1 = p; | |
| 5824 | |
| 5825 /* We need to skip no_op's before we look for the | |
| 5826 start_memory in case this on_failure_jump is happening as | |
| 5827 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 | |
| 5828 against aba. */ | |
| 5829 while (p1 < pend && (re_opcode_t) *p1 == no_op) | |
| 5830 p1++; | |
| 5831 | |
| 5832 if (p1 < pend && (re_opcode_t) *p1 == start_memory) | |
| 5833 { | |
| 5834 /* We have a new highest active register now. This will | |
| 5835 get reset at the start_memory we are about to get to, | |
| 5836 but we will have saved all the registers relevant to | |
| 5837 this repetition op, as described above. */ | |
| 5838 highest_active_reg = *(p1 + 1) + *(p1 + 2); | |
| 5839 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) | |
| 5840 lowest_active_reg = *(p1 + 1); | |
| 5841 } | |
| 5842 | |
| 5041 | 5843 DEBUG_MATCH_PRINT1 (":\n"); |
| 428 | 5844 PUSH_FAILURE_POINT (p + mcnt, d, -2); |
| 5845 break; | |
| 5846 | |
| 5847 | |
| 5848 /* A smart repeat ends with `maybe_pop_jump'. | |
| 5849 We change it to either `pop_failure_jump' or `jump'. */ | |
| 5850 case maybe_pop_jump: | |
| 5851 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 5041 | 5852 DEBUG_MATCH_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); |
| 428 | 5853 { |
| 5854 REGISTER unsigned char *p2 = p; | |
| 5855 | |
| 5856 /* Compare the beginning of the repeat with what in the | |
| 5857 pattern follows its end. If we can establish that there | |
| 5858 is nothing that they would both match, i.e., that we | |
| 5859 would have to backtrack because of (as in, e.g., `a*a') | |
| 5860 then we can change to pop_failure_jump, because we'll | |
| 5861 never have to backtrack. | |
| 5862 | |
| 5863 This is not true in the case of alternatives: in | |
| 5864 `(a|ab)*' we do need to backtrack to the `ab' alternative | |
| 5865 (e.g., if the string was `ab'). But instead of trying to | |
| 5866 detect that here, the alternative has put on a dummy | |
| 5867 failure point which is what we will end up popping. */ | |
| 5868 | |
| 5869 /* Skip over open/close-group commands. | |
| 5870 If what follows this loop is a ...+ construct, | |
| 5871 look at what begins its body, since we will have to | |
| 5872 match at least one of that. */ | |
| 5873 while (1) | |
| 5874 { | |
| 5875 if (p2 + 2 < pend | |
| 5876 && ((re_opcode_t) *p2 == stop_memory | |
| 5877 || (re_opcode_t) *p2 == start_memory)) | |
| 5878 p2 += 3; | |
| 5879 else if (p2 + 6 < pend | |
| 5880 && (re_opcode_t) *p2 == dummy_failure_jump) | |
| 5881 p2 += 6; | |
| 5882 else | |
| 5883 break; | |
| 5884 } | |
| 5885 | |
| 5886 p1 = p + mcnt; | |
| 5887 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding | |
| 5888 to the `maybe_finalize_jump' of this case. Examine what | |
| 5889 follows. */ | |
| 5890 | |
| 5891 /* If we're at the end of the pattern, we can change. */ | |
| 5892 if (p2 == pend) | |
| 5893 { | |
| 5894 /* Consider what happens when matching ":\(.*\)" | |
| 5895 against ":/". I don't really understand this code | |
| 5896 yet. */ | |
| 5897 p[-3] = (unsigned char) pop_failure_jump; | |
| 5041 | 5898 DEBUG_MATCH_PRINT1 |
| 428 | 5899 (" End of pattern: change to `pop_failure_jump'.\n"); |
| 5900 } | |
| 5901 | |
| 5902 else if ((re_opcode_t) *p2 == exactn | |
| 5903 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) | |
| 5904 { | |
| 5905 REGISTER unsigned char c | |
| 5906 = *p2 == (unsigned char) endline ? '\n' : p2[2]; | |
| 5907 | |
| 5908 if ((re_opcode_t) p1[3] == exactn && p1[5] != c) | |
| 5909 { | |
| 5910 p[-3] = (unsigned char) pop_failure_jump; | |
| 5041 | 5911 DEBUG_MATCH_PRINT3 (" %c != %c => pop_failure_jump.\n", |
| 428 | 5912 c, p1[5]); |
| 5913 } | |
| 5914 | |
| 5915 else if ((re_opcode_t) p1[3] == charset | |
| 5916 || (re_opcode_t) p1[3] == charset_not) | |
| 5917 { | |
| 458 | 5918 int not_p = (re_opcode_t) p1[3] == charset_not; |
| 428 | 5919 |
| 5920 if (c < (unsigned char) (p1[4] * BYTEWIDTH) | |
| 5921 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) | |
| 458 | 5922 not_p = !not_p; |
| 5923 | |
| 5924 /* `not_p' is equal to 1 if c would match, which means | |
| 428 | 5925 that we can't change to pop_failure_jump. */ |
| 458 | 5926 if (!not_p) |
| 428 | 5927 { |
| 5928 p[-3] = (unsigned char) pop_failure_jump; | |
| 5041 | 5929 DEBUG_MATCH_PRINT1 (" No match => pop_failure_jump.\n"); |
| 428 | 5930 } |
| 5931 } | |
| 5932 } | |
| 5933 else if ((re_opcode_t) *p2 == charset) | |
| 5934 { | |
| 5935 #ifdef DEBUG | |
| 5936 REGISTER unsigned char c | |
| 5937 = *p2 == (unsigned char) endline ? '\n' : p2[2]; | |
| 5938 #endif | |
| 5939 | |
| 5940 if ((re_opcode_t) p1[3] == exactn | |
| 5941 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5] | |
| 5942 && (p2[2 + p1[5] / BYTEWIDTH] | |
| 5943 & (1 << (p1[5] % BYTEWIDTH))))) | |
| 5944 { | |
| 5945 p[-3] = (unsigned char) pop_failure_jump; | |
| 5041 | 5946 DEBUG_MATCH_PRINT3 (" %c != %c => pop_failure_jump.\n", |
| 428 | 5947 c, p1[5]); |
| 5948 } | |
| 5949 | |
| 5950 else if ((re_opcode_t) p1[3] == charset_not) | |
| 5951 { | |
| 5952 int idx; | |
| 5953 /* We win if the charset_not inside the loop | |
| 5954 lists every character listed in the charset after. */ | |
| 5955 for (idx = 0; idx < (int) p2[1]; idx++) | |
| 5956 if (! (p2[2 + idx] == 0 | |
| 5957 || (idx < (int) p1[4] | |
| 5958 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0)))) | |
| 5959 break; | |
| 5960 | |
| 5961 if (idx == p2[1]) | |
| 5962 { | |
| 5963 p[-3] = (unsigned char) pop_failure_jump; | |
| 5041 | 5964 DEBUG_MATCH_PRINT1 (" No match => pop_failure_jump.\n"); |
| 428 | 5965 } |
| 5966 } | |
| 5967 else if ((re_opcode_t) p1[3] == charset) | |
| 5968 { | |
| 5969 int idx; | |
| 5970 /* We win if the charset inside the loop | |
| 5971 has no overlap with the one after the loop. */ | |
| 5972 for (idx = 0; | |
| 5973 idx < (int) p2[1] && idx < (int) p1[4]; | |
| 5974 idx++) | |
| 5975 if ((p2[2 + idx] & p1[5 + idx]) != 0) | |
| 5976 break; | |
| 5977 | |
| 5978 if (idx == p2[1] || idx == p1[4]) | |
| 5979 { | |
| 5980 p[-3] = (unsigned char) pop_failure_jump; | |
| 5041 | 5981 DEBUG_MATCH_PRINT1 (" No match => pop_failure_jump.\n"); |
| 428 | 5982 } |
| 5983 } | |
| 5984 } | |
| 5985 } | |
| 5986 p -= 2; /* Point at relative address again. */ | |
| 5987 if ((re_opcode_t) p[-1] != pop_failure_jump) | |
| 5988 { | |
| 5989 p[-1] = (unsigned char) jump; | |
| 5041 | 5990 DEBUG_MATCH_PRINT1 (" Match => jump.\n"); |
| 428 | 5991 goto unconditional_jump; |
| 5992 } | |
| 5993 /* Note fall through. */ | |
| 5994 | |
| 5995 | |
| 5996 /* The end of a simple repeat has a pop_failure_jump back to | |
| 5997 its matching on_failure_jump, where the latter will push a | |
| 5998 failure point. The pop_failure_jump takes off failure | |
| 5999 points put on by this pop_failure_jump's matching | |
| 6000 on_failure_jump; we got through the pattern to here from the | |
| 6001 matching on_failure_jump, so didn't fail. */ | |
| 6002 case pop_failure_jump: | |
| 6003 { | |
| 6004 /* We need to pass separate storage for the lowest and | |
| 6005 highest registers, even though we don't care about the | |
| 6006 actual values. Otherwise, we will restore only one | |
| 6007 register from the stack, since lowest will == highest in | |
| 6008 `pop_failure_point'. */ | |
| 647 | 6009 int dummy_low_reg, dummy_high_reg; |
| 428 | 6010 unsigned char *pdummy; |
| 446 | 6011 re_char *sdummy = NULL; |
| 428 | 6012 |
| 5041 | 6013 DEBUG_MATCH_PRINT1 ("EXECUTING pop_failure_jump.\n"); |
| 428 | 6014 POP_FAILURE_POINT (sdummy, pdummy, |
| 6015 dummy_low_reg, dummy_high_reg, | |
| 6016 reg_dummy, reg_dummy, reg_info_dummy); | |
| 6017 } | |
| 6018 /* Note fall through. */ | |
| 6019 | |
| 6020 | |
| 6021 /* Unconditionally jump (without popping any failure points). */ | |
| 6022 case jump: | |
| 6023 unconditional_jump: | |
| 6024 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ | |
| 5041 | 6025 DEBUG_MATCH_PRINT2 ("EXECUTING jump %d ", mcnt); |
| 428 | 6026 p += mcnt; /* Do the jump. */ |
| 5041 | 6027 DEBUG_MATCH_PRINT2 ("(to 0x%lx).\n", (long) p); |
| 428 | 6028 break; |
| 6029 | |
| 6030 | |
| 6031 /* We need this opcode so we can detect where alternatives end | |
| 6032 in `group_match_null_string_p' et al. */ | |
| 6033 case jump_past_alt: | |
| 5041 | 6034 DEBUG_MATCH_PRINT1 ("EXECUTING jump_past_alt.\n"); |
| 428 | 6035 goto unconditional_jump; |
| 6036 | |
| 6037 | |
| 6038 /* Normally, the on_failure_jump pushes a failure point, which | |
| 6039 then gets popped at pop_failure_jump. We will end up at | |
| 6040 pop_failure_jump, also, and with a pattern of, say, `a+', we | |
| 6041 are skipping over the on_failure_jump, so we have to push | |
| 6042 something meaningless for pop_failure_jump to pop. */ | |
| 6043 case dummy_failure_jump: | |
| 5041 | 6044 DEBUG_MATCH_PRINT1 ("EXECUTING dummy_failure_jump.\n"); |
| 428 | 6045 /* It doesn't matter what we push for the string here. What |
| 6046 the code at `fail' tests is the value for the pattern. */ | |
| 446 | 6047 PUSH_FAILURE_POINT ((unsigned char *) 0, (unsigned char *) 0, -2); |
| 428 | 6048 goto unconditional_jump; |
| 6049 | |
| 6050 | |
| 6051 /* At the end of an alternative, we need to push a dummy failure | |
| 6052 point in case we are followed by a `pop_failure_jump', because | |
| 6053 we don't want the failure point for the alternative to be | |
| 6054 popped. For example, matching `(a|ab)*' against `aab' | |
| 6055 requires that we match the `ab' alternative. */ | |
| 6056 case push_dummy_failure: | |
| 5041 | 6057 DEBUG_MATCH_PRINT1 ("EXECUTING push_dummy_failure.\n"); |
| 428 | 6058 /* See comments just above at `dummy_failure_jump' about the |
| 6059 two zeroes. */ | |
| 446 | 6060 PUSH_FAILURE_POINT ((unsigned char *) 0, (unsigned char *) 0, -2); |
| 428 | 6061 break; |
| 6062 | |
| 6063 /* Have to succeed matching what follows at least n times. | |
| 6064 After that, handle like `on_failure_jump'. */ | |
| 6065 case succeed_n: | |
| 6066 EXTRACT_NUMBER (mcnt, p + 2); | |
| 5041 | 6067 DEBUG_MATCH_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); |
| 428 | 6068 |
| 6069 assert (mcnt >= 0); | |
| 6070 /* Originally, this is how many times we HAVE to succeed. */ | |
| 6071 if (mcnt > 0) | |
| 6072 { | |
| 6073 mcnt--; | |
| 6074 p += 2; | |
| 6075 STORE_NUMBER_AND_INCR (p, mcnt); | |
| 5041 | 6076 DEBUG_MATCH_PRINT3 (" Setting 0x%lx to %d.\n", (long) p, mcnt); |
| 428 | 6077 } |
| 6078 else if (mcnt == 0) | |
| 6079 { | |
| 5041 | 6080 DEBUG_MATCH_PRINT2 (" Setting two bytes from 0x%lx to no_op.\n", |
| 428 | 6081 (long) (p+2)); |
| 6082 p[2] = (unsigned char) no_op; | |
| 6083 p[3] = (unsigned char) no_op; | |
| 6084 goto on_failure; | |
| 6085 } | |
| 6086 break; | |
| 6087 | |
| 6088 case jump_n: | |
| 6089 EXTRACT_NUMBER (mcnt, p + 2); | |
| 5041 | 6090 DEBUG_MATCH_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); |
| 428 | 6091 |
| 6092 /* Originally, this is how many times we CAN jump. */ | |
| 6093 if (mcnt) | |
| 6094 { | |
| 6095 mcnt--; | |
| 6096 STORE_NUMBER (p + 2, mcnt); | |
| 6097 goto unconditional_jump; | |
| 6098 } | |
| 6099 /* If don't have to jump any more, skip over the rest of command. */ | |
| 6100 else | |
| 6101 p += 4; | |
| 6102 break; | |
| 6103 | |
| 6104 case set_number_at: | |
| 6105 { | |
| 5041 | 6106 DEBUG_MATCH_PRINT1 ("EXECUTING set_number_at.\n"); |
| 428 | 6107 |
| 6108 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 6109 p1 = p + mcnt; | |
| 6110 EXTRACT_NUMBER_AND_INCR (mcnt, p); | |
| 5041 | 6111 DEBUG_MATCH_PRINT3 (" Setting 0x%lx to %d.\n", (long) p1, mcnt); |
| 428 | 6112 STORE_NUMBER (p1, mcnt); |
| 6113 break; | |
| 6114 } | |
| 6115 | |
| 6116 case wordbound: | |
| 5041 | 6117 DEBUG_MATCH_PRINT1 ("EXECUTING wordbound.\n"); |
| 428 | 6118 should_succeed = 1; |
| 6119 matchwordbound: | |
| 6120 { | |
| 6121 /* XEmacs change */ | |
| 1377 | 6122 /* Straightforward and (I hope) correct implementation. |
| 6123 Probably should be optimized by arranging to compute | |
| 1497 | 6124 charpos only once. */ |
| 1377 | 6125 /* emch1 is the character before d, syn1 is the syntax of |
| 6126 emch1, emch2 is the character at d, and syn2 is the | |
| 6127 syntax of emch2. */ | |
| 6128 Ichar emch1, emch2; | |
| 1468 | 6129 int syn1 = 0, |
| 6130 syn2 = 0; | |
| 1377 | 6131 re_char *d_before, *d_after; |
| 6132 int result, | |
| 6133 at_beg = AT_STRINGS_BEG (d), | |
| 6134 at_end = AT_STRINGS_END (d); | |
| 6135 #ifdef emacs | |
| 1497 | 6136 Charxpos charpos; |
| 1377 | 6137 #endif |
| 6138 | |
| 6139 if (at_beg && at_end) | |
| 6140 { | |
| 6141 result = 0; | |
| 6142 } | |
| 428 | 6143 else |
| 6144 { | |
| 1377 | 6145 if (!at_beg) |
| 6146 { | |
| 6147 d_before = POS_BEFORE_GAP_UNSAFE (d); | |
| 6148 DEC_IBYTEPTR_FMT (d_before, fmt); | |
| 6149 emch1 = itext_ichar_fmt (d_before, fmt, lispobj); | |
| 460 | 6150 #ifdef emacs |
| 1497 | 6151 charpos = offset_to_charxpos (lispobj, |
| 6152 PTR_TO_OFFSET (d)) - 1; | |
| 1377 | 6153 BEGIN_REGEX_MALLOC_OK (); |
| 1497 | 6154 UPDATE_SYNTAX_CACHE (scache, charpos); |
| 460 | 6155 #endif |
| 1377 | 6156 syn1 = SYNTAX_FROM_CACHE (scache, emch1); |
| 6157 END_REGEX_MALLOC_OK (); | |
| 6158 } | |
| 6159 if (!at_end) | |
| 6160 { | |
| 6161 d_after = POS_AFTER_GAP_UNSAFE (d); | |
| 6162 emch2 = itext_ichar_fmt (d_after, fmt, lispobj); | |
| 460 | 6163 #ifdef emacs |
| 1497 | 6164 charpos = offset_to_charxpos (lispobj, PTR_TO_OFFSET (d)); |
| 1377 | 6165 BEGIN_REGEX_MALLOC_OK (); |
| 1497 | 6166 UPDATE_SYNTAX_CACHE_FORWARD (scache, charpos); |
| 460 | 6167 #endif |
| 1377 | 6168 syn2 = SYNTAX_FROM_CACHE (scache, emch2); |
| 6169 END_REGEX_MALLOC_OK (); | |
| 6170 } | |
| 1333 | 6171 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); |
| 1377 | 6172 |
| 6173 if (at_beg) | |
| 6174 result = (syn2 == Sword); | |
| 6175 else if (at_end) | |
| 6176 result = (syn1 == Sword); | |
| 6177 else | |
| 6178 result = ((syn1 == Sword) != (syn2 == Sword)); | |
| 428 | 6179 } |
| 1377 | 6180 |
| 428 | 6181 if (result == should_succeed) |
| 6182 break; | |
| 6183 goto fail; | |
| 6184 } | |
| 6185 | |
| 6186 case notwordbound: | |
| 5041 | 6187 DEBUG_MATCH_PRINT1 ("EXECUTING notwordbound.\n"); |
| 428 | 6188 should_succeed = 0; |
| 6189 goto matchwordbound; | |
| 6190 | |
| 6191 case wordbeg: | |
| 5041 | 6192 DEBUG_MATCH_PRINT1 ("EXECUTING wordbeg.\n"); |
| 460 | 6193 if (AT_STRINGS_END (d)) |
| 6194 goto fail; | |
| 428 | 6195 { |
| 6196 /* XEmacs: this originally read: | |
| 6197 | |
| 6198 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1))) | |
| 6199 break; | |
| 6200 | |
| 6201 */ | |
| 460 | 6202 re_char *dtmp = POS_AFTER_GAP_UNSAFE (d); |
| 867 | 6203 Ichar emch = itext_ichar_fmt (dtmp, fmt, lispobj); |
| 1333 | 6204 int tempres; |
| 1347 | 6205 #ifdef emacs |
| 6206 Charxpos charpos = offset_to_charxpos (lispobj, PTR_TO_OFFSET (d)); | |
| 6207 #endif | |
| 1333 | 6208 BEGIN_REGEX_MALLOC_OK (); |
| 460 | 6209 #ifdef emacs |
| 826 | 6210 UPDATE_SYNTAX_CACHE (scache, charpos); |
| 460 | 6211 #endif |
| 1333 | 6212 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); |
| 6213 END_REGEX_MALLOC_OK (); | |
| 6214 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6215 if (tempres) | |
| 428 | 6216 goto fail; |
| 6217 if (AT_STRINGS_BEG (d)) | |
| 6218 break; | |
| 460 | 6219 dtmp = POS_BEFORE_GAP_UNSAFE (d); |
| 867 | 6220 DEC_IBYTEPTR_FMT (dtmp, fmt); |
| 6221 emch = itext_ichar_fmt (dtmp, fmt, lispobj); | |
| 1333 | 6222 BEGIN_REGEX_MALLOC_OK (); |
| 460 | 6223 #ifdef emacs |
| 826 | 6224 UPDATE_SYNTAX_CACHE_BACKWARD (scache, charpos - 1); |
| 460 | 6225 #endif |
| 1333 | 6226 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); |
| 6227 END_REGEX_MALLOC_OK (); | |
| 6228 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6229 if (tempres) | |
| 428 | 6230 break; |
| 6231 goto fail; | |
| 6232 } | |
| 6233 | |
| 6234 case wordend: | |
| 5041 | 6235 DEBUG_MATCH_PRINT1 ("EXECUTING wordend.\n"); |
| 460 | 6236 if (AT_STRINGS_BEG (d)) |
| 6237 goto fail; | |
| 428 | 6238 { |
| 6239 /* XEmacs: this originally read: | |
| 6240 | |
| 6241 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1) | |
| 6242 && (!WORDCHAR_P (d) || AT_STRINGS_END (d))) | |
| 6243 break; | |
| 6244 | |
| 6245 The or condition is incorrect (reversed). | |
| 6246 */ | |
| 460 | 6247 re_char *dtmp; |
| 867 | 6248 Ichar emch; |
| 1333 | 6249 int tempres; |
| 460 | 6250 #ifdef emacs |
| 826 | 6251 Charxpos charpos = offset_to_charxpos (lispobj, PTR_TO_OFFSET (d)); |
| 1347 | 6252 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 6253 UPDATE_SYNTAX_CACHE (scache, charpos); |
| 1333 | 6254 END_REGEX_MALLOC_OK (); |
| 6255 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 1347 | 6256 #endif |
| 460 | 6257 dtmp = POS_BEFORE_GAP_UNSAFE (d); |
| 867 | 6258 DEC_IBYTEPTR_FMT (dtmp, fmt); |
| 6259 emch = itext_ichar_fmt (dtmp, fmt, lispobj); | |
| 1333 | 6260 BEGIN_REGEX_MALLOC_OK (); |
| 6261 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); | |
| 6262 END_REGEX_MALLOC_OK (); | |
| 6263 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6264 if (tempres) | |
| 428 | 6265 goto fail; |
| 6266 if (AT_STRINGS_END (d)) | |
| 6267 break; | |
| 460 | 6268 dtmp = POS_AFTER_GAP_UNSAFE (d); |
| 867 | 6269 emch = itext_ichar_fmt (dtmp, fmt, lispobj); |
| 1333 | 6270 BEGIN_REGEX_MALLOC_OK (); |
| 460 | 6271 #ifdef emacs |
| 826 | 6272 UPDATE_SYNTAX_CACHE_FORWARD (scache, charpos + 1); |
| 460 | 6273 #endif |
| 1333 | 6274 tempres = (SYNTAX_FROM_CACHE (scache, emch) != Sword); |
| 6275 END_REGEX_MALLOC_OK (); | |
| 6276 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 6277 if (tempres) | |
| 428 | 6278 break; |
| 6279 goto fail; | |
| 6280 } | |
| 6281 | |
| 6282 #ifdef emacs | |
| 6283 case before_dot: | |
| 5041 | 6284 DEBUG_MATCH_PRINT1 ("EXECUTING before_dot.\n"); |
| 826 | 6285 if (!BUFFERP (lispobj) |
| 6286 || (BUF_PTR_BYTE_POS (XBUFFER (lispobj), (unsigned char *) d) | |
| 6287 >= BUF_PT (XBUFFER (lispobj)))) | |
| 428 | 6288 goto fail; |
| 6289 break; | |
| 6290 | |
| 6291 case at_dot: | |
| 5041 | 6292 DEBUG_MATCH_PRINT1 ("EXECUTING at_dot.\n"); |
| 826 | 6293 if (!BUFFERP (lispobj) |
| 6294 || (BUF_PTR_BYTE_POS (XBUFFER (lispobj), (unsigned char *) d) | |
| 6295 != BUF_PT (XBUFFER (lispobj)))) | |
| 428 | 6296 goto fail; |
| 6297 break; | |
| 6298 | |
| 6299 case after_dot: | |
| 5041 | 6300 DEBUG_MATCH_PRINT1 ("EXECUTING after_dot.\n"); |
| 826 | 6301 if (!BUFFERP (lispobj) |
| 6302 || (BUF_PTR_BYTE_POS (XBUFFER (lispobj), (unsigned char *) d) | |
| 6303 <= BUF_PT (XBUFFER (lispobj)))) | |
| 428 | 6304 goto fail; |
| 6305 break; | |
| 6306 | |
| 6307 case syntaxspec: | |
| 5041 | 6308 DEBUG_MATCH_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); |
| 428 | 6309 mcnt = *p++; |
| 6310 goto matchsyntax; | |
| 6311 | |
| 6312 case wordchar: | |
| 5041 | 6313 DEBUG_MATCH_PRINT1 ("EXECUTING Emacs wordchar.\n"); |
| 428 | 6314 mcnt = (int) Sword; |
| 6315 matchsyntax: | |
| 6316 should_succeed = 1; | |
| 6317 matchornotsyntax: | |
| 6318 { | |
| 6319 int matches; | |
| 867 | 6320 Ichar emch; |
| 428 | 6321 |
| 450 | 6322 REGEX_PREFETCH (); |
| 1333 | 6323 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 6324 UPDATE_SYNTAX_CACHE |
| 6325 (scache, offset_to_charxpos (lispobj, PTR_TO_OFFSET (d))); | |
| 1333 | 6326 END_REGEX_MALLOC_OK (); |
| 6327 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 826 | 6328 |
| 867 | 6329 emch = itext_ichar_fmt (d, fmt, lispobj); |
| 1333 | 6330 BEGIN_REGEX_MALLOC_OK (); |
| 826 | 6331 matches = (SYNTAX_FROM_CACHE (scache, emch) == |
| 6332 (enum syntaxcode) mcnt); | |
| 1333 | 6333 END_REGEX_MALLOC_OK (); |
| 6334 RE_MATCH_RELOCATE_MOVEABLE_DATA_POINTERS (); | |
| 867 | 6335 INC_IBYTEPTR_FMT (d, fmt); |
| 428 | 6336 if (matches != should_succeed) |
| 6337 goto fail; | |
| 6338 SET_REGS_MATCHED (); | |
| 6339 } | |
| 6340 break; | |
| 6341 | |
| 6342 case notsyntaxspec: | |
| 5041 | 6343 DEBUG_MATCH_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); |
| 428 | 6344 mcnt = *p++; |
| 6345 goto matchnotsyntax; | |
| 6346 | |
| 6347 case notwordchar: | |
| 5041 | 6348 DEBUG_MATCH_PRINT1 ("EXECUTING Emacs notwordchar.\n"); |
| 428 | 6349 mcnt = (int) Sword; |
| 6350 matchnotsyntax: | |
| 6351 should_succeed = 0; | |
| 6352 goto matchornotsyntax; | |
| 6353 | |
| 6354 #ifdef MULE | |
| 6355 /* 97/2/17 jhod Mule category code patch */ | |
| 6356 case categoryspec: | |
| 6357 should_succeed = 1; | |
| 6358 matchornotcategory: | |
| 6359 { | |
| 867 | 6360 Ichar emch; |
| 428 | 6361 |
| 6362 mcnt = *p++; | |
| 450 | 6363 REGEX_PREFETCH (); |
| 867 | 6364 emch = itext_ichar_fmt (d, fmt, lispobj); |
| 6365 INC_IBYTEPTR_FMT (d, fmt); | |
| 826 | 6366 if (check_category_char (emch, BUFFER_CATEGORY_TABLE (lispbuf), |
| 6367 mcnt, should_succeed)) | |
| 428 | 6368 goto fail; |
| 6369 SET_REGS_MATCHED (); | |
| 6370 } | |
| 6371 break; | |
| 6372 | |
| 6373 case notcategoryspec: | |
| 6374 should_succeed = 0; | |
| 6375 goto matchornotcategory; | |
| 6376 /* end of category patch */ | |
| 6377 #endif /* MULE */ | |
| 6378 #else /* not emacs */ | |
| 6379 case wordchar: | |
| 5041 | 6380 DEBUG_MATCH_PRINT1 ("EXECUTING non-Emacs wordchar.\n"); |
| 450 | 6381 REGEX_PREFETCH (); |
| 826 | 6382 if (!WORDCHAR_P ((int) (*d))) |
| 428 | 6383 goto fail; |
| 6384 SET_REGS_MATCHED (); | |
| 6385 d++; | |
| 6386 break; | |
| 6387 | |
| 6388 case notwordchar: | |
| 5041 | 6389 DEBUG_MATCH_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); |
| 450 | 6390 REGEX_PREFETCH (); |
| 826 | 6391 if (!WORDCHAR_P ((int) (*d))) |
| 428 | 6392 goto fail; |
| 6393 SET_REGS_MATCHED (); | |
| 6394 d++; | |
| 6395 break; | |
| 446 | 6396 #endif /* emacs */ |
| 428 | 6397 |
| 6398 default: | |
| 2500 | 6399 ABORT (); |
| 428 | 6400 } |
| 6401 continue; /* Successfully executed one pattern command; keep going. */ | |
| 6402 | |
| 6403 | |
| 6404 /* We goto here if a matching operation fails. */ | |
| 6405 fail: | |
| 6406 if (!FAIL_STACK_EMPTY ()) | |
| 6407 { /* A restart point is known. Restore to that state. */ | |
| 5041 | 6408 DEBUG_MATCH_PRINT1 ("\nFAIL:\n"); |
| 428 | 6409 POP_FAILURE_POINT (d, p, |
| 6410 lowest_active_reg, highest_active_reg, | |
| 6411 regstart, regend, reg_info); | |
| 6412 | |
| 6413 /* If this failure point is a dummy, try the next one. */ | |
| 6414 if (!p) | |
| 6415 goto fail; | |
| 6416 | |
| 6417 /* If we failed to the end of the pattern, don't examine *p. */ | |
| 6418 assert (p <= pend); | |
| 6419 if (p < pend) | |
| 6420 { | |
| 460 | 6421 re_bool is_a_jump_n = false; |
| 428 | 6422 |
| 6423 /* If failed to a backwards jump that's part of a repetition | |
| 6424 loop, need to pop this failure point and use the next one. */ | |
| 6425 switch ((re_opcode_t) *p) | |
| 6426 { | |
| 6427 case jump_n: | |
| 6428 is_a_jump_n = true; | |
| 6429 case maybe_pop_jump: | |
| 6430 case pop_failure_jump: | |
| 6431 case jump: | |
| 6432 p1 = p + 1; | |
| 6433 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6434 p1 += mcnt; | |
| 6435 | |
| 6436 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) | |
| 6437 || (!is_a_jump_n | |
| 6438 && (re_opcode_t) *p1 == on_failure_jump)) | |
| 6439 goto fail; | |
| 6440 break; | |
| 6441 default: | |
| 6442 /* do nothing */ ; | |
| 6443 } | |
| 6444 } | |
| 6445 | |
| 6446 if (d >= string1 && d <= end1) | |
| 6447 dend = end_match_1; | |
| 6448 } | |
| 6449 else | |
| 6450 break; /* Matching at this starting point really fails. */ | |
| 6451 } /* for (;;) */ | |
| 6452 | |
| 6453 if (best_regs_set) | |
| 6454 goto restore_best_regs; | |
| 6455 | |
| 6456 FREE_VARIABLES (); | |
| 6457 | |
| 6458 return -1; /* Failure to match. */ | |
| 1333 | 6459 } /* re_match_2_internal */ |
| 428 | 6460 |
| 6461 /* Subroutine definitions for re_match_2. */ | |
| 6462 | |
| 6463 | |
| 6464 /* We are passed P pointing to a register number after a start_memory. | |
| 6465 | |
| 6466 Return true if the pattern up to the corresponding stop_memory can | |
| 6467 match the empty string, and false otherwise. | |
| 6468 | |
| 6469 If we find the matching stop_memory, sets P to point to one past its number. | |
| 6470 Otherwise, sets P to an undefined byte less than or equal to END. | |
| 6471 | |
| 6472 We don't handle duplicates properly (yet). */ | |
| 6473 | |
| 460 | 6474 static re_bool |
| 428 | 6475 group_match_null_string_p (unsigned char **p, unsigned char *end, |
| 6476 register_info_type *reg_info) | |
| 6477 { | |
| 6478 int mcnt; | |
| 6479 /* Point to after the args to the start_memory. */ | |
| 6480 unsigned char *p1 = *p + 2; | |
| 6481 | |
| 6482 while (p1 < end) | |
| 6483 { | |
| 6484 /* Skip over opcodes that can match nothing, and return true or | |
| 6485 false, as appropriate, when we get to one that can't, or to the | |
| 6486 matching stop_memory. */ | |
| 6487 | |
| 6488 switch ((re_opcode_t) *p1) | |
| 6489 { | |
| 6490 /* Could be either a loop or a series of alternatives. */ | |
| 6491 case on_failure_jump: | |
| 6492 p1++; | |
| 6493 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6494 | |
| 6495 /* If the next operation is not a jump backwards in the | |
| 6496 pattern. */ | |
| 6497 | |
| 6498 if (mcnt >= 0) | |
| 6499 { | |
| 6500 /* Go through the on_failure_jumps of the alternatives, | |
| 6501 seeing if any of the alternatives cannot match nothing. | |
| 6502 The last alternative starts with only a jump, | |
| 6503 whereas the rest start with on_failure_jump and end | |
| 6504 with a jump, e.g., here is the pattern for `a|b|c': | |
| 6505 | |
| 6506 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 | |
| 6507 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 | |
| 6508 /exactn/1/c | |
| 6509 | |
| 6510 So, we have to first go through the first (n-1) | |
| 6511 alternatives and then deal with the last one separately. */ | |
| 6512 | |
| 6513 | |
| 6514 /* Deal with the first (n-1) alternatives, which start | |
| 6515 with an on_failure_jump (see above) that jumps to right | |
| 6516 past a jump_past_alt. */ | |
| 6517 | |
| 6518 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt) | |
| 6519 { | |
| 6520 /* `mcnt' holds how many bytes long the alternative | |
| 6521 is, including the ending `jump_past_alt' and | |
| 6522 its number. */ | |
| 6523 | |
| 6524 if (!alt_match_null_string_p (p1, p1 + mcnt - 3, | |
| 6525 reg_info)) | |
| 6526 return false; | |
| 6527 | |
| 6528 /* Move to right after this alternative, including the | |
| 6529 jump_past_alt. */ | |
| 6530 p1 += mcnt; | |
| 6531 | |
| 6532 /* Break if it's the beginning of an n-th alternative | |
| 6533 that doesn't begin with an on_failure_jump. */ | |
| 6534 if ((re_opcode_t) *p1 != on_failure_jump) | |
| 6535 break; | |
| 6536 | |
| 6537 /* Still have to check that it's not an n-th | |
| 6538 alternative that starts with an on_failure_jump. */ | |
| 6539 p1++; | |
| 6540 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6541 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt) | |
| 6542 { | |
| 6543 /* Get to the beginning of the n-th alternative. */ | |
| 6544 p1 -= 3; | |
| 6545 break; | |
| 6546 } | |
| 6547 } | |
| 6548 | |
| 6549 /* Deal with the last alternative: go back and get number | |
| 6550 of the `jump_past_alt' just before it. `mcnt' contains | |
| 6551 the length of the alternative. */ | |
| 6552 EXTRACT_NUMBER (mcnt, p1 - 2); | |
| 6553 | |
| 6554 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info)) | |
| 6555 return false; | |
| 6556 | |
| 6557 p1 += mcnt; /* Get past the n-th alternative. */ | |
| 6558 } /* if mcnt > 0 */ | |
| 6559 break; | |
| 6560 | |
| 6561 | |
| 6562 case stop_memory: | |
| 6563 assert (p1[1] == **p); | |
| 6564 *p = p1 + 2; | |
| 6565 return true; | |
| 6566 | |
| 6567 | |
| 6568 default: | |
| 6569 if (!common_op_match_null_string_p (&p1, end, reg_info)) | |
| 6570 return false; | |
| 6571 } | |
| 6572 } /* while p1 < end */ | |
| 6573 | |
| 6574 return false; | |
| 6575 } /* group_match_null_string_p */ | |
| 6576 | |
| 6577 | |
| 6578 /* Similar to group_match_null_string_p, but doesn't deal with alternatives: | |
| 6579 It expects P to be the first byte of a single alternative and END one | |
| 6580 byte past the last. The alternative can contain groups. */ | |
| 6581 | |
| 460 | 6582 static re_bool |
| 428 | 6583 alt_match_null_string_p (unsigned char *p, unsigned char *end, |
| 6584 register_info_type *reg_info) | |
| 6585 { | |
| 6586 int mcnt; | |
| 6587 unsigned char *p1 = p; | |
| 6588 | |
| 6589 while (p1 < end) | |
| 6590 { | |
| 6591 /* Skip over opcodes that can match nothing, and break when we get | |
| 6592 to one that can't. */ | |
| 6593 | |
| 6594 switch ((re_opcode_t) *p1) | |
| 6595 { | |
| 6596 /* It's a loop. */ | |
| 6597 case on_failure_jump: | |
| 6598 p1++; | |
| 6599 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6600 p1 += mcnt; | |
| 6601 break; | |
| 6602 | |
| 6603 default: | |
| 6604 if (!common_op_match_null_string_p (&p1, end, reg_info)) | |
| 6605 return false; | |
| 6606 } | |
| 6607 } /* while p1 < end */ | |
| 6608 | |
| 6609 return true; | |
| 6610 } /* alt_match_null_string_p */ | |
| 6611 | |
| 6612 | |
| 6613 /* Deals with the ops common to group_match_null_string_p and | |
| 6614 alt_match_null_string_p. | |
| 6615 | |
| 6616 Sets P to one after the op and its arguments, if any. */ | |
| 6617 | |
| 460 | 6618 static re_bool |
| 428 | 6619 common_op_match_null_string_p (unsigned char **p, unsigned char *end, |
| 6620 register_info_type *reg_info) | |
| 6621 { | |
| 6622 int mcnt; | |
| 460 | 6623 re_bool ret; |
| 428 | 6624 int reg_no; |
| 6625 unsigned char *p1 = *p; | |
| 6626 | |
| 6627 switch ((re_opcode_t) *p1++) | |
| 6628 { | |
| 6629 case no_op: | |
| 6630 case begline: | |
| 6631 case endline: | |
| 6632 case begbuf: | |
| 6633 case endbuf: | |
| 6634 case wordbeg: | |
| 6635 case wordend: | |
| 6636 case wordbound: | |
| 6637 case notwordbound: | |
| 6638 #ifdef emacs | |
| 6639 case before_dot: | |
| 6640 case at_dot: | |
| 6641 case after_dot: | |
| 6642 #endif | |
| 6643 break; | |
| 6644 | |
| 6645 case start_memory: | |
| 6646 reg_no = *p1; | |
| 6647 assert (reg_no > 0 && reg_no <= MAX_REGNUM); | |
| 6648 ret = group_match_null_string_p (&p1, end, reg_info); | |
| 6649 | |
| 6650 /* Have to set this here in case we're checking a group which | |
| 6651 contains a group and a back reference to it. */ | |
| 6652 | |
| 6653 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE) | |
| 6654 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret; | |
| 6655 | |
| 6656 if (!ret) | |
| 6657 return false; | |
| 6658 break; | |
| 6659 | |
| 6660 /* If this is an optimized succeed_n for zero times, make the jump. */ | |
| 6661 case jump: | |
| 6662 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6663 if (mcnt >= 0) | |
| 6664 p1 += mcnt; | |
| 6665 else | |
| 6666 return false; | |
| 6667 break; | |
| 6668 | |
| 6669 case succeed_n: | |
| 6670 /* Get to the number of times to succeed. */ | |
| 6671 p1 += 2; | |
| 6672 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6673 | |
| 6674 if (mcnt == 0) | |
| 6675 { | |
| 6676 p1 -= 4; | |
| 6677 EXTRACT_NUMBER_AND_INCR (mcnt, p1); | |
| 6678 p1 += mcnt; | |
| 6679 } | |
| 6680 else | |
| 6681 return false; | |
| 6682 break; | |
| 6683 | |
| 6684 case duplicate: | |
| 6685 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) | |
| 6686 return false; | |
| 6687 break; | |
| 6688 | |
| 6689 case set_number_at: | |
| 6690 p1 += 4; | |
| 6691 | |
| 6692 default: | |
| 6693 /* All other opcodes mean we cannot match the empty string. */ | |
| 6694 return false; | |
| 6695 } | |
| 6696 | |
| 6697 *p = p1; | |
| 6698 return true; | |
| 6699 } /* common_op_match_null_string_p */ | |
| 6700 | |
| 6701 | |
| 6702 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN | |
| 6703 bytes; nonzero otherwise. */ | |
| 6704 | |
| 6705 static int | |
| 446 | 6706 bcmp_translate (re_char *s1, re_char *s2, |
| 826 | 6707 REGISTER int len, RE_TRANSLATE_TYPE translate |
| 6708 #ifdef emacs | |
| 2333 | 6709 , Internal_Format USED_IF_MULE (fmt), |
| 6710 Lisp_Object USED_IF_MULE (lispobj) | |
| 826 | 6711 #endif |
| 6712 ) | |
| 428 | 6713 { |
| 826 | 6714 REGISTER re_char *p1 = s1, *p2 = s2; |
| 446 | 6715 #ifdef MULE |
| 826 | 6716 re_char *p1_end = s1 + len; |
| 6717 re_char *p2_end = s2 + len; | |
| 446 | 6718 |
| 6719 while (p1 != p1_end && p2 != p2_end) | |
| 6720 { | |
| 867 | 6721 Ichar p1_ch, p2_ch; |
| 6722 | |
| 6723 p1_ch = itext_ichar_fmt (p1, fmt, lispobj); | |
| 6724 p2_ch = itext_ichar_fmt (p2, fmt, lispobj); | |
| 826 | 6725 |
| 6726 if (RE_TRANSLATE_1 (p1_ch) | |
| 6727 != RE_TRANSLATE_1 (p2_ch)) | |
| 446 | 6728 return 1; |
| 867 | 6729 INC_IBYTEPTR_FMT (p1, fmt); |
| 6730 INC_IBYTEPTR_FMT (p2, fmt); | |
| 446 | 6731 } |
| 6732 #else /* not MULE */ | |
| 428 | 6733 while (len) |
| 6734 { | |
| 826 | 6735 if (RE_TRANSLATE_1 (*p1++) != RE_TRANSLATE_1 (*p2++)) return 1; |
| 428 | 6736 len--; |
| 6737 } | |
| 446 | 6738 #endif /* MULE */ |
| 428 | 6739 return 0; |
| 6740 } | |
| 6741 | |
| 6742 /* Entry points for GNU code. */ | |
| 6743 | |
| 6744 /* re_compile_pattern is the GNU regular expression compiler: it | |
| 6745 compiles PATTERN (of length SIZE) and puts the result in BUFP. | |
| 6746 Returns 0 if the pattern was valid, otherwise an error string. | |
| 6747 | |
| 6748 Assumes the `allocated' (and perhaps `buffer') and `translate' fields | |
| 6749 are set in BUFP on entry. | |
| 6750 | |
| 6751 We call regex_compile to do the actual compilation. */ | |
| 6752 | |
| 442 | 6753 const char * |
| 6754 re_compile_pattern (const char *pattern, int length, | |
| 428 | 6755 struct re_pattern_buffer *bufp) |
| 6756 { | |
| 6757 reg_errcode_t ret; | |
| 6758 | |
| 6759 /* GNU code is written to assume at least RE_NREGS registers will be set | |
| 6760 (and at least one extra will be -1). */ | |
| 6761 bufp->regs_allocated = REGS_UNALLOCATED; | |
| 6762 | |
| 6763 /* And GNU code determines whether or not to get register information | |
| 6764 by passing null for the REGS argument to re_match, etc., not by | |
| 6765 setting no_sub. */ | |
| 6766 bufp->no_sub = 0; | |
| 6767 | |
| 6768 /* Match anchors at newline. */ | |
| 6769 bufp->newline_anchor = 1; | |
| 6770 | |
| 826 | 6771 ret = regex_compile ((unsigned char *) pattern, length, re_syntax_options, |
| 6772 bufp); | |
| 428 | 6773 |
| 6774 if (!ret) | |
| 6775 return NULL; | |
| 6776 return gettext (re_error_msgid[(int) ret]); | |
| 6777 } | |
| 6778 | |
| 6779 /* Entry points compatible with 4.2 BSD regex library. We don't define | |
| 6780 them unless specifically requested. */ | |
| 6781 | |
| 6782 #ifdef _REGEX_RE_COMP | |
| 6783 | |
| 6784 /* BSD has one and only one pattern buffer. */ | |
| 6785 static struct re_pattern_buffer re_comp_buf; | |
| 6786 | |
| 6787 char * | |
| 442 | 6788 re_comp (const char *s) |
| 428 | 6789 { |
| 6790 reg_errcode_t ret; | |
| 6791 | |
| 6792 if (!s) | |
| 6793 { | |
| 6794 if (!re_comp_buf.buffer) | |
| 6795 return gettext ("No previous regular expression"); | |
| 6796 return 0; | |
| 6797 } | |
| 6798 | |
| 6799 if (!re_comp_buf.buffer) | |
| 6800 { | |
| 1333 | 6801 re_comp_buf.buffer = (unsigned char *) xmalloc (200); |
| 428 | 6802 if (re_comp_buf.buffer == NULL) |
| 6803 return gettext (re_error_msgid[(int) REG_ESPACE]); | |
| 6804 re_comp_buf.allocated = 200; | |
| 6805 | |
| 1333 | 6806 re_comp_buf.fastmap = (char *) xmalloc (1 << BYTEWIDTH); |
| 428 | 6807 if (re_comp_buf.fastmap == NULL) |
| 6808 return gettext (re_error_msgid[(int) REG_ESPACE]); | |
| 6809 } | |
| 6810 | |
| 6811 /* Since `re_exec' always passes NULL for the `regs' argument, we | |
| 6812 don't need to initialize the pattern buffer fields which affect it. */ | |
| 6813 | |
| 6814 /* Match anchors at newlines. */ | |
| 6815 re_comp_buf.newline_anchor = 1; | |
| 6816 | |
| 826 | 6817 ret = regex_compile ((unsigned char *)s, strlen (s), re_syntax_options, |
| 6818 &re_comp_buf); | |
| 428 | 6819 |
| 6820 if (!ret) | |
| 6821 return NULL; | |
| 6822 | |
| 442 | 6823 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */ |
| 428 | 6824 return (char *) gettext (re_error_msgid[(int) ret]); |
| 6825 } | |
| 6826 | |
| 6827 | |
| 6828 int | |
| 442 | 6829 re_exec (const char *s) |
| 428 | 6830 { |
| 442 | 6831 const int len = strlen (s); |
| 428 | 6832 return |
| 6833 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0); | |
| 6834 } | |
| 6835 #endif /* _REGEX_RE_COMP */ | |
| 6836 | |
| 6837 /* POSIX.2 functions. Don't define these for Emacs. */ | |
| 6838 | |
| 6839 #ifndef emacs | |
| 6840 | |
| 6841 /* regcomp takes a regular expression as a string and compiles it. | |
| 6842 | |
| 6843 PREG is a regex_t *. We do not expect any fields to be initialized, | |
| 6844 since POSIX says we shouldn't. Thus, we set | |
| 6845 | |
| 6846 `buffer' to the compiled pattern; | |
| 6847 `used' to the length of the compiled pattern; | |
| 6848 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the | |
| 6849 REG_EXTENDED bit in CFLAGS is set; otherwise, to | |
| 6850 RE_SYNTAX_POSIX_BASIC; | |
| 6851 `newline_anchor' to REG_NEWLINE being set in CFLAGS; | |
| 6852 `fastmap' and `fastmap_accurate' to zero; | |
| 6853 `re_nsub' to the number of subexpressions in PATTERN. | |
| 502 | 6854 (non-shy of course. POSIX probably doesn't know about |
| 6855 shy ones, and in any case they should be invisible.) | |
| 428 | 6856 |
| 6857 PATTERN is the address of the pattern string. | |
| 6858 | |
| 6859 CFLAGS is a series of bits which affect compilation. | |
| 6860 | |
| 6861 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we | |
| 6862 use POSIX basic syntax. | |
| 6863 | |
| 6864 If REG_NEWLINE is set, then . and [^...] don't match newline. | |
| 6865 Also, regexec will try a match beginning after every newline. | |
| 6866 | |
| 6867 If REG_ICASE is set, then we considers upper- and lowercase | |
| 6868 versions of letters to be equivalent when matching. | |
| 6869 | |
| 6870 If REG_NOSUB is set, then when PREG is passed to regexec, that | |
| 6871 routine will report only success or failure, and nothing about the | |
| 6872 registers. | |
| 6873 | |
| 6874 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for | |
| 6875 the return codes and their meanings.) */ | |
| 6876 | |
| 6877 int | |
| 442 | 6878 regcomp (regex_t *preg, const char *pattern, int cflags) |
| 428 | 6879 { |
| 6880 reg_errcode_t ret; | |
| 647 | 6881 unsigned int syntax |
| 428 | 6882 = (cflags & REG_EXTENDED) ? |
| 6883 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC; | |
| 6884 | |
| 6885 /* regex_compile will allocate the space for the compiled pattern. */ | |
| 6886 preg->buffer = 0; | |
| 6887 preg->allocated = 0; | |
| 6888 preg->used = 0; | |
| 6889 | |
| 6890 /* Don't bother to use a fastmap when searching. This simplifies the | |
| 6891 REG_NEWLINE case: if we used a fastmap, we'd have to put all the | |
| 6892 characters after newlines into the fastmap. This way, we just try | |
| 6893 every character. */ | |
| 6894 preg->fastmap = 0; | |
| 6895 | |
| 6896 if (cflags & REG_ICASE) | |
| 6897 { | |
| 647 | 6898 int i; |
| 428 | 6899 |
| 1333 | 6900 preg->translate = (char *) xmalloc (CHAR_SET_SIZE); |
| 428 | 6901 if (preg->translate == NULL) |
| 6902 return (int) REG_ESPACE; | |
| 6903 | |
| 6904 /* Map uppercase characters to corresponding lowercase ones. */ | |
| 6905 for (i = 0; i < CHAR_SET_SIZE; i++) | |
| 6906 preg->translate[i] = ISUPPER (i) ? tolower (i) : i; | |
| 6907 } | |
| 6908 else | |
| 6909 preg->translate = NULL; | |
| 6910 | |
| 6911 /* If REG_NEWLINE is set, newlines are treated differently. */ | |
| 6912 if (cflags & REG_NEWLINE) | |
| 6913 { /* REG_NEWLINE implies neither . nor [^...] match newline. */ | |
| 6914 syntax &= ~RE_DOT_NEWLINE; | |
| 6915 syntax |= RE_HAT_LISTS_NOT_NEWLINE; | |
| 6916 /* It also changes the matching behavior. */ | |
| 6917 preg->newline_anchor = 1; | |
| 6918 } | |
| 6919 else | |
| 6920 preg->newline_anchor = 0; | |
| 6921 | |
| 6922 preg->no_sub = !!(cflags & REG_NOSUB); | |
| 6923 | |
| 6924 /* POSIX says a null character in the pattern terminates it, so we | |
| 6925 can use strlen here in compiling the pattern. */ | |
| 446 | 6926 ret = regex_compile ((unsigned char *) pattern, strlen (pattern), syntax, preg); |
| 428 | 6927 |
| 6928 /* POSIX doesn't distinguish between an unmatched open-group and an | |
| 6929 unmatched close-group: both are REG_EPAREN. */ | |
| 6930 if (ret == REG_ERPAREN) ret = REG_EPAREN; | |
| 6931 | |
| 6932 return (int) ret; | |
| 6933 } | |
| 6934 | |
| 6935 | |
| 6936 /* regexec searches for a given pattern, specified by PREG, in the | |
| 6937 string STRING. | |
| 6938 | |
| 6939 If NMATCH is zero or REG_NOSUB was set in the cflags argument to | |
| 6940 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at | |
| 6941 least NMATCH elements, and we set them to the offsets of the | |
| 6942 corresponding matched substrings. | |
| 6943 | |
| 6944 EFLAGS specifies `execution flags' which affect matching: if | |
| 6945 REG_NOTBOL is set, then ^ does not match at the beginning of the | |
| 6946 string; if REG_NOTEOL is set, then $ does not match at the end. | |
| 6947 | |
| 6948 We return 0 if we find a match and REG_NOMATCH if not. */ | |
| 6949 | |
| 6950 int | |
| 442 | 6951 regexec (const regex_t *preg, const char *string, size_t nmatch, |
| 428 | 6952 regmatch_t pmatch[], int eflags) |
| 6953 { | |
| 6954 int ret; | |
| 6955 struct re_registers regs; | |
| 6956 regex_t private_preg; | |
| 6957 int len = strlen (string); | |
| 460 | 6958 re_bool want_reg_info = !preg->no_sub && nmatch > 0; |
| 428 | 6959 |
| 6960 private_preg = *preg; | |
| 6961 | |
| 6962 private_preg.not_bol = !!(eflags & REG_NOTBOL); | |
| 6963 private_preg.not_eol = !!(eflags & REG_NOTEOL); | |
| 6964 | |
| 6965 /* The user has told us exactly how many registers to return | |
| 6966 information about, via `nmatch'. We have to pass that on to the | |
| 6967 matching routines. */ | |
| 6968 private_preg.regs_allocated = REGS_FIXED; | |
| 6969 | |
| 6970 if (want_reg_info) | |
| 6971 { | |
| 647 | 6972 regs.num_regs = (int) nmatch; |
| 6973 regs.start = TALLOC ((int) nmatch, regoff_t); | |
| 6974 regs.end = TALLOC ((int) nmatch, regoff_t); | |
| 428 | 6975 if (regs.start == NULL || regs.end == NULL) |
| 6976 return (int) REG_NOMATCH; | |
| 6977 } | |
| 6978 | |
| 6979 /* Perform the searching operation. */ | |
| 6980 ret = re_search (&private_preg, string, len, | |
| 6981 /* start: */ 0, /* range: */ len, | |
| 6982 want_reg_info ? ®s : (struct re_registers *) 0); | |
| 6983 | |
| 6984 /* Copy the register information to the POSIX structure. */ | |
| 6985 if (want_reg_info) | |
| 6986 { | |
| 6987 if (ret >= 0) | |
| 6988 { | |
| 647 | 6989 int r; |
| 6990 | |
| 6991 for (r = 0; r < (int) nmatch; r++) | |
| 428 | 6992 { |
| 6993 pmatch[r].rm_so = regs.start[r]; | |
| 6994 pmatch[r].rm_eo = regs.end[r]; | |
| 6995 } | |
| 6996 } | |
| 6997 | |
| 6998 /* If we needed the temporary register info, free the space now. */ | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
6999 xfree (regs.start); |
|
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
7000 xfree (regs.end); |
| 428 | 7001 } |
| 7002 | |
| 7003 /* We want zero return to mean success, unlike `re_search'. */ | |
| 7004 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH; | |
| 7005 } | |
| 7006 | |
| 7007 | |
| 7008 /* Returns a message corresponding to an error code, ERRCODE, returned | |
| 7009 from either regcomp or regexec. We don't use PREG here. */ | |
| 7010 | |
| 7011 size_t | |
| 2286 | 7012 regerror (int errcode, const regex_t *UNUSED (preg), char *errbuf, |
| 647 | 7013 size_t errbuf_size) |
| 428 | 7014 { |
| 442 | 7015 const char *msg; |
| 665 | 7016 Bytecount msg_size; |
| 428 | 7017 |
| 7018 if (errcode < 0 | |
| 647 | 7019 || errcode >= (int) (sizeof (re_error_msgid) / |
| 7020 sizeof (re_error_msgid[0]))) | |
| 428 | 7021 /* Only error codes returned by the rest of the code should be passed |
| 7022 to this routine. If we are given anything else, or if other regex | |
| 7023 code generates an invalid error code, then the program has a bug. | |
| 7024 Dump core so we can fix it. */ | |
| 2500 | 7025 ABORT (); |
| 428 | 7026 |
| 7027 msg = gettext (re_error_msgid[errcode]); | |
| 7028 | |
| 7029 msg_size = strlen (msg) + 1; /* Includes the null. */ | |
| 7030 | |
| 7031 if (errbuf_size != 0) | |
| 7032 { | |
| 665 | 7033 if (msg_size > (Bytecount) errbuf_size) |
| 428 | 7034 { |
| 7035 strncpy (errbuf, msg, errbuf_size - 1); | |
| 7036 errbuf[errbuf_size - 1] = 0; | |
| 7037 } | |
| 7038 else | |
| 7039 strcpy (errbuf, msg); | |
| 7040 } | |
| 7041 | |
| 647 | 7042 return (size_t) msg_size; |
| 428 | 7043 } |
| 7044 | |
| 7045 | |
| 7046 /* Free dynamically allocated space used by PREG. */ | |
| 7047 | |
| 7048 void | |
| 7049 regfree (regex_t *preg) | |
| 7050 { | |
| 7051 if (preg->buffer != NULL) | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
7052 xfree (preg->buffer); |
| 428 | 7053 preg->buffer = NULL; |
| 7054 | |
| 7055 preg->allocated = 0; | |
| 7056 preg->used = 0; | |
| 7057 | |
| 7058 if (preg->fastmap != NULL) | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
7059 xfree (preg->fastmap); |
| 428 | 7060 preg->fastmap = NULL; |
| 7061 preg->fastmap_accurate = 0; | |
| 7062 | |
| 7063 if (preg->translate != NULL) | |
|
4976
16112448d484
Rename xfree(FOO, TYPE) -> xfree(FOO)
Ben Wing <ben@xemacs.org>
parents:
4832
diff
changeset
|
7064 xfree (preg->translate); |
| 428 | 7065 preg->translate = NULL; |
| 7066 } | |
| 7067 | |
| 7068 #endif /* not emacs */ | |
| 7069 |