1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993-1999, 2000, 2001 Free Software Foundation, Inc.
6 This file is part of the GNU C Library.
8 The GNU C Library is free software; you can redistribute it and/or
9 modify it under the terms of the GNU Lesser General Public
10 License as published by the Free Software Foundation; either
11 version 2.1 of the License, or (at your option) any later version.
13 The GNU C Library is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 Lesser General Public License for more details.
18 You should have received a copy of the GNU Lesser General Public
19 License along with the GNU C Library; if not, write to the Free
20 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
23 /* AIX requires this to be the first thing in the file. */
24 #if defined _AIX && !defined REGEX_MALLOC
36 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
37 # define PARAMS(args) args
39 # define PARAMS(args) ()
41 #endif /* Not PARAMS. */
43 #ifndef INSIDE_RECURSION
45 # if defined STDC_HEADERS && !defined emacs
48 /* We need this for `regex.h', and perhaps for the Emacs include files. */
49 # include <sys/types.h>
52 # define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
54 /* For platform which support the ISO C amendement 1 functionality we
55 support user defined character classes. */
56 # if defined _LIBC || WIDE_CHAR_SUPPORT
57 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
63 /* We have to keep the namespace clean. */
64 # define regfree(preg) __regfree (preg)
65 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
66 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
67 # define regerror(errcode, preg, errbuf, errbuf_size) \
68 __regerror(errcode, preg, errbuf, errbuf_size)
69 # define re_set_registers(bu, re, nu, st, en) \
70 __re_set_registers (bu, re, nu, st, en)
71 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
72 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
73 # define re_match(bufp, string, size, pos, regs) \
74 __re_match (bufp, string, size, pos, regs)
75 # define re_search(bufp, string, size, startpos, range, regs) \
76 __re_search (bufp, string, size, startpos, range, regs)
77 # define re_compile_pattern(pattern, length, bufp) \
78 __re_compile_pattern (pattern, length, bufp)
79 # define re_set_syntax(syntax) __re_set_syntax (syntax)
80 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
81 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
82 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
84 # define btowc __btowc
86 /* We are also using some library internals. */
87 # include <locale/localeinfo.h>
88 # include <locale/elem-hash.h>
89 # include <langinfo.h>
90 # include <locale/coll-lookup.h>
93 /* This is for other GNU distributions with internationalized messages. */
94 # if HAVE_LIBINTL_H || defined _LIBC
98 # define gettext(msgid) __dcgettext ("libc", msgid, LC_MESSAGES)
101 # define gettext(msgid) (msgid)
104 # ifndef gettext_noop
105 /* This define is so xgettext can find the internationalizable
107 # define gettext_noop(String) String
110 /* The `emacs' switch turns on certain matching commands
111 that make sense only in Emacs. */
118 # else /* not emacs */
120 /* If we are not linking with Emacs proper,
121 we can't use the relocating allocator
122 even if config.h says that we can. */
125 # if defined STDC_HEADERS || defined _LIBC
132 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
133 If nothing else has been done, use the method below. */
134 # ifdef INHIBIT_STRING_HEADER
135 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
136 # if !defined bzero && !defined bcopy
137 # undef INHIBIT_STRING_HEADER
142 /* This is the normal way of making sure we have a bcopy and a bzero.
143 This is used in most programs--a few other programs avoid this
144 by defining INHIBIT_STRING_HEADER. */
145 # ifndef INHIBIT_STRING_HEADER
146 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
150 # define bzero(s, n) (memset (s, '\0', n), (s))
152 # define bzero(s, n) __bzero (s, n)
156 # include <strings.h>
158 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
161 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
166 /* Define the syntax stuff for \<, \>, etc. */
168 /* This must be nonzero for the wordchar and notwordchar pattern
169 commands in re_match_2. */
174 # ifdef SWITCH_ENUM_BUG
175 # define SWITCH_ENUM_CAST(x) ((int)(x))
177 # define SWITCH_ENUM_CAST(x) (x)
180 # endif /* not emacs */
182 # if defined _LIBC || HAVE_LIMITS_H
187 # define MB_LEN_MAX 1
190 /* Get the interface, including the syntax bits. */
193 /* isalpha etc. are used for the character classes. */
196 /* Jim Meyering writes:
198 "... Some ctype macros are valid only for character codes that
199 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
200 using /bin/cc or gcc but without giving an ansi option). So, all
201 ctype uses should be through macros like ISPRINT... If
202 STDC_HEADERS is defined, then autoconf has verified that the ctype
203 macros don't need to be guarded with references to isascii. ...
204 Defining isascii to 1 should let any compiler worth its salt
205 eliminate the && through constant folding."
206 Solaris defines some of these symbols so we must undefine them first. */
209 # if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
210 # define ISASCII(c) 1
212 # define ISASCII(c) isascii(c)
216 # define ISBLANK(c) (ISASCII (c) && isblank (c))
218 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
221 # define ISGRAPH(c) (ISASCII (c) && isgraph (c))
223 # define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
227 # define ISPRINT(c) (ISASCII (c) && isprint (c))
228 # define ISDIGIT(c) (ISASCII (c) && isdigit (c))
229 # define ISALNUM(c) (ISASCII (c) && isalnum (c))
230 # define ISALPHA(c) (ISASCII (c) && isalpha (c))
231 # define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
232 # define ISLOWER(c) (ISASCII (c) && islower (c))
233 # define ISPUNCT(c) (ISASCII (c) && ispunct (c))
234 # define ISSPACE(c) (ISASCII (c) && isspace (c))
235 # define ISUPPER(c) (ISASCII (c) && isupper (c))
236 # define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
239 # define TOLOWER(c) _tolower(c)
241 # define TOLOWER(c) tolower(c)
245 # define NULL (void *)0
248 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
249 since ours (we hope) works properly with all combinations of
250 machines, compilers, `char' and `unsigned char' argument types.
251 (Per Bothner suggested the basic approach.) */
252 # undef SIGN_EXTEND_CHAR
254 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
255 # else /* not __STDC__ */
256 /* As in Harbison and Steele. */
257 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
261 /* How many characters in the character set. */
262 # define CHAR_SET_SIZE 256
266 extern char *re_syntax_table;
268 # else /* not SYNTAX_TABLE */
270 static char re_syntax_table[CHAR_SET_SIZE];
272 static void init_syntax_once PARAMS ((void));
282 bzero (re_syntax_table, sizeof re_syntax_table);
284 for (c = 0; c < CHAR_SET_SIZE; ++c)
286 re_syntax_table[c] = Sword;
288 re_syntax_table['_'] = Sword;
293 # endif /* not SYNTAX_TABLE */
295 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
299 /* Integer type for pointers. */
301 typedef unsigned long int uintptr_t;
304 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
305 use `alloca' instead of `malloc'. This is because using malloc in
306 re_search* or re_match* could cause memory leaks when C-g is used in
307 Emacs; also, malloc is slower and causes storage fragmentation. On
308 the other hand, malloc is more portable, and easier to debug.
310 Because we sometimes use alloca, some routines have to be macros,
311 not functions -- `alloca'-allocated space disappears at the end of the
312 function it is called in. */
316 # define REGEX_ALLOCATE malloc
317 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
318 # define REGEX_FREE free
320 # else /* not REGEX_MALLOC */
322 /* Emacs already defines alloca, sometimes. */
325 /* Make alloca work the best possible way. */
327 # define alloca __builtin_alloca
328 # else /* not __GNUC__ */
331 # endif /* HAVE_ALLOCA_H */
332 # endif /* not __GNUC__ */
334 # endif /* not alloca */
336 # define REGEX_ALLOCATE alloca
338 /* Assumes a `char *destination' variable. */
339 # define REGEX_REALLOCATE(source, osize, nsize) \
340 (destination = (char *) alloca (nsize), \
341 memcpy (destination, source, osize))
343 /* No need to do anything to free, after alloca. */
344 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
346 # endif /* not REGEX_MALLOC */
348 /* Define how to allocate the failure stack. */
350 # if defined REL_ALLOC && defined REGEX_MALLOC
352 # define REGEX_ALLOCATE_STACK(size) \
353 r_alloc (&failure_stack_ptr, (size))
354 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
355 r_re_alloc (&failure_stack_ptr, (nsize))
356 # define REGEX_FREE_STACK(ptr) \
357 r_alloc_free (&failure_stack_ptr)
359 # else /* not using relocating allocator */
363 # define REGEX_ALLOCATE_STACK malloc
364 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
365 # define REGEX_FREE_STACK free
367 # else /* not REGEX_MALLOC */
369 # define REGEX_ALLOCATE_STACK alloca
371 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
372 REGEX_REALLOCATE (source, osize, nsize)
373 /* No need to explicitly free anything. */
374 # define REGEX_FREE_STACK(arg)
376 # endif /* not REGEX_MALLOC */
377 # endif /* not using relocating allocator */
380 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
381 `string1' or just past its end. This works if PTR is NULL, which is
383 # define FIRST_STRING_P(ptr) \
384 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
386 /* (Re)Allocate N items of type T using malloc, or fail. */
387 # define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
388 # define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
389 # define RETALLOC_IF(addr, n, t) \
390 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
391 # define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
393 # define BYTEWIDTH 8 /* In bits. */
395 # define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
399 # define MAX(a, b) ((a) > (b) ? (a) : (b))
400 # define MIN(a, b) ((a) < (b) ? (a) : (b))
402 typedef char boolean;
406 static reg_errcode_t byte_regex_compile _RE_ARGS ((const char *pattern, size_t size,
408 struct re_pattern_buffer *bufp));
410 static int byte_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
411 const char *string1, int size1,
412 const char *string2, int size2,
414 struct re_registers *regs,
416 static int byte_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
417 const char *string1, int size1,
418 const char *string2, int size2,
419 int startpos, int range,
420 struct re_registers *regs, int stop));
421 static int byte_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
424 static reg_errcode_t wcs_regex_compile _RE_ARGS ((const char *pattern, size_t size,
426 struct re_pattern_buffer *bufp));
429 static int wcs_re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
430 const char *cstring1, int csize1,
431 const char *cstring2, int csize2,
433 struct re_registers *regs,
435 wchar_t *string1, int size1,
436 wchar_t *string2, int size2,
437 int *mbs_offset1, int *mbs_offset2));
438 static int wcs_re_search_2 PARAMS ((struct re_pattern_buffer *bufp,
439 const char *string1, int size1,
440 const char *string2, int size2,
441 int startpos, int range,
442 struct re_registers *regs, int stop));
443 static int wcs_re_compile_fastmap PARAMS ((struct re_pattern_buffer *bufp));
446 /* These are the command codes that appear in compiled regular
447 expressions. Some opcodes are followed by argument bytes. A
448 command code can specify any interpretation whatsoever for its
449 arguments. Zero bytes may appear in the compiled regular expression. */
455 /* Succeed right away--no more backtracking. */
458 /* Followed by one byte giving n, then by n literal bytes. */
462 /* Same as exactn, but contains binary data. */
466 /* Matches any (more or less) character. */
469 /* Matches any one char belonging to specified set. First
470 following byte is number of bitmap bytes. Then come bytes
471 for a bitmap saying which chars are in. Bits in each byte
472 are ordered low-bit-first. A character is in the set if its
473 bit is 1. A character too large to have a bit in the map is
474 automatically not in the set. */
475 /* ifdef MBS_SUPPORT, following element is length of character
476 classes, length of collating symbols, length of equivalence
477 classes, length of character ranges, and length of characters.
478 Next, character class element, collating symbols elements,
479 equivalence class elements, range elements, and character
481 See regex_compile function. */
484 /* Same parameters as charset, but match any character that is
485 not one of those specified. */
488 /* Start remembering the text that is matched, for storing in a
489 register. Followed by one byte with the register number, in
490 the range 0 to one less than the pattern buffer's re_nsub
491 field. Then followed by one byte with the number of groups
492 inner to this one. (This last has to be part of the
493 start_memory only because we need it in the on_failure_jump
497 /* Stop remembering the text that is matched and store it in a
498 memory register. Followed by one byte with the register
499 number, in the range 0 to one less than `re_nsub' in the
500 pattern buffer, and one byte with the number of inner groups,
501 just like `start_memory'. (We need the number of inner
502 groups here because we don't have any easy way of finding the
503 corresponding start_memory when we're at a stop_memory.) */
506 /* Match a duplicate of something remembered. Followed by one
507 byte containing the register number. */
510 /* Fail unless at beginning of line. */
513 /* Fail unless at end of line. */
516 /* Succeeds if at beginning of buffer (if emacs) or at beginning
517 of string to be matched (if not). */
520 /* Analogously, for end of buffer/string. */
523 /* Followed by two byte relative address to which to jump. */
526 /* Same as jump, but marks the end of an alternative. */
529 /* Followed by two-byte relative address of place to resume at
530 in case of failure. */
531 /* ifdef MBS_SUPPORT, the size of address is 1. */
534 /* Like on_failure_jump, but pushes a placeholder instead of the
535 current string position when executed. */
536 on_failure_keep_string_jump,
538 /* Throw away latest failure point and then jump to following
539 two-byte relative address. */
540 /* ifdef MBS_SUPPORT, the size of address is 1. */
543 /* Change to pop_failure_jump if know won't have to backtrack to
544 match; otherwise change to jump. This is used to jump
545 back to the beginning of a repeat. If what follows this jump
546 clearly won't match what the repeat does, such that we can be
547 sure that there is no use backtracking out of repetitions
548 already matched, then we change it to a pop_failure_jump.
549 Followed by two-byte address. */
550 /* ifdef MBS_SUPPORT, the size of address is 1. */
553 /* Jump to following two-byte address, and push a dummy failure
554 point. This failure point will be thrown away if an attempt
555 is made to use it for a failure. A `+' construct makes this
556 before the first repeat. Also used as an intermediary kind
557 of jump when compiling an alternative. */
558 /* ifdef MBS_SUPPORT, the size of address is 1. */
561 /* Push a dummy failure point and continue. Used at the end of
565 /* Followed by two-byte relative address and two-byte number n.
566 After matching N times, jump to the address upon failure. */
567 /* ifdef MBS_SUPPORT, the size of address is 1. */
570 /* Followed by two-byte relative address, and two-byte number n.
571 Jump to the address N times, then fail. */
572 /* ifdef MBS_SUPPORT, the size of address is 1. */
575 /* Set the following two-byte relative address to the
576 subsequent two-byte number. The address *includes* the two
578 /* ifdef MBS_SUPPORT, the size of address is 1. */
581 wordchar, /* Matches any word-constituent character. */
582 notwordchar, /* Matches any char that is not a word-constituent. */
584 wordbeg, /* Succeeds if at word beginning. */
585 wordend, /* Succeeds if at word end. */
587 wordbound, /* Succeeds if at a word boundary. */
588 notwordbound /* Succeeds if not at a word boundary. */
591 ,before_dot, /* Succeeds if before point. */
592 at_dot, /* Succeeds if at point. */
593 after_dot, /* Succeeds if after point. */
595 /* Matches any character whose syntax is specified. Followed by
596 a byte which contains a syntax code, e.g., Sword. */
599 /* Matches any character whose syntax is not that specified. */
603 #endif /* not INSIDE_RECURSION */
608 # define UCHAR_T unsigned char
609 # define COMPILED_BUFFER_VAR bufp->buffer
610 # define OFFSET_ADDRESS_SIZE 2
611 # define PREFIX(name) byte_##name
612 # define ARG_PREFIX(name) name
613 # define PUT_CHAR(c) putchar (c)
616 # define CHAR_T wchar_t
617 # define UCHAR_T wchar_t
618 # define COMPILED_BUFFER_VAR wc_buffer
619 # define OFFSET_ADDRESS_SIZE 1 /* the size which STORE_NUMBER macro use */
620 # define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
621 # define PREFIX(name) wcs_##name
622 # define ARG_PREFIX(name) c##name
623 /* Should we use wide stream?? */
624 # define PUT_CHAR(c) printf ("%C", c);
630 # define INSIDE_RECURSION
632 # undef INSIDE_RECURSION
635 # define INSIDE_RECURSION
637 # undef INSIDE_RECURSION
641 #ifdef INSIDE_RECURSION
642 /* Common operations on the compiled pattern. */
644 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
645 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
648 # define STORE_NUMBER(destination, number) \
650 *(destination) = (UCHAR_T)(number); \
653 # define STORE_NUMBER(destination, number) \
655 (destination)[0] = (number) & 0377; \
656 (destination)[1] = (number) >> 8; \
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 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
665 # define STORE_NUMBER_AND_INCR(destination, number) \
667 STORE_NUMBER (destination, number); \
668 (destination) += OFFSET_ADDRESS_SIZE; \
671 /* Put into DESTINATION a number stored in two contiguous bytes starting
673 /* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
676 # define EXTRACT_NUMBER(destination, source) \
678 (destination) = *(source); \
681 # define EXTRACT_NUMBER(destination, source) \
683 (destination) = *(source) & 0377; \
684 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
689 static void PREFIX(extract_number) _RE_ARGS ((int *dest, UCHAR_T *source));
691 PREFIX(extract_number) (dest, source)
698 int temp = SIGN_EXTEND_CHAR (*(source + 1));
699 *dest = *source & 0377;
704 # ifndef EXTRACT_MACROS /* To debug the macros. */
705 # undef EXTRACT_NUMBER
706 # define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
707 # endif /* not EXTRACT_MACROS */
711 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
712 SOURCE must be an lvalue. */
714 # define EXTRACT_NUMBER_AND_INCR(destination, source) \
716 EXTRACT_NUMBER (destination, source); \
717 (source) += OFFSET_ADDRESS_SIZE; \
721 static void PREFIX(extract_number_and_incr) _RE_ARGS ((int *destination,
724 PREFIX(extract_number_and_incr) (destination, source)
728 PREFIX(extract_number) (destination, *source);
729 *source += OFFSET_ADDRESS_SIZE;
732 # ifndef EXTRACT_MACROS
733 # undef EXTRACT_NUMBER_AND_INCR
734 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
735 PREFIX(extract_number_and_incr) (&dest, &src)
736 # endif /* not EXTRACT_MACROS */
742 /* If DEBUG is defined, Regex prints many voluminous messages about what
743 it is doing (if the variable `debug' is nonzero). If linked with the
744 main program in `iregex.c', you can enter patterns and strings
745 interactively. And if linked with the main program in `main.c' and
746 the other test files, you can run the already-written tests. */
750 # ifndef DEFINED_ONCE
752 /* We use standard I/O for debugging. */
755 /* It is useful to test things that ``must'' be true when debugging. */
760 # define DEBUG_STATEMENT(e) e
761 # define DEBUG_PRINT1(x) if (debug) printf (x)
762 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
763 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
764 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
765 # endif /* not DEFINED_ONCE */
767 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
768 if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
769 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
770 if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
773 /* Print the fastmap in human-readable form. */
775 # ifndef DEFINED_ONCE
777 print_fastmap (fastmap)
780 unsigned was_a_range = 0;
783 while (i < (1 << BYTEWIDTH))
789 while (i < (1 << BYTEWIDTH) && fastmap[i])
803 # endif /* not DEFINED_ONCE */
806 /* Print a compiled pattern string in human-readable form, starting at
807 the START pointer into it and ending just before the pointer END. */
810 PREFIX(print_partial_compiled_pattern) (start, end)
825 /* Loop over pattern commands. */
829 printf ("%td:\t", p - start);
831 printf ("%ld:\t", (long int) (p - start));
834 switch ((re_opcode_t) *p++)
842 printf ("/exactn/%d", mcnt);
854 printf ("/exactn_bin/%d", mcnt);
857 printf("/%lx", (long int) *p++);
861 # endif /* MBS_SUPPORT */
865 printf ("/start_memory/%d/%ld", mcnt, (long int) *p++);
870 printf ("/stop_memory/%d/%ld", mcnt, (long int) *p++);
874 printf ("/duplicate/%ld", (long int) *p++);
887 printf ("/charset [%s",
888 (re_opcode_t) *(workp - 1) == charset_not ? "^" : "");
890 length = *workp++; /* the length of char_classes */
891 for (i=0 ; i<length ; i++)
892 printf("[:%lx:]", (long int) *p++);
893 length = *workp++; /* the length of collating_symbol */
894 for (i=0 ; i<length ;)
898 PUT_CHAR((i++,*p++));
902 length = *workp++; /* the length of equivalence_class */
903 for (i=0 ; i<length ;)
907 PUT_CHAR((i++,*p++));
911 length = *workp++; /* the length of char_range */
912 for (i=0 ; i<length ; i++)
914 wchar_t range_start = *p++;
915 wchar_t range_end = *p++;
916 printf("%C-%C", range_start, range_end);
918 length = *workp++; /* the length of char */
919 for (i=0 ; i<length ; i++)
923 register int c, last = -100;
924 register int in_range = 0;
926 printf ("/charset [%s",
927 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
929 assert (p + *p < pend);
931 for (c = 0; c < 256; c++)
933 && (p[1 + (c/8)] & (1 << (c % 8))))
935 /* Are we starting a range? */
936 if (last + 1 == c && ! in_range)
941 /* Have we broken a range? */
942 else if (last + 1 != c && in_range)
972 case on_failure_jump:
973 PREFIX(extract_number_and_incr) (&mcnt, &p);
975 printf ("/on_failure_jump to %td", p + mcnt - start);
977 printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
981 case on_failure_keep_string_jump:
982 PREFIX(extract_number_and_incr) (&mcnt, &p);
984 printf ("/on_failure_keep_string_jump to %td", p + mcnt - start);
986 printf ("/on_failure_keep_string_jump to %ld",
987 (long int) (p + mcnt - start));
991 case dummy_failure_jump:
992 PREFIX(extract_number_and_incr) (&mcnt, &p);
994 printf ("/dummy_failure_jump to %td", p + mcnt - start);
996 printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
1000 case push_dummy_failure:
1001 printf ("/push_dummy_failure");
1004 case maybe_pop_jump:
1005 PREFIX(extract_number_and_incr) (&mcnt, &p);
1007 printf ("/maybe_pop_jump to %td", p + mcnt - start);
1009 printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
1013 case pop_failure_jump:
1014 PREFIX(extract_number_and_incr) (&mcnt, &p);
1016 printf ("/pop_failure_jump to %td", p + mcnt - start);
1018 printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
1023 PREFIX(extract_number_and_incr) (&mcnt, &p);
1025 printf ("/jump_past_alt to %td", p + mcnt - start);
1027 printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
1032 PREFIX(extract_number_and_incr) (&mcnt, &p);
1034 printf ("/jump to %td", p + mcnt - start);
1036 printf ("/jump to %ld", (long int) (p + mcnt - start));
1041 PREFIX(extract_number_and_incr) (&mcnt, &p);
1043 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1045 printf ("/succeed_n to %td, %d times", p1 - start, mcnt2);
1047 printf ("/succeed_n to %ld, %d times",
1048 (long int) (p1 - start), mcnt2);
1053 PREFIX(extract_number_and_incr) (&mcnt, &p);
1055 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1056 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
1060 PREFIX(extract_number_and_incr) (&mcnt, &p);
1062 PREFIX(extract_number_and_incr) (&mcnt2, &p);
1064 printf ("/set_number_at location %td to %d", p1 - start, mcnt2);
1066 printf ("/set_number_at location %ld to %d",
1067 (long int) (p1 - start), mcnt2);
1072 printf ("/wordbound");
1076 printf ("/notwordbound");
1080 printf ("/wordbeg");
1084 printf ("/wordend");
1089 printf ("/before_dot");
1097 printf ("/after_dot");
1101 printf ("/syntaxspec");
1103 printf ("/%d", mcnt);
1107 printf ("/notsyntaxspec");
1109 printf ("/%d", mcnt);
1114 printf ("/wordchar");
1118 printf ("/notwordchar");
1130 printf ("?%ld", (long int) *(p-1));
1137 printf ("%td:\tend of pattern.\n", p - start);
1139 printf ("%ld:\tend of pattern.\n", (long int) (p - start));
1145 PREFIX(print_compiled_pattern) (bufp)
1146 struct re_pattern_buffer *bufp;
1148 UCHAR_T *buffer = (UCHAR_T*) bufp->buffer;
1150 PREFIX(print_partial_compiled_pattern) (buffer, buffer
1151 + bufp->used / sizeof(UCHAR_T));
1152 printf ("%ld bytes used/%ld bytes allocated.\n",
1153 bufp->used, bufp->allocated);
1155 if (bufp->fastmap_accurate && bufp->fastmap)
1157 printf ("fastmap: ");
1158 print_fastmap (bufp->fastmap);
1162 printf ("re_nsub: %Zd\t", bufp->re_nsub);
1164 printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
1166 printf ("regs_alloc: %d\t", bufp->regs_allocated);
1167 printf ("can_be_null: %d\t", bufp->can_be_null);
1168 printf ("newline_anchor: %d\n", bufp->newline_anchor);
1169 printf ("no_sub: %d\t", bufp->no_sub);
1170 printf ("not_bol: %d\t", bufp->not_bol);
1171 printf ("not_eol: %d\t", bufp->not_eol);
1172 printf ("syntax: %lx\n", bufp->syntax);
1173 /* Perhaps we should print the translate table? */
1178 PREFIX(print_double_string) (where, string1, size1, string2, size2)
1179 const CHAR_T *where;
1180 const CHAR_T *string1;
1181 const CHAR_T *string2;
1193 if (FIRST_STRING_P (where))
1195 for (this_char = where - string1; this_char < size1; this_char++)
1196 PUT_CHAR (string1[this_char]);
1202 for (this_char = where - string2; this_char < size2; this_char++)
1204 PUT_CHAR (string2[this_char]);
1207 fputs ("...", stdout);
1214 # ifndef DEFINED_ONCE
1223 # else /* not DEBUG */
1225 # ifndef DEFINED_ONCE
1229 # define DEBUG_STATEMENT(e)
1230 # define DEBUG_PRINT1(x)
1231 # define DEBUG_PRINT2(x1, x2)
1232 # define DEBUG_PRINT3(x1, x2, x3)
1233 # define DEBUG_PRINT4(x1, x2, x3, x4)
1234 # endif /* not DEFINED_ONCE */
1235 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1236 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1238 # endif /* not DEBUG */
1243 /* This convert a multibyte string to a wide character string.
1244 And write their correspondances to offset_buffer(see below)
1245 and write whether each wchar_t is binary data to is_binary.
1246 This assume invalid multibyte sequences as binary data.
1247 We assume offset_buffer and is_binary is already allocated
1250 static size_t convert_mbs_to_wcs (CHAR_T *dest, const unsigned char* src,
1251 size_t len, int *offset_buffer,
1254 convert_mbs_to_wcs (dest, src, len, offset_buffer, is_binary)
1256 const unsigned char* src;
1257 size_t len; /* the length of multibyte string. */
1259 /* It hold correspondances between src(char string) and
1260 dest(wchar_t string) for optimization.
1262 dest = {'X', 'Y', 'Z'}
1263 (each "xxx", "y" and "zz" represent one multibyte character
1264 corresponding to 'X', 'Y' and 'Z'.)
1265 offset_buffer = {0, 0+3("xxx"), 0+3+1("y"), 0+3+1+2("zz")}
1271 wchar_t *pdest = dest;
1272 const unsigned char *psrc = src;
1273 size_t wc_count = 0;
1277 size_t mb_remain = len;
1278 size_t mb_count = 0;
1280 /* Initialize the conversion state. */
1281 memset (&mbs, 0, sizeof (mbstate_t));
1283 offset_buffer[0] = 0;
1284 for( ; mb_remain > 0 ; ++wc_count, ++pdest, mb_remain -= consumed,
1288 consumed = __mbrtowc (pdest, psrc, mb_remain, &mbs);
1290 consumed = mbrtowc (pdest, psrc, mb_remain, &mbs);
1294 /* failed to convert. maybe src contains binary data.
1295 So we consume 1 byte manualy. */
1299 is_binary[wc_count] = TRUE;
1302 is_binary[wc_count] = FALSE;
1303 /* In sjis encoding, we use yen sign as escape character in
1304 place of reverse solidus. So we convert 0x5c(yen sign in
1305 sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
1306 solidus in UCS2). */
1307 if (consumed == 1 && (int) *psrc == 0x5c && (int) *pdest == 0xa5)
1308 *pdest = (wchar_t) *psrc;
1310 offset_buffer[wc_count + 1] = mb_count += consumed;
1313 /* Fill remain of the buffer with sentinel. */
1314 for (i = wc_count + 1 ; i <= len ; i++)
1315 offset_buffer[i] = mb_count + 1;
1322 #else /* not INSIDE_RECURSION */
1324 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1325 also be assigned to arbitrarily: each pattern buffer stores its own
1326 syntax, so it can be changed between regex compilations. */
1327 /* This has no initializer because initialized variables in Emacs
1328 become read-only after dumping. */
1329 reg_syntax_t re_syntax_options;
1332 /* Specify the precise syntax of regexps for compilation. This provides
1333 for compatibility for various utilities which historically have
1334 different, incompatible syntaxes.
1336 The argument SYNTAX is a bit mask comprised of the various bits
1337 defined in regex.h. We return the old syntax. */
1340 re_set_syntax (syntax)
1341 reg_syntax_t syntax;
1343 reg_syntax_t ret = re_syntax_options;
1345 re_syntax_options = syntax;
1347 if (syntax & RE_DEBUG)
1349 else if (debug) /* was on but now is not */
1355 weak_alias (__re_set_syntax, re_set_syntax)
1358 /* This table gives an error message for each of the error codes listed
1359 in regex.h. Obviously the order here has to be same as there.
1360 POSIX doesn't require that we do anything for REG_NOERROR,
1361 but why not be nice? */
1363 static const char re_error_msgid[] =
1365 # define REG_NOERROR_IDX 0
1366 gettext_noop ("Success") /* REG_NOERROR */
1368 # define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
1369 gettext_noop ("No match") /* REG_NOMATCH */
1371 # define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
1372 gettext_noop ("Invalid regular expression") /* REG_BADPAT */
1374 # define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
1375 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
1377 # define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
1378 gettext_noop ("Invalid character class name") /* REG_ECTYPE */
1380 # define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
1381 gettext_noop ("Trailing backslash") /* REG_EESCAPE */
1383 # define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
1384 gettext_noop ("Invalid back reference") /* REG_ESUBREG */
1386 # define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
1387 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
1389 # define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
1390 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
1392 # define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
1393 gettext_noop ("Unmatched \\{") /* REG_EBRACE */
1395 # define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
1396 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
1398 # define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
1399 gettext_noop ("Invalid range end") /* REG_ERANGE */
1401 # define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
1402 gettext_noop ("Memory exhausted") /* REG_ESPACE */
1404 # define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
1405 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
1407 # define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
1408 gettext_noop ("Premature end of regular expression") /* REG_EEND */
1410 # define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
1411 gettext_noop ("Regular expression too big") /* REG_ESIZE */
1413 # define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
1414 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
1417 static const size_t re_error_msgid_idx[] =
1438 #endif /* INSIDE_RECURSION */
1440 #ifndef DEFINED_ONCE
1441 /* Avoiding alloca during matching, to placate r_alloc. */
1443 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1444 searching and matching functions should not call alloca. On some
1445 systems, alloca is implemented in terms of malloc, and if we're
1446 using the relocating allocator routines, then malloc could cause a
1447 relocation, which might (if the strings being searched are in the
1448 ralloc heap) shift the data out from underneath the regexp
1451 Here's another reason to avoid allocation: Emacs
1452 processes input from X in a signal handler; processing X input may
1453 call malloc; if input arrives while a matching routine is calling
1454 malloc, then we're scrod. But Emacs can't just block input while
1455 calling matching routines; then we don't notice interrupts when
1456 they come in. So, Emacs blocks input around all regexp calls
1457 except the matching calls, which it leaves unprotected, in the
1458 faith that they will not malloc. */
1460 /* Normally, this is fine. */
1461 # define MATCH_MAY_ALLOCATE
1463 /* When using GNU C, we are not REALLY using the C alloca, no matter
1464 what config.h may say. So don't take precautions for it. */
1469 /* The match routines may not allocate if (1) they would do it with malloc
1470 and (2) it's not safe for them to use malloc.
1471 Note that if REL_ALLOC is defined, matching would not use malloc for the
1472 failure stack, but we would still use it for the register vectors;
1473 so REL_ALLOC should not affect this. */
1474 # if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1475 # undef MATCH_MAY_ALLOCATE
1477 #endif /* not DEFINED_ONCE */
1479 #ifdef INSIDE_RECURSION
1480 /* Failure stack declarations and macros; both re_compile_fastmap and
1481 re_match_2 use a failure stack. These have to be macros because of
1482 REGEX_ALLOCATE_STACK. */
1485 /* Number of failure points for which to initially allocate space
1486 when matching. If this number is exceeded, we allocate more
1487 space, so it is not a hard limit. */
1488 # ifndef INIT_FAILURE_ALLOC
1489 # define INIT_FAILURE_ALLOC 5
1492 /* Roughly the maximum number of failure points on the stack. Would be
1493 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1494 This is a variable only so users of regex can assign to it; we never
1495 change it ourselves. */
1497 # ifdef INT_IS_16BIT
1499 # ifndef DEFINED_ONCE
1500 # if defined MATCH_MAY_ALLOCATE
1501 /* 4400 was enough to cause a crash on Alpha OSF/1,
1502 whose default stack limit is 2mb. */
1503 long int re_max_failures = 4000;
1505 long int re_max_failures = 2000;
1509 union PREFIX(fail_stack_elt)
1515 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1519 PREFIX(fail_stack_elt_t) *stack;
1520 unsigned long int size;
1521 unsigned long int avail; /* Offset of next open position. */
1522 } PREFIX(fail_stack_type);
1524 # else /* not INT_IS_16BIT */
1526 # ifndef DEFINED_ONCE
1527 # if defined MATCH_MAY_ALLOCATE
1528 /* 4400 was enough to cause a crash on Alpha OSF/1,
1529 whose default stack limit is 2mb. */
1530 int re_max_failures = 4000;
1532 int re_max_failures = 2000;
1536 union PREFIX(fail_stack_elt)
1542 typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
1546 PREFIX(fail_stack_elt_t) *stack;
1548 unsigned avail; /* Offset of next open position. */
1549 } PREFIX(fail_stack_type);
1551 # endif /* INT_IS_16BIT */
1553 # ifndef DEFINED_ONCE
1554 # define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1555 # define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1556 # define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1560 /* Define macros to initialize and free the failure stack.
1561 Do `return -2' if the alloc fails. */
1563 # ifdef MATCH_MAY_ALLOCATE
1564 # define INIT_FAIL_STACK() \
1566 fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
1567 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
1569 if (fail_stack.stack == NULL) \
1572 fail_stack.size = INIT_FAILURE_ALLOC; \
1573 fail_stack.avail = 0; \
1576 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1578 # define INIT_FAIL_STACK() \
1580 fail_stack.avail = 0; \
1583 # define RESET_FAIL_STACK()
1587 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1589 Return 1 if succeeds, and 0 if either ran out of memory
1590 allocating space for it or it was already too large.
1592 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1594 # define DOUBLE_FAIL_STACK(fail_stack) \
1595 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1597 : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
1598 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1599 (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
1600 ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
1602 (fail_stack).stack == NULL \
1604 : ((fail_stack).size <<= 1, \
1608 /* Push pointer POINTER on FAIL_STACK.
1609 Return 1 if was able to do so and 0 if ran out of memory allocating
1611 # define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1612 ((FAIL_STACK_FULL () \
1613 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1615 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1618 /* Push a pointer value onto the failure stack.
1619 Assumes the variable `fail_stack'. Probably should only
1620 be called from within `PUSH_FAILURE_POINT'. */
1621 # define PUSH_FAILURE_POINTER(item) \
1622 fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
1624 /* This pushes an integer-valued item onto the failure stack.
1625 Assumes the variable `fail_stack'. Probably should only
1626 be called from within `PUSH_FAILURE_POINT'. */
1627 # define PUSH_FAILURE_INT(item) \
1628 fail_stack.stack[fail_stack.avail++].integer = (item)
1630 /* Push a fail_stack_elt_t value onto the failure stack.
1631 Assumes the variable `fail_stack'. Probably should only
1632 be called from within `PUSH_FAILURE_POINT'. */
1633 # define PUSH_FAILURE_ELT(item) \
1634 fail_stack.stack[fail_stack.avail++] = (item)
1636 /* These three POP... operations complement the three PUSH... operations.
1637 All assume that `fail_stack' is nonempty. */
1638 # define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1639 # define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1640 # define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1642 /* Used to omit pushing failure point id's when we're not debugging. */
1644 # define DEBUG_PUSH PUSH_FAILURE_INT
1645 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1647 # define DEBUG_PUSH(item)
1648 # define DEBUG_POP(item_addr)
1652 /* Push the information about the state we will need
1653 if we ever fail back to it.
1655 Requires variables fail_stack, regstart, regend, reg_info, and
1656 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1659 Does `return FAILURE_CODE' if runs out of memory. */
1661 # define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1663 char *destination; \
1664 /* Must be int, so when we don't save any registers, the arithmetic \
1665 of 0 + -1 isn't done as unsigned. */ \
1666 /* Can't be int, since there is not a shred of a guarantee that int \
1667 is wide enough to hold a value of something to which pointer can \
1669 active_reg_t this_reg; \
1671 DEBUG_STATEMENT (failure_id++); \
1672 DEBUG_STATEMENT (nfailure_points_pushed++); \
1673 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1674 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1675 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1677 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1678 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1680 /* Ensure we have enough space allocated for what we will push. */ \
1681 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1683 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1684 return failure_code; \
1686 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1687 (fail_stack).size); \
1688 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1691 /* Push the info, starting with the registers. */ \
1692 DEBUG_PRINT1 ("\n"); \
1695 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1698 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1699 DEBUG_STATEMENT (num_regs_pushed++); \
1701 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1702 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1704 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1705 PUSH_FAILURE_POINTER (regend[this_reg]); \
1707 DEBUG_PRINT2 (" info: %p\n ", \
1708 reg_info[this_reg].word.pointer); \
1709 DEBUG_PRINT2 (" match_null=%d", \
1710 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1711 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1712 DEBUG_PRINT2 (" matched_something=%d", \
1713 MATCHED_SOMETHING (reg_info[this_reg])); \
1714 DEBUG_PRINT2 (" ever_matched=%d", \
1715 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1716 DEBUG_PRINT1 ("\n"); \
1717 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1720 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1721 PUSH_FAILURE_INT (lowest_active_reg); \
1723 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1724 PUSH_FAILURE_INT (highest_active_reg); \
1726 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1727 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1728 PUSH_FAILURE_POINTER (pattern_place); \
1730 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1731 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1733 DEBUG_PRINT1 ("'\n"); \
1734 PUSH_FAILURE_POINTER (string_place); \
1736 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1737 DEBUG_PUSH (failure_id); \
1740 # ifndef DEFINED_ONCE
1741 /* This is the number of items that are pushed and popped on the stack
1742 for each register. */
1743 # define NUM_REG_ITEMS 3
1745 /* Individual items aside from the registers. */
1747 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1749 # define NUM_NONREG_ITEMS 4
1752 /* We push at most this many items on the stack. */
1753 /* We used to use (num_regs - 1), which is the number of registers
1754 this regexp will save; but that was changed to 5
1755 to avoid stack overflow for a regexp with lots of parens. */
1756 # define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1758 /* We actually push this many items. */
1759 # define NUM_FAILURE_ITEMS \
1761 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1765 /* How many items can still be added to the stack without overflowing it. */
1766 # define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1767 # endif /* not DEFINED_ONCE */
1770 /* Pops what PUSH_FAIL_STACK pushes.
1772 We restore into the parameters, all of which should be lvalues:
1773 STR -- the saved data position.
1774 PAT -- the saved pattern position.
1775 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1776 REGSTART, REGEND -- arrays of string positions.
1777 REG_INFO -- array of information about each subexpression.
1779 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1780 `pend', `string1', `size1', `string2', and `size2'. */
1781 # define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1783 DEBUG_STATEMENT (unsigned failure_id;) \
1784 active_reg_t this_reg; \
1785 const UCHAR_T *string_temp; \
1787 assert (!FAIL_STACK_EMPTY ()); \
1789 /* Remove failure points and point to how many regs pushed. */ \
1790 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1791 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1792 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1794 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1796 DEBUG_POP (&failure_id); \
1797 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1799 /* If the saved string location is NULL, it came from an \
1800 on_failure_keep_string_jump opcode, and we want to throw away the \
1801 saved NULL, thus retaining our current position in the string. */ \
1802 string_temp = POP_FAILURE_POINTER (); \
1803 if (string_temp != NULL) \
1804 str = (const CHAR_T *) string_temp; \
1806 DEBUG_PRINT2 (" Popping string %p: `", str); \
1807 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1808 DEBUG_PRINT1 ("'\n"); \
1810 pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
1811 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1812 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1814 /* Restore register info. */ \
1815 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1816 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1818 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1819 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1822 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1824 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1826 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1827 DEBUG_PRINT2 (" info: %p\n", \
1828 reg_info[this_reg].word.pointer); \
1830 regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1831 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1833 regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
1834 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1838 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1840 reg_info[this_reg].word.integer = 0; \
1841 regend[this_reg] = 0; \
1842 regstart[this_reg] = 0; \
1844 highest_active_reg = high_reg; \
1847 set_regs_matched_done = 0; \
1848 DEBUG_STATEMENT (nfailure_points_popped++); \
1849 } /* POP_FAILURE_POINT */
1851 /* Structure for per-register (a.k.a. per-group) information.
1852 Other register information, such as the
1853 starting and ending positions (which are addresses), and the list of
1854 inner groups (which is a bits list) are maintained in separate
1857 We are making a (strictly speaking) nonportable assumption here: that
1858 the compiler will pack our bit fields into something that fits into
1859 the type of `word', i.e., is something that fits into one item on the
1863 /* Declarations and macros for re_match_2. */
1867 PREFIX(fail_stack_elt_t) word;
1870 /* This field is one if this group can match the empty string,
1871 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1872 # define MATCH_NULL_UNSET_VALUE 3
1873 unsigned match_null_string_p : 2;
1874 unsigned is_active : 1;
1875 unsigned matched_something : 1;
1876 unsigned ever_matched_something : 1;
1878 } PREFIX(register_info_type);
1880 # ifndef DEFINED_ONCE
1881 # define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1882 # define IS_ACTIVE(R) ((R).bits.is_active)
1883 # define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1884 # define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1887 /* Call this when have matched a real character; it sets `matched' flags
1888 for the subexpressions which we are currently inside. Also records
1889 that those subexprs have matched. */
1890 # define SET_REGS_MATCHED() \
1893 if (!set_regs_matched_done) \
1896 set_regs_matched_done = 1; \
1897 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1899 MATCHED_SOMETHING (reg_info[r]) \
1900 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1906 # endif /* not DEFINED_ONCE */
1908 /* Registers are set to a sentinel when they haven't yet matched. */
1909 static CHAR_T PREFIX(reg_unset_dummy);
1910 # define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
1911 # define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1913 /* Subroutine declarations and macros for regex_compile. */
1914 static void PREFIX(store_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc, int arg));
1915 static void PREFIX(store_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1916 int arg1, int arg2));
1917 static void PREFIX(insert_op1) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1918 int arg, UCHAR_T *end));
1919 static void PREFIX(insert_op2) _RE_ARGS ((re_opcode_t op, UCHAR_T *loc,
1920 int arg1, int arg2, UCHAR_T *end));
1921 static boolean PREFIX(at_begline_loc_p) _RE_ARGS ((const CHAR_T *pattern,
1923 reg_syntax_t syntax));
1924 static boolean PREFIX(at_endline_loc_p) _RE_ARGS ((const CHAR_T *p,
1926 reg_syntax_t syntax));
1928 static reg_errcode_t wcs_compile_range _RE_ARGS ((CHAR_T range_start,
1929 const CHAR_T **p_ptr,
1932 reg_syntax_t syntax,
1935 static void insert_space _RE_ARGS ((int num, CHAR_T *loc, CHAR_T *end));
1937 static reg_errcode_t byte_compile_range _RE_ARGS ((unsigned int range_start,
1941 reg_syntax_t syntax,
1945 /* Fetch the next character in the uncompiled pattern---translating it
1946 if necessary. Also cast from a signed character in the constant
1947 string passed to us by the user to an unsigned char that we can use
1948 as an array index (in, e.g., `translate'). */
1949 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1950 because it is impossible to allocate 4GB array for some encodings
1951 which have 4 byte character_set like UCS4. */
1954 # define PATFETCH(c) \
1955 do {if (p == pend) return REG_EEND; \
1956 c = (UCHAR_T) *p++; \
1957 if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
1960 # define PATFETCH(c) \
1961 do {if (p == pend) return REG_EEND; \
1962 c = (unsigned char) *p++; \
1963 if (translate) c = (unsigned char) translate[c]; \
1968 /* Fetch the next character in the uncompiled pattern, with no
1970 # define PATFETCH_RAW(c) \
1971 do {if (p == pend) return REG_EEND; \
1972 c = (UCHAR_T) *p++; \
1975 /* Go backwards one character in the pattern. */
1976 # define PATUNFETCH p--
1979 /* If `translate' is non-null, return translate[D], else just D. We
1980 cast the subscript to translate because some data is declared as
1981 `char *', to avoid warnings when a string constant is passed. But
1982 when we use a character as a subscript we must make it unsigned. */
1983 /* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
1984 because it is impossible to allocate 4GB array for some encodings
1985 which have 4 byte character_set like UCS4. */
1989 # define TRANSLATE(d) \
1990 ((translate && ((UCHAR_T) (d)) <= 0xff) \
1991 ? (char) translate[(unsigned char) (d)] : (d))
1993 # define TRANSLATE(d) \
1994 (translate ? (char) translate[(unsigned char) (d)] : (d))
1999 /* Macros for outputting the compiled pattern into `buffer'. */
2001 /* If the buffer isn't allocated when it comes in, use this. */
2002 # define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
2004 /* Make sure we have at least N more bytes of space in buffer. */
2006 # define GET_BUFFER_SPACE(n) \
2007 while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
2008 + (n)*sizeof(CHAR_T)) > bufp->allocated) \
2011 # define GET_BUFFER_SPACE(n) \
2012 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
2016 /* Make sure we have one more byte of buffer space and then add C to it. */
2017 # define BUF_PUSH(c) \
2019 GET_BUFFER_SPACE (1); \
2020 *b++ = (UCHAR_T) (c); \
2024 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
2025 # define BUF_PUSH_2(c1, c2) \
2027 GET_BUFFER_SPACE (2); \
2028 *b++ = (UCHAR_T) (c1); \
2029 *b++ = (UCHAR_T) (c2); \
2033 /* As with BUF_PUSH_2, except for three bytes. */
2034 # define BUF_PUSH_3(c1, c2, c3) \
2036 GET_BUFFER_SPACE (3); \
2037 *b++ = (UCHAR_T) (c1); \
2038 *b++ = (UCHAR_T) (c2); \
2039 *b++ = (UCHAR_T) (c3); \
2042 /* Store a jump with opcode OP at LOC to location TO. We store a
2043 relative address offset by the three bytes the jump itself occupies. */
2044 # define STORE_JUMP(op, loc, to) \
2045 PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
2047 /* Likewise, for a two-argument jump. */
2048 # define STORE_JUMP2(op, loc, to, arg) \
2049 PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
2051 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
2052 # define INSERT_JUMP(op, loc, to) \
2053 PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
2055 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
2056 # define INSERT_JUMP2(op, loc, to, arg) \
2057 PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
2060 /* This is not an arbitrary limit: the arguments which represent offsets
2061 into the pattern are two bytes long. So if 2^16 bytes turns out to
2062 be too small, many things would have to change. */
2063 /* Any other compiler which, like MSC, has allocation limit below 2^16
2064 bytes will have to use approach similar to what was done below for
2065 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
2066 reallocating to 0 bytes. Such thing is not going to work too well.
2067 You have been warned!! */
2068 # ifndef DEFINED_ONCE
2069 # if defined _MSC_VER && !defined WIN32
2070 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
2071 The REALLOC define eliminates a flurry of conversion warnings,
2072 but is not required. */
2073 # define MAX_BUF_SIZE 65500L
2074 # define REALLOC(p,s) realloc ((p), (size_t) (s))
2076 # define MAX_BUF_SIZE (1L << 16)
2077 # define REALLOC(p,s) realloc ((p), (s))
2080 /* Extend the buffer by twice its current size via realloc and
2081 reset the pointers that pointed into the old block to point to the
2082 correct places in the new one. If extending the buffer results in it
2083 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
2084 # if __BOUNDED_POINTERS__
2085 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
2086 # define MOVE_BUFFER_POINTER(P) \
2087 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
2088 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
2091 SET_HIGH_BOUND (b); \
2092 SET_HIGH_BOUND (begalt); \
2093 if (fixup_alt_jump) \
2094 SET_HIGH_BOUND (fixup_alt_jump); \
2096 SET_HIGH_BOUND (laststart); \
2097 if (pending_exact) \
2098 SET_HIGH_BOUND (pending_exact); \
2101 # define MOVE_BUFFER_POINTER(P) (P) += incr
2102 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
2104 # endif /* not DEFINED_ONCE */
2107 # define EXTEND_BUFFER() \
2109 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2111 if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
2113 bufp->allocated <<= 1; \
2114 if (bufp->allocated > MAX_BUF_SIZE) \
2115 bufp->allocated = MAX_BUF_SIZE; \
2116 /* How many characters the new buffer can have? */ \
2117 wchar_count = bufp->allocated / sizeof(UCHAR_T); \
2118 if (wchar_count == 0) wchar_count = 1; \
2119 /* Truncate the buffer to CHAR_T align. */ \
2120 bufp->allocated = wchar_count * sizeof(UCHAR_T); \
2121 RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
2122 bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
2123 if (COMPILED_BUFFER_VAR == NULL) \
2124 return REG_ESPACE; \
2125 /* If the buffer moved, move all the pointers into it. */ \
2126 if (old_buffer != COMPILED_BUFFER_VAR) \
2128 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2129 MOVE_BUFFER_POINTER (b); \
2130 MOVE_BUFFER_POINTER (begalt); \
2131 if (fixup_alt_jump) \
2132 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2134 MOVE_BUFFER_POINTER (laststart); \
2135 if (pending_exact) \
2136 MOVE_BUFFER_POINTER (pending_exact); \
2138 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2141 # define EXTEND_BUFFER() \
2143 UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
2144 if (bufp->allocated == MAX_BUF_SIZE) \
2146 bufp->allocated <<= 1; \
2147 if (bufp->allocated > MAX_BUF_SIZE) \
2148 bufp->allocated = MAX_BUF_SIZE; \
2149 bufp->buffer = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, \
2151 if (COMPILED_BUFFER_VAR == NULL) \
2152 return REG_ESPACE; \
2153 /* If the buffer moved, move all the pointers into it. */ \
2154 if (old_buffer != COMPILED_BUFFER_VAR) \
2156 int incr = COMPILED_BUFFER_VAR - old_buffer; \
2157 MOVE_BUFFER_POINTER (b); \
2158 MOVE_BUFFER_POINTER (begalt); \
2159 if (fixup_alt_jump) \
2160 MOVE_BUFFER_POINTER (fixup_alt_jump); \
2162 MOVE_BUFFER_POINTER (laststart); \
2163 if (pending_exact) \
2164 MOVE_BUFFER_POINTER (pending_exact); \
2166 ELSE_EXTEND_BUFFER_HIGH_BOUND \
2170 # ifndef DEFINED_ONCE
2171 /* Since we have one byte reserved for the register number argument to
2172 {start,stop}_memory, the maximum number of groups we can report
2173 things about is what fits in that byte. */
2174 # define MAX_REGNUM 255
2176 /* But patterns can have more than `MAX_REGNUM' registers. We just
2177 ignore the excess. */
2178 typedef unsigned regnum_t;
2181 /* Macros for the compile stack. */
2183 /* Since offsets can go either forwards or backwards, this type needs to
2184 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
2185 /* int may be not enough when sizeof(int) == 2. */
2186 typedef long pattern_offset_t;
2190 pattern_offset_t begalt_offset;
2191 pattern_offset_t fixup_alt_jump;
2192 pattern_offset_t inner_group_offset;
2193 pattern_offset_t laststart_offset;
2195 } compile_stack_elt_t;
2200 compile_stack_elt_t *stack;
2202 unsigned avail; /* Offset of next open position. */
2203 } compile_stack_type;
2206 # define INIT_COMPILE_STACK_SIZE 32
2208 # define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
2209 # define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
2211 /* The next available element. */
2212 # define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
2214 # endif /* not DEFINED_ONCE */
2216 /* Set the bit for character C in a list. */
2217 # ifndef DEFINED_ONCE
2218 # define SET_LIST_BIT(c) \
2219 (b[((unsigned char) (c)) / BYTEWIDTH] \
2220 |= 1 << (((unsigned char) c) % BYTEWIDTH))
2221 # endif /* DEFINED_ONCE */
2223 /* Get the next unsigned number in the uncompiled pattern. */
2224 # define GET_UNSIGNED_NUMBER(num) \
2229 if (c < '0' || c > '9') \
2231 if (num <= RE_DUP_MAX) \
2235 num = num * 10 + c - '0'; \
2240 # ifndef DEFINED_ONCE
2241 # if defined _LIBC || WIDE_CHAR_SUPPORT
2242 /* The GNU C library provides support for user-defined character classes
2243 and the functions from ISO C amendement 1. */
2244 # ifdef CHARCLASS_NAME_MAX
2245 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
2247 /* This shouldn't happen but some implementation might still have this
2248 problem. Use a reasonable default value. */
2249 # define CHAR_CLASS_MAX_LENGTH 256
2253 # define IS_CHAR_CLASS(string) __wctype (string)
2255 # define IS_CHAR_CLASS(string) wctype (string)
2258 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
2260 # define IS_CHAR_CLASS(string) \
2261 (STREQ (string, "alpha") || STREQ (string, "upper") \
2262 || STREQ (string, "lower") || STREQ (string, "digit") \
2263 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
2264 || STREQ (string, "space") || STREQ (string, "print") \
2265 || STREQ (string, "punct") || STREQ (string, "graph") \
2266 || STREQ (string, "cntrl") || STREQ (string, "blank"))
2268 # endif /* DEFINED_ONCE */
2270 # ifndef MATCH_MAY_ALLOCATE
2272 /* If we cannot allocate large objects within re_match_2_internal,
2273 we make the fail stack and register vectors global.
2274 The fail stack, we grow to the maximum size when a regexp
2276 The register vectors, we adjust in size each time we
2277 compile a regexp, according to the number of registers it needs. */
2279 static PREFIX(fail_stack_type) fail_stack;
2281 /* Size with which the following vectors are currently allocated.
2282 That is so we can make them bigger as needed,
2283 but never make them smaller. */
2284 # ifdef DEFINED_ONCE
2285 static int regs_allocated_size;
2287 static const char ** regstart, ** regend;
2288 static const char ** old_regstart, ** old_regend;
2289 static const char **best_regstart, **best_regend;
2290 static const char **reg_dummy;
2291 # endif /* DEFINED_ONCE */
2293 static PREFIX(register_info_type) *PREFIX(reg_info);
2294 static PREFIX(register_info_type) *PREFIX(reg_info_dummy);
2296 /* Make the register vectors big enough for NUM_REGS registers,
2297 but don't make them smaller. */
2300 PREFIX(regex_grow_registers) (num_regs)
2303 if (num_regs > regs_allocated_size)
2305 RETALLOC_IF (regstart, num_regs, const char *);
2306 RETALLOC_IF (regend, num_regs, const char *);
2307 RETALLOC_IF (old_regstart, num_regs, const char *);
2308 RETALLOC_IF (old_regend, num_regs, const char *);
2309 RETALLOC_IF (best_regstart, num_regs, const char *);
2310 RETALLOC_IF (best_regend, num_regs, const char *);
2311 RETALLOC_IF (PREFIX(reg_info), num_regs, PREFIX(register_info_type));
2312 RETALLOC_IF (reg_dummy, num_regs, const char *);
2313 RETALLOC_IF (PREFIX(reg_info_dummy), num_regs, PREFIX(register_info_type));
2315 regs_allocated_size = num_regs;
2319 # endif /* not MATCH_MAY_ALLOCATE */
2321 # ifndef DEFINED_ONCE
2322 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
2325 # endif /* not DEFINED_ONCE */
2327 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
2328 Returns one of error codes defined in `regex.h', or zero for success.
2330 Assumes the `allocated' (and perhaps `buffer') and `translate'
2331 fields are set in BUFP on entry.
2333 If it succeeds, results are put in BUFP (if it returns an error, the
2334 contents of BUFP are undefined):
2335 `buffer' is the compiled pattern;
2336 `syntax' is set to SYNTAX;
2337 `used' is set to the length of the compiled pattern;
2338 `fastmap_accurate' is zero;
2339 `re_nsub' is the number of subexpressions in PATTERN;
2340 `not_bol' and `not_eol' are zero;
2342 The `fastmap' and `newline_anchor' fields are neither
2343 examined nor set. */
2345 /* Return, freeing storage we allocated. */
2347 # define FREE_STACK_RETURN(value) \
2348 return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
2350 # define FREE_STACK_RETURN(value) \
2351 return (free (compile_stack.stack), value)
2354 static reg_errcode_t
2355 PREFIX(regex_compile) (ARG_PREFIX(pattern), ARG_PREFIX(size), syntax, bufp)
2356 const char *ARG_PREFIX(pattern);
2357 size_t ARG_PREFIX(size);
2358 reg_syntax_t syntax;
2359 struct re_pattern_buffer *bufp;
2361 /* We fetch characters from PATTERN here. Even though PATTERN is
2362 `char *' (i.e., signed), we declare these variables as unsigned, so
2363 they can be reliably used as array indices. */
2364 register UCHAR_T c, c1;
2367 /* A temporary space to keep wchar_t pattern and compiled pattern. */
2368 CHAR_T *pattern, *COMPILED_BUFFER_VAR;
2370 /* offset buffer for optimization. See convert_mbs_to_wc. */
2371 int *mbs_offset = NULL;
2372 /* It hold whether each wchar_t is binary data or not. */
2373 char *is_binary = NULL;
2374 /* A flag whether exactn is handling binary data or not. */
2375 char is_exactn_bin = FALSE;
2378 /* A random temporary spot in PATTERN. */
2381 /* Points to the end of the buffer, where we should append. */
2382 register UCHAR_T *b;
2384 /* Keeps track of unclosed groups. */
2385 compile_stack_type compile_stack;
2387 /* Points to the current (ending) position in the pattern. */
2392 const CHAR_T *p = pattern;
2393 const CHAR_T *pend = pattern + size;
2396 /* How to translate the characters in the pattern. */
2397 RE_TRANSLATE_TYPE translate = bufp->translate;
2399 /* Address of the count-byte of the most recently inserted `exactn'
2400 command. This makes it possible to tell if a new exact-match
2401 character can be added to that command or if the character requires
2402 a new `exactn' command. */
2403 UCHAR_T *pending_exact = 0;
2405 /* Address of start of the most recently finished expression.
2406 This tells, e.g., postfix * where to find the start of its
2407 operand. Reset at the beginning of groups and alternatives. */
2408 UCHAR_T *laststart = 0;
2410 /* Address of beginning of regexp, or inside of last group. */
2413 /* Address of the place where a forward jump should go to the end of
2414 the containing expression. Each alternative of an `or' -- except the
2415 last -- ends with a forward jump of this sort. */
2416 UCHAR_T *fixup_alt_jump = 0;
2418 /* Counts open-groups as they are encountered. Remembered for the
2419 matching close-group on the compile stack, so the same register
2420 number is put in the stop_memory as the start_memory. */
2421 regnum_t regnum = 0;
2424 /* Initialize the wchar_t PATTERN and offset_buffer. */
2425 p = pend = pattern = TALLOC(csize + 1, CHAR_T);
2426 mbs_offset = TALLOC(csize + 1, int);
2427 is_binary = TALLOC(csize + 1, char);
2428 if (pattern == NULL || mbs_offset == NULL || is_binary == NULL)
2435 pattern[csize] = L'\0'; /* sentinel */
2436 size = convert_mbs_to_wcs(pattern, cpattern, csize, mbs_offset, is_binary);
2448 DEBUG_PRINT1 ("\nCompiling pattern: ");
2451 unsigned debug_count;
2453 for (debug_count = 0; debug_count < size; debug_count++)
2454 PUT_CHAR (pattern[debug_count]);
2459 /* Initialize the compile stack. */
2460 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2461 if (compile_stack.stack == NULL)
2471 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2472 compile_stack.avail = 0;
2474 /* Initialize the pattern buffer. */
2475 bufp->syntax = syntax;
2476 bufp->fastmap_accurate = 0;
2477 bufp->not_bol = bufp->not_eol = 0;
2479 /* Set `used' to zero, so that if we return an error, the pattern
2480 printer (for debugging) will think there's no pattern. We reset it
2484 /* Always count groups, whether or not bufp->no_sub is set. */
2487 #if !defined emacs && !defined SYNTAX_TABLE
2488 /* Initialize the syntax table. */
2489 init_syntax_once ();
2492 if (bufp->allocated == 0)
2495 { /* If zero allocated, but buffer is non-null, try to realloc
2496 enough space. This loses if buffer's address is bogus, but
2497 that is the user's responsibility. */
2499 /* Free bufp->buffer and allocate an array for wchar_t pattern
2502 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE/sizeof(UCHAR_T),
2505 RETALLOC (COMPILED_BUFFER_VAR, INIT_BUF_SIZE, UCHAR_T);
2509 { /* Caller did not allocate a buffer. Do it for them. */
2510 COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE / sizeof(UCHAR_T),
2514 if (!COMPILED_BUFFER_VAR) FREE_STACK_RETURN (REG_ESPACE);
2516 bufp->buffer = (char*)COMPILED_BUFFER_VAR;
2518 bufp->allocated = INIT_BUF_SIZE;
2522 COMPILED_BUFFER_VAR = (UCHAR_T*) bufp->buffer;
2525 begalt = b = COMPILED_BUFFER_VAR;
2527 /* Loop through the uncompiled pattern until we're at the end. */
2536 if ( /* If at start of pattern, it's an operator. */
2538 /* If context independent, it's an operator. */
2539 || syntax & RE_CONTEXT_INDEP_ANCHORS
2540 /* Otherwise, depends on what's come before. */
2541 || PREFIX(at_begline_loc_p) (pattern, p, syntax))
2551 if ( /* If at end of pattern, it's an operator. */
2553 /* If context independent, it's an operator. */
2554 || syntax & RE_CONTEXT_INDEP_ANCHORS
2555 /* Otherwise, depends on what's next. */
2556 || PREFIX(at_endline_loc_p) (p, pend, syntax))
2566 if ((syntax & RE_BK_PLUS_QM)
2567 || (syntax & RE_LIMITED_OPS))
2571 /* If there is no previous pattern... */
2574 if (syntax & RE_CONTEXT_INVALID_OPS)
2575 FREE_STACK_RETURN (REG_BADRPT);
2576 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2581 /* Are we optimizing this jump? */
2582 boolean keep_string_p = false;
2584 /* 1 means zero (many) matches is allowed. */
2585 char zero_times_ok = 0, many_times_ok = 0;
2587 /* If there is a sequence of repetition chars, collapse it
2588 down to just one (the right one). We can't combine
2589 interval operators with these because of, e.g., `a{2}*',
2590 which should only match an even number of `a's. */
2594 zero_times_ok |= c != '+';
2595 many_times_ok |= c != '?';
2603 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2606 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2608 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2611 if (!(c1 == '+' || c1 == '?'))
2626 /* If we get here, we found another repeat character. */
2629 /* Star, etc. applied to an empty pattern is equivalent
2630 to an empty pattern. */
2634 /* Now we know whether or not zero matches is allowed
2635 and also whether or not two or more matches is allowed. */
2637 { /* More than one repetition is allowed, so put in at the
2638 end a backward relative jump from `b' to before the next
2639 jump we're going to put in below (which jumps from
2640 laststart to after this jump).
2642 But if we are at the `*' in the exact sequence `.*\n',
2643 insert an unconditional jump backwards to the .,
2644 instead of the beginning of the loop. This way we only
2645 push a failure point once, instead of every time
2646 through the loop. */
2647 assert (p - 1 > pattern);
2649 /* Allocate the space for the jump. */
2650 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2652 /* We know we are not at the first character of the pattern,
2653 because laststart was nonzero. And we've already
2654 incremented `p', by the way, to be the character after
2655 the `*'. Do we have to do something analogous here
2656 for null bytes, because of RE_DOT_NOT_NULL? */
2657 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2659 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2660 && !(syntax & RE_DOT_NEWLINE))
2661 { /* We have .*\n. */
2662 STORE_JUMP (jump, b, laststart);
2663 keep_string_p = true;
2666 /* Anything else. */
2667 STORE_JUMP (maybe_pop_jump, b, laststart -
2668 (1 + OFFSET_ADDRESS_SIZE));
2670 /* We've added more stuff to the buffer. */
2671 b += 1 + OFFSET_ADDRESS_SIZE;
2674 /* On failure, jump from laststart to b + 3, which will be the
2675 end of the buffer after this jump is inserted. */
2676 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
2678 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2679 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2681 laststart, b + 1 + OFFSET_ADDRESS_SIZE);
2683 b += 1 + OFFSET_ADDRESS_SIZE;
2687 /* At least one repetition is required, so insert a
2688 `dummy_failure_jump' before the initial
2689 `on_failure_jump' instruction of the loop. This
2690 effects a skip over that instruction the first time
2691 we hit that loop. */
2692 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
2693 INSERT_JUMP (dummy_failure_jump, laststart, laststart +
2694 2 + 2 * OFFSET_ADDRESS_SIZE);
2695 b += 1 + OFFSET_ADDRESS_SIZE;
2709 boolean had_char_class = false;
2711 CHAR_T range_start = 0xffffffff;
2713 unsigned int range_start = 0xffffffff;
2715 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2718 /* We assume a charset(_not) structure as a wchar_t array.
2719 charset[0] = (re_opcode_t) charset(_not)
2720 charset[1] = l (= length of char_classes)
2721 charset[2] = m (= length of collating_symbols)
2722 charset[3] = n (= length of equivalence_classes)
2723 charset[4] = o (= length of char_ranges)
2724 charset[5] = p (= length of chars)
2726 charset[6] = char_class (wctype_t)
2727 charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
2729 charset[l+5] = char_class (wctype_t)
2731 charset[l+6] = collating_symbol (wchar_t)
2733 charset[l+m+5] = collating_symbol (wchar_t)
2734 ifdef _LIBC we use the index if
2735 _NL_COLLATE_SYMB_EXTRAMB instead of
2738 charset[l+m+6] = equivalence_classes (wchar_t)
2740 charset[l+m+n+5] = equivalence_classes (wchar_t)
2741 ifdef _LIBC we use the index in
2742 _NL_COLLATE_WEIGHT instead of
2745 charset[l+m+n+6] = range_start
2746 charset[l+m+n+7] = range_end
2748 charset[l+m+n+2o+4] = range_start
2749 charset[l+m+n+2o+5] = range_end
2750 ifdef _LIBC we use the value looked up
2751 in _NL_COLLATE_COLLSEQ instead of
2754 charset[l+m+n+2o+6] = char
2756 charset[l+m+n+2o+p+5] = char
2760 /* We need at least 6 spaces: the opcode, the length of
2761 char_classes, the length of collating_symbols, the length of
2762 equivalence_classes, the length of char_ranges, the length of
2764 GET_BUFFER_SPACE (6);
2766 /* Save b as laststart. And We use laststart as the pointer
2767 to the first element of the charset here.
2768 In other words, laststart[i] indicates charset[i]. */
2771 /* We test `*p == '^' twice, instead of using an if
2772 statement, so we only need one BUF_PUSH. */
2773 BUF_PUSH (*p == '^' ? charset_not : charset);
2777 /* Push the length of char_classes, the length of
2778 collating_symbols, the length of equivalence_classes, the
2779 length of char_ranges and the length of chars. */
2780 BUF_PUSH_3 (0, 0, 0);
2783 /* Remember the first position in the bracket expression. */
2786 /* charset_not matches newline according to a syntax bit. */
2787 if ((re_opcode_t) b[-6] == charset_not
2788 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2791 laststart[5]++; /* Update the length of characters */
2794 /* Read in characters and ranges, setting map bits. */
2797 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2801 /* \ might escape characters inside [...] and [^...]. */
2802 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2804 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2808 laststart[5]++; /* Update the length of chars */
2813 /* Could be the end of the bracket expression. If it's
2814 not (i.e., when the bracket expression is `[]' so
2815 far), the ']' character bit gets set way below. */
2816 if (c == ']' && p != p1 + 1)
2819 /* Look ahead to see if it's a range when the last thing
2820 was a character class. */
2821 if (had_char_class && c == '-' && *p != ']')
2822 FREE_STACK_RETURN (REG_ERANGE);
2824 /* Look ahead to see if it's a range when the last thing
2825 was a character: if this is a hyphen not at the
2826 beginning or the end of a list, then it's the range
2829 && !(p - 2 >= pattern && p[-2] == '[')
2830 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2834 /* Allocate the space for range_start and range_end. */
2835 GET_BUFFER_SPACE (2);
2836 /* Update the pointer to indicate end of buffer. */
2838 ret = wcs_compile_range (range_start, &p, pend, translate,
2839 syntax, b, laststart);
2840 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2841 range_start = 0xffffffff;
2843 else if (p[0] == '-' && p[1] != ']')
2844 { /* This handles ranges made up of characters only. */
2847 /* Move past the `-'. */
2849 /* Allocate the space for range_start and range_end. */
2850 GET_BUFFER_SPACE (2);
2851 /* Update the pointer to indicate end of buffer. */
2853 ret = wcs_compile_range (c, &p, pend, translate, syntax, b,
2855 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2856 range_start = 0xffffffff;
2859 /* See if we're at the beginning of a possible character
2861 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2862 { /* Leave room for the null. */
2863 char str[CHAR_CLASS_MAX_LENGTH + 1];
2868 /* If pattern is `[[:'. */
2869 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2874 if ((c == ':' && *p == ']') || p == pend)
2876 if (c1 < CHAR_CLASS_MAX_LENGTH)
2879 /* This is in any case an invalid class name. */
2884 /* If isn't a word bracketed by `[:' and `:]':
2885 undo the ending character, the letters, and leave
2886 the leading `:' and `[' (but store them as character). */
2887 if (c == ':' && *p == ']')
2892 /* Query the character class as wctype_t. */
2893 wt = IS_CHAR_CLASS (str);
2895 FREE_STACK_RETURN (REG_ECTYPE);
2897 /* Throw away the ] at the end of the character
2901 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2903 /* Allocate the space for character class. */
2904 GET_BUFFER_SPACE(CHAR_CLASS_SIZE);
2905 /* Update the pointer to indicate end of buffer. */
2906 b += CHAR_CLASS_SIZE;
2907 /* Move data which follow character classes
2908 not to violate the data. */
2909 insert_space(CHAR_CLASS_SIZE,
2910 laststart + 6 + laststart[1],
2912 alignedp = ((uintptr_t)(laststart + 6 + laststart[1])
2913 + __alignof__(wctype_t) - 1)
2914 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
2915 /* Store the character class. */
2916 *((wctype_t*)alignedp) = wt;
2917 /* Update length of char_classes */
2918 laststart[1] += CHAR_CLASS_SIZE;
2920 had_char_class = true;
2929 laststart[5] += 2; /* Update the length of characters */
2931 had_char_class = false;
2934 else if (syntax & RE_CHAR_CLASSES && c == '[' && (*p == '='
2937 CHAR_T str[128]; /* Should be large enough. */
2938 CHAR_T delim = *p; /* '=' or '.' */
2941 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2946 /* If pattern is `[[=' or '[[.'. */
2947 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2952 if ((c == delim && *p == ']') || p == pend)
2954 if (c1 < sizeof (str) - 1)
2957 /* This is in any case an invalid class name. */
2962 if (c == delim && *p == ']' && str[0] != '\0')
2964 unsigned int i, offset;
2965 /* If we have no collation data we use the default
2966 collation in which each character is in a class
2967 by itself. It also means that ASCII is the
2968 character set and therefore we cannot have character
2969 with more than one byte in the multibyte
2972 /* If not defined _LIBC, we push the name and
2973 `\0' for the sake of matching performance. */
2974 int datasize = c1 + 1;
2982 FREE_STACK_RETURN (REG_ECOLLATE);
2987 const int32_t *table;
2988 const int32_t *weights;
2989 const int32_t *extra;
2990 const int32_t *indirect;
2993 /* This #include defines a local function! */
2994 # include <locale/weightwc.h>
2998 /* We push the index for equivalence class. */
3001 table = (const int32_t *)
3002 _NL_CURRENT (LC_COLLATE,
3003 _NL_COLLATE_TABLEWC);
3004 weights = (const int32_t *)
3005 _NL_CURRENT (LC_COLLATE,
3006 _NL_COLLATE_WEIGHTWC);
3007 extra = (const int32_t *)
3008 _NL_CURRENT (LC_COLLATE,
3009 _NL_COLLATE_EXTRAWC);
3010 indirect = (const int32_t *)
3011 _NL_CURRENT (LC_COLLATE,
3012 _NL_COLLATE_INDIRECTWC);
3014 idx = findidx ((const wint_t**)&cp);
3015 if (idx == 0 || cp < (wint_t*) str + c1)
3016 /* This is no valid character. */
3017 FREE_STACK_RETURN (REG_ECOLLATE);
3019 str[0] = (wchar_t)idx;
3021 else /* delim == '.' */
3023 /* We push collation sequence value
3024 for collating symbol. */
3026 const int32_t *symb_table;
3027 const unsigned char *extra;
3034 /* We have to convert the name to a single-byte
3035 string. This is possible since the names
3036 consist of ASCII characters and the internal
3037 representation is UCS4. */
3038 for (i = 0; i < c1; ++i)
3039 char_str[i] = str[i];
3042 _NL_CURRENT_WORD (LC_COLLATE,
3043 _NL_COLLATE_SYMB_HASH_SIZEMB);
3044 symb_table = (const int32_t *)
3045 _NL_CURRENT (LC_COLLATE,
3046 _NL_COLLATE_SYMB_TABLEMB);
3047 extra = (const unsigned char *)
3048 _NL_CURRENT (LC_COLLATE,
3049 _NL_COLLATE_SYMB_EXTRAMB);
3051 /* Locate the character in the hashing table. */
3052 hash = elem_hash (char_str, c1);
3055 elem = hash % table_size;
3056 second = hash % (table_size - 2);
3057 while (symb_table[2 * elem] != 0)
3059 /* First compare the hashing value. */
3060 if (symb_table[2 * elem] == hash
3061 && c1 == extra[symb_table[2 * elem + 1]]
3062 && memcmp (char_str,
3063 &extra[symb_table[2 * elem + 1]
3066 /* Yep, this is the entry. */
3067 idx = symb_table[2 * elem + 1];
3068 idx += 1 + extra[idx];
3076 if (symb_table[2 * elem] != 0)
3078 /* Compute the index of the byte sequence
3080 idx += 1 + extra[idx];
3081 /* Adjust for the alignment. */
3082 idx = (idx + 3) & ~3;
3084 str[0] = (wchar_t) idx + 4;
3086 else if (symb_table[2 * elem] == 0 && c1 == 1)
3088 /* No valid character. Match it as a
3089 single byte character. */
3090 had_char_class = false;
3092 /* Update the length of characters */
3094 range_start = str[0];
3096 /* Throw away the ] at the end of the
3097 collating symbol. */
3099 /* exit from the switch block. */
3103 FREE_STACK_RETURN (REG_ECOLLATE);
3108 /* Throw away the ] at the end of the equivalence
3109 class (or collating symbol). */
3112 /* Allocate the space for the equivalence class
3113 (or collating symbol) (and '\0' if needed). */
3114 GET_BUFFER_SPACE(datasize);
3115 /* Update the pointer to indicate end of buffer. */
3119 { /* equivalence class */
3120 /* Calculate the offset of char_ranges,
3121 which is next to equivalence_classes. */
3122 offset = laststart[1] + laststart[2]
3125 insert_space(datasize, laststart + offset, b - 1);
3127 /* Write the equivalence_class and \0. */
3128 for (i = 0 ; i < datasize ; i++)
3129 laststart[offset + i] = str[i];
3131 /* Update the length of equivalence_classes. */
3132 laststart[3] += datasize;
3133 had_char_class = true;
3135 else /* delim == '.' */
3136 { /* collating symbol */
3137 /* Calculate the offset of the equivalence_classes,
3138 which is next to collating_symbols. */
3139 offset = laststart[1] + laststart[2] + 6;
3140 /* Insert space and write the collationg_symbol
3142 insert_space(datasize, laststart + offset, b-1);
3143 for (i = 0 ; i < datasize ; i++)
3144 laststart[offset + i] = str[i];
3146 /* In re_match_2_internal if range_start < -1, we
3147 assume -range_start is the offset of the
3148 collating symbol which is specified as
3149 the character of the range start. So we assign
3150 -(laststart[1] + laststart[2] + 6) to
3152 range_start = -(laststart[1] + laststart[2] + 6);
3153 /* Update the length of collating_symbol. */
3154 laststart[2] += datasize;
3155 had_char_class = false;
3165 laststart[5] += 2; /* Update the length of characters */
3166 range_start = delim;
3167 had_char_class = false;
3172 had_char_class = false;
3174 laststart[5]++; /* Update the length of characters */
3180 /* Ensure that we have enough space to push a charset: the
3181 opcode, the length count, and the bitset; 34 bytes in all. */
3182 GET_BUFFER_SPACE (34);
3186 /* We test `*p == '^' twice, instead of using an if
3187 statement, so we only need one BUF_PUSH. */
3188 BUF_PUSH (*p == '^' ? charset_not : charset);
3192 /* Remember the first position in the bracket expression. */
3195 /* Push the number of bytes in the bitmap. */
3196 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
3198 /* Clear the whole map. */
3199 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
3201 /* charset_not matches newline according to a syntax bit. */
3202 if ((re_opcode_t) b[-2] == charset_not
3203 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
3204 SET_LIST_BIT ('\n');
3206 /* Read in characters and ranges, setting map bits. */
3209 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3213 /* \ might escape characters inside [...] and [^...]. */
3214 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
3216 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3224 /* Could be the end of the bracket expression. If it's
3225 not (i.e., when the bracket expression is `[]' so
3226 far), the ']' character bit gets set way below. */
3227 if (c == ']' && p != p1 + 1)
3230 /* Look ahead to see if it's a range when the last thing
3231 was a character class. */
3232 if (had_char_class && c == '-' && *p != ']')
3233 FREE_STACK_RETURN (REG_ERANGE);
3235 /* Look ahead to see if it's a range when the last thing
3236 was a character: if this is a hyphen not at the
3237 beginning or the end of a list, then it's the range
3240 && !(p - 2 >= pattern && p[-2] == '[')
3241 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
3245 = byte_compile_range (range_start, &p, pend, translate,
3247 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3248 range_start = 0xffffffff;
3251 else if (p[0] == '-' && p[1] != ']')
3252 { /* This handles ranges made up of characters only. */
3255 /* Move past the `-'. */
3258 ret = byte_compile_range (c, &p, pend, translate, syntax, b);
3259 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
3260 range_start = 0xffffffff;
3263 /* See if we're at the beginning of a possible character
3266 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
3267 { /* Leave room for the null. */
3268 char str[CHAR_CLASS_MAX_LENGTH + 1];
3273 /* If pattern is `[[:'. */
3274 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3279 if ((c == ':' && *p == ']') || p == pend)
3281 if (c1 < CHAR_CLASS_MAX_LENGTH)
3284 /* This is in any case an invalid class name. */
3289 /* If isn't a word bracketed by `[:' and `:]':
3290 undo the ending character, the letters, and leave
3291 the leading `:' and `[' (but set bits for them). */
3292 if (c == ':' && *p == ']')
3294 # if defined _LIBC || WIDE_CHAR_SUPPORT
3295 boolean is_lower = STREQ (str, "lower");
3296 boolean is_upper = STREQ (str, "upper");
3300 wt = IS_CHAR_CLASS (str);
3302 FREE_STACK_RETURN (REG_ECTYPE);
3304 /* Throw away the ] at the end of the character
3308 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3310 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
3313 if (__iswctype (__btowc (ch), wt))
3316 if (iswctype (btowc (ch), wt))
3320 if (translate && (is_upper || is_lower)
3321 && (ISUPPER (ch) || ISLOWER (ch)))
3325 had_char_class = true;
3328 boolean is_alnum = STREQ (str, "alnum");
3329 boolean is_alpha = STREQ (str, "alpha");
3330 boolean is_blank = STREQ (str, "blank");
3331 boolean is_cntrl = STREQ (str, "cntrl");
3332 boolean is_digit = STREQ (str, "digit");
3333 boolean is_graph = STREQ (str, "graph");
3334 boolean is_lower = STREQ (str, "lower");
3335 boolean is_print = STREQ (str, "print");
3336 boolean is_punct = STREQ (str, "punct");
3337 boolean is_space = STREQ (str, "space");
3338 boolean is_upper = STREQ (str, "upper");
3339 boolean is_xdigit = STREQ (str, "xdigit");
3341 if (!IS_CHAR_CLASS (str))
3342 FREE_STACK_RETURN (REG_ECTYPE);
3344 /* Throw away the ] at the end of the character
3348 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3350 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
3352 /* This was split into 3 if's to
3353 avoid an arbitrary limit in some compiler. */
3354 if ( (is_alnum && ISALNUM (ch))
3355 || (is_alpha && ISALPHA (ch))
3356 || (is_blank && ISBLANK (ch))
3357 || (is_cntrl && ISCNTRL (ch)))
3359 if ( (is_digit && ISDIGIT (ch))
3360 || (is_graph && ISGRAPH (ch))
3361 || (is_lower && ISLOWER (ch))
3362 || (is_print && ISPRINT (ch)))
3364 if ( (is_punct && ISPUNCT (ch))
3365 || (is_space && ISSPACE (ch))
3366 || (is_upper && ISUPPER (ch))
3367 || (is_xdigit && ISXDIGIT (ch)))
3369 if ( translate && (is_upper || is_lower)
3370 && (ISUPPER (ch) || ISLOWER (ch)))
3373 had_char_class = true;
3374 # endif /* libc || wctype.h */
3384 had_char_class = false;
3387 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
3389 unsigned char str[MB_LEN_MAX + 1];
3392 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3398 /* If pattern is `[[='. */
3399 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3404 if ((c == '=' && *p == ']') || p == pend)
3406 if (c1 < MB_LEN_MAX)
3409 /* This is in any case an invalid class name. */
3414 if (c == '=' && *p == ']' && str[0] != '\0')
3416 /* If we have no collation data we use the default
3417 collation in which each character is in a class
3418 by itself. It also means that ASCII is the
3419 character set and therefore we cannot have character
3420 with more than one byte in the multibyte
3427 FREE_STACK_RETURN (REG_ECOLLATE);
3429 /* Throw away the ] at the end of the equivalence
3433 /* Set the bit for the character. */
3434 SET_LIST_BIT (str[0]);
3439 /* Try to match the byte sequence in `str' against
3440 those known to the collate implementation.
3441 First find out whether the bytes in `str' are
3442 actually from exactly one character. */
3443 const int32_t *table;
3444 const unsigned char *weights;
3445 const unsigned char *extra;
3446 const int32_t *indirect;
3448 const unsigned char *cp = str;
3451 /* This #include defines a local function! */
3452 # include <locale/weight.h>
3454 table = (const int32_t *)
3455 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
3456 weights = (const unsigned char *)
3457 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
3458 extra = (const unsigned char *)
3459 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
3460 indirect = (const int32_t *)
3461 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
3463 idx = findidx (&cp);
3464 if (idx == 0 || cp < str + c1)
3465 /* This is no valid character. */
3466 FREE_STACK_RETURN (REG_ECOLLATE);
3468 /* Throw away the ] at the end of the equivalence
3472 /* Now we have to go throught the whole table
3473 and find all characters which have the same
3476 XXX Note that this is not entirely correct.
3477 we would have to match multibyte sequences
3478 but this is not possible with the current
3480 for (ch = 1; ch < 256; ++ch)
3481 /* XXX This test would have to be changed if we
3482 would allow matching multibyte sequences. */
3485 int32_t idx2 = table[ch];
3486 size_t len = weights[idx2];
3488 /* Test whether the lenghts match. */
3489 if (weights[idx] == len)
3491 /* They do. New compare the bytes of
3496 && (weights[idx + 1 + cnt]
3497 == weights[idx2 + 1 + cnt]))
3501 /* They match. Mark the character as
3508 had_char_class = true;
3518 had_char_class = false;
3521 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
3523 unsigned char str[128]; /* Should be large enough. */
3526 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
3532 /* If pattern is `[[.'. */
3533 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
3538 if ((c == '.' && *p == ']') || p == pend)
3540 if (c1 < sizeof (str))
3543 /* This is in any case an invalid class name. */
3548 if (c == '.' && *p == ']' && str[0] != '\0')
3550 /* If we have no collation data we use the default
3551 collation in which each character is the name
3552 for its own class which contains only the one
3553 character. It also means that ASCII is the
3554 character set and therefore we cannot have character
3555 with more than one byte in the multibyte
3562 FREE_STACK_RETURN (REG_ECOLLATE);
3564 /* Throw away the ] at the end of the equivalence
3568 /* Set the bit for the character. */
3569 SET_LIST_BIT (str[0]);
3570 range_start = ((const unsigned char *) str)[0];
3575 /* Try to match the byte sequence in `str' against
3576 those known to the collate implementation.
3577 First find out whether the bytes in `str' are
3578 actually from exactly one character. */
3580 const int32_t *symb_table;
3581 const unsigned char *extra;
3588 _NL_CURRENT_WORD (LC_COLLATE,
3589 _NL_COLLATE_SYMB_HASH_SIZEMB);
3590 symb_table = (const int32_t *)
3591 _NL_CURRENT (LC_COLLATE,
3592 _NL_COLLATE_SYMB_TABLEMB);
3593 extra = (const unsigned char *)
3594 _NL_CURRENT (LC_COLLATE,
3595 _NL_COLLATE_SYMB_EXTRAMB);
3597 /* Locate the character in the hashing table. */
3598 hash = elem_hash (str, c1);
3601 elem = hash % table_size;
3602 second = hash % (table_size - 2);
3603 while (symb_table[2 * elem] != 0)
3605 /* First compare the hashing value. */
3606 if (symb_table[2 * elem] == hash
3607 && c1 == extra[symb_table[2 * elem + 1]]
3609 &extra[symb_table[2 * elem + 1]
3613 /* Yep, this is the entry. */
3614 idx = symb_table[2 * elem + 1];
3615 idx += 1 + extra[idx];
3623 if (symb_table[2 * elem] == 0)
3624 /* This is no valid character. */
3625 FREE_STACK_RETURN (REG_ECOLLATE);
3627 /* Throw away the ] at the end of the equivalence
3631 /* Now add the multibyte character(s) we found
3634 XXX Note that this is not entirely correct.
3635 we would have to match multibyte sequences
3636 but this is not possible with the current
3637 implementation. Also, we have to match
3638 collating symbols, which expand to more than
3639 one file, as a whole and not allow the
3640 individual bytes. */
3643 range_start = extra[idx];
3646 SET_LIST_BIT (extra[idx]);
3651 had_char_class = false;
3661 had_char_class = false;
3666 had_char_class = false;
3672 /* Discard any (non)matching list bytes that are all 0 at the
3673 end of the map. Decrease the map-length byte too. */
3674 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
3683 if (syntax & RE_NO_BK_PARENS)
3690 if (syntax & RE_NO_BK_PARENS)
3697 if (syntax & RE_NEWLINE_ALT)
3704 if (syntax & RE_NO_BK_VBAR)
3711 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
3712 goto handle_interval;
3718 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
3720 /* Do not translate the character after the \, so that we can
3721 distinguish, e.g., \B from \b, even if we normally would
3722 translate, e.g., B to b. */
3728 if (syntax & RE_NO_BK_PARENS)
3729 goto normal_backslash;
3735 if (COMPILE_STACK_FULL)
3737 RETALLOC (compile_stack.stack, compile_stack.size << 1,
3738 compile_stack_elt_t);
3739 if (compile_stack.stack == NULL) return REG_ESPACE;
3741 compile_stack.size <<= 1;
3744 /* These are the values to restore when we hit end of this
3745 group. They are all relative offsets, so that if the
3746 whole pattern moves because of realloc, they will still
3748 COMPILE_STACK_TOP.begalt_offset = begalt - COMPILED_BUFFER_VAR;
3749 COMPILE_STACK_TOP.fixup_alt_jump
3750 = fixup_alt_jump ? fixup_alt_jump - COMPILED_BUFFER_VAR + 1 : 0;
3751 COMPILE_STACK_TOP.laststart_offset = b - COMPILED_BUFFER_VAR;
3752 COMPILE_STACK_TOP.regnum = regnum;
3754 /* We will eventually replace the 0 with the number of
3755 groups inner to this one. But do not push a
3756 start_memory for groups beyond the last one we can
3757 represent in the compiled pattern. */
3758 if (regnum <= MAX_REGNUM)
3760 COMPILE_STACK_TOP.inner_group_offset = b
3761 - COMPILED_BUFFER_VAR + 2;
3762 BUF_PUSH_3 (start_memory, regnum, 0);
3765 compile_stack.avail++;
3770 /* If we've reached MAX_REGNUM groups, then this open
3771 won't actually generate any code, so we'll have to
3772 clear pending_exact explicitly. */
3778 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
3780 if (COMPILE_STACK_EMPTY)
3782 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3783 goto normal_backslash;
3785 FREE_STACK_RETURN (REG_ERPAREN);
3790 { /* Push a dummy failure point at the end of the
3791 alternative for a possible future
3792 `pop_failure_jump' to pop. See comments at
3793 `push_dummy_failure' in `re_match_2'. */
3794 BUF_PUSH (push_dummy_failure);
3796 /* We allocated space for this jump when we assigned
3797 to `fixup_alt_jump', in the `handle_alt' case below. */
3798 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
3801 /* See similar code for backslashed left paren above. */
3802 if (COMPILE_STACK_EMPTY)
3804 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
3807 FREE_STACK_RETURN (REG_ERPAREN);
3810 /* Since we just checked for an empty stack above, this
3811 ``can't happen''. */
3812 assert (compile_stack.avail != 0);
3814 /* We don't just want to restore into `regnum', because
3815 later groups should continue to be numbered higher,
3816 as in `(ab)c(de)' -- the second group is #2. */
3817 regnum_t this_group_regnum;
3819 compile_stack.avail--;
3820 begalt = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.begalt_offset;
3822 = COMPILE_STACK_TOP.fixup_alt_jump
3823 ? COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.fixup_alt_jump - 1
3825 laststart = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.laststart_offset;
3826 this_group_regnum = COMPILE_STACK_TOP.regnum;
3827 /* If we've reached MAX_REGNUM groups, then this open
3828 won't actually generate any code, so we'll have to
3829 clear pending_exact explicitly. */
3832 /* We're at the end of the group, so now we know how many
3833 groups were inside this one. */
3834 if (this_group_regnum <= MAX_REGNUM)
3836 UCHAR_T *inner_group_loc
3837 = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.inner_group_offset;
3839 *inner_group_loc = regnum - this_group_regnum;
3840 BUF_PUSH_3 (stop_memory, this_group_regnum,
3841 regnum - this_group_regnum);
3847 case '|': /* `\|'. */
3848 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
3849 goto normal_backslash;
3851 if (syntax & RE_LIMITED_OPS)
3854 /* Insert before the previous alternative a jump which
3855 jumps to this alternative if the former fails. */
3856 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3857 INSERT_JUMP (on_failure_jump, begalt,
3858 b + 2 + 2 * OFFSET_ADDRESS_SIZE);
3860 b += 1 + OFFSET_ADDRESS_SIZE;
3862 /* The alternative before this one has a jump after it
3863 which gets executed if it gets matched. Adjust that
3864 jump so it will jump to this alternative's analogous
3865 jump (put in below, which in turn will jump to the next
3866 (if any) alternative's such jump, etc.). The last such
3867 jump jumps to the correct final destination. A picture:
3873 If we are at `b', then fixup_alt_jump right now points to a
3874 three-byte space after `a'. We'll put in the jump, set
3875 fixup_alt_jump to right after `b', and leave behind three
3876 bytes which we'll fill in when we get to after `c'. */
3879 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
3881 /* Mark and leave space for a jump after this alternative,
3882 to be filled in later either by next alternative or
3883 when know we're at the end of a series of alternatives. */
3885 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3886 b += 1 + OFFSET_ADDRESS_SIZE;
3894 /* If \{ is a literal. */
3895 if (!(syntax & RE_INTERVALS)
3896 /* If we're at `\{' and it's not the open-interval
3898 || (syntax & RE_NO_BK_BRACES))
3899 goto normal_backslash;
3903 /* If got here, then the syntax allows intervals. */
3905 /* At least (most) this many matches must be made. */
3906 int lower_bound = -1, upper_bound = -1;
3908 /* Place in the uncompiled pattern (i.e., just after
3909 the '{') to go back to if the interval is invalid. */
3910 const CHAR_T *beg_interval = p;
3913 goto invalid_interval;
3915 GET_UNSIGNED_NUMBER (lower_bound);
3919 GET_UNSIGNED_NUMBER (upper_bound);
3920 if (upper_bound < 0)
3921 upper_bound = RE_DUP_MAX;
3924 /* Interval such as `{1}' => match exactly once. */
3925 upper_bound = lower_bound;
3927 if (! (0 <= lower_bound && lower_bound <= upper_bound))
3928 goto invalid_interval;
3930 if (!(syntax & RE_NO_BK_BRACES))
3932 if (c != '\\' || p == pend)
3933 goto invalid_interval;
3938 goto invalid_interval;
3940 /* If it's invalid to have no preceding re. */
3943 if (syntax & RE_CONTEXT_INVALID_OPS
3944 && !(syntax & RE_INVALID_INTERVAL_ORD))
3945 FREE_STACK_RETURN (REG_BADRPT);
3946 else if (syntax & RE_CONTEXT_INDEP_OPS)
3949 goto unfetch_interval;
3952 /* We just parsed a valid interval. */
3954 if (RE_DUP_MAX < upper_bound)
3955 FREE_STACK_RETURN (REG_BADBR);
3957 /* If the upper bound is zero, don't want to succeed at
3958 all; jump from `laststart' to `b + 3', which will be
3959 the end of the buffer after we insert the jump. */
3960 /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
3961 instead of 'b + 3'. */
3962 if (upper_bound == 0)
3964 GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
3965 INSERT_JUMP (jump, laststart, b + 1
3966 + OFFSET_ADDRESS_SIZE);
3967 b += 1 + OFFSET_ADDRESS_SIZE;
3970 /* Otherwise, we have a nontrivial interval. When
3971 we're all done, the pattern will look like:
3972 set_number_at <jump count> <upper bound>
3973 set_number_at <succeed_n count> <lower bound>
3974 succeed_n <after jump addr> <succeed_n count>
3976 jump_n <succeed_n addr> <jump count>
3977 (The upper bound and `jump_n' are omitted if
3978 `upper_bound' is 1, though.) */
3980 { /* If the upper bound is > 1, we need to insert
3981 more at the end of the loop. */
3982 unsigned nbytes = 2 + 4 * OFFSET_ADDRESS_SIZE +
3983 (upper_bound > 1) * (2 + 4 * OFFSET_ADDRESS_SIZE);
3985 GET_BUFFER_SPACE (nbytes);
3987 /* Initialize lower bound of the `succeed_n', even
3988 though it will be set during matching by its
3989 attendant `set_number_at' (inserted next),
3990 because `re_compile_fastmap' needs to know.
3991 Jump to the `jump_n' we might insert below. */
3992 INSERT_JUMP2 (succeed_n, laststart,
3993 b + 1 + 2 * OFFSET_ADDRESS_SIZE
3994 + (upper_bound > 1) * (1 + 2 * OFFSET_ADDRESS_SIZE)
3996 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
3998 /* Code to initialize the lower bound. Insert
3999 before the `succeed_n'. The `5' is the last two
4000 bytes of this `set_number_at', plus 3 bytes of
4001 the following `succeed_n'. */
4002 /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
4003 is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
4004 of the following `succeed_n'. */
4005 PREFIX(insert_op2) (set_number_at, laststart, 1
4006 + 2 * OFFSET_ADDRESS_SIZE, lower_bound, b);
4007 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4009 if (upper_bound > 1)
4010 { /* More than one repetition is allowed, so
4011 append a backward jump to the `succeed_n'
4012 that starts this interval.
4014 When we've reached this during matching,
4015 we'll have matched the interval once, so
4016 jump back only `upper_bound - 1' times. */
4017 STORE_JUMP2 (jump_n, b, laststart
4018 + 2 * OFFSET_ADDRESS_SIZE + 1,
4020 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4022 /* The location we want to set is the second
4023 parameter of the `jump_n'; that is `b-2' as
4024 an absolute address. `laststart' will be
4025 the `set_number_at' we're about to insert;
4026 `laststart+3' the number to set, the source
4027 for the relative address. But we are
4028 inserting into the middle of the pattern --
4029 so everything is getting moved up by 5.
4030 Conclusion: (b - 2) - (laststart + 3) + 5,
4031 i.e., b - laststart.
4033 We insert this at the beginning of the loop
4034 so that if we fail during matching, we'll
4035 reinitialize the bounds. */
4036 PREFIX(insert_op2) (set_number_at, laststart,
4038 upper_bound - 1, b);
4039 b += 1 + 2 * OFFSET_ADDRESS_SIZE;
4046 if (!(syntax & RE_INVALID_INTERVAL_ORD))
4047 FREE_STACK_RETURN (p == pend ? REG_EBRACE : REG_BADBR);
4049 /* Match the characters as literals. */
4052 if (syntax & RE_NO_BK_BRACES)
4055 goto normal_backslash;
4059 /* There is no way to specify the before_dot and after_dot
4060 operators. rms says this is ok. --karl */
4068 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
4074 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
4080 if (syntax & RE_NO_GNU_OPS)
4083 BUF_PUSH (wordchar);
4088 if (syntax & RE_NO_GNU_OPS)
4091 BUF_PUSH (notwordchar);
4096 if (syntax & RE_NO_GNU_OPS)
4102 if (syntax & RE_NO_GNU_OPS)
4108 if (syntax & RE_NO_GNU_OPS)
4110 BUF_PUSH (wordbound);
4114 if (syntax & RE_NO_GNU_OPS)
4116 BUF_PUSH (notwordbound);
4120 if (syntax & RE_NO_GNU_OPS)
4126 if (syntax & RE_NO_GNU_OPS)
4131 case '1': case '2': case '3': case '4': case '5':
4132 case '6': case '7': case '8': case '9':
4133 if (syntax & RE_NO_BK_REFS)
4139 FREE_STACK_RETURN (REG_ESUBREG);
4141 /* Can't back reference to a subexpression if inside of it. */
4142 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
4146 BUF_PUSH_2 (duplicate, c1);
4152 if (syntax & RE_BK_PLUS_QM)
4155 goto normal_backslash;
4159 /* You might think it would be useful for \ to mean
4160 not to translate; but if we don't translate it
4161 it will never match anything. */
4169 /* Expects the character in `c'. */
4171 /* If no exactn currently being built. */
4174 /* If last exactn handle binary(or character) and
4175 new exactn handle character(or binary). */
4176 || is_exactn_bin != is_binary[p - 1 - pattern]
4179 /* If last exactn not at current position. */
4180 || pending_exact + *pending_exact + 1 != b
4182 /* We have only one byte following the exactn for the count. */
4183 || *pending_exact == (1 << BYTEWIDTH) - 1
4185 /* If followed by a repetition operator. */
4186 || *p == '*' || *p == '^'
4187 || ((syntax & RE_BK_PLUS_QM)
4188 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
4189 : (*p == '+' || *p == '?'))
4190 || ((syntax & RE_INTERVALS)
4191 && ((syntax & RE_NO_BK_BRACES)
4193 : (p[0] == '\\' && p[1] == '{'))))
4195 /* Start building a new exactn. */
4200 /* Is this exactn binary data or character? */
4201 is_exactn_bin = is_binary[p - 1 - pattern];
4203 BUF_PUSH_2 (exactn_bin, 0);
4205 BUF_PUSH_2 (exactn, 0);
4207 BUF_PUSH_2 (exactn, 0);
4209 pending_exact = b - 1;
4216 } /* while p != pend */
4219 /* Through the pattern now. */
4222 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
4224 if (!COMPILE_STACK_EMPTY)
4225 FREE_STACK_RETURN (REG_EPAREN);
4227 /* If we don't want backtracking, force success
4228 the first time we reach the end of the compiled pattern. */
4229 if (syntax & RE_NO_POSIX_BACKTRACKING)
4237 free (compile_stack.stack);
4239 /* We have succeeded; set the length of the buffer. */
4241 bufp->used = (uintptr_t) b - (uintptr_t) COMPILED_BUFFER_VAR;
4243 bufp->used = b - bufp->buffer;
4249 DEBUG_PRINT1 ("\nCompiled pattern: \n");
4250 PREFIX(print_compiled_pattern) (bufp);
4254 #ifndef MATCH_MAY_ALLOCATE
4255 /* Initialize the failure stack to the largest possible stack. This
4256 isn't necessary unless we're trying to avoid calling alloca in
4257 the search and match routines. */
4259 int num_regs = bufp->re_nsub + 1;
4261 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
4262 is strictly greater than re_max_failures, the largest possible stack
4263 is 2 * re_max_failures failure points. */
4264 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
4266 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
4269 if (! fail_stack.stack)
4271 = (PREFIX(fail_stack_elt_t) *) xmalloc (fail_stack.size
4272 * sizeof (PREFIX(fail_stack_elt_t)));
4275 = (PREFIX(fail_stack_elt_t) *) xrealloc (fail_stack.stack,
4277 * sizeof (PREFIX(fail_stack_elt_t))));
4278 # else /* not emacs */
4279 if (! fail_stack.stack)
4281 = (PREFIX(fail_stack_elt_t) *) malloc (fail_stack.size
4282 * sizeof (PREFIX(fail_stack_elt_t)));
4285 = (PREFIX(fail_stack_elt_t) *) realloc (fail_stack.stack,
4287 * sizeof (PREFIX(fail_stack_elt_t))));
4288 # endif /* not emacs */
4291 PREFIX(regex_grow_registers) (num_regs);
4293 #endif /* not MATCH_MAY_ALLOCATE */
4296 } /* regex_compile */
4298 /* Subroutines for `regex_compile'. */
4300 /* Store OP at LOC followed by two-byte integer parameter ARG. */
4301 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4304 PREFIX(store_op1) (op, loc, arg)
4309 *loc = (UCHAR_T) op;
4310 STORE_NUMBER (loc + 1, arg);
4314 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
4315 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4318 PREFIX(store_op2) (op, loc, arg1, arg2)
4323 *loc = (UCHAR_T) op;
4324 STORE_NUMBER (loc + 1, arg1);
4325 STORE_NUMBER (loc + 1 + OFFSET_ADDRESS_SIZE, arg2);
4329 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
4330 for OP followed by two-byte integer parameter ARG. */
4331 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4334 PREFIX(insert_op1) (op, loc, arg, end)
4340 register UCHAR_T *pfrom = end;
4341 register UCHAR_T *pto = end + 1 + OFFSET_ADDRESS_SIZE;
4343 while (pfrom != loc)
4346 PREFIX(store_op1) (op, loc, arg);
4350 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
4351 /* ifdef WCHAR, integer parameter is 1 wchar_t. */
4354 PREFIX(insert_op2) (op, loc, arg1, arg2, end)
4360 register UCHAR_T *pfrom = end;
4361 register UCHAR_T *pto = end + 1 + 2 * OFFSET_ADDRESS_SIZE;
4363 while (pfrom != loc)
4366 PREFIX(store_op2) (op, loc, arg1, arg2);
4370 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
4371 after an alternative or a begin-subexpression. We assume there is at
4372 least one character before the ^. */
4375 PREFIX(at_begline_loc_p) (pattern, p, syntax)
4376 const CHAR_T *pattern, *p;
4377 reg_syntax_t syntax;
4379 const CHAR_T *prev = p - 2;
4380 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
4383 /* After a subexpression? */
4384 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
4385 /* After an alternative? */
4386 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
4390 /* The dual of at_begline_loc_p. This one is for $. We assume there is
4391 at least one character after the $, i.e., `P < PEND'. */
4394 PREFIX(at_endline_loc_p) (p, pend, syntax)
4395 const CHAR_T *p, *pend;
4396 reg_syntax_t syntax;
4398 const CHAR_T *next = p;
4399 boolean next_backslash = *next == '\\';
4400 const CHAR_T *next_next = p + 1 < pend ? p + 1 : 0;
4403 /* Before a subexpression? */
4404 (syntax & RE_NO_BK_PARENS ? *next == ')'
4405 : next_backslash && next_next && *next_next == ')')
4406 /* Before an alternative? */
4407 || (syntax & RE_NO_BK_VBAR ? *next == '|'
4408 : next_backslash && next_next && *next_next == '|');
4411 #else /* not INSIDE_RECURSION */
4413 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
4414 false if it's not. */
4417 group_in_compile_stack (compile_stack, regnum)
4418 compile_stack_type compile_stack;
4423 for (this_element = compile_stack.avail - 1;
4426 if (compile_stack.stack[this_element].regnum == regnum)
4431 #endif /* not INSIDE_RECURSION */
4433 #ifdef INSIDE_RECURSION
4436 /* This insert space, which size is "num", into the pattern at "loc".
4437 "end" must point the end of the allocated buffer. */
4439 insert_space (num, loc, end)
4444 register CHAR_T *pto = end;
4445 register CHAR_T *pfrom = end - num;
4447 while (pfrom >= loc)
4453 static reg_errcode_t
4454 wcs_compile_range (range_start_char, p_ptr, pend, translate, syntax, b,
4456 CHAR_T range_start_char;
4457 const CHAR_T **p_ptr, *pend;
4458 CHAR_T *char_set, *b;
4459 RE_TRANSLATE_TYPE translate;
4460 reg_syntax_t syntax;
4462 const CHAR_T *p = *p_ptr;
4463 CHAR_T range_start, range_end;
4467 uint32_t start_val, end_val;
4473 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
4476 const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE,
4477 _NL_COLLATE_COLLSEQWC);
4478 const unsigned char *extra = (const unsigned char *)
4479 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
4481 if (range_start_char < -1)
4483 /* range_start is a collating symbol. */
4485 /* Retreive the index and get collation sequence value. */
4486 wextra = (int32_t*)(extra + char_set[-range_start_char]);
4487 start_val = wextra[1 + *wextra];
4490 start_val = collseq_table_lookup(collseq, TRANSLATE(range_start_char));
4492 end_val = collseq_table_lookup (collseq, TRANSLATE (p[0]));
4494 /* Report an error if the range is empty and the syntax prohibits
4496 ret = ((syntax & RE_NO_EMPTY_RANGES)
4497 && (start_val > end_val))? REG_ERANGE : REG_NOERROR;
4499 /* Insert space to the end of the char_ranges. */
4500 insert_space(2, b - char_set[5] - 2, b - 1);
4501 *(b - char_set[5] - 2) = (wchar_t)start_val;
4502 *(b - char_set[5] - 1) = (wchar_t)end_val;
4503 char_set[4]++; /* ranges_index */
4508 range_start = (range_start_char >= 0)? TRANSLATE (range_start_char):
4510 range_end = TRANSLATE (p[0]);
4511 /* Report an error if the range is empty and the syntax prohibits
4513 ret = ((syntax & RE_NO_EMPTY_RANGES)
4514 && (range_start > range_end))? REG_ERANGE : REG_NOERROR;
4516 /* Insert space to the end of the char_ranges. */
4517 insert_space(2, b - char_set[5] - 2, b - 1);
4518 *(b - char_set[5] - 2) = range_start;
4519 *(b - char_set[5] - 1) = range_end;
4520 char_set[4]++; /* ranges_index */
4522 /* Have to increment the pointer into the pattern string, so the
4523 caller isn't still at the ending character. */
4529 /* Read the ending character of a range (in a bracket expression) from the
4530 uncompiled pattern *P_PTR (which ends at PEND). We assume the
4531 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
4532 Then we set the translation of all bits between the starting and
4533 ending characters (inclusive) in the compiled pattern B.
4535 Return an error code.
4537 We use these short variable names so we can use the same macros as
4538 `regex_compile' itself. */
4540 static reg_errcode_t
4541 byte_compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
4542 unsigned int range_start_char;
4543 const char **p_ptr, *pend;
4544 RE_TRANSLATE_TYPE translate;
4545 reg_syntax_t syntax;
4549 const char *p = *p_ptr;
4552 const unsigned char *collseq;
4553 unsigned int start_colseq;
4554 unsigned int end_colseq;
4562 /* Have to increment the pointer into the pattern string, so the
4563 caller isn't still at the ending character. */
4566 /* Report an error if the range is empty and the syntax prohibits this. */
4567 ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
4570 collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
4571 _NL_COLLATE_COLLSEQMB);
4573 start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
4574 end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
4575 for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
4577 unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
4579 if (start_colseq <= this_colseq && this_colseq <= end_colseq)
4581 SET_LIST_BIT (TRANSLATE (this_char));
4586 /* Here we see why `this_char' has to be larger than an `unsigned
4587 char' -- we would otherwise go into an infinite loop, since all
4588 characters <= 0xff. */
4589 range_start_char = TRANSLATE (range_start_char);
4590 /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
4591 and some compilers cast it to int implicitly, so following for_loop
4592 may fall to (almost) infinite loop.
4593 e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
4594 To avoid this, we cast p[0] to unsigned int and truncate it. */
4595 end_char = ((unsigned)TRANSLATE(p[0]) & ((1 << BYTEWIDTH) - 1));
4597 for (this_char = range_start_char; this_char <= end_char; ++this_char)
4599 SET_LIST_BIT (TRANSLATE (this_char));
4608 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
4609 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
4610 characters can start a string that matches the pattern. This fastmap
4611 is used by re_search to skip quickly over impossible starting points.
4613 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
4614 area as BUFP->fastmap.
4616 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
4619 Returns 0 if we succeed, -2 if an internal error. */
4622 /* local function for re_compile_fastmap.
4623 truncate wchar_t character to char. */
4624 static unsigned char truncate_wchar (CHAR_T c);
4626 static unsigned char
4630 unsigned char buf[MB_LEN_MAX];
4631 int retval = wctomb(buf, c);
4632 return retval > 0 ? buf[0] : (unsigned char)c;
4637 PREFIX(re_compile_fastmap) (bufp)
4638 struct re_pattern_buffer *bufp;
4641 #ifdef MATCH_MAY_ALLOCATE
4642 PREFIX(fail_stack_type) fail_stack;
4644 #ifndef REGEX_MALLOC
4648 register char *fastmap = bufp->fastmap;
4651 /* We need to cast pattern to (wchar_t*), because we casted this compiled
4652 pattern to (char*) in regex_compile. */
4653 UCHAR_T *pattern = (UCHAR_T*)bufp->buffer;
4654 register UCHAR_T *pend = (UCHAR_T*) (bufp->buffer + bufp->used);
4656 UCHAR_T *pattern = bufp->buffer;
4657 register UCHAR_T *pend = pattern + bufp->used;
4659 UCHAR_T *p = pattern;
4662 /* This holds the pointer to the failure stack, when
4663 it is allocated relocatably. */
4664 fail_stack_elt_t *failure_stack_ptr;
4667 /* Assume that each path through the pattern can be null until
4668 proven otherwise. We set this false at the bottom of switch
4669 statement, to which we get only if a particular path doesn't
4670 match the empty string. */
4671 boolean path_can_be_null = true;
4673 /* We aren't doing a `succeed_n' to begin with. */
4674 boolean succeed_n_p = false;
4676 assert (fastmap != NULL && p != NULL);
4679 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
4680 bufp->fastmap_accurate = 1; /* It will be when we're done. */
4681 bufp->can_be_null = 0;
4685 if (p == pend || *p == succeed)
4687 /* We have reached the (effective) end of pattern. */
4688 if (!FAIL_STACK_EMPTY ())
4690 bufp->can_be_null |= path_can_be_null;
4692 /* Reset for next path. */
4693 path_can_be_null = true;
4695 p = fail_stack.stack[--fail_stack.avail].pointer;
4703 /* We should never be about to go beyond the end of the pattern. */
4706 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4709 /* I guess the idea here is to simply not bother with a fastmap
4710 if a backreference is used, since it's too hard to figure out
4711 the fastmap for the corresponding group. Setting
4712 `can_be_null' stops `re_search_2' from using the fastmap, so
4713 that is all we do. */
4715 bufp->can_be_null = 1;
4719 /* Following are the cases which match a character. These end
4724 fastmap[truncate_wchar(p[1])] = 1;
4738 /* It is hard to distinguish fastmap from (multi byte) characters
4739 which depends on current locale. */
4744 bufp->can_be_null = 1;
4748 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4749 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
4755 /* Chars beyond end of map must be allowed. */
4756 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
4759 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
4760 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
4766 for (j = 0; j < (1 << BYTEWIDTH); j++)
4767 if (SYNTAX (j) == Sword)
4773 for (j = 0; j < (1 << BYTEWIDTH); j++)
4774 if (SYNTAX (j) != Sword)
4781 int fastmap_newline = fastmap['\n'];
4783 /* `.' matches anything ... */
4784 for (j = 0; j < (1 << BYTEWIDTH); j++)
4787 /* ... except perhaps newline. */
4788 if (!(bufp->syntax & RE_DOT_NEWLINE))
4789 fastmap['\n'] = fastmap_newline;
4791 /* Return if we have already set `can_be_null'; if we have,
4792 then the fastmap is irrelevant. Something's wrong here. */
4793 else if (bufp->can_be_null)
4796 /* Otherwise, have to check alternative paths. */
4803 for (j = 0; j < (1 << BYTEWIDTH); j++)
4804 if (SYNTAX (j) == (enum syntaxcode) k)
4811 for (j = 0; j < (1 << BYTEWIDTH); j++)
4812 if (SYNTAX (j) != (enum syntaxcode) k)
4817 /* All cases after this match the empty string. These end with
4837 case push_dummy_failure:
4842 case pop_failure_jump:
4843 case maybe_pop_jump:
4846 case dummy_failure_jump:
4847 EXTRACT_NUMBER_AND_INCR (j, p);
4852 /* Jump backward implies we just went through the body of a
4853 loop and matched nothing. Opcode jumped to should be
4854 `on_failure_jump' or `succeed_n'. Just treat it like an
4855 ordinary jump. For a * loop, it has pushed its failure
4856 point already; if so, discard that as redundant. */
4857 if ((re_opcode_t) *p != on_failure_jump
4858 && (re_opcode_t) *p != succeed_n)
4862 EXTRACT_NUMBER_AND_INCR (j, p);
4865 /* If what's on the stack is where we are now, pop it. */
4866 if (!FAIL_STACK_EMPTY ()
4867 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
4873 case on_failure_jump:
4874 case on_failure_keep_string_jump:
4875 handle_on_failure_jump:
4876 EXTRACT_NUMBER_AND_INCR (j, p);
4878 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
4879 end of the pattern. We don't want to push such a point,
4880 since when we restore it above, entering the switch will
4881 increment `p' past the end of the pattern. We don't need
4882 to push such a point since we obviously won't find any more
4883 fastmap entries beyond `pend'. Such a pattern can match
4884 the null string, though. */
4887 if (!PUSH_PATTERN_OP (p + j, fail_stack))
4889 RESET_FAIL_STACK ();
4894 bufp->can_be_null = 1;
4898 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
4899 succeed_n_p = false;
4906 /* Get to the number of times to succeed. */
4907 p += OFFSET_ADDRESS_SIZE;
4909 /* Increment p past the n for when k != 0. */
4910 EXTRACT_NUMBER_AND_INCR (k, p);
4913 p -= 2 * OFFSET_ADDRESS_SIZE;
4914 succeed_n_p = true; /* Spaghetti code alert. */
4915 goto handle_on_failure_jump;
4921 p += 2 * OFFSET_ADDRESS_SIZE;
4932 abort (); /* We have listed all the cases. */
4935 /* Getting here means we have found the possible starting
4936 characters for one path of the pattern -- and that the empty
4937 string does not match. We need not follow this path further.
4938 Instead, look at the next alternative (remembered on the
4939 stack), or quit if no more. The test at the top of the loop
4940 does these things. */
4941 path_can_be_null = false;
4945 /* Set `can_be_null' for the last path (also the first path, if the
4946 pattern is empty). */
4947 bufp->can_be_null |= path_can_be_null;
4950 RESET_FAIL_STACK ();
4954 #else /* not INSIDE_RECURSION */
4957 re_compile_fastmap (bufp)
4958 struct re_pattern_buffer *bufp;
4961 if (MB_CUR_MAX != 1)
4962 return wcs_re_compile_fastmap(bufp);
4965 return byte_re_compile_fastmap(bufp);
4966 } /* re_compile_fastmap */
4968 weak_alias (__re_compile_fastmap, re_compile_fastmap)
4972 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
4973 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
4974 this memory for recording register information. STARTS and ENDS
4975 must be allocated using the malloc library routine, and must each
4976 be at least NUM_REGS * sizeof (regoff_t) bytes long.
4978 If NUM_REGS == 0, then subsequent matches should allocate their own
4981 Unless this function is called, the first search or match using
4982 PATTERN_BUFFER will allocate its own register data, without
4983 freeing the old data. */
4986 re_set_registers (bufp, regs, num_regs, starts, ends)
4987 struct re_pattern_buffer *bufp;
4988 struct re_registers *regs;
4990 regoff_t *starts, *ends;
4994 bufp->regs_allocated = REGS_REALLOCATE;
4995 regs->num_regs = num_regs;
4996 regs->start = starts;
5001 bufp->regs_allocated = REGS_UNALLOCATED;
5003 regs->start = regs->end = (regoff_t *) 0;
5007 weak_alias (__re_set_registers, re_set_registers)
5010 /* Searching routines. */
5012 /* Like re_search_2, below, but only one string is specified, and
5013 doesn't let you say where to stop matching. */
5016 re_search (bufp, string, size, startpos, range, regs)
5017 struct re_pattern_buffer *bufp;
5019 int size, startpos, range;
5020 struct re_registers *regs;
5022 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
5026 weak_alias (__re_search, re_search)
5030 /* Using the compiled pattern in BUFP->buffer, first tries to match the
5031 virtual concatenation of STRING1 and STRING2, starting first at index
5032 STARTPOS, then at STARTPOS + 1, and so on.
5034 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
5036 RANGE is how far to scan while trying to match. RANGE = 0 means try
5037 only at STARTPOS; in general, the last start tried is STARTPOS +
5040 In REGS, return the indices of the virtual concatenation of STRING1
5041 and STRING2 that matched the entire BUFP->buffer and its contained
5044 Do not consider matching one past the index STOP in the virtual
5045 concatenation of STRING1 and STRING2.
5047 We return either the position in the strings at which the match was
5048 found, -1 if no match, or -2 if error (such as failure
5052 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
5053 struct re_pattern_buffer *bufp;
5054 const char *string1, *string2;
5058 struct re_registers *regs;
5062 if (MB_CUR_MAX != 1)
5063 return wcs_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5067 return byte_re_search_2 (bufp, string1, size1, string2, size2, startpos,
5071 weak_alias (__re_search_2, re_search_2)
5074 #endif /* not INSIDE_RECURSION */
5076 #ifdef INSIDE_RECURSION
5078 #ifdef MATCH_MAY_ALLOCATE
5079 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
5081 # define FREE_VAR(var) if (var) free (var); var = NULL
5085 # define MAX_ALLOCA_SIZE 2000
5087 # define FREE_WCS_BUFFERS() \
5089 if (size1 > MAX_ALLOCA_SIZE) \
5091 free (wcs_string1); \
5092 free (mbs_offset1); \
5096 FREE_VAR (wcs_string1); \
5097 FREE_VAR (mbs_offset1); \
5099 if (size2 > MAX_ALLOCA_SIZE) \
5101 free (wcs_string2); \
5102 free (mbs_offset2); \
5106 FREE_VAR (wcs_string2); \
5107 FREE_VAR (mbs_offset2); \
5115 PREFIX(re_search_2) (bufp, string1, size1, string2, size2, startpos, range,
5117 struct re_pattern_buffer *bufp;
5118 const char *string1, *string2;
5122 struct re_registers *regs;
5126 register char *fastmap = bufp->fastmap;
5127 register RE_TRANSLATE_TYPE translate = bufp->translate;
5128 int total_size = size1 + size2;
5129 int endpos = startpos + range;
5131 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5132 wchar_t *wcs_string1 = NULL, *wcs_string2 = NULL;
5133 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5134 int wcs_size1 = 0, wcs_size2 = 0;
5135 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5136 int *mbs_offset1 = NULL, *mbs_offset2 = NULL;
5137 /* They hold whether each wchar_t is binary data or not. */
5138 char *is_binary = NULL;
5141 /* Check for out-of-range STARTPOS. */
5142 if (startpos < 0 || startpos > total_size)
5145 /* Fix up RANGE if it might eventually take us outside
5146 the virtual concatenation of STRING1 and STRING2.
5147 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
5149 range = 0 - startpos;
5150 else if (endpos > total_size)
5151 range = total_size - startpos;
5153 /* If the search isn't to be a backwards one, don't waste time in a
5154 search for a pattern that must be anchored. */
5155 if (bufp->used > 0 && range > 0
5156 && ((re_opcode_t) bufp->buffer[0] == begbuf
5157 /* `begline' is like `begbuf' if it cannot match at newlines. */
5158 || ((re_opcode_t) bufp->buffer[0] == begline
5159 && !bufp->newline_anchor)))
5168 /* In a forward search for something that starts with \=.
5169 don't keep searching past point. */
5170 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
5172 range = PT - startpos;
5178 /* Update the fastmap now if not correct already. */
5179 if (fastmap && !bufp->fastmap_accurate)
5180 if (re_compile_fastmap (bufp) == -2)
5184 /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
5185 fill them with converted string. */
5188 if (size1 > MAX_ALLOCA_SIZE)
5190 wcs_string1 = TALLOC (size1 + 1, CHAR_T);
5191 mbs_offset1 = TALLOC (size1 + 1, int);
5192 is_binary = TALLOC (size1 + 1, char);
5196 wcs_string1 = REGEX_TALLOC (size1 + 1, CHAR_T);
5197 mbs_offset1 = REGEX_TALLOC (size1 + 1, int);
5198 is_binary = REGEX_TALLOC (size1 + 1, char);
5200 if (!wcs_string1 || !mbs_offset1 || !is_binary)
5202 if (size1 > MAX_ALLOCA_SIZE)
5210 FREE_VAR (wcs_string1);
5211 FREE_VAR (mbs_offset1);
5212 FREE_VAR (is_binary);
5216 wcs_size1 = convert_mbs_to_wcs(wcs_string1, string1, size1,
5217 mbs_offset1, is_binary);
5218 wcs_string1[wcs_size1] = L'\0'; /* for a sentinel */
5219 if (size1 > MAX_ALLOCA_SIZE)
5222 FREE_VAR (is_binary);
5226 if (size2 > MAX_ALLOCA_SIZE)
5228 wcs_string2 = TALLOC (size2 + 1, CHAR_T);
5229 mbs_offset2 = TALLOC (size2 + 1, int);
5230 is_binary = TALLOC (size2 + 1, char);
5234 wcs_string2 = REGEX_TALLOC (size2 + 1, CHAR_T);
5235 mbs_offset2 = REGEX_TALLOC (size2 + 1, int);
5236 is_binary = REGEX_TALLOC (size2 + 1, char);
5238 if (!wcs_string2 || !mbs_offset2 || !is_binary)
5240 FREE_WCS_BUFFERS ();
5241 if (size2 > MAX_ALLOCA_SIZE)
5244 FREE_VAR (is_binary);
5247 wcs_size2 = convert_mbs_to_wcs(wcs_string2, string2, size2,
5248 mbs_offset2, is_binary);
5249 wcs_string2[wcs_size2] = L'\0'; /* for a sentinel */
5250 if (size2 > MAX_ALLOCA_SIZE)
5253 FREE_VAR (is_binary);
5258 /* Loop through the string, looking for a place to start matching. */
5261 /* If a fastmap is supplied, skip quickly over characters that
5262 cannot be the start of a match. If the pattern can match the
5263 null string, however, we don't need to skip characters; we want
5264 the first null string. */
5265 if (fastmap && startpos < total_size && !bufp->can_be_null)
5267 if (range > 0) /* Searching forwards. */
5269 register const char *d;
5270 register int lim = 0;
5273 if (startpos < size1 && startpos + range >= size1)
5274 lim = range - (size1 - startpos);
5276 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
5278 /* Written out as an if-else to avoid testing `translate'
5282 && !fastmap[(unsigned char)
5283 translate[(unsigned char) *d++]])
5286 while (range > lim && !fastmap[(unsigned char) *d++])
5289 startpos += irange - range;
5291 else /* Searching backwards. */
5293 register CHAR_T c = (size1 == 0 || startpos >= size1
5294 ? string2[startpos - size1]
5295 : string1[startpos]);
5297 if (!fastmap[(unsigned char) TRANSLATE (c)])
5302 /* If can't match the null string, and that's all we have left, fail. */
5303 if (range >= 0 && startpos == total_size && fastmap
5304 && !bufp->can_be_null)
5307 FREE_WCS_BUFFERS ();
5313 val = wcs_re_match_2_internal (bufp, string1, size1, string2,
5314 size2, startpos, regs, stop,
5315 wcs_string1, wcs_size1,
5316 wcs_string2, wcs_size2,
5317 mbs_offset1, mbs_offset2);
5319 val = byte_re_match_2_internal (bufp, string1, size1, string2,
5320 size2, startpos, regs, stop);
5323 #ifndef REGEX_MALLOC
5332 FREE_WCS_BUFFERS ();
5340 FREE_WCS_BUFFERS ();
5360 FREE_WCS_BUFFERS ();
5366 /* This converts PTR, a pointer into one of the search wchar_t strings
5367 `string1' and `string2' into an multibyte string offset from the
5368 beginning of that string. We use mbs_offset to optimize.
5369 See convert_mbs_to_wcs. */
5370 # define POINTER_TO_OFFSET(ptr) \
5371 (FIRST_STRING_P (ptr) \
5372 ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
5373 : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
5376 /* This converts PTR, a pointer into one of the search strings `string1'
5377 and `string2' into an offset from the beginning of that string. */
5378 # define POINTER_TO_OFFSET(ptr) \
5379 (FIRST_STRING_P (ptr) \
5380 ? ((regoff_t) ((ptr) - string1)) \
5381 : ((regoff_t) ((ptr) - string2 + size1)))
5384 /* Macros for dealing with the split strings in re_match_2. */
5386 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
5388 /* Call before fetching a character with *d. This switches over to
5389 string2 if necessary. */
5390 #define PREFETCH() \
5393 /* End of string2 => fail. */ \
5394 if (dend == end_match_2) \
5396 /* End of string1 => advance to string2. */ \
5398 dend = end_match_2; \
5401 /* Test if at very beginning or at very end of the virtual concatenation
5402 of `string1' and `string2'. If only one string, it's `string2'. */
5403 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
5404 #define AT_STRINGS_END(d) ((d) == end2)
5407 /* Test if D points to a character which is word-constituent. We have
5408 two special cases to check for: if past the end of string1, look at
5409 the first character in string2; and if before the beginning of
5410 string2, look at the last character in string1. */
5412 /* Use internationalized API instead of SYNTAX. */
5413 # define WORDCHAR_P(d) \
5414 (iswalnum ((wint_t)((d) == end1 ? *string2 \
5415 : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
5416 || ((d) == end1 ? *string2 \
5417 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == L'_')
5419 # define WORDCHAR_P(d) \
5420 (SYNTAX ((d) == end1 ? *string2 \
5421 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
5425 /* Disabled due to a compiler bug -- see comment at case wordbound */
5427 /* Test if the character before D and the one at D differ with respect
5428 to being word-constituent. */
5429 #define AT_WORD_BOUNDARY(d) \
5430 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
5431 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
5434 /* Free everything we malloc. */
5435 #ifdef MATCH_MAY_ALLOCATE
5437 # define FREE_VARIABLES() \
5439 REGEX_FREE_STACK (fail_stack.stack); \
5440 FREE_VAR (regstart); \
5441 FREE_VAR (regend); \
5442 FREE_VAR (old_regstart); \
5443 FREE_VAR (old_regend); \
5444 FREE_VAR (best_regstart); \
5445 FREE_VAR (best_regend); \
5446 FREE_VAR (reg_info); \
5447 FREE_VAR (reg_dummy); \
5448 FREE_VAR (reg_info_dummy); \
5449 if (!cant_free_wcs_buf) \
5451 FREE_VAR (string1); \
5452 FREE_VAR (string2); \
5453 FREE_VAR (mbs_offset1); \
5454 FREE_VAR (mbs_offset2); \
5458 # define FREE_VARIABLES() \
5460 REGEX_FREE_STACK (fail_stack.stack); \
5461 FREE_VAR (regstart); \
5462 FREE_VAR (regend); \
5463 FREE_VAR (old_regstart); \
5464 FREE_VAR (old_regend); \
5465 FREE_VAR (best_regstart); \
5466 FREE_VAR (best_regend); \
5467 FREE_VAR (reg_info); \
5468 FREE_VAR (reg_dummy); \
5469 FREE_VAR (reg_info_dummy); \
5474 # define FREE_VARIABLES() \
5476 if (!cant_free_wcs_buf) \
5478 FREE_VAR (string1); \
5479 FREE_VAR (string2); \
5480 FREE_VAR (mbs_offset1); \
5481 FREE_VAR (mbs_offset2); \
5485 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
5487 #endif /* not MATCH_MAY_ALLOCATE */
5489 /* These values must meet several constraints. They must not be valid
5490 register values; since we have a limit of 255 registers (because
5491 we use only one byte in the pattern for the register number), we can
5492 use numbers larger than 255. They must differ by 1, because of
5493 NUM_FAILURE_ITEMS above. And the value for the lowest register must
5494 be larger than the value for the highest register, so we do not try
5495 to actually save any registers when none are active. */
5496 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
5497 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
5499 #else /* not INSIDE_RECURSION */
5500 /* Matching routines. */
5502 #ifndef emacs /* Emacs never uses this. */
5503 /* re_match is like re_match_2 except it takes only a single string. */
5506 re_match (bufp, string, size, pos, regs)
5507 struct re_pattern_buffer *bufp;
5510 struct re_registers *regs;
5514 if (MB_CUR_MAX != 1)
5515 result = wcs_re_match_2_internal (bufp, NULL, 0, string, size,
5517 NULL, 0, NULL, 0, NULL, NULL);
5520 result = byte_re_match_2_internal (bufp, NULL, 0, string, size,
5522 # ifndef REGEX_MALLOC
5530 weak_alias (__re_match, re_match)
5532 #endif /* not emacs */
5534 #endif /* not INSIDE_RECURSION */
5536 #ifdef INSIDE_RECURSION
5537 static boolean PREFIX(group_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5539 PREFIX(register_info_type) *reg_info));
5540 static boolean PREFIX(alt_match_null_string_p) _RE_ARGS ((UCHAR_T *p,
5542 PREFIX(register_info_type) *reg_info));
5543 static boolean PREFIX(common_op_match_null_string_p) _RE_ARGS ((UCHAR_T **p,
5545 PREFIX(register_info_type) *reg_info));
5546 static int PREFIX(bcmp_translate) _RE_ARGS ((const CHAR_T *s1, const CHAR_T *s2,
5547 int len, char *translate));
5548 #else /* not INSIDE_RECURSION */
5550 /* re_match_2 matches the compiled pattern in BUFP against the
5551 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
5552 and SIZE2, respectively). We start matching at POS, and stop
5555 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
5556 store offsets for the substring each group matched in REGS. See the
5557 documentation for exactly how many groups we fill.
5559 We return -1 if no match, -2 if an internal error (such as the
5560 failure stack overflowing). Otherwise, we return the length of the
5561 matched substring. */
5564 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
5565 struct re_pattern_buffer *bufp;
5566 const char *string1, *string2;
5569 struct re_registers *regs;
5574 if (MB_CUR_MAX != 1)
5575 result = wcs_re_match_2_internal (bufp, string1, size1, string2, size2,
5577 NULL, 0, NULL, 0, NULL, NULL);
5580 result = byte_re_match_2_internal (bufp, string1, size1, string2, size2,
5583 #ifndef REGEX_MALLOC
5591 weak_alias (__re_match_2, re_match_2)
5594 #endif /* not INSIDE_RECURSION */
5596 #ifdef INSIDE_RECURSION
5599 static int count_mbs_length PARAMS ((int *, int));
5601 /* This check the substring (from 0, to length) of the multibyte string,
5602 to which offset_buffer correspond. And count how many wchar_t_characters
5603 the substring occupy. We use offset_buffer to optimization.
5604 See convert_mbs_to_wcs. */
5607 count_mbs_length(offset_buffer, length)
5613 /* Check whether the size is valid. */
5617 if (offset_buffer == NULL)
5620 /* If there are no multibyte character, offset_buffer[i] == i.
5621 Optmize for this case. */
5622 if (offset_buffer[length] == length)
5625 /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
5631 int middle = (lower + upper) / 2;
5632 if (middle == lower || middle == upper)
5634 if (offset_buffer[middle] > length)
5636 else if (offset_buffer[middle] < length)
5646 /* This is a separate function so that we can force an alloca cleanup
5650 wcs_re_match_2_internal (bufp, cstring1, csize1, cstring2, csize2, pos,
5651 regs, stop, string1, size1, string2, size2,
5652 mbs_offset1, mbs_offset2)
5653 struct re_pattern_buffer *bufp;
5654 const char *cstring1, *cstring2;
5657 struct re_registers *regs;
5659 /* string1 == string2 == NULL means string1/2, size1/2 and
5660 mbs_offset1/2 need seting up in this function. */
5661 /* We need wchar_t* buffers correspond to cstring1, cstring2. */
5662 wchar_t *string1, *string2;
5663 /* We need the size of wchar_t buffers correspond to csize1, csize2. */
5665 /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
5666 int *mbs_offset1, *mbs_offset2;
5669 byte_re_match_2_internal (bufp, string1, size1,string2, size2, pos,
5671 struct re_pattern_buffer *bufp;
5672 const char *string1, *string2;
5675 struct re_registers *regs;
5679 /* General temporaries. */
5683 /* They hold whether each wchar_t is binary data or not. */
5684 char *is_binary = NULL;
5685 /* If true, we can't free string1/2, mbs_offset1/2. */
5686 int cant_free_wcs_buf = 1;
5689 /* Just past the end of the corresponding string. */
5690 const CHAR_T *end1, *end2;
5692 /* Pointers into string1 and string2, just past the last characters in
5693 each to consider matching. */
5694 const CHAR_T *end_match_1, *end_match_2;
5696 /* Where we are in the data, and the end of the current string. */
5697 const CHAR_T *d, *dend;
5699 /* Where we are in the pattern, and the end of the pattern. */
5701 UCHAR_T *pattern, *p;
5702 register UCHAR_T *pend;
5704 UCHAR_T *p = bufp->buffer;
5705 register UCHAR_T *pend = p + bufp->used;
5708 /* Mark the opcode just after a start_memory, so we can test for an
5709 empty subpattern when we get to the stop_memory. */
5710 UCHAR_T *just_past_start_mem = 0;
5712 /* We use this to map every character in the string. */
5713 RE_TRANSLATE_TYPE translate = bufp->translate;
5715 /* Failure point stack. Each place that can handle a failure further
5716 down the line pushes a failure point on this stack. It consists of
5717 restart, regend, and reg_info for all registers corresponding to
5718 the subexpressions we're currently inside, plus the number of such
5719 registers, and, finally, two char *'s. The first char * is where
5720 to resume scanning the pattern; the second one is where to resume
5721 scanning the strings. If the latter is zero, the failure point is
5722 a ``dummy''; if a failure happens and the failure point is a dummy,
5723 it gets discarded and the next next one is tried. */
5724 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5725 PREFIX(fail_stack_type) fail_stack;
5728 static unsigned failure_id;
5729 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
5733 /* This holds the pointer to the failure stack, when
5734 it is allocated relocatably. */
5735 fail_stack_elt_t *failure_stack_ptr;
5738 /* We fill all the registers internally, independent of what we
5739 return, for use in backreferences. The number here includes
5740 an element for register zero. */
5741 size_t num_regs = bufp->re_nsub + 1;
5743 /* The currently active registers. */
5744 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
5745 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
5747 /* Information on the contents of registers. These are pointers into
5748 the input strings; they record just what was matched (on this
5749 attempt) by a subexpression part of the pattern, that is, the
5750 regnum-th regstart pointer points to where in the pattern we began
5751 matching and the regnum-th regend points to right after where we
5752 stopped matching the regnum-th subexpression. (The zeroth register
5753 keeps track of what the whole pattern matches.) */
5754 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5755 const CHAR_T **regstart, **regend;
5758 /* If a group that's operated upon by a repetition operator fails to
5759 match anything, then the register for its start will need to be
5760 restored because it will have been set to wherever in the string we
5761 are when we last see its open-group operator. Similarly for a
5763 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5764 const CHAR_T **old_regstart, **old_regend;
5767 /* The is_active field of reg_info helps us keep track of which (possibly
5768 nested) subexpressions we are currently in. The matched_something
5769 field of reg_info[reg_num] helps us tell whether or not we have
5770 matched any of the pattern so far this time through the reg_num-th
5771 subexpression. These two fields get reset each time through any
5772 loop their register is in. */
5773 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
5774 PREFIX(register_info_type) *reg_info;
5777 /* The following record the register info as found in the above
5778 variables when we find a match better than any we've seen before.
5779 This happens as we backtrack through the failure points, which in
5780 turn happens only if we have not yet matched the entire string. */
5781 unsigned best_regs_set = false;
5782 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5783 const CHAR_T **best_regstart, **best_regend;
5786 /* Logically, this is `best_regend[0]'. But we don't want to have to
5787 allocate space for that if we're not allocating space for anything
5788 else (see below). Also, we never need info about register 0 for
5789 any of the other register vectors, and it seems rather a kludge to
5790 treat `best_regend' differently than the rest. So we keep track of
5791 the end of the best match so far in a separate variable. We
5792 initialize this to NULL so that when we backtrack the first time
5793 and need to test it, it's not garbage. */
5794 const CHAR_T *match_end = NULL;
5796 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
5797 int set_regs_matched_done = 0;
5799 /* Used when we pop values we don't care about. */
5800 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
5801 const CHAR_T **reg_dummy;
5802 PREFIX(register_info_type) *reg_info_dummy;
5806 /* Counts the total number of registers pushed. */
5807 unsigned num_regs_pushed = 0;
5810 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
5814 #ifdef MATCH_MAY_ALLOCATE
5815 /* Do not bother to initialize all the register variables if there are
5816 no groups in the pattern, as it takes a fair amount of time. If
5817 there are groups, we include space for register 0 (the whole
5818 pattern), even though we never use it, since it simplifies the
5819 array indexing. We should fix this. */
5822 regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5823 regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5824 old_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5825 old_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5826 best_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
5827 best_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
5828 reg_info = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5829 reg_dummy = REGEX_TALLOC (num_regs, const CHAR_T *);
5830 reg_info_dummy = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
5832 if (!(regstart && regend && old_regstart && old_regend && reg_info
5833 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
5841 /* We must initialize all our variables to NULL, so that
5842 `FREE_VARIABLES' doesn't try to free them. */
5843 regstart = regend = old_regstart = old_regend = best_regstart
5844 = best_regend = reg_dummy = NULL;
5845 reg_info = reg_info_dummy = (PREFIX(register_info_type) *) NULL;
5847 #endif /* MATCH_MAY_ALLOCATE */
5849 /* The starting position is bogus. */
5851 if (pos < 0 || pos > csize1 + csize2)
5853 if (pos < 0 || pos > size1 + size2)
5861 /* Allocate wchar_t array for string1 and string2 and
5862 fill them with converted string. */
5863 if (string1 == NULL && string2 == NULL)
5865 /* We need seting up buffers here. */
5867 /* We must free wcs buffers in this function. */
5868 cant_free_wcs_buf = 0;
5872 string1 = REGEX_TALLOC (csize1 + 1, CHAR_T);
5873 mbs_offset1 = REGEX_TALLOC (csize1 + 1, int);
5874 is_binary = REGEX_TALLOC (csize1 + 1, char);
5875 if (!string1 || !mbs_offset1 || !is_binary)
5878 FREE_VAR (mbs_offset1);
5879 FREE_VAR (is_binary);
5885 string2 = REGEX_TALLOC (csize2 + 1, CHAR_T);
5886 mbs_offset2 = REGEX_TALLOC (csize2 + 1, int);
5887 is_binary = REGEX_TALLOC (csize2 + 1, char);
5888 if (!string2 || !mbs_offset2 || !is_binary)
5891 FREE_VAR (mbs_offset1);
5893 FREE_VAR (mbs_offset2);
5894 FREE_VAR (is_binary);
5897 size2 = convert_mbs_to_wcs(string2, cstring2, csize2,
5898 mbs_offset2, is_binary);
5899 string2[size2] = L'\0'; /* for a sentinel */
5900 FREE_VAR (is_binary);
5904 /* We need to cast pattern to (wchar_t*), because we casted this compiled
5905 pattern to (char*) in regex_compile. */
5906 p = pattern = (CHAR_T*)bufp->buffer;
5907 pend = (CHAR_T*)(bufp->buffer + bufp->used);
5911 /* Initialize subexpression text positions to -1 to mark ones that no
5912 start_memory/stop_memory has been seen for. Also initialize the
5913 register information struct. */
5914 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
5916 regstart[mcnt] = regend[mcnt]
5917 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
5919 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
5920 IS_ACTIVE (reg_info[mcnt]) = 0;
5921 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5922 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
5925 /* We move `string1' into `string2' if the latter's empty -- but not if
5926 `string1' is null. */
5927 if (size2 == 0 && string1 != NULL)
5934 mbs_offset2 = mbs_offset1;
5940 end1 = string1 + size1;
5941 end2 = string2 + size2;
5943 /* Compute where to stop matching, within the two strings. */
5947 mcnt = count_mbs_length(mbs_offset1, stop);
5948 end_match_1 = string1 + mcnt;
5949 end_match_2 = string2;
5953 if (stop > csize1 + csize2)
5954 stop = csize1 + csize2;
5956 mcnt = count_mbs_length(mbs_offset2, stop-csize1);
5957 end_match_2 = string2 + mcnt;
5960 { /* count_mbs_length return error. */
5967 end_match_1 = string1 + stop;
5968 end_match_2 = string2;
5973 end_match_2 = string2 + stop - size1;
5977 /* `p' scans through the pattern as `d' scans through the data.
5978 `dend' is the end of the input string that `d' points within. `d'
5979 is advanced into the following input string whenever necessary, but
5980 this happens before fetching; therefore, at the beginning of the
5981 loop, `d' can be pointing at the end of a string, but it cannot
5984 if (size1 > 0 && pos <= csize1)
5986 mcnt = count_mbs_length(mbs_offset1, pos);
5992 mcnt = count_mbs_length(mbs_offset2, pos-csize1);
5998 { /* count_mbs_length return error. */
6003 if (size1 > 0 && pos <= size1)
6010 d = string2 + pos - size1;
6015 DEBUG_PRINT1 ("The compiled pattern is:\n");
6016 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
6017 DEBUG_PRINT1 ("The string to match is: `");
6018 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
6019 DEBUG_PRINT1 ("'\n");
6021 /* This loops over pattern commands. It exits by returning from the
6022 function if the match is complete, or it drops through if the match
6023 fails at this starting point in the input data. */
6027 DEBUG_PRINT2 ("\n%p: ", p);
6029 DEBUG_PRINT2 ("\n0x%x: ", p);
6033 { /* End of pattern means we might have succeeded. */
6034 DEBUG_PRINT1 ("end of pattern ... ");
6036 /* If we haven't matched the entire string, and we want the
6037 longest match, try backtracking. */
6038 if (d != end_match_2)
6040 /* 1 if this match ends in the same string (string1 or string2)
6041 as the best previous match. */
6042 boolean same_str_p = (FIRST_STRING_P (match_end)
6043 == MATCHING_IN_FIRST_STRING);
6044 /* 1 if this match is the best seen so far. */
6045 boolean best_match_p;
6047 /* AIX compiler got confused when this was combined
6048 with the previous declaration. */
6050 best_match_p = d > match_end;
6052 best_match_p = !MATCHING_IN_FIRST_STRING;
6054 DEBUG_PRINT1 ("backtracking.\n");
6056 if (!FAIL_STACK_EMPTY ())
6057 { /* More failure points to try. */
6059 /* If exceeds best match so far, save it. */
6060 if (!best_regs_set || best_match_p)
6062 best_regs_set = true;
6065 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
6067 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6069 best_regstart[mcnt] = regstart[mcnt];
6070 best_regend[mcnt] = regend[mcnt];
6076 /* If no failure points, don't restore garbage. And if
6077 last match is real best match, don't restore second
6079 else if (best_regs_set && !best_match_p)
6082 /* Restore best match. It may happen that `dend ==
6083 end_match_1' while the restored d is in string2.
6084 For example, the pattern `x.*y.*z' against the
6085 strings `x-' and `y-z-', if the two strings are
6086 not consecutive in memory. */
6087 DEBUG_PRINT1 ("Restoring best registers.\n");
6090 dend = ((d >= string1 && d <= end1)
6091 ? end_match_1 : end_match_2);
6093 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
6095 regstart[mcnt] = best_regstart[mcnt];
6096 regend[mcnt] = best_regend[mcnt];
6099 } /* d != end_match_2 */
6102 DEBUG_PRINT1 ("Accepting match.\n");
6103 /* If caller wants register contents data back, do it. */
6104 if (regs && !bufp->no_sub)
6106 /* Have the register data arrays been allocated? */
6107 if (bufp->regs_allocated == REGS_UNALLOCATED)
6108 { /* No. So allocate them with malloc. We need one
6109 extra element beyond `num_regs' for the `-1' marker
6111 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
6112 regs->start = TALLOC (regs->num_regs, regoff_t);
6113 regs->end = TALLOC (regs->num_regs, regoff_t);
6114 if (regs->start == NULL || regs->end == NULL)
6119 bufp->regs_allocated = REGS_REALLOCATE;
6121 else if (bufp->regs_allocated == REGS_REALLOCATE)
6122 { /* Yes. If we need more elements than were already
6123 allocated, reallocate them. If we need fewer, just
6125 if (regs->num_regs < num_regs + 1)
6127 regs->num_regs = num_regs + 1;
6128 RETALLOC (regs->start, regs->num_regs, regoff_t);
6129 RETALLOC (regs->end, regs->num_regs, regoff_t);
6130 if (regs->start == NULL || regs->end == NULL)
6139 /* These braces fend off a "empty body in an else-statement"
6140 warning under GCC when assert expands to nothing. */
6141 assert (bufp->regs_allocated == REGS_FIXED);
6144 /* Convert the pointer data in `regstart' and `regend' to
6145 indices. Register zero has to be set differently,
6146 since we haven't kept track of any info for it. */
6147 if (regs->num_regs > 0)
6149 regs->start[0] = pos;
6151 if (MATCHING_IN_FIRST_STRING)
6152 regs->end[0] = mbs_offset1 != NULL ?
6153 mbs_offset1[d-string1] : 0;
6155 regs->end[0] = csize1 + (mbs_offset2 != NULL ?
6156 mbs_offset2[d-string2] : 0);
6158 regs->end[0] = (MATCHING_IN_FIRST_STRING
6159 ? ((regoff_t) (d - string1))
6160 : ((regoff_t) (d - string2 + size1)));
6164 /* Go through the first `min (num_regs, regs->num_regs)'
6165 registers, since that is all we initialized. */
6166 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
6169 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
6170 regs->start[mcnt] = regs->end[mcnt] = -1;
6174 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
6176 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
6180 /* If the regs structure we return has more elements than
6181 were in the pattern, set the extra elements to -1. If
6182 we (re)allocated the registers, this is the case,
6183 because we always allocate enough to have at least one
6185 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
6186 regs->start[mcnt] = regs->end[mcnt] = -1;
6187 } /* regs && !bufp->no_sub */
6189 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
6190 nfailure_points_pushed, nfailure_points_popped,
6191 nfailure_points_pushed - nfailure_points_popped);
6192 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
6195 if (MATCHING_IN_FIRST_STRING)
6196 mcnt = mbs_offset1 != NULL ? mbs_offset1[d-string1] : 0;
6198 mcnt = (mbs_offset2 != NULL ? mbs_offset2[d-string2] : 0) +
6202 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
6207 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
6213 /* Otherwise match next pattern command. */
6214 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
6216 /* Ignore these. Used to ignore the n of succeed_n's which
6217 currently have n == 0. */
6219 DEBUG_PRINT1 ("EXECUTING no_op.\n");
6223 DEBUG_PRINT1 ("EXECUTING succeed.\n");
6226 /* Match the next n pattern characters exactly. The following
6227 byte in the pattern defines n, and the n bytes after that
6228 are the characters to match. */
6234 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
6236 /* This is written out as an if-else so we don't waste time
6237 testing `translate' inside the loop. */
6246 if ((UCHAR_T) translate[(unsigned char) *d++]
6252 if (*d++ != (CHAR_T) *p++)
6256 if ((UCHAR_T) translate[(unsigned char) *d++]
6268 if (*d++ != (CHAR_T) *p++) goto fail;
6272 SET_REGS_MATCHED ();
6276 /* Match any character except possibly a newline or a null. */
6278 DEBUG_PRINT1 ("EXECUTING anychar.\n");
6282 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
6283 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
6286 SET_REGS_MATCHED ();
6287 DEBUG_PRINT2 (" Matched `%ld'.\n", (long int) *d);
6297 unsigned int i, char_class_length, coll_symbol_length,
6298 equiv_class_length, ranges_length, chars_length, length;
6299 CHAR_T *workp, *workp2, *charset_top;
6300 #define WORK_BUFFER_SIZE 128
6301 CHAR_T str_buf[WORK_BUFFER_SIZE];
6306 boolean not = (re_opcode_t) *(p - 1) == charset_not;
6308 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
6310 c = TRANSLATE (*d); /* The character to match. */
6313 nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
6315 charset_top = p - 1;
6316 char_class_length = *p++;
6317 coll_symbol_length = *p++;
6318 equiv_class_length = *p++;
6319 ranges_length = *p++;
6320 chars_length = *p++;
6321 /* p points charset[6], so the address of the next instruction
6322 (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
6323 where l=length of char_classes, m=length of collating_symbol,
6324 n=equivalence_class, o=length of char_range,
6325 p'=length of character. */
6327 /* Update p to indicate the next instruction. */
6328 p += char_class_length + coll_symbol_length+ equiv_class_length +
6329 2*ranges_length + chars_length;
6331 /* match with char_class? */
6332 for (i = 0; i < char_class_length ; i += CHAR_CLASS_SIZE)
6335 uintptr_t alignedp = ((uintptr_t)workp
6336 + __alignof__(wctype_t) - 1)
6337 & ~(uintptr_t)(__alignof__(wctype_t) - 1);
6338 wctype = *((wctype_t*)alignedp);
6339 workp += CHAR_CLASS_SIZE;
6341 if (__iswctype((wint_t)c, wctype))
6342 goto char_set_matched;
6344 if (iswctype((wint_t)c, wctype))
6345 goto char_set_matched;
6349 /* match with collating_symbol? */
6353 const unsigned char *extra = (const unsigned char *)
6354 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
6356 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;
6360 wextra = (int32_t*)(extra + *workp++);
6361 for (i = 0; i < *wextra; ++i)
6362 if (TRANSLATE(d[i]) != wextra[1 + i])
6367 /* Update d, however d will be incremented at
6368 char_set_matched:, we decrement d here. */
6370 goto char_set_matched;
6374 else /* (nrules == 0) */
6376 /* If we can't look up collation data, we use wcscoll
6379 for (workp2 = workp + coll_symbol_length ; workp < workp2 ;)
6381 const CHAR_T *backup_d = d, *backup_dend = dend;
6383 length = __wcslen (workp);
6385 length = wcslen (workp);
6388 /* If wcscoll(the collating symbol, whole string) > 0,
6389 any substring of the string never match with the
6390 collating symbol. */
6391 if (__wcscoll (workp, d) > 0)
6393 workp += length + 1;
6397 /* First, we compare the collating symbol with
6398 the first character of the string.
6399 If it don't match, we add the next character to
6400 the compare buffer in turn. */
6401 for (i = 0 ; i < WORK_BUFFER_SIZE-1 ; i++, d++)
6406 if (dend == end_match_2)
6412 /* add next character to the compare buffer. */
6413 str_buf[i] = TRANSLATE(*d);
6414 str_buf[i+1] = '\0';
6417 match = __wcscoll (workp, str_buf);
6419 match = wcscoll (workp, str_buf);
6422 goto char_set_matched;
6425 /* (str_buf > workp) indicate (str_buf + X > workp),
6426 because for all X (str_buf + X > str_buf).
6427 So we don't need continue this loop. */
6430 /* Otherwise(str_buf < workp),
6431 (str_buf+next_character) may equals (workp).
6432 So we continue this loop. */
6437 workp += length + 1;
6440 /* match with equivalence_class? */
6444 const CHAR_T *backup_d = d, *backup_dend = dend;
6445 /* Try to match the equivalence class against
6446 those known to the collate implementation. */
6447 const int32_t *table;
6448 const int32_t *weights;
6449 const int32_t *extra;
6450 const int32_t *indirect;
6455 /* This #include defines a local function! */
6456 # include <locale/weightwc.h>
6458 table = (const int32_t *)
6459 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC);
6460 weights = (const wint_t *)
6461 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC);
6462 extra = (const wint_t *)
6463 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC);
6464 indirect = (const int32_t *)
6465 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC);
6467 /* Write 1 collating element to str_buf, and
6471 for (i = 0 ; idx2 == 0 && i < WORK_BUFFER_SIZE - 1; i++)
6473 cp = (wint_t*)str_buf;
6476 if (dend == end_match_2)
6481 str_buf[i] = TRANSLATE(*(d+i));
6482 str_buf[i+1] = '\0'; /* sentinel */
6483 idx2 = findidx ((const wint_t**)&cp);
6486 /* Update d, however d will be incremented at
6487 char_set_matched:, we decrement d here. */
6488 d = backup_d + ((wchar_t*)cp - (wchar_t*)str_buf - 1);
6491 if (dend == end_match_2)
6500 len = weights[idx2];
6502 for (workp2 = workp + equiv_class_length ; workp < workp2 ;
6505 idx = (int32_t)*workp;
6506 /* We already checked idx != 0 in regex_compile. */
6508 if (idx2 != 0 && len == weights[idx])
6511 while (cnt < len && (weights[idx + 1 + cnt]
6512 == weights[idx2 + 1 + cnt]))
6516 goto char_set_matched;
6523 else /* (nrules == 0) */
6525 /* If we can't look up collation data, we use wcscoll
6528 for (workp2 = workp + equiv_class_length ; workp < workp2 ;)
6530 const CHAR_T *backup_d = d, *backup_dend = dend;
6532 length = __wcslen (workp);
6534 length = wcslen (workp);
6537 /* If wcscoll(the collating symbol, whole string) > 0,
6538 any substring of the string never match with the
6539 collating symbol. */
6541 if (__wcscoll (workp, d) > 0)
6543 if (wcscoll (workp, d) > 0)
6546 workp += length + 1;
6550 /* First, we compare the equivalence class with
6551 the first character of the string.
6552 If it don't match, we add the next character to
6553 the compare buffer in turn. */
6554 for (i = 0 ; i < WORK_BUFFER_SIZE - 1 ; i++, d++)
6559 if (dend == end_match_2)
6565 /* add next character to the compare buffer. */
6566 str_buf[i] = TRANSLATE(*d);
6567 str_buf[i+1] = '\0';
6570 match = __wcscoll (workp, str_buf);
6572 match = wcscoll (workp, str_buf);
6576 goto char_set_matched;
6579 /* (str_buf > workp) indicate (str_buf + X > workp),
6580 because for all X (str_buf + X > str_buf).
6581 So we don't need continue this loop. */
6584 /* Otherwise(str_buf < workp),
6585 (str_buf+next_character) may equals (workp).
6586 So we continue this loop. */
6591 workp += length + 1;
6595 /* match with char_range? */
6599 uint32_t collseqval;
6600 const char *collseq = (const char *)
6601 _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC);
6603 collseqval = collseq_table_lookup (collseq, c);
6605 for (; workp < p - chars_length ;)
6607 uint32_t start_val, end_val;
6609 /* We already compute the collation sequence value
6610 of the characters (or collating symbols). */
6611 start_val = (uint32_t) *workp++; /* range_start */
6612 end_val = (uint32_t) *workp++; /* range_end */
6614 if (start_val <= collseqval && collseqval <= end_val)
6615 goto char_set_matched;
6621 /* We set range_start_char at str_buf[0], range_end_char
6622 at str_buf[4], and compared char at str_buf[2]. */
6627 for (; workp < p - chars_length ;)
6629 wchar_t *range_start_char, *range_end_char;
6631 /* match if (range_start_char <= c <= range_end_char). */
6633 /* If range_start(or end) < 0, we assume -range_start(end)
6634 is the offset of the collating symbol which is specified
6635 as the character of the range start(end). */
6639 range_start_char = charset_top - (*workp++);
6642 str_buf[0] = *workp++;
6643 range_start_char = str_buf;
6648 range_end_char = charset_top - (*workp++);
6651 str_buf[4] = *workp++;
6652 range_end_char = str_buf + 4;
6656 if (__wcscoll (range_start_char, str_buf+2) <= 0
6657 && __wcscoll (str_buf+2, range_end_char) <= 0)
6659 if (wcscoll (range_start_char, str_buf+2) <= 0
6660 && wcscoll (str_buf+2, range_end_char) <= 0)
6662 goto char_set_matched;
6666 /* match with char? */
6667 for (; workp < p ; workp++)
6669 goto char_set_matched;
6676 /* Cast to `unsigned' instead of `unsigned char' in case the
6677 bit list is a full 32 bytes long. */
6678 if (c < (unsigned) (*p * BYTEWIDTH)
6679 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
6684 if (!not) goto fail;
6685 #undef WORK_BUFFER_SIZE
6687 SET_REGS_MATCHED ();
6693 /* The beginning of a group is represented by start_memory.
6694 The arguments are the register number in the next byte, and the
6695 number of groups inner to this one in the next. The text
6696 matched within the group is recorded (in the internal
6697 registers data structure) under the register number. */
6699 DEBUG_PRINT3 ("EXECUTING start_memory %ld (%ld):\n",
6700 (long int) *p, (long int) p[1]);
6702 /* Find out if this group can match the empty string. */
6703 p1 = p; /* To send to group_match_null_string_p. */
6705 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
6706 REG_MATCH_NULL_STRING_P (reg_info[*p])
6707 = PREFIX(group_match_null_string_p) (&p1, pend, reg_info);
6709 /* Save the position in the string where we were the last time
6710 we were at this open-group operator in case the group is
6711 operated upon by a repetition operator, e.g., with `(a*)*b'
6712 against `ab'; then we want to ignore where we are now in
6713 the string in case this attempt to match fails. */
6714 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6715 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
6717 DEBUG_PRINT2 (" old_regstart: %d\n",
6718 POINTER_TO_OFFSET (old_regstart[*p]));
6721 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
6723 IS_ACTIVE (reg_info[*p]) = 1;
6724 MATCHED_SOMETHING (reg_info[*p]) = 0;
6726 /* Clear this whenever we change the register activity status. */
6727 set_regs_matched_done = 0;
6729 /* This is the new highest active register. */
6730 highest_active_reg = *p;
6732 /* If nothing was active before, this is the new lowest active
6734 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
6735 lowest_active_reg = *p;
6737 /* Move past the register number and inner group count. */
6739 just_past_start_mem = p;
6744 /* The stop_memory opcode represents the end of a group. Its
6745 arguments are the same as start_memory's: the register
6746 number, and the number of inner groups. */
6748 DEBUG_PRINT3 ("EXECUTING stop_memory %ld (%ld):\n",
6749 (long int) *p, (long int) p[1]);
6751 /* We need to save the string position the last time we were at
6752 this close-group operator in case the group is operated
6753 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
6754 against `aba'; then we want to ignore where we are now in
6755 the string in case this attempt to match fails. */
6756 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
6757 ? REG_UNSET (regend[*p]) ? d : regend[*p]
6759 DEBUG_PRINT2 (" old_regend: %d\n",
6760 POINTER_TO_OFFSET (old_regend[*p]));
6763 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
6765 /* This register isn't active anymore. */
6766 IS_ACTIVE (reg_info[*p]) = 0;
6768 /* Clear this whenever we change the register activity status. */
6769 set_regs_matched_done = 0;
6771 /* If this was the only register active, nothing is active
6773 if (lowest_active_reg == highest_active_reg)
6775 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6776 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6779 { /* We must scan for the new highest active register, since
6780 it isn't necessarily one less than now: consider
6781 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
6782 new highest active register is 1. */
6784 while (r > 0 && !IS_ACTIVE (reg_info[r]))
6787 /* If we end up at register zero, that means that we saved
6788 the registers as the result of an `on_failure_jump', not
6789 a `start_memory', and we jumped to past the innermost
6790 `stop_memory'. For example, in ((.)*) we save
6791 registers 1 and 2 as a result of the *, but when we pop
6792 back to the second ), we are at the stop_memory 1.
6793 Thus, nothing is active. */
6796 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
6797 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
6800 highest_active_reg = r;
6803 /* If just failed to match something this time around with a
6804 group that's operated on by a repetition operator, try to
6805 force exit from the ``loop'', and restore the register
6806 information for this group that we had before trying this
6808 if ((!MATCHED_SOMETHING (reg_info[*p])
6809 || just_past_start_mem == p - 1)
6812 boolean is_a_jump_n = false;
6816 switch ((re_opcode_t) *p1++)
6820 case pop_failure_jump:
6821 case maybe_pop_jump:
6823 case dummy_failure_jump:
6824 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6826 p1 += OFFSET_ADDRESS_SIZE;
6834 /* If the next operation is a jump backwards in the pattern
6835 to an on_failure_jump right before the start_memory
6836 corresponding to this stop_memory, exit from the loop
6837 by forcing a failure after pushing on the stack the
6838 on_failure_jump's jump in the pattern, and d. */
6839 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
6840 && (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == start_memory
6841 && p1[2+OFFSET_ADDRESS_SIZE] == *p)
6843 /* If this group ever matched anything, then restore
6844 what its registers were before trying this last
6845 failed match, e.g., with `(a*)*b' against `ab' for
6846 regstart[1], and, e.g., with `((a*)*(b*)*)*'
6847 against `aba' for regend[3].
6849 Also restore the registers for inner groups for,
6850 e.g., `((a*)(b*))*' against `aba' (register 3 would
6851 otherwise get trashed). */
6853 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
6857 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
6859 /* Restore this and inner groups' (if any) registers. */
6860 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
6863 regstart[r] = old_regstart[r];
6865 /* xx why this test? */
6866 if (old_regend[r] >= regstart[r])
6867 regend[r] = old_regend[r];
6871 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
6872 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
6878 /* Move past the register number and the inner group count. */
6883 /* \<digit> has been turned into a `duplicate' command which is
6884 followed by the numeric value of <digit> as the register number. */
6887 register const CHAR_T *d2, *dend2;
6888 int regno = *p++; /* Get which register to match against. */
6889 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
6891 /* Can't back reference a group which we've never matched. */
6892 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
6895 /* Where in input to try to start matching. */
6896 d2 = regstart[regno];
6898 /* Where to stop matching; if both the place to start and
6899 the place to stop matching are in the same string, then
6900 set to the place to stop, otherwise, for now have to use
6901 the end of the first string. */
6903 dend2 = ((FIRST_STRING_P (regstart[regno])
6904 == FIRST_STRING_P (regend[regno]))
6905 ? regend[regno] : end_match_1);
6908 /* If necessary, advance to next segment in register
6912 if (dend2 == end_match_2) break;
6913 if (dend2 == regend[regno]) break;
6915 /* End of string1 => advance to string2. */
6917 dend2 = regend[regno];
6919 /* At end of register contents => success */
6920 if (d2 == dend2) break;
6922 /* If necessary, advance to next segment in data. */
6925 /* How many characters left in this segment to match. */
6928 /* Want how many consecutive characters we can match in
6929 one shot, so, if necessary, adjust the count. */
6930 if (mcnt > dend2 - d2)
6933 /* Compare that many; failure if mismatch, else move
6936 ? PREFIX(bcmp_translate) (d, d2, mcnt, translate)
6937 : memcmp (d, d2, mcnt*sizeof(UCHAR_T)))
6939 d += mcnt, d2 += mcnt;
6941 /* Do this because we've match some characters. */
6942 SET_REGS_MATCHED ();
6948 /* begline matches the empty string at the beginning of the string
6949 (unless `not_bol' is set in `bufp'), and, if
6950 `newline_anchor' is set, after newlines. */
6952 DEBUG_PRINT1 ("EXECUTING begline.\n");
6954 if (AT_STRINGS_BEG (d))
6956 if (!bufp->not_bol) break;
6958 else if (d[-1] == '\n' && bufp->newline_anchor)
6962 /* In all other cases, we fail. */
6966 /* endline is the dual of begline. */
6968 DEBUG_PRINT1 ("EXECUTING endline.\n");
6970 if (AT_STRINGS_END (d))
6972 if (!bufp->not_eol) break;
6975 /* We have to ``prefetch'' the next character. */
6976 else if ((d == end1 ? *string2 : *d) == '\n'
6977 && bufp->newline_anchor)
6984 /* Match at the very beginning of the data. */
6986 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
6987 if (AT_STRINGS_BEG (d))
6992 /* Match at the very end of the data. */
6994 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
6995 if (AT_STRINGS_END (d))
7000 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
7001 pushes NULL as the value for the string on the stack. Then
7002 `pop_failure_point' will keep the current value for the
7003 string, instead of restoring it. To see why, consider
7004 matching `foo\nbar' against `.*\n'. The .* matches the foo;
7005 then the . fails against the \n. But the next thing we want
7006 to do is match the \n against the \n; if we restored the
7007 string value, we would be back at the foo.
7009 Because this is used only in specific cases, we don't need to
7010 check all the things that `on_failure_jump' does, to make
7011 sure the right things get saved on the stack. Hence we don't
7012 share its code. The only reason to push anything on the
7013 stack at all is that otherwise we would have to change
7014 `anychar's code to do something besides goto fail in this
7015 case; that seems worse than this. */
7016 case on_failure_keep_string_jump:
7017 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
7019 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7021 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
7023 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
7026 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
7030 /* Uses of on_failure_jump:
7032 Each alternative starts with an on_failure_jump that points
7033 to the beginning of the next alternative. Each alternative
7034 except the last ends with a jump that in effect jumps past
7035 the rest of the alternatives. (They really jump to the
7036 ending jump of the following alternative, because tensioning
7037 these jumps is a hassle.)
7039 Repeats start with an on_failure_jump that points past both
7040 the repetition text and either the following jump or
7041 pop_failure_jump back to this on_failure_jump. */
7042 case on_failure_jump:
7044 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
7046 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7048 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
7050 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
7053 /* If this on_failure_jump comes right before a group (i.e.,
7054 the original * applied to a group), save the information
7055 for that group and all inner ones, so that if we fail back
7056 to this point, the group's information will be correct.
7057 For example, in \(a*\)*\1, we need the preceding group,
7058 and in \(zz\(a*\)b*\)\2, we need the inner group. */
7060 /* We can't use `p' to check ahead because we push
7061 a failure point to `p + mcnt' after we do this. */
7064 /* We need to skip no_op's before we look for the
7065 start_memory in case this on_failure_jump is happening as
7066 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
7068 while (p1 < pend && (re_opcode_t) *p1 == no_op)
7071 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
7073 /* We have a new highest active register now. This will
7074 get reset at the start_memory we are about to get to,
7075 but we will have saved all the registers relevant to
7076 this repetition op, as described above. */
7077 highest_active_reg = *(p1 + 1) + *(p1 + 2);
7078 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
7079 lowest_active_reg = *(p1 + 1);
7082 DEBUG_PRINT1 (":\n");
7083 PUSH_FAILURE_POINT (p + mcnt, d, -2);
7087 /* A smart repeat ends with `maybe_pop_jump'.
7088 We change it to either `pop_failure_jump' or `jump'. */
7089 case maybe_pop_jump:
7090 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7091 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
7093 register UCHAR_T *p2 = p;
7095 /* Compare the beginning of the repeat with what in the
7096 pattern follows its end. If we can establish that there
7097 is nothing that they would both match, i.e., that we
7098 would have to backtrack because of (as in, e.g., `a*a')
7099 then we can change to pop_failure_jump, because we'll
7100 never have to backtrack.
7102 This is not true in the case of alternatives: in
7103 `(a|ab)*' we do need to backtrack to the `ab' alternative
7104 (e.g., if the string was `ab'). But instead of trying to
7105 detect that here, the alternative has put on a dummy
7106 failure point which is what we will end up popping. */
7108 /* Skip over open/close-group commands.
7109 If what follows this loop is a ...+ construct,
7110 look at what begins its body, since we will have to
7111 match at least one of that. */
7115 && ((re_opcode_t) *p2 == stop_memory
7116 || (re_opcode_t) *p2 == start_memory))
7118 else if (p2 + 2 + 2 * OFFSET_ADDRESS_SIZE < pend
7119 && (re_opcode_t) *p2 == dummy_failure_jump)
7120 p2 += 2 + 2 * OFFSET_ADDRESS_SIZE;
7126 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
7127 to the `maybe_finalize_jump' of this case. Examine what
7130 /* If we're at the end of the pattern, we can change. */
7133 /* Consider what happens when matching ":\(.*\)"
7134 against ":/". I don't really understand this code
7136 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7139 (" End of pattern: change to `pop_failure_jump'.\n");
7142 else if ((re_opcode_t) *p2 == exactn
7144 || (re_opcode_t) *p2 == exactn_bin
7146 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
7149 = *p2 == (UCHAR_T) endline ? '\n' : p2[2];
7151 if (((re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn
7153 || (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn_bin
7155 ) && p1[3+OFFSET_ADDRESS_SIZE] != c)
7157 p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
7160 DEBUG_PRINT3 (" %C != %C => pop_failure_jump.\n",
7162 (wint_t) p1[3+OFFSET_ADDRESS_SIZE]);
7164 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
7166 (char) p1[3+OFFSET_ADDRESS_SIZE]);
7171 else if ((re_opcode_t) p1[3] == charset
7172 || (re_opcode_t) p1[3] == charset_not)
7174 int not = (re_opcode_t) p1[3] == charset_not;
7176 if (c < (unsigned) (p1[4] * BYTEWIDTH)
7177 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
7180 /* `not' is equal to 1 if c would match, which means
7181 that we can't change to pop_failure_jump. */
7184 p[-3] = (unsigned char) pop_failure_jump;
7185 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7188 #endif /* not WCHAR */
7191 else if ((re_opcode_t) *p2 == charset)
7193 /* We win if the first character of the loop is not part
7195 if ((re_opcode_t) p1[3] == exactn
7196 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
7197 && (p2[2 + p1[5] / BYTEWIDTH]
7198 & (1 << (p1[5] % BYTEWIDTH)))))
7200 p[-3] = (unsigned char) pop_failure_jump;
7201 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7204 else if ((re_opcode_t) p1[3] == charset_not)
7207 /* We win if the charset_not inside the loop
7208 lists every character listed in the charset after. */
7209 for (idx = 0; idx < (int) p2[1]; idx++)
7210 if (! (p2[2 + idx] == 0
7211 || (idx < (int) p1[4]
7212 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
7217 p[-3] = (unsigned char) pop_failure_jump;
7218 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7221 else if ((re_opcode_t) p1[3] == charset)
7224 /* We win if the charset inside the loop
7225 has no overlap with the one after the loop. */
7227 idx < (int) p2[1] && idx < (int) p1[4];
7229 if ((p2[2 + idx] & p1[5 + idx]) != 0)
7232 if (idx == p2[1] || idx == p1[4])
7234 p[-3] = (unsigned char) pop_failure_jump;
7235 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
7239 #endif /* not WCHAR */
7241 p -= OFFSET_ADDRESS_SIZE; /* Point at relative address again. */
7242 if ((re_opcode_t) p[-1] != pop_failure_jump)
7244 p[-1] = (UCHAR_T) jump;
7245 DEBUG_PRINT1 (" Match => jump.\n");
7246 goto unconditional_jump;
7248 /* Note fall through. */
7251 /* The end of a simple repeat has a pop_failure_jump back to
7252 its matching on_failure_jump, where the latter will push a
7253 failure point. The pop_failure_jump takes off failure
7254 points put on by this pop_failure_jump's matching
7255 on_failure_jump; we got through the pattern to here from the
7256 matching on_failure_jump, so didn't fail. */
7257 case pop_failure_jump:
7259 /* We need to pass separate storage for the lowest and
7260 highest registers, even though we don't care about the
7261 actual values. Otherwise, we will restore only one
7262 register from the stack, since lowest will == highest in
7263 `pop_failure_point'. */
7264 active_reg_t dummy_low_reg, dummy_high_reg;
7265 UCHAR_T *pdummy = NULL;
7266 const CHAR_T *sdummy = NULL;
7268 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
7269 POP_FAILURE_POINT (sdummy, pdummy,
7270 dummy_low_reg, dummy_high_reg,
7271 reg_dummy, reg_dummy, reg_info_dummy);
7273 /* Note fall through. */
7277 DEBUG_PRINT2 ("\n%p: ", p);
7279 DEBUG_PRINT2 ("\n0x%x: ", p);
7281 /* Note fall through. */
7283 /* Unconditionally jump (without popping any failure points). */
7285 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
7286 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
7287 p += mcnt; /* Do the jump. */
7289 DEBUG_PRINT2 ("(to %p).\n", p);
7291 DEBUG_PRINT2 ("(to 0x%x).\n", p);
7296 /* We need this opcode so we can detect where alternatives end
7297 in `group_match_null_string_p' et al. */
7299 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
7300 goto unconditional_jump;
7303 /* Normally, the on_failure_jump pushes a failure point, which
7304 then gets popped at pop_failure_jump. We will end up at
7305 pop_failure_jump, also, and with a pattern of, say, `a+', we
7306 are skipping over the on_failure_jump, so we have to push
7307 something meaningless for pop_failure_jump to pop. */
7308 case dummy_failure_jump:
7309 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
7310 /* It doesn't matter what we push for the string here. What
7311 the code at `fail' tests is the value for the pattern. */
7312 PUSH_FAILURE_POINT (NULL, NULL, -2);
7313 goto unconditional_jump;
7316 /* At the end of an alternative, we need to push a dummy failure
7317 point in case we are followed by a `pop_failure_jump', because
7318 we don't want the failure point for the alternative to be
7319 popped. For example, matching `(a|ab)*' against `aab'
7320 requires that we match the `ab' alternative. */
7321 case push_dummy_failure:
7322 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
7323 /* See comments just above at `dummy_failure_jump' about the
7325 PUSH_FAILURE_POINT (NULL, NULL, -2);
7328 /* Have to succeed matching what follows at least n times.
7329 After that, handle like `on_failure_jump'. */
7331 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7332 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
7335 /* Originally, this is how many times we HAVE to succeed. */
7339 p += OFFSET_ADDRESS_SIZE;
7340 STORE_NUMBER_AND_INCR (p, mcnt);
7342 DEBUG_PRINT3 (" Setting %p to %d.\n", p - OFFSET_ADDRESS_SIZE
7345 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - OFFSET_ADDRESS_SIZE
7352 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n",
7353 p + OFFSET_ADDRESS_SIZE);
7355 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n",
7356 p + OFFSET_ADDRESS_SIZE);
7360 p[1] = (UCHAR_T) no_op;
7362 p[2] = (UCHAR_T) no_op;
7363 p[3] = (UCHAR_T) no_op;
7370 EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
7371 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
7373 /* Originally, this is how many times we CAN jump. */
7377 STORE_NUMBER (p + OFFSET_ADDRESS_SIZE, mcnt);
7380 DEBUG_PRINT3 (" Setting %p to %d.\n", p + OFFSET_ADDRESS_SIZE,
7383 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + OFFSET_ADDRESS_SIZE,
7386 goto unconditional_jump;
7388 /* If don't have to jump any more, skip over the rest of command. */
7390 p += 2 * OFFSET_ADDRESS_SIZE;
7395 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
7397 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7399 EXTRACT_NUMBER_AND_INCR (mcnt, p);
7401 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
7403 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
7405 STORE_NUMBER (p1, mcnt);
7410 /* The DEC Alpha C compiler 3.x generates incorrect code for the
7411 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
7412 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
7413 macro and introducing temporary variables works around the bug. */
7416 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7417 if (AT_WORD_BOUNDARY (d))
7422 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7423 if (AT_WORD_BOUNDARY (d))
7429 boolean prevchar, thischar;
7431 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
7432 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7435 prevchar = WORDCHAR_P (d - 1);
7436 thischar = WORDCHAR_P (d);
7437 if (prevchar != thischar)
7444 boolean prevchar, thischar;
7446 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
7447 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
7450 prevchar = WORDCHAR_P (d - 1);
7451 thischar = WORDCHAR_P (d);
7452 if (prevchar != thischar)
7459 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
7460 if (!AT_STRINGS_END (d) && WORDCHAR_P (d)
7461 && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
7466 DEBUG_PRINT1 ("EXECUTING wordend.\n");
7467 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
7468 && (AT_STRINGS_END (d) || !WORDCHAR_P (d)))
7474 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
7475 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
7480 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
7481 if (PTR_CHAR_POS ((unsigned char *) d) != point)
7486 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
7487 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
7492 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
7497 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
7501 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7503 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
7505 SET_REGS_MATCHED ();
7509 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
7511 goto matchnotsyntax;
7514 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
7518 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
7520 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
7522 SET_REGS_MATCHED ();
7525 #else /* not emacs */
7527 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
7529 if (!WORDCHAR_P (d))
7531 SET_REGS_MATCHED ();
7536 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
7540 SET_REGS_MATCHED ();
7543 #endif /* not emacs */
7548 continue; /* Successfully executed one pattern command; keep going. */
7551 /* We goto here if a matching operation fails. */
7553 if (!FAIL_STACK_EMPTY ())
7554 { /* A restart point is known. Restore to that state. */
7555 DEBUG_PRINT1 ("\nFAIL:\n");
7556 POP_FAILURE_POINT (d, p,
7557 lowest_active_reg, highest_active_reg,
7558 regstart, regend, reg_info);
7560 /* If this failure point is a dummy, try the next one. */
7564 /* If we failed to the end of the pattern, don't examine *p. */
7568 boolean is_a_jump_n = false;
7570 /* If failed to a backwards jump that's part of a repetition
7571 loop, need to pop this failure point and use the next one. */
7572 switch ((re_opcode_t) *p)
7576 case maybe_pop_jump:
7577 case pop_failure_jump:
7580 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7583 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
7585 && (re_opcode_t) *p1 == on_failure_jump))
7593 if (d >= string1 && d <= end1)
7597 break; /* Matching at this starting point really fails. */
7601 goto restore_best_regs;
7605 return -1; /* Failure to match. */
7608 /* Subroutine definitions for re_match_2. */
7611 /* We are passed P pointing to a register number after a start_memory.
7613 Return true if the pattern up to the corresponding stop_memory can
7614 match the empty string, and false otherwise.
7616 If we find the matching stop_memory, sets P to point to one past its number.
7617 Otherwise, sets P to an undefined byte less than or equal to END.
7619 We don't handle duplicates properly (yet). */
7622 PREFIX(group_match_null_string_p) (p, end, reg_info)
7624 PREFIX(register_info_type) *reg_info;
7627 /* Point to after the args to the start_memory. */
7628 UCHAR_T *p1 = *p + 2;
7632 /* Skip over opcodes that can match nothing, and return true or
7633 false, as appropriate, when we get to one that can't, or to the
7634 matching stop_memory. */
7636 switch ((re_opcode_t) *p1)
7638 /* Could be either a loop or a series of alternatives. */
7639 case on_failure_jump:
7641 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7643 /* If the next operation is not a jump backwards in the
7648 /* Go through the on_failure_jumps of the alternatives,
7649 seeing if any of the alternatives cannot match nothing.
7650 The last alternative starts with only a jump,
7651 whereas the rest start with on_failure_jump and end
7652 with a jump, e.g., here is the pattern for `a|b|c':
7654 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
7655 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
7658 So, we have to first go through the first (n-1)
7659 alternatives and then deal with the last one separately. */
7662 /* Deal with the first (n-1) alternatives, which start
7663 with an on_failure_jump (see above) that jumps to right
7664 past a jump_past_alt. */
7666 while ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] ==
7669 /* `mcnt' holds how many bytes long the alternative
7670 is, including the ending `jump_past_alt' and
7673 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt -
7674 (1 + OFFSET_ADDRESS_SIZE),
7678 /* Move to right after this alternative, including the
7682 /* Break if it's the beginning of an n-th alternative
7683 that doesn't begin with an on_failure_jump. */
7684 if ((re_opcode_t) *p1 != on_failure_jump)
7687 /* Still have to check that it's not an n-th
7688 alternative that starts with an on_failure_jump. */
7690 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7691 if ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] !=
7694 /* Get to the beginning of the n-th alternative. */
7695 p1 -= 1 + OFFSET_ADDRESS_SIZE;
7700 /* Deal with the last alternative: go back and get number
7701 of the `jump_past_alt' just before it. `mcnt' contains
7702 the length of the alternative. */
7703 EXTRACT_NUMBER (mcnt, p1 - OFFSET_ADDRESS_SIZE);
7705 if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt, reg_info))
7708 p1 += mcnt; /* Get past the n-th alternative. */
7714 assert (p1[1] == **p);
7720 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7723 } /* while p1 < end */
7726 } /* group_match_null_string_p */
7729 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
7730 It expects P to be the first byte of a single alternative and END one
7731 byte past the last. The alternative can contain groups. */
7734 PREFIX(alt_match_null_string_p) (p, end, reg_info)
7736 PREFIX(register_info_type) *reg_info;
7743 /* Skip over opcodes that can match nothing, and break when we get
7744 to one that can't. */
7746 switch ((re_opcode_t) *p1)
7749 case on_failure_jump:
7751 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7756 if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
7759 } /* while p1 < end */
7762 } /* alt_match_null_string_p */
7765 /* Deals with the ops common to group_match_null_string_p and
7766 alt_match_null_string_p.
7768 Sets P to one after the op and its arguments, if any. */
7771 PREFIX(common_op_match_null_string_p) (p, end, reg_info)
7773 PREFIX(register_info_type) *reg_info;
7780 switch ((re_opcode_t) *p1++)
7800 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
7801 ret = PREFIX(group_match_null_string_p) (&p1, end, reg_info);
7803 /* Have to set this here in case we're checking a group which
7804 contains a group and a back reference to it. */
7806 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
7807 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
7813 /* If this is an optimized succeed_n for zero times, make the jump. */
7815 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7823 /* Get to the number of times to succeed. */
7824 p1 += OFFSET_ADDRESS_SIZE;
7825 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7829 p1 -= 2 * OFFSET_ADDRESS_SIZE;
7830 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
7838 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
7843 p1 += 2 * OFFSET_ADDRESS_SIZE;
7846 /* All other opcodes mean we cannot match the empty string. */
7852 } /* common_op_match_null_string_p */
7855 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
7856 bytes; nonzero otherwise. */
7859 PREFIX(bcmp_translate) (s1, s2, len, translate)
7860 const CHAR_T *s1, *s2;
7862 RE_TRANSLATE_TYPE translate;
7864 register const UCHAR_T *p1 = (const UCHAR_T *) s1;
7865 register const UCHAR_T *p2 = (const UCHAR_T *) s2;
7869 if (((*p1<=0xff)?translate[*p1++]:*p1++)
7870 != ((*p2<=0xff)?translate[*p2++]:*p2++))
7873 if (translate[*p1++] != translate[*p2++]) return 1;
7881 #else /* not INSIDE_RECURSION */
7883 /* Entry points for GNU code. */
7885 /* re_compile_pattern is the GNU regular expression compiler: it
7886 compiles PATTERN (of length SIZE) and puts the result in BUFP.
7887 Returns 0 if the pattern was valid, otherwise an error string.
7889 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
7890 are set in BUFP on entry.
7892 We call regex_compile to do the actual compilation. */
7895 re_compile_pattern (pattern, length, bufp)
7896 const char *pattern;
7898 struct re_pattern_buffer *bufp;
7902 /* GNU code is written to assume at least RE_NREGS registers will be set
7903 (and at least one extra will be -1). */
7904 bufp->regs_allocated = REGS_UNALLOCATED;
7906 /* And GNU code determines whether or not to get register information
7907 by passing null for the REGS argument to re_match, etc., not by
7911 /* Match anchors at newline. */
7912 bufp->newline_anchor = 1;
7915 if (MB_CUR_MAX != 1)
7916 ret = wcs_regex_compile (pattern, length, re_syntax_options, bufp);
7919 ret = byte_regex_compile (pattern, length, re_syntax_options, bufp);
7923 return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7926 weak_alias (__re_compile_pattern, re_compile_pattern)
7929 /* Entry points compatible with 4.2 BSD regex library. We don't define
7930 them unless specifically requested. */
7932 #if defined _REGEX_RE_COMP || defined _LIBC
7934 /* BSD has one and only one pattern buffer. */
7935 static struct re_pattern_buffer re_comp_buf;
7939 /* Make these definitions weak in libc, so POSIX programs can redefine
7940 these names if they don't use our functions, and still use
7941 regcomp/regexec below without link errors. */
7951 if (!re_comp_buf.buffer)
7952 return gettext ("No previous regular expression");
7956 if (!re_comp_buf.buffer)
7958 re_comp_buf.buffer = (unsigned char *) malloc (200);
7959 if (re_comp_buf.buffer == NULL)
7960 return (char *) gettext (re_error_msgid
7961 + re_error_msgid_idx[(int) REG_ESPACE]);
7962 re_comp_buf.allocated = 200;
7964 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
7965 if (re_comp_buf.fastmap == NULL)
7966 return (char *) gettext (re_error_msgid
7967 + re_error_msgid_idx[(int) REG_ESPACE]);
7970 /* Since `re_exec' always passes NULL for the `regs' argument, we
7971 don't need to initialize the pattern buffer fields which affect it. */
7973 /* Match anchors at newlines. */
7974 re_comp_buf.newline_anchor = 1;
7977 if (MB_CUR_MAX != 1)
7978 ret = wcs_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7981 ret = byte_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
7986 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
7987 return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
7998 const int len = strlen (s);
8000 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
8003 #endif /* _REGEX_RE_COMP */
8005 /* POSIX.2 functions. Don't define these for Emacs. */
8009 /* regcomp takes a regular expression as a string and compiles it.
8011 PREG is a regex_t *. We do not expect any fields to be initialized,
8012 since POSIX says we shouldn't. Thus, we set
8014 `buffer' to the compiled pattern;
8015 `used' to the length of the compiled pattern;
8016 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
8017 REG_EXTENDED bit in CFLAGS is set; otherwise, to
8018 RE_SYNTAX_POSIX_BASIC;
8019 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
8020 `fastmap' to an allocated space for the fastmap;
8021 `fastmap_accurate' to zero;
8022 `re_nsub' to the number of subexpressions in PATTERN.
8024 PATTERN is the address of the pattern string.
8026 CFLAGS is a series of bits which affect compilation.
8028 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
8029 use POSIX basic syntax.
8031 If REG_NEWLINE is set, then . and [^...] don't match newline.
8032 Also, regexec will try a match beginning after every newline.
8034 If REG_ICASE is set, then we considers upper- and lowercase
8035 versions of letters to be equivalent when matching.
8037 If REG_NOSUB is set, then when PREG is passed to regexec, that
8038 routine will report only success or failure, and nothing about the
8041 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
8042 the return codes and their meanings.) */
8045 regcomp (preg, pattern, cflags)
8047 const char *pattern;
8052 = (cflags & REG_EXTENDED) ?
8053 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
8055 /* regex_compile will allocate the space for the compiled pattern. */
8057 preg->allocated = 0;
8060 /* Try to allocate space for the fastmap. */
8061 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
8063 if (cflags & REG_ICASE)
8068 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
8069 * sizeof (*(RE_TRANSLATE_TYPE)0));
8070 if (preg->translate == NULL)
8071 return (int) REG_ESPACE;
8073 /* Map uppercase characters to corresponding lowercase ones. */
8074 for (i = 0; i < CHAR_SET_SIZE; i++)
8075 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
8078 preg->translate = NULL;
8080 /* If REG_NEWLINE is set, newlines are treated differently. */
8081 if (cflags & REG_NEWLINE)
8082 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
8083 syntax &= ~RE_DOT_NEWLINE;
8084 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
8085 /* It also changes the matching behavior. */
8086 preg->newline_anchor = 1;
8089 preg->newline_anchor = 0;
8091 preg->no_sub = !!(cflags & REG_NOSUB);
8093 /* POSIX says a null character in the pattern terminates it, so we
8094 can use strlen here in compiling the pattern. */
8096 if (MB_CUR_MAX != 1)
8097 ret = wcs_regex_compile (pattern, strlen (pattern), syntax, preg);
8100 ret = byte_regex_compile (pattern, strlen (pattern), syntax, preg);
8102 /* POSIX doesn't distinguish between an unmatched open-group and an
8103 unmatched close-group: both are REG_EPAREN. */
8104 if (ret == REG_ERPAREN) ret = REG_EPAREN;
8106 if (ret == REG_NOERROR && preg->fastmap)
8108 /* Compute the fastmap now, since regexec cannot modify the pattern
8110 if (re_compile_fastmap (preg) == -2)
8112 /* Some error occurred while computing the fastmap, just forget
8114 free (preg->fastmap);
8115 preg->fastmap = NULL;
8122 weak_alias (__regcomp, regcomp)
8126 /* regexec searches for a given pattern, specified by PREG, in the
8129 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
8130 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
8131 least NMATCH elements, and we set them to the offsets of the
8132 corresponding matched substrings.
8134 EFLAGS specifies `execution flags' which affect matching: if
8135 REG_NOTBOL is set, then ^ does not match at the beginning of the
8136 string; if REG_NOTEOL is set, then $ does not match at the end.
8138 We return 0 if we find a match and REG_NOMATCH if not. */
8141 regexec (preg, string, nmatch, pmatch, eflags)
8142 const regex_t *preg;
8145 regmatch_t pmatch[];
8149 struct re_registers regs;
8150 regex_t private_preg;
8151 int len = strlen (string);
8152 boolean want_reg_info = !preg->no_sub && nmatch > 0;
8154 private_preg = *preg;
8156 private_preg.not_bol = !!(eflags & REG_NOTBOL);
8157 private_preg.not_eol = !!(eflags & REG_NOTEOL);
8159 /* The user has told us exactly how many registers to return
8160 information about, via `nmatch'. We have to pass that on to the
8161 matching routines. */
8162 private_preg.regs_allocated = REGS_FIXED;
8166 regs.num_regs = nmatch;
8167 regs.start = TALLOC (nmatch * 2, regoff_t);
8168 if (regs.start == NULL)
8169 return (int) REG_NOMATCH;
8170 regs.end = regs.start + nmatch;
8173 /* Perform the searching operation. */
8174 ret = re_search (&private_preg, string, len,
8175 /* start: */ 0, /* range: */ len,
8176 want_reg_info ? ®s : (struct re_registers *) 0);
8178 /* Copy the register information to the POSIX structure. */
8185 for (r = 0; r < nmatch; r++)
8187 pmatch[r].rm_so = regs.start[r];
8188 pmatch[r].rm_eo = regs.end[r];
8192 /* If we needed the temporary register info, free the space now. */
8196 /* We want zero return to mean success, unlike `re_search'. */
8197 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
8200 weak_alias (__regexec, regexec)
8204 /* Returns a message corresponding to an error code, ERRCODE, returned
8205 from either regcomp or regexec. We don't use PREG here. */
8208 regerror (errcode, preg, errbuf, errbuf_size)
8210 const regex_t *preg;
8218 || errcode >= (int) (sizeof (re_error_msgid_idx)
8219 / sizeof (re_error_msgid_idx[0])))
8220 /* Only error codes returned by the rest of the code should be passed
8221 to this routine. If we are given anything else, or if other regex
8222 code generates an invalid error code, then the program has a bug.
8223 Dump core so we can fix it. */
8226 msg = gettext (re_error_msgid + re_error_msgid_idx[errcode]);
8228 msg_size = strlen (msg) + 1; /* Includes the null. */
8230 if (errbuf_size != 0)
8232 if (msg_size > errbuf_size)
8234 #if defined HAVE_MEMPCPY || defined _LIBC
8235 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
8237 memcpy (errbuf, msg, errbuf_size - 1);
8238 errbuf[errbuf_size - 1] = 0;
8242 memcpy (errbuf, msg, msg_size);
8248 weak_alias (__regerror, regerror)
8252 /* Free dynamically allocated space used by PREG. */
8258 if (preg->buffer != NULL)
8259 free (preg->buffer);
8260 preg->buffer = NULL;
8262 preg->allocated = 0;
8265 if (preg->fastmap != NULL)
8266 free (preg->fastmap);
8267 preg->fastmap = NULL;
8268 preg->fastmap_accurate = 0;
8270 if (preg->translate != NULL)
8271 free (preg->translate);
8272 preg->translate = NULL;
8275 weak_alias (__regfree, regfree)
8278 #endif /* not emacs */
8280 #endif /* not INSIDE_RECURSION */
8284 #undef STORE_NUMBER_AND_INCR
8285 #undef EXTRACT_NUMBER
8286 #undef EXTRACT_NUMBER_AND_INCR
8288 #undef DEBUG_PRINT_COMPILED_PATTERN
8289 #undef DEBUG_PRINT_DOUBLE_STRING
8291 #undef INIT_FAIL_STACK
8292 #undef RESET_FAIL_STACK
8293 #undef DOUBLE_FAIL_STACK
8294 #undef PUSH_PATTERN_OP
8295 #undef PUSH_FAILURE_POINTER
8296 #undef PUSH_FAILURE_INT
8297 #undef PUSH_FAILURE_ELT
8298 #undef POP_FAILURE_POINTER
8299 #undef POP_FAILURE_INT
8300 #undef POP_FAILURE_ELT
8303 #undef PUSH_FAILURE_POINT
8304 #undef POP_FAILURE_POINT
8306 #undef REG_UNSET_VALUE
8314 #undef INIT_BUF_SIZE
8315 #undef GET_BUFFER_SPACE
8323 #undef EXTEND_BUFFER
8324 #undef GET_UNSIGNED_NUMBER
8325 #undef FREE_STACK_RETURN
8327 # undef POINTER_TO_OFFSET
8328 # undef MATCHING_IN_FRST_STRING
8330 # undef AT_STRINGS_BEG
8331 # undef AT_STRINGS_END
8334 # undef FREE_VARIABLES
8335 # undef NO_HIGHEST_ACTIVE_REG
8336 # undef NO_LOWEST_ACTIVE_REG
8340 # undef COMPILED_BUFFER_VAR
8341 # undef OFFSET_ADDRESS_SIZE
8342 # undef CHAR_CLASS_SIZE
8349 # define DEFINED_ONCE
projects / platform / upstream / glibc.git / blob