version 0.12.
(Implements POSIX draft P1003.2/D11.2, except for some of the
internationalization features.)
- Copyright (C) 1993, 94, 95, 96, 97, 98 Free Software Foundation, Inc.
- NOTE: The canonical source of this file is maintained with the GNU C Library.
- Bugs can be reported to bug-glibc@prep.ai.mit.edu.
+ Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
+ 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
- This program is free software; you can redistribute it and/or modify it
- under the terms of the GNU General Public License as published by the
- Free Software Foundation; either version 2, or (at your option) any
- later version.
+ This program is free software; you can redistribute it and/or modify
+ it under the terms of the GNU General Public License as published by
+ the Free Software Foundation; either version 2, or (at your option)
+ any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
- You should have received a copy of the GNU General Public License
- along with this program; if not, write to the Free Software
- Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
- USA. */
+ You should have received a copy of the GNU General Public License along
+ with this program; if not, write to the Free Software Foundation,
+ Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. */
/* AIX requires this to be the first thing in the file. */
#if defined _AIX && !defined REGEX_MALLOC
# include <config.h>
#endif
-#ifndef PARAMS
-# if defined __GNUC__ || (defined __STDC__ && __STDC__)
-# define PARAMS(args) args
-# else
-# define PARAMS(args) ()
-# endif /* GCC. */
-#endif /* Not PARAMS. */
+#ifndef INSIDE_RECURSION
-#if defined STDC_HEADERS && !defined emacs
# include <stddef.h>
-#else
-/* We need this for `regex.h', and perhaps for the Emacs include files. */
-# include <sys/types.h>
-#endif
-#define WIDE_CHAR_SUPPORT \
- defined _LIBC || (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
+# define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
/* For platform which support the ISO C amendement 1 functionality we
support user defined character classes. */
-#if WIDE_CHAR_SUPPORT
+# if defined _LIBC || WIDE_CHAR_SUPPORT
+/* Tru64 with Desktop Toolkit C has a bug: <stdio.h> must be included
+ before <wchar.h>. */
+# include <stdio.h>
/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
-# include <wchar.h>
-# include <wctype.h>
-#endif
+# include <wchar.h>
+# include <wctype.h>
+# endif
-#ifdef _LIBC
+# ifdef _LIBC
/* We have to keep the namespace clean. */
-# define regfree(preg) __regfree (preg)
-# define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
-# define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
-# define regerror(errcode, preg, errbuf, errbuf_size) \
+# define regfree(preg) __regfree (preg)
+# define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
+# define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
+# define regerror(errcode, preg, errbuf, errbuf_size) \
__regerror(errcode, preg, errbuf, errbuf_size)
-# define re_set_registers(bu, re, nu, st, en) \
+# define re_set_registers(bu, re, nu, st, en) \
__re_set_registers (bu, re, nu, st, en)
-# define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
+# define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
__re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
-# define re_match(bufp, string, size, pos, regs) \
+# define re_match(bufp, string, size, pos, regs) \
__re_match (bufp, string, size, pos, regs)
-# define re_search(bufp, string, size, startpos, range, regs) \
+# define re_search(bufp, string, size, startpos, range, regs) \
__re_search (bufp, string, size, startpos, range, regs)
-# define re_compile_pattern(pattern, length, bufp) \
+# define re_compile_pattern(pattern, length, bufp) \
__re_compile_pattern (pattern, length, bufp)
-# define re_set_syntax(syntax) __re_set_syntax (syntax)
-# define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
+# define re_set_syntax(syntax) __re_set_syntax (syntax)
+# define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
__re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
-# define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
-
-#define btowc __btowc
-#endif
+# define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
+
+# define btowc __btowc
+# define iswctype __iswctype
+# define mbrtowc __mbrtowc
+# define wcslen __wcslen
+# define wcscoll __wcscoll
+# define wcrtomb __wcrtomb
+
+/* We are also using some library internals. */
+# include <locale/localeinfo.h>
+# include <locale/elem-hash.h>
+# include <langinfo.h>
+# include <locale/coll-lookup.h>
+# endif
+# ifdef _LIBC
+# include <libintl.h>
+# undef gettext
+# define gettext(msgid) __dcgettext ("libc", msgid, LC_MESSAGES)
+ /* This define is so xgettext can find the internationalizable strings. */
+# define gettext_noop(msgid) msgid
+# else
/* This is for other GNU distributions with internationalized messages. */
-#if HAVE_LIBINTL_H || defined _LIBC
-# include <libintl.h>
-#else
-# define gettext(msgid) (msgid)
-#endif
+# include "gettext.h"
+# endif
-#ifndef gettext_noop
-/* This define is so xgettext can find the internationalizable
- strings. */
-# define gettext_noop(String) String
-#endif
+/* Support for bounded pointers. */
+# if !defined _LIBC && !defined __BOUNDED_POINTERS__
+# define __bounded /* nothing */
+# define __unbounded /* nothing */
+# define __ptrvalue /* nothing */
+# endif
/* The `emacs' switch turns on certain matching commands
that make sense only in Emacs. */
-#ifdef emacs
+# ifdef emacs
-# include "lisp.h"
-# include "buffer.h"
-# include "syntax.h"
+# include "lisp.h"
+# include "buffer.h"
+# include "syntax.h"
-#else /* not emacs */
+# else /* not emacs */
/* If we are not linking with Emacs proper,
we can't use the relocating allocator
even if config.h says that we can. */
-# undef REL_ALLOC
+# undef REL_ALLOC
-# if defined STDC_HEADERS || defined _LIBC
# include <stdlib.h>
-# else
-char *malloc ();
-char *realloc ();
-# endif
/* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
If nothing else has been done, use the method below. */
-# ifdef INHIBIT_STRING_HEADER
-# if !(defined HAVE_BZERO && defined HAVE_BCOPY)
-# if !defined bzero && !defined bcopy
-# undef INHIBIT_STRING_HEADER
+# ifdef INHIBIT_STRING_HEADER
+# if !(defined HAVE_BZERO && defined HAVE_BCOPY)
+# if !defined bzero && !defined bcopy
+# undef INHIBIT_STRING_HEADER
+# endif
# endif
# endif
-# endif
/* This is the normal way of making sure we have a bcopy and a bzero.
This is used in most programs--a few other programs avoid this
by defining INHIBIT_STRING_HEADER. */
-# ifndef INHIBIT_STRING_HEADER
-# if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
+# ifndef INHIBIT_STRING_HEADER
# include <string.h>
# ifndef bzero
# ifndef _LIBC
# define bzero(s, n) __bzero (s, n)
# endif
# endif
-# else
-# include <strings.h>
-# ifndef memcmp
-# define memcmp(s1, s2, n) bcmp (s1, s2, n)
-# endif
-# ifndef memcpy
-# define memcpy(d, s, n) (bcopy (s, d, n), (d))
-# endif
# endif
-# endif
/* Define the syntax stuff for \<, \>, etc. */
/* This must be nonzero for the wordchar and notwordchar pattern
commands in re_match_2. */
-# ifndef Sword
-# define Sword 1
+# ifndef Sword
+# define Sword 1
+# endif
+
+# ifdef SWITCH_ENUM_BUG
+# define SWITCH_ENUM_CAST(x) ((int)(x))
+# else
+# define SWITCH_ENUM_CAST(x) (x)
+# endif
+
+# endif /* not emacs */
+
+# include <limits.h>
+
+# ifndef MB_LEN_MAX
+# define MB_LEN_MAX 1
# endif
+\f
+/* Get the interface, including the syntax bits. */
+# include <regex.h>
+
+/* isalpha etc. are used for the character classes. */
+# include <ctype.h>
-# ifdef SWITCH_ENUM_BUG
-# define SWITCH_ENUM_CAST(x) ((int)(x))
+/* Jim Meyering writes:
+
+ "... Some ctype macros are valid only for character codes that
+ isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
+ using /bin/cc or gcc but without giving an ansi option). So, all
+ ctype uses should be through macros like ISPRINT... If
+ STDC_HEADERS is defined, then autoconf has verified that the ctype
+ macros don't need to be guarded with references to isascii. ...
+ Defining isascii to 1 should let any compiler worth its salt
+ eliminate the && through constant folding."
+ Solaris defines some of these symbols so we must undefine them first. */
+
+# if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
+# define IN_CTYPE_DOMAIN(c) 1
# else
-# define SWITCH_ENUM_CAST(x) (x)
+# define IN_CTYPE_DOMAIN(c) isascii(c)
# endif
+# ifdef isblank
+# define ISBLANK(c) (IN_CTYPE_DOMAIN (c) && isblank (c))
+# else
+# define ISBLANK(c) ((c) == ' ' || (c) == '\t')
+# endif
+# ifdef isgraph
+# define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isgraph (c))
+# else
+# define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isprint (c) && !isspace (c))
+# endif
+
+# undef ISPRINT
+# define ISPRINT(c) (IN_CTYPE_DOMAIN (c) && isprint (c))
+# define ISDIGIT(c) (IN_CTYPE_DOMAIN (c) && isdigit (c))
+# define ISALNUM(c) (IN_CTYPE_DOMAIN (c) && isalnum (c))
+# define ISALPHA(c) (IN_CTYPE_DOMAIN (c) && isalpha (c))
+# define ISCNTRL(c) (IN_CTYPE_DOMAIN (c) && iscntrl (c))
+# define ISLOWER(c) (IN_CTYPE_DOMAIN (c) && islower (c))
+# define ISPUNCT(c) (IN_CTYPE_DOMAIN (c) && ispunct (c))
+# define ISSPACE(c) (IN_CTYPE_DOMAIN (c) && isspace (c))
+# define ISUPPER(c) (IN_CTYPE_DOMAIN (c) && isupper (c))
+# define ISXDIGIT(c) (IN_CTYPE_DOMAIN (c) && isxdigit (c))
+
+# ifdef _tolower
+# define TOLOWER(c) _tolower(c)
+# else
+# define TOLOWER(c) tolower(c)
+# endif
+\f
+# ifndef emacs
/* How many characters in the character set. */
-# define CHAR_SET_SIZE 256
+# define CHAR_SET_SIZE 256
-# ifdef SYNTAX_TABLE
+# ifdef SYNTAX_TABLE
extern char *re_syntax_table;
-# else /* not SYNTAX_TABLE */
+# else /* not SYNTAX_TABLE */
static char re_syntax_table[CHAR_SET_SIZE];
static void
-init_syntax_once ()
+init_syntax_once (void)
{
register int c;
static int done = 0;
if (done)
return;
-
bzero (re_syntax_table, sizeof re_syntax_table);
- for (c = 'a'; c <= 'z'; c++)
- re_syntax_table[c] = Sword;
-
- for (c = 'A'; c <= 'Z'; c++)
- re_syntax_table[c] = Sword;
-
- for (c = '0'; c <= '9'; c++)
- re_syntax_table[c] = Sword;
+ for (c = 0; c < CHAR_SET_SIZE; ++c)
+ if (ISALNUM (c))
+ re_syntax_table[c] = Sword;
re_syntax_table['_'] = Sword;
done = 1;
}
-# endif /* not SYNTAX_TABLE */
+# endif /* not SYNTAX_TABLE */
-# define SYNTAX(c) re_syntax_table[c]
+# define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
-#endif /* not emacs */
-\f
-/* Get the interface, including the syntax bits. */
-#include "regex.h"
-
-/* isalpha etc. are used for the character classes. */
-#include <ctype.h>
-
-/* Jim Meyering writes:
-
- "... Some ctype macros are valid only for character codes that
- isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
- using /bin/cc or gcc but without giving an ansi option). So, all
- ctype uses should be through macros like ISPRINT... If
- STDC_HEADERS is defined, then autoconf has verified that the ctype
- macros don't need to be guarded with references to isascii. ...
- Defining isascii to 1 should let any compiler worth its salt
- eliminate the && through constant folding."
- Solaris defines some of these symbols so we must undefine them first. */
-
-#undef ISASCII
-#if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
-# define ISASCII(c) 1
-#else
-# define ISASCII(c) isascii(c)
-#endif
-
-#ifdef isblank
-# define ISBLANK(c) (ISASCII (c) && isblank (c))
-#else
-# define ISBLANK(c) ((c) == ' ' || (c) == '\t')
-#endif
-#ifdef isgraph
-# define ISGRAPH(c) (ISASCII (c) && isgraph (c))
-#else
-# define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
-#endif
-
-#undef ISPRINT
-#define ISPRINT(c) (ISASCII (c) && isprint (c))
-#define ISDIGIT(c) (ISASCII (c) && isdigit (c))
-#define ISALNUM(c) (ISASCII (c) && isalnum (c))
-#define ISALPHA(c) (ISASCII (c) && isalpha (c))
-#define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
-#define ISLOWER(c) (ISASCII (c) && islower (c))
-#define ISPUNCT(c) (ISASCII (c) && ispunct (c))
-#define ISSPACE(c) (ISASCII (c) && isspace (c))
-#define ISUPPER(c) (ISASCII (c) && isupper (c))
-#define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
-
-#ifndef NULL
-# define NULL (void *)0
-#endif
-
-/* We remove any previous definition of `SIGN_EXTEND_CHAR',
- since ours (we hope) works properly with all combinations of
- machines, compilers, `char' and `unsigned char' argument types.
- (Per Bothner suggested the basic approach.) */
-#undef SIGN_EXTEND_CHAR
-#if __STDC__
-# define SIGN_EXTEND_CHAR(c) ((signed char) (c))
-#else /* not __STDC__ */
-/* As in Harbison and Steele. */
-# define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
-#endif
+# endif /* emacs */
\f
/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
use `alloca' instead of `malloc'. This is because using malloc in
not functions -- `alloca'-allocated space disappears at the end of the
function it is called in. */
-#ifdef REGEX_MALLOC
+# ifdef REGEX_MALLOC
-# define REGEX_ALLOCATE malloc
-# define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
-# define REGEX_FREE free
+# define REGEX_ALLOCATE malloc
+# define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
+# define REGEX_FREE free
-#else /* not REGEX_MALLOC */
+# else /* not REGEX_MALLOC */
/* Emacs already defines alloca, sometimes. */
-# ifndef alloca
+# ifndef alloca
/* Make alloca work the best possible way. */
-# ifdef __GNUC__
-# define alloca __builtin_alloca
-# else /* not __GNUC__ */
-# if HAVE_ALLOCA_H
-# include <alloca.h>
-# endif /* HAVE_ALLOCA_H */
-# endif /* not __GNUC__ */
+# include <alloca.h>
-# endif /* not alloca */
+# endif /* not alloca */
-# define REGEX_ALLOCATE alloca
+# define REGEX_ALLOCATE alloca
/* Assumes a `char *destination' variable. */
-# define REGEX_REALLOCATE(source, osize, nsize) \
+# define REGEX_REALLOCATE(source, osize, nsize) \
(destination = (char *) alloca (nsize), \
memcpy (destination, source, osize))
/* No need to do anything to free, after alloca. */
-# define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
+# define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
-#endif /* not REGEX_MALLOC */
+# endif /* not REGEX_MALLOC */
/* Define how to allocate the failure stack. */
-#if defined REL_ALLOC && defined REGEX_MALLOC
+# if defined REL_ALLOC && defined REGEX_MALLOC
-# define REGEX_ALLOCATE_STACK(size) \
+# define REGEX_ALLOCATE_STACK(size) \
r_alloc (&failure_stack_ptr, (size))
-# define REGEX_REALLOCATE_STACK(source, osize, nsize) \
+# define REGEX_REALLOCATE_STACK(source, osize, nsize) \
r_re_alloc (&failure_stack_ptr, (nsize))
-# define REGEX_FREE_STACK(ptr) \
+# define REGEX_FREE_STACK(ptr) \
r_alloc_free (&failure_stack_ptr)
-#else /* not using relocating allocator */
+# else /* not using relocating allocator */
-# ifdef REGEX_MALLOC
+# ifdef REGEX_MALLOC
-# define REGEX_ALLOCATE_STACK malloc
-# define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
-# define REGEX_FREE_STACK free
+# define REGEX_ALLOCATE_STACK malloc
+# define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
+# define REGEX_FREE_STACK free
-# else /* not REGEX_MALLOC */
+# else /* not REGEX_MALLOC */
-# define REGEX_ALLOCATE_STACK alloca
+# define REGEX_ALLOCATE_STACK alloca
-# define REGEX_REALLOCATE_STACK(source, osize, nsize) \
+# define REGEX_REALLOCATE_STACK(source, osize, nsize) \
REGEX_REALLOCATE (source, osize, nsize)
/* No need to explicitly free anything. */
-# define REGEX_FREE_STACK(arg)
+# define REGEX_FREE_STACK(arg)
-# endif /* not REGEX_MALLOC */
-#endif /* not using relocating allocator */
+# endif /* not REGEX_MALLOC */
+# endif /* not using relocating allocator */
/* True if `size1' is non-NULL and PTR is pointing anywhere inside
`string1' or just past its end. This works if PTR is NULL, which is
a good thing. */
-#define FIRST_STRING_P(ptr) \
+# define FIRST_STRING_P(ptr) \
(size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
/* (Re)Allocate N items of type T using malloc, or fail. */
-#define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
-#define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
-#define RETALLOC_IF(addr, n, t) \
+# define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
+# define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
+# define RETALLOC_IF(addr, n, t) \
if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
-#define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
+# define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
-#define BYTEWIDTH 8 /* In bits. */
+# define BYTEWIDTH 8 /* In bits. */
-#define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
+# define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
-#undef MAX
-#undef MIN
-#define MAX(a, b) ((a) > (b) ? (a) : (b))
-#define MIN(a, b) ((a) < (b) ? (a) : (b))
+# undef MAX
+# undef MIN
+# define MAX(a, b) ((a) > (b) ? (a) : (b))
+# define MIN(a, b) ((a) < (b) ? (a) : (b))
typedef char boolean;
-#define false 0
-#define true 1
-
-static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
- const char *string1, int size1,
- const char *string2, int size2,
- int pos,
- struct re_registers *regs,
- int stop));
+# define false 0
+# define true 1
+
+static reg_errcode_t byte_regex_compile (const char *pattern, size_t size,
+ reg_syntax_t syntax,
+ struct re_pattern_buffer *bufp);
+
+static int byte_re_match_2_internal (struct re_pattern_buffer *bufp,
+ const char *string1, int size1,
+ const char *string2, int size2,
+ int pos,
+ struct re_registers *regs,
+ int stop);
+static int byte_re_search_2 (struct re_pattern_buffer *bufp,
+ const char *string1, int size1,
+ const char *string2, int size2,
+ int startpos, int range,
+ struct re_registers *regs, int stop);
+static int byte_re_compile_fastmap (struct re_pattern_buffer *bufp);
+
+#ifdef MBS_SUPPORT
+static reg_errcode_t wcs_regex_compile (const char *pattern, size_t size,
+ reg_syntax_t syntax,
+ struct re_pattern_buffer *bufp);
+
+
+static int wcs_re_match_2_internal (struct re_pattern_buffer *bufp,
+ const char *cstring1, int csize1,
+ const char *cstring2, int csize2,
+ int pos,
+ struct re_registers *regs,
+ int stop,
+ wchar_t *string1, int size1,
+ wchar_t *string2, int size2,
+ int *mbs_offset1, int *mbs_offset2);
+static int wcs_re_search_2 (struct re_pattern_buffer *bufp,
+ const char *string1, int size1,
+ const char *string2, int size2,
+ int startpos, int range,
+ struct re_registers *regs, int stop);
+static int wcs_re_compile_fastmap (struct re_pattern_buffer *bufp);
+#endif
\f
/* These are the command codes that appear in compiled regular
expressions. Some opcodes are followed by argument bytes. A
/* Followed by one byte giving n, then by n literal bytes. */
exactn,
+# ifdef MBS_SUPPORT
+ /* Same as exactn, but contains binary data. */
+ exactn_bin,
+# endif
+
/* Matches any (more or less) character. */
anychar,
are ordered low-bit-first. A character is in the set if its
bit is 1. A character too large to have a bit in the map is
automatically not in the set. */
+ /* ifdef MBS_SUPPORT, following element is length of character
+ classes, length of collating symbols, length of equivalence
+ classes, length of character ranges, and length of characters.
+ Next, character class element, collating symbols elements,
+ equivalence class elements, range elements, and character
+ elements follow.
+ See regex_compile function. */
charset,
/* Same parameters as charset, but match any character that is
/* Followed by two-byte relative address of place to resume at
in case of failure. */
+ /* ifdef MBS_SUPPORT, the size of address is 1. */
on_failure_jump,
/* Like on_failure_jump, but pushes a placeholder instead of the
/* Throw away latest failure point and then jump to following
two-byte relative address. */
+ /* ifdef MBS_SUPPORT, the size of address is 1. */
pop_failure_jump,
/* Change to pop_failure_jump if know won't have to backtrack to
sure that there is no use backtracking out of repetitions
already matched, then we change it to a pop_failure_jump.
Followed by two-byte address. */
+ /* ifdef MBS_SUPPORT, the size of address is 1. */
maybe_pop_jump,
/* Jump to following two-byte address, and push a dummy failure
is made to use it for a failure. A `+' construct makes this
before the first repeat. Also used as an intermediary kind
of jump when compiling an alternative. */
+ /* ifdef MBS_SUPPORT, the size of address is 1. */
dummy_failure_jump,
/* Push a dummy failure point and continue. Used at the end of
/* Followed by two-byte relative address and two-byte number n.
After matching N times, jump to the address upon failure. */
+ /* ifdef MBS_SUPPORT, the size of address is 1. */
succeed_n,
/* Followed by two-byte relative address, and two-byte number n.
Jump to the address N times, then fail. */
+ /* ifdef MBS_SUPPORT, the size of address is 1. */
jump_n,
/* Set the following two-byte relative address to the
subsequent two-byte number. The address *includes* the two
bytes of number. */
+ /* ifdef MBS_SUPPORT, the size of address is 1. */
set_number_at,
wordchar, /* Matches any word-constituent character. */
wordbound, /* Succeeds if at a word boundary. */
notwordbound /* Succeeds if not at a word boundary. */
-#ifdef emacs
+# ifdef emacs
,before_dot, /* Succeeds if before point. */
at_dot, /* Succeeds if at point. */
after_dot, /* Succeeds if after point. */
/* Matches any character whose syntax is not that specified. */
notsyntaxspec
-#endif /* emacs */
+# endif /* emacs */
} re_opcode_t;
+#endif /* not INSIDE_RECURSION */
\f
+
+#ifdef BYTE
+# define CHAR_T char
+# define UCHAR_T unsigned char
+# define COMPILED_BUFFER_VAR bufp->buffer
+# define OFFSET_ADDRESS_SIZE 2
+# define PREFIX(name) byte_##name
+# define ARG_PREFIX(name) name
+# define PUT_CHAR(c) putchar (c)
+#else
+# ifdef WCHAR
+# define CHAR_T wchar_t
+# define UCHAR_T wchar_t
+# define COMPILED_BUFFER_VAR wc_buffer
+# define OFFSET_ADDRESS_SIZE 1 /* the size which STORE_NUMBER macro use */
+# define CHAR_CLASS_SIZE ((__alignof__(wctype_t)+sizeof(wctype_t))/sizeof(CHAR_T)+1)
+# define PREFIX(name) wcs_##name
+# define ARG_PREFIX(name) c##name
+/* Should we use wide stream?? */
+# define PUT_CHAR(c) printf ("%C", c);
+# define TRUE 1
+# define FALSE 0
+# else
+# ifdef MBS_SUPPORT
+# define WCHAR
+# define INSIDE_RECURSION
+# include "regex.c"
+# undef INSIDE_RECURSION
+# endif
+# define BYTE
+# define INSIDE_RECURSION
+# include "regex.c"
+# undef INSIDE_RECURSION
+# endif
+#endif
+
+#if USE_UNLOCKED_IO
+# include "unlocked-io.h"
+#endif
+
+#ifdef INSIDE_RECURSION
/* Common operations on the compiled pattern. */
/* Store NUMBER in two contiguous bytes starting at DESTINATION. */
+/* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
-#define STORE_NUMBER(destination, number) \
+# ifdef WCHAR
+# define STORE_NUMBER(destination, number) \
+ do { \
+ *(destination) = (UCHAR_T)(number); \
+ } while (0)
+# else /* BYTE */
+# define STORE_NUMBER(destination, number) \
do { \
(destination)[0] = (number) & 0377; \
(destination)[1] = (number) >> 8; \
} while (0)
+# endif /* WCHAR */
/* Same as STORE_NUMBER, except increment DESTINATION to
the byte after where the number is stored. Therefore, DESTINATION
must be an lvalue. */
+/* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
-#define STORE_NUMBER_AND_INCR(destination, number) \
+# define STORE_NUMBER_AND_INCR(destination, number) \
do { \
STORE_NUMBER (destination, number); \
- (destination) += 2; \
+ (destination) += OFFSET_ADDRESS_SIZE; \
} while (0)
/* Put into DESTINATION a number stored in two contiguous bytes starting
at SOURCE. */
+/* ifdef MBS_SUPPORT, we store NUMBER in 1 element. */
-#define EXTRACT_NUMBER(destination, source) \
+# ifdef WCHAR
+# define EXTRACT_NUMBER(destination, source) \
+ do { \
+ (destination) = *(source); \
+ } while (0)
+# else /* BYTE */
+# define EXTRACT_NUMBER(destination, source) \
do { \
(destination) = *(source) & 0377; \
- (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
+ (destination) += (signed char) (*((source) + 1)) << 8; \
} while (0)
+# endif
-#ifdef DEBUG
-static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
+# ifdef DEBUG
static void
-extract_number (dest, source)
- int *dest;
- unsigned char *source;
+PREFIX(extract_number) (int *dest, UCHAR_T *source)
{
- int temp = SIGN_EXTEND_CHAR (*(source + 1));
+# ifdef WCHAR
+ *dest = *source;
+# else /* BYTE */
+ signed char temp = source[1];
*dest = *source & 0377;
*dest += temp << 8;
+# endif
}
-# ifndef EXTRACT_MACROS /* To debug the macros. */
-# undef EXTRACT_NUMBER
-# define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
-# endif /* not EXTRACT_MACROS */
+# ifndef EXTRACT_MACROS /* To debug the macros. */
+# undef EXTRACT_NUMBER
+# define EXTRACT_NUMBER(dest, src) PREFIX(extract_number) (&dest, src)
+# endif /* not EXTRACT_MACROS */
-#endif /* DEBUG */
+# endif /* DEBUG */
/* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
SOURCE must be an lvalue. */
-#define EXTRACT_NUMBER_AND_INCR(destination, source) \
+# define EXTRACT_NUMBER_AND_INCR(destination, source) \
do { \
EXTRACT_NUMBER (destination, source); \
- (source) += 2; \
+ (source) += OFFSET_ADDRESS_SIZE; \
} while (0)
-#ifdef DEBUG
-static void extract_number_and_incr _RE_ARGS ((int *destination,
- unsigned char **source));
+# ifdef DEBUG
static void
-extract_number_and_incr (destination, source)
- int *destination;
- unsigned char **source;
+PREFIX(extract_number_and_incr) (int *destination, UCHAR_T **source)
{
- extract_number (destination, *source);
- *source += 2;
+ PREFIX(extract_number) (destination, *source);
+ *source += OFFSET_ADDRESS_SIZE;
}
-# ifndef EXTRACT_MACROS
-# undef EXTRACT_NUMBER_AND_INCR
-# define EXTRACT_NUMBER_AND_INCR(dest, src) \
- extract_number_and_incr (&dest, &src)
-# endif /* not EXTRACT_MACROS */
+# ifndef EXTRACT_MACROS
+# undef EXTRACT_NUMBER_AND_INCR
+# define EXTRACT_NUMBER_AND_INCR(dest, src) \
+ PREFIX(extract_number_and_incr) (&dest, &src)
+# endif /* not EXTRACT_MACROS */
+
+# endif /* DEBUG */
-#endif /* DEBUG */
\f
+
/* If DEBUG is defined, Regex prints many voluminous messages about what
it is doing (if the variable `debug' is nonzero). If linked with the
main program in `iregex.c', you can enter patterns and strings
interactively. And if linked with the main program in `main.c' and
the other test files, you can run the already-written tests. */
-#ifdef DEBUG
+# ifdef DEBUG
+
+# ifndef DEFINED_ONCE
/* We use standard I/O for debugging. */
-# include <stdio.h>
+# include <stdio.h>
/* It is useful to test things that ``must'' be true when debugging. */
-# include <assert.h>
+# include <assert.h>
-static int debug = 0;
+static int debug;
-# define DEBUG_STATEMENT(e) e
-# define DEBUG_PRINT1(x) if (debug) printf (x)
-# define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
-# define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
-# define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
-# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
- if (debug) print_partial_compiled_pattern (s, e)
-# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
- if (debug) print_double_string (w, s1, sz1, s2, sz2)
+# define DEBUG_STATEMENT(e) e
+# define DEBUG_PRINT1(x) if (debug) printf (x)
+# define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
+# define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
+# define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
+# endif /* not DEFINED_ONCE */
+
+# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
+ if (debug) PREFIX(print_partial_compiled_pattern) (s, e)
+# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
+ if (debug) PREFIX(print_double_string) (w, s1, sz1, s2, sz2)
/* Print the fastmap in human-readable form. */
+# ifndef DEFINED_ONCE
void
-print_fastmap (fastmap)
- char *fastmap;
+print_fastmap (char *fastmap)
{
unsigned was_a_range = 0;
unsigned i = 0;
}
putchar ('\n');
}
+# endif /* not DEFINED_ONCE */
/* Print a compiled pattern string in human-readable form, starting at
the START pointer into it and ending just before the pointer END. */
void
-print_partial_compiled_pattern (start, end)
- unsigned char *start;
- unsigned char *end;
+PREFIX(print_partial_compiled_pattern) (UCHAR_T *start, UCHAR_T *end)
{
int mcnt, mcnt2;
- unsigned char *p1;
- unsigned char *p = start;
- unsigned char *pend = end;
+ UCHAR_T *p1;
+ UCHAR_T *p = start;
+ UCHAR_T *pend = end;
if (start == NULL)
{
/* Loop over pattern commands. */
while (p < pend)
{
- printf ("%d:\t", p - start);
+# ifdef _LIBC
+ printf ("%td:\t", p - start);
+# else
+ printf ("%ld:\t", (long int) (p - start));
+# endif
switch ((re_opcode_t) *p++)
{
do
{
putchar ('/');
- putchar (*p++);
+ PUT_CHAR (*p++);
+ }
+ while (--mcnt);
+ break;
+
+# ifdef MBS_SUPPORT
+ case exactn_bin:
+ mcnt = *p++;
+ printf ("/exactn_bin/%d", mcnt);
+ do
+ {
+ printf("/%lx", (long int) *p++);
}
while (--mcnt);
break;
+# endif /* MBS_SUPPORT */
case start_memory:
mcnt = *p++;
- printf ("/start_memory/%d/%d", mcnt, *p++);
+ printf ("/start_memory/%d/%ld", mcnt, (long int) *p++);
break;
case stop_memory:
mcnt = *p++;
- printf ("/stop_memory/%d/%d", mcnt, *p++);
+ printf ("/stop_memory/%d/%ld", mcnt, (long int) *p++);
break;
case duplicate:
- printf ("/duplicate/%d", *p++);
+ printf ("/duplicate/%ld", (long int) *p++);
break;
case anychar:
case charset:
case charset_not:
{
+# ifdef WCHAR
+ int i, length;
+ wchar_t *workp = p;
+ printf ("/charset [%s",
+ (re_opcode_t) *(workp - 1) == charset_not ? "^" : "");
+ p += 5;
+ length = *workp++; /* the length of char_classes */
+ for (i=0 ; i<length ; i++)
+ printf("[:%lx:]", (long int) *p++);
+ length = *workp++; /* the length of collating_symbol */
+ for (i=0 ; i<length ;)
+ {
+ printf("[.");
+ while(*p != 0)
+ PUT_CHAR((i++,*p++));
+ i++,p++;
+ printf(".]");
+ }
+ length = *workp++; /* the length of equivalence_class */
+ for (i=0 ; i<length ;)
+ {
+ printf("[=");
+ while(*p != 0)
+ PUT_CHAR((i++,*p++));
+ i++,p++;
+ printf("=]");
+ }
+ length = *workp++; /* the length of char_range */
+ for (i=0 ; i<length ; i++)
+ {
+ wchar_t range_start = *p++;
+ wchar_t range_end = *p++;
+ printf("%C-%C", range_start, range_end);
+ }
+ length = *workp++; /* the length of char */
+ for (i=0 ; i<length ; i++)
+ printf("%C", *p++);
+ putchar (']');
+# else
register int c, last = -100;
register int in_range = 0;
putchar (']');
p += 1 + *p;
+# endif /* WCHAR */
}
break;
break;
case on_failure_jump:
- extract_number_and_incr (&mcnt, &p);
- printf ("/on_failure_jump to %d", p + mcnt - start);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
+# ifdef _LIBC
+ printf ("/on_failure_jump to %td", p + mcnt - start);
+# else
+ printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
+# endif
break;
case on_failure_keep_string_jump:
- extract_number_and_incr (&mcnt, &p);
- printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
+# ifdef _LIBC
+ printf ("/on_failure_keep_string_jump to %td", p + mcnt - start);
+# else
+ printf ("/on_failure_keep_string_jump to %ld",
+ (long int) (p + mcnt - start));
+# endif
break;
case dummy_failure_jump:
- extract_number_and_incr (&mcnt, &p);
- printf ("/dummy_failure_jump to %d", p + mcnt - start);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
+# ifdef _LIBC
+ printf ("/dummy_failure_jump to %td", p + mcnt - start);
+# else
+ printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
+# endif
break;
case push_dummy_failure:
break;
case maybe_pop_jump:
- extract_number_and_incr (&mcnt, &p);
- printf ("/maybe_pop_jump to %d", p + mcnt - start);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
+# ifdef _LIBC
+ printf ("/maybe_pop_jump to %td", p + mcnt - start);
+# else
+ printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
+# endif
break;
case pop_failure_jump:
- extract_number_and_incr (&mcnt, &p);
- printf ("/pop_failure_jump to %d", p + mcnt - start);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
+# ifdef _LIBC
+ printf ("/pop_failure_jump to %td", p + mcnt - start);
+# else
+ printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
+# endif
break;
case jump_past_alt:
- extract_number_and_incr (&mcnt, &p);
- printf ("/jump_past_alt to %d", p + mcnt - start);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
+# ifdef _LIBC
+ printf ("/jump_past_alt to %td", p + mcnt - start);
+# else
+ printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
+# endif
break;
case jump:
- extract_number_and_incr (&mcnt, &p);
- printf ("/jump to %d", p + mcnt - start);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
+# ifdef _LIBC
+ printf ("/jump to %td", p + mcnt - start);
+# else
+ printf ("/jump to %ld", (long int) (p + mcnt - start));
+# endif
break;
case succeed_n:
- extract_number_and_incr (&mcnt, &p);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
p1 = p + mcnt;
- extract_number_and_incr (&mcnt2, &p);
- printf ("/succeed_n to %d, %d times", p1 - start, mcnt2);
+ PREFIX(extract_number_and_incr) (&mcnt2, &p);
+# ifdef _LIBC
+ printf ("/succeed_n to %td, %d times", p1 - start, mcnt2);
+# else
+ printf ("/succeed_n to %ld, %d times",
+ (long int) (p1 - start), mcnt2);
+# endif
break;
case jump_n:
- extract_number_and_incr (&mcnt, &p);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
p1 = p + mcnt;
- extract_number_and_incr (&mcnt2, &p);
+ PREFIX(extract_number_and_incr) (&mcnt2, &p);
printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
break;
case set_number_at:
- extract_number_and_incr (&mcnt, &p);
+ PREFIX(extract_number_and_incr) (&mcnt, &p);
p1 = p + mcnt;
- extract_number_and_incr (&mcnt2, &p);
- printf ("/set_number_at location %d to %d", p1 - start, mcnt2);
+ PREFIX(extract_number_and_incr) (&mcnt2, &p);
+# ifdef _LIBC
+ printf ("/set_number_at location %td to %d", p1 - start, mcnt2);
+# else
+ printf ("/set_number_at location %ld to %d",
+ (long int) (p1 - start), mcnt2);
+# endif
break;
case wordbound:
case wordend:
printf ("/wordend");
+ break;
-# ifdef emacs
+# ifdef emacs
case before_dot:
printf ("/before_dot");
break;
mcnt = *p++;
printf ("/%d", mcnt);
break;
-# endif /* emacs */
+# endif /* emacs */
case wordchar:
printf ("/wordchar");
break;
default:
- printf ("?%d", *(p-1));
+ printf ("?%ld", (long int) *(p-1));
}
putchar ('\n');
}
- printf ("%d:\tend of pattern.\n", p - start);
+# ifdef _LIBC
+ printf ("%td:\tend of pattern.\n", p - start);
+# else
+ printf ("%ld:\tend of pattern.\n", (long int) (p - start));
+# endif
}
void
-print_compiled_pattern (bufp)
- struct re_pattern_buffer *bufp;
+PREFIX(print_compiled_pattern) (struct re_pattern_buffer *bufp)
{
- unsigned char *buffer = bufp->buffer;
+ UCHAR_T *buffer = (UCHAR_T*) bufp->buffer;
- print_partial_compiled_pattern (buffer, buffer + bufp->used);
+ PREFIX(print_partial_compiled_pattern) (buffer, buffer
+ + bufp->used / sizeof(UCHAR_T));
printf ("%ld bytes used/%ld bytes allocated.\n",
bufp->used, bufp->allocated);
print_fastmap (bufp->fastmap);
}
- printf ("re_nsub: %d\t", bufp->re_nsub);
+# ifdef _LIBC
+ printf ("re_nsub: %Zd\t", bufp->re_nsub);
+# else
+ printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
+# endif
printf ("regs_alloc: %d\t", bufp->regs_allocated);
printf ("can_be_null: %d\t", bufp->can_be_null);
printf ("newline_anchor: %d\n", bufp->newline_anchor);
void
-print_double_string (where, string1, size1, string2, size2)
- const char *where;
- const char *string1;
- const char *string2;
- int size1;
- int size2;
+PREFIX(print_double_string) (const CHAR_T *where,
+ const CHAR_T *string1,
+ const CHAR_T *string2,
+ int size1,
+ int size2)
{
int this_char;
printf ("(null)");
else
{
+ int cnt;
+
if (FIRST_STRING_P (where))
{
for (this_char = where - string1; this_char < size1; this_char++)
- putchar (string1[this_char]);
+ PUT_CHAR (string1[this_char]);
where = string2;
}
+ cnt = 0;
for (this_char = where - string2; this_char < size2; this_char++)
- putchar (string2[this_char]);
+ {
+ PUT_CHAR (string2[this_char]);
+ if (++cnt > 100)
+ {
+ fputs ("...", stdout);
+ break;
+ }
+ }
}
}
+# ifndef DEFINED_ONCE
void
printchar (c)
int c;
{
putc (c, stderr);
}
+# endif
+
+# else /* not DEBUG */
-#else /* not DEBUG */
+# ifndef DEFINED_ONCE
+# undef assert
+# define assert(e)
-# undef assert
-# define assert(e)
+# define DEBUG_STATEMENT(e)
+# define DEBUG_PRINT1(x)
+# define DEBUG_PRINT2(x1, x2)
+# define DEBUG_PRINT3(x1, x2, x3)
+# define DEBUG_PRINT4(x1, x2, x3, x4)
+# endif /* not DEFINED_ONCE */
+# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
+# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
-# define DEBUG_STATEMENT(e)
-# define DEBUG_PRINT1(x)
-# define DEBUG_PRINT2(x1, x2)
-# define DEBUG_PRINT3(x1, x2, x3)
-# define DEBUG_PRINT4(x1, x2, x3, x4)
-# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
-# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
+# endif /* not DEBUG */
-#endif /* not DEBUG */
\f
+
+# ifdef WCHAR
+/* This convert a multibyte string to a wide character string.
+ And write their correspondances to offset_buffer(see below)
+ and write whether each wchar_t is binary data to is_binary.
+ This assume invalid multibyte sequences as binary data.
+ We assume offset_buffer and is_binary is already allocated
+ enough space. */
+
+static size_t
+convert_mbs_to_wcs (CHAR_T *dest,
+ const unsigned char* src,
+
+ /* The length of multibyte string. */
+ size_t len,
+
+ /* Correspondences between src(char string) and
+ dest(wchar_t string) for optimization. E.g.:
+ src = "xxxyzz"
+ dest = {'X', 'Y', 'Z'}
+ (each "xxx", "y" and "zz" represent one
+ multibyte character corresponding to 'X',
+ 'Y' and 'Z'.)
+ offset_buffer = {0, 0+3("xxx"), 0+3+1("y"),
+ 0+3+1+2("zz")}
+ = {0, 3, 4, 6} */
+ int *offset_buffer,
+
+ char *is_binary)
+{
+ wchar_t *pdest = dest;
+ const unsigned char *psrc = src;
+ size_t wc_count = 0;
+
+ mbstate_t mbs;
+ int i, consumed;
+ size_t mb_remain = len;
+ size_t mb_count = 0;
+
+ /* Initialize the conversion state. */
+ memset (&mbs, 0, sizeof (mbstate_t));
+
+ offset_buffer[0] = 0;
+ for( ; mb_remain > 0 ; ++wc_count, ++pdest, mb_remain -= consumed,
+ psrc += consumed)
+ {
+ consumed = mbrtowc (pdest, psrc, mb_remain, &mbs);
+
+ if (consumed <= 0)
+ /* failed to convert. maybe src contains binary data.
+ So we consume 1 byte manualy. */
+ {
+ *pdest = *psrc;
+ consumed = 1;
+ is_binary[wc_count] = TRUE;
+ }
+ else
+ is_binary[wc_count] = FALSE;
+ /* In sjis encoding, we use yen sign as escape character in
+ place of reverse solidus. So we convert 0x5c(yen sign in
+ sjis) to not 0xa5(yen sign in UCS2) but 0x5c(reverse
+ solidus in UCS2). */
+ if (consumed == 1 && (int) *psrc == 0x5c && (int) *pdest == 0xa5)
+ *pdest = (wchar_t) *psrc;
+
+ offset_buffer[wc_count + 1] = mb_count += consumed;
+ }
+
+ /* Fill remain of the buffer with sentinel. */
+ for (i = wc_count + 1 ; i <= len ; i++)
+ offset_buffer[i] = mb_count + 1;
+
+ return wc_count;
+}
+
+# endif /* WCHAR */
+
+#else /* not INSIDE_RECURSION */
+
/* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
also be assigned to arbitrarily: each pattern buffer stores its own
syntax, so it can be changed between regex compilations. */
defined in regex.h. We return the old syntax. */
reg_syntax_t
-re_set_syntax (syntax)
- reg_syntax_t syntax;
+re_set_syntax (reg_syntax_t syntax)
{
reg_syntax_t ret = re_syntax_options;
re_syntax_options = syntax;
-#ifdef DEBUG
+# ifdef DEBUG
if (syntax & RE_DEBUG)
debug = 1;
else if (debug) /* was on but now is not */
debug = 0;
-#endif /* DEBUG */
+# endif /* DEBUG */
return ret;
}
-#ifdef _LIBC
+# ifdef _LIBC
weak_alias (__re_set_syntax, re_set_syntax)
-#endif
+# endif
\f
/* This table gives an error message for each of the error codes listed
in regex.h. Obviously the order here has to be same as there.
POSIX doesn't require that we do anything for REG_NOERROR,
but why not be nice? */
-static const char *re_error_msgid[] =
+static const char re_error_msgid[] =
{
- gettext_noop ("Success"), /* REG_NOERROR */
- gettext_noop ("No match"), /* REG_NOMATCH */
- gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
- gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
- gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
- gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
- gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
- gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
- gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
- gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
- gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
- gettext_noop ("Invalid range end"), /* REG_ERANGE */
- gettext_noop ("Memory exhausted"), /* REG_ESPACE */
- gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
- gettext_noop ("Premature end of regular expression"), /* REG_EEND */
- gettext_noop ("Regular expression too big"), /* REG_ESIZE */
- gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
+# define REG_NOERROR_IDX 0
+ gettext_noop ("Success") /* REG_NOERROR */
+ "\0"
+# define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
+ gettext_noop ("No match") /* REG_NOMATCH */
+ "\0"
+# define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
+ gettext_noop ("Invalid regular expression") /* REG_BADPAT */
+ "\0"
+# define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
+ gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
+ "\0"
+# define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
+ gettext_noop ("Invalid character class name") /* REG_ECTYPE */
+ "\0"
+# define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
+ gettext_noop ("Trailing backslash") /* REG_EESCAPE */
+ "\0"
+# define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
+ gettext_noop ("Invalid back reference") /* REG_ESUBREG */
+ "\0"
+# define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
+ gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
+ "\0"
+# define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
+ gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
+ "\0"
+# define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
+ gettext_noop ("Unmatched \\{") /* REG_EBRACE */
+ "\0"
+# define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
+ gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
+ "\0"
+# define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
+ gettext_noop ("Invalid range end") /* REG_ERANGE */
+ "\0"
+# define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
+ gettext_noop ("Memory exhausted") /* REG_ESPACE */
+ "\0"
+# define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
+ gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
+ "\0"
+# define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
+ gettext_noop ("Premature end of regular expression") /* REG_EEND */
+ "\0"
+# define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
+ gettext_noop ("Regular expression too big") /* REG_ESIZE */
+ "\0"
+# define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
+ gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
+ };
+
+static const size_t re_error_msgid_idx[] =
+ {
+ REG_NOERROR_IDX,
+ REG_NOMATCH_IDX,
+ REG_BADPAT_IDX,
+ REG_ECOLLATE_IDX,
+ REG_ECTYPE_IDX,
+ REG_EESCAPE_IDX,
+ REG_ESUBREG_IDX,
+ REG_EBRACK_IDX,
+ REG_EPAREN_IDX,
+ REG_EBRACE_IDX,
+ REG_BADBR_IDX,
+ REG_ERANGE_IDX,
+ REG_ESPACE_IDX,
+ REG_BADRPT_IDX,
+ REG_EEND_IDX,
+ REG_ESIZE_IDX,
+ REG_ERPAREN_IDX
};
\f
+#endif /* INSIDE_RECURSION */
+
+#ifndef DEFINED_ONCE
/* Avoiding alloca during matching, to placate r_alloc. */
/* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
faith that they will not malloc. */
/* Normally, this is fine. */
-#define MATCH_MAY_ALLOCATE
+# define MATCH_MAY_ALLOCATE
/* When using GNU C, we are not REALLY using the C alloca, no matter
what config.h may say. So don't take precautions for it. */
-#ifdef __GNUC__
-# undef C_ALLOCA
-#endif
+# ifdef __GNUC__
+# undef C_ALLOCA
+# endif
/* The match routines may not allocate if (1) they would do it with malloc
and (2) it's not safe for them to use malloc.
Note that if REL_ALLOC is defined, matching would not use malloc for the
failure stack, but we would still use it for the register vectors;
so REL_ALLOC should not affect this. */
-#if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
-# undef MATCH_MAY_ALLOCATE
-#endif
-
+# if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
+# undef MATCH_MAY_ALLOCATE
+# endif
+#endif /* not DEFINED_ONCE */
\f
+#ifdef INSIDE_RECURSION
/* Failure stack declarations and macros; both re_compile_fastmap and
re_match_2 use a failure stack. These have to be macros because of
REGEX_ALLOCATE_STACK. */
/* Number of failure points for which to initially allocate space
when matching. If this number is exceeded, we allocate more
space, so it is not a hard limit. */
-#ifndef INIT_FAILURE_ALLOC
-# define INIT_FAILURE_ALLOC 5
-#endif
+# ifndef INIT_FAILURE_ALLOC
+# define INIT_FAILURE_ALLOC 5
+# endif
/* Roughly the maximum number of failure points on the stack. Would be
exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
This is a variable only so users of regex can assign to it; we never
change it ourselves. */
-#ifdef INT_IS_16BIT
+# ifdef INT_IS_16BIT
-# if defined MATCH_MAY_ALLOCATE
+# ifndef DEFINED_ONCE
+# if defined MATCH_MAY_ALLOCATE
/* 4400 was enough to cause a crash on Alpha OSF/1,
whose default stack limit is 2mb. */
long int re_max_failures = 4000;
-# else
+# else
long int re_max_failures = 2000;
-# endif
+# endif
+# endif
-union fail_stack_elt
+union PREFIX(fail_stack_elt)
{
- unsigned char *pointer;
+ UCHAR_T *pointer;
long int integer;
};
-typedef union fail_stack_elt fail_stack_elt_t;
+typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
typedef struct
{
- fail_stack_elt_t *stack;
+ PREFIX(fail_stack_elt_t) *stack;
unsigned long int size;
unsigned long int avail; /* Offset of next open position. */
-} fail_stack_type;
+} PREFIX(fail_stack_type);
-#else /* not INT_IS_16BIT */
+# else /* not INT_IS_16BIT */
-# if defined MATCH_MAY_ALLOCATE
+# ifndef DEFINED_ONCE
+# if defined MATCH_MAY_ALLOCATE
/* 4400 was enough to cause a crash on Alpha OSF/1,
whose default stack limit is 2mb. */
-int re_max_failures = 20000;
-# else
+int re_max_failures = 4000;
+# else
int re_max_failures = 2000;
-# endif
+# endif
+# endif
-union fail_stack_elt
+union PREFIX(fail_stack_elt)
{
- unsigned char *pointer;
+ UCHAR_T *pointer;
int integer;
};
-typedef union fail_stack_elt fail_stack_elt_t;
+typedef union PREFIX(fail_stack_elt) PREFIX(fail_stack_elt_t);
typedef struct
{
- fail_stack_elt_t *stack;
+ PREFIX(fail_stack_elt_t) *stack;
unsigned size;
unsigned avail; /* Offset of next open position. */
-} fail_stack_type;
+} PREFIX(fail_stack_type);
-#endif /* INT_IS_16BIT */
+# endif /* INT_IS_16BIT */
-#define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
-#define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
-#define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
+# ifndef DEFINED_ONCE
+# define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
+# define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
+# define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
+# endif
/* Define macros to initialize and free the failure stack.
Do `return -2' if the alloc fails. */
-#ifdef MATCH_MAY_ALLOCATE
-# define INIT_FAIL_STACK() \
+# ifdef MATCH_MAY_ALLOCATE
+# define INIT_FAIL_STACK() \
do { \
- fail_stack.stack = (fail_stack_elt_t *) \
- REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
+ fail_stack.stack = (PREFIX(fail_stack_elt_t) *) \
+ REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (PREFIX(fail_stack_elt_t))); \
\
- if (fail_stack.stack == NULL) \
+ if (fail_stack.stack == NULL) \
return -2; \
\
- fail_stack.size = INIT_FAILURE_ALLOC; \
- fail_stack.avail = 0; \
+ fail_stack.size = INIT_FAILURE_ALLOC; \
+ fail_stack.avail = 0; \
} while (0)
-# define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
-#else
-# define INIT_FAIL_STACK() \
+# define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
+# else
+# define INIT_FAIL_STACK() \
do { \
- fail_stack.avail = 0; \
+ fail_stack.avail = 0; \
} while (0)
-# define RESET_FAIL_STACK()
-#endif
+# define RESET_FAIL_STACK()
+# endif
/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
REGEX_REALLOCATE_STACK requires `destination' be declared. */
-#define DOUBLE_FAIL_STACK(fail_stack) \
+# define DOUBLE_FAIL_STACK(fail_stack) \
((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
? 0 \
- : ((fail_stack).stack = (fail_stack_elt_t *) \
+ : ((fail_stack).stack = (PREFIX(fail_stack_elt_t) *) \
REGEX_REALLOCATE_STACK ((fail_stack).stack, \
- (fail_stack).size * sizeof (fail_stack_elt_t), \
- ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
+ (fail_stack).size * sizeof (PREFIX(fail_stack_elt_t)), \
+ ((fail_stack).size << 1) * sizeof (PREFIX(fail_stack_elt_t))),\
\
(fail_stack).stack == NULL \
? 0 \
/* Push pointer POINTER on FAIL_STACK.
Return 1 if was able to do so and 0 if ran out of memory allocating
space to do so. */
-#define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
+# define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
((FAIL_STACK_FULL () \
&& !DOUBLE_FAIL_STACK (FAIL_STACK)) \
? 0 \
/* Push a pointer value onto the failure stack.
Assumes the variable `fail_stack'. Probably should only
be called from within `PUSH_FAILURE_POINT'. */
-#define PUSH_FAILURE_POINTER(item) \
- fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
+# define PUSH_FAILURE_POINTER(item) \
+ fail_stack.stack[fail_stack.avail++].pointer = (UCHAR_T *) (item)
/* This pushes an integer-valued item onto the failure stack.
Assumes the variable `fail_stack'. Probably should only
be called from within `PUSH_FAILURE_POINT'. */
-#define PUSH_FAILURE_INT(item) \
+# define PUSH_FAILURE_INT(item) \
fail_stack.stack[fail_stack.avail++].integer = (item)
/* Push a fail_stack_elt_t value onto the failure stack.
Assumes the variable `fail_stack'. Probably should only
be called from within `PUSH_FAILURE_POINT'. */
-#define PUSH_FAILURE_ELT(item) \
+# define PUSH_FAILURE_ELT(item) \
fail_stack.stack[fail_stack.avail++] = (item)
/* These three POP... operations complement the three PUSH... operations.
All assume that `fail_stack' is nonempty. */
-#define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
-#define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
-#define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
+# define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
+# define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
+# define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
/* Used to omit pushing failure point id's when we're not debugging. */
-#ifdef DEBUG
-# define DEBUG_PUSH PUSH_FAILURE_INT
-# define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
-#else
-# define DEBUG_PUSH(item)
-# define DEBUG_POP(item_addr)
-#endif
+# ifdef DEBUG
+# define DEBUG_PUSH PUSH_FAILURE_INT
+# define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
+# else
+# define DEBUG_PUSH(item)
+# define DEBUG_POP(item_addr)
+# endif
/* Push the information about the state we will need
Does `return FAILURE_CODE' if runs out of memory. */
-#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
+# define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
do { \
char *destination; \
/* Must be int, so when we don't save any registers, the arithmetic \
DEBUG_PUSH (failure_id); \
} while (0)
+# ifndef DEFINED_ONCE
/* This is the number of items that are pushed and popped on the stack
for each register. */
-#define NUM_REG_ITEMS 3
+# define NUM_REG_ITEMS 3
/* Individual items aside from the registers. */
-#ifdef DEBUG
-# define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
-#else
-# define NUM_NONREG_ITEMS 4
-#endif
+# ifdef DEBUG
+# define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
+# else
+# define NUM_NONREG_ITEMS 4
+# endif
/* We push at most this many items on the stack. */
/* We used to use (num_regs - 1), which is the number of registers
this regexp will save; but that was changed to 5
to avoid stack overflow for a regexp with lots of parens. */
-#define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
+# define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
/* We actually push this many items. */
-#define NUM_FAILURE_ITEMS \
+# define NUM_FAILURE_ITEMS \
(((0 \
? 0 : highest_active_reg - lowest_active_reg + 1) \
* NUM_REG_ITEMS) \
+ NUM_NONREG_ITEMS)
/* How many items can still be added to the stack without overflowing it. */
-#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
+# define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
+# endif /* not DEFINED_ONCE */
/* Pops what PUSH_FAIL_STACK pushes.
Also assumes the variables `fail_stack' and (if debugging), `bufp',
`pend', `string1', `size1', `string2', and `size2'. */
-
-#define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
+# define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
{ \
DEBUG_STATEMENT (unsigned failure_id;) \
active_reg_t this_reg; \
- const unsigned char *string_temp; \
+ const UCHAR_T *string_temp; \
\
assert (!FAIL_STACK_EMPTY ()); \
\
saved NULL, thus retaining our current position in the string. */ \
string_temp = POP_FAILURE_POINTER (); \
if (string_temp != NULL) \
- str = (const char *) string_temp; \
+ str = (const CHAR_T *) string_temp; \
\
DEBUG_PRINT2 (" Popping string %p: `", str); \
DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
DEBUG_PRINT1 ("'\n"); \
\
- pat = (unsigned char *) POP_FAILURE_POINTER (); \
+ pat = (UCHAR_T *) POP_FAILURE_POINTER (); \
DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
\
DEBUG_PRINT2 (" info: %p\n", \
reg_info[this_reg].word.pointer); \
\
- regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
+ regend[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
\
- regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
+ regstart[this_reg] = (const CHAR_T *) POP_FAILURE_POINTER (); \
DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
} \
else \
set_regs_matched_done = 0; \
DEBUG_STATEMENT (nfailure_points_popped++); \
} /* POP_FAILURE_POINT */
-
-
\f
/* Structure for per-register (a.k.a. per-group) information.
Other register information, such as the
typedef union
{
- fail_stack_elt_t word;
+ PREFIX(fail_stack_elt_t) word;
struct
{
/* This field is one if this group can match the empty string,
zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
-#define MATCH_NULL_UNSET_VALUE 3
+# define MATCH_NULL_UNSET_VALUE 3
unsigned match_null_string_p : 2;
unsigned is_active : 1;
unsigned matched_something : 1;
unsigned ever_matched_something : 1;
} bits;
-} register_info_type;
+} PREFIX(register_info_type);
-#define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
-#define IS_ACTIVE(R) ((R).bits.is_active)
-#define MATCHED_SOMETHING(R) ((R).bits.matched_something)
-#define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
+# ifndef DEFINED_ONCE
+# define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
+# define IS_ACTIVE(R) ((R).bits.is_active)
+# define MATCHED_SOMETHING(R) ((R).bits.matched_something)
+# define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
/* Call this when have matched a real character; it sets `matched' flags
for the subexpressions which we are currently inside. Also records
that those subexprs have matched. */
-#define SET_REGS_MATCHED() \
+# define SET_REGS_MATCHED() \
do \
{ \
if (!set_regs_matched_done) \
} \
} \
while (0)
+# endif /* not DEFINED_ONCE */
/* Registers are set to a sentinel when they haven't yet matched. */
-static char reg_unset_dummy;
-#define REG_UNSET_VALUE (®_unset_dummy)
-#define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
-\f
-/* Subroutine declarations and macros for regex_compile. */
+static CHAR_T PREFIX(reg_unset_dummy);
+# define REG_UNSET_VALUE (&PREFIX(reg_unset_dummy))
+# define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
-static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
- reg_syntax_t syntax,
- struct re_pattern_buffer *bufp));
-static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
-static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
- int arg1, int arg2));
-static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
- int arg, unsigned char *end));
-static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
- int arg1, int arg2, unsigned char *end));
-static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
- reg_syntax_t syntax));
-static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
- reg_syntax_t syntax));
-static reg_errcode_t compile_range _RE_ARGS ((const char **p_ptr,
- const char *pend,
- char *translate,
- reg_syntax_t syntax,
- unsigned char *b));
+/* Subroutine declarations and macros for regex_compile. */
+static void PREFIX(store_op1) (re_opcode_t op, UCHAR_T *loc, int arg);
+static void PREFIX(store_op2) (re_opcode_t op, UCHAR_T *loc,
+ int arg1, int arg2);
+static void PREFIX(insert_op1) (re_opcode_t op, UCHAR_T *loc,
+ int arg, UCHAR_T *end);
+static void PREFIX(insert_op2) (re_opcode_t op, UCHAR_T *loc,
+ int arg1, int arg2, UCHAR_T *end);
+static boolean PREFIX(at_begline_loc_p) (const CHAR_T *pattern,
+ const CHAR_T *p,
+ reg_syntax_t syntax);
+static boolean PREFIX(at_endline_loc_p) (const CHAR_T *p,
+ const CHAR_T *pend,
+ reg_syntax_t syntax);
+# ifdef WCHAR
+static reg_errcode_t wcs_compile_range (CHAR_T range_start,
+ const CHAR_T **p_ptr,
+ const CHAR_T *pend,
+ char *translate,
+ reg_syntax_t syntax,
+ UCHAR_T *b,
+ CHAR_T *char_set);
+static void insert_space (int num, CHAR_T *loc, CHAR_T *end);
+# else /* BYTE */
+static reg_errcode_t byte_compile_range (unsigned int range_start,
+ const char **p_ptr,
+ const char *pend,
+ char *translate,
+ reg_syntax_t syntax,
+ unsigned char *b);
+# endif /* WCHAR */
/* Fetch the next character in the uncompiled pattern---translating it
if necessary. Also cast from a signed character in the constant
string passed to us by the user to an unsigned char that we can use
as an array index (in, e.g., `translate'). */
-#ifndef PATFETCH
-# define PATFETCH(c) \
+/* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
+ because it is impossible to allocate 4GB array for some encodings
+ which have 4 byte character_set like UCS4. */
+# ifndef PATFETCH
+# ifdef WCHAR
+# define PATFETCH(c) \
+ do {if (p == pend) return REG_EEND; \
+ c = (UCHAR_T) *p++; \
+ if (translate && (c <= 0xff)) c = (UCHAR_T) translate[c]; \
+ } while (0)
+# else /* BYTE */
+# define PATFETCH(c) \
do {if (p == pend) return REG_EEND; \
c = (unsigned char) *p++; \
if (translate) c = (unsigned char) translate[c]; \
} while (0)
-#endif
+# endif /* WCHAR */
+# endif
/* Fetch the next character in the uncompiled pattern, with no
translation. */
-#define PATFETCH_RAW(c) \
+# define PATFETCH_RAW(c) \
do {if (p == pend) return REG_EEND; \
- c = (unsigned char) *p++; \
+ c = (UCHAR_T) *p++; \
} while (0)
/* Go backwards one character in the pattern. */
-#define PATUNFETCH p--
+# define PATUNFETCH p--
/* If `translate' is non-null, return translate[D], else just D. We
cast the subscript to translate because some data is declared as
`char *', to avoid warnings when a string constant is passed. But
when we use a character as a subscript we must make it unsigned. */
-#ifndef TRANSLATE
-# define TRANSLATE(d) \
+/* ifdef MBS_SUPPORT, we translate only if character <= 0xff,
+ because it is impossible to allocate 4GB array for some encodings
+ which have 4 byte character_set like UCS4. */
+
+# ifndef TRANSLATE
+# ifdef WCHAR
+# define TRANSLATE(d) \
+ ((translate && ((UCHAR_T) (d)) <= 0xff) \
+ ? (char) translate[(unsigned char) (d)] : (d))
+# else /* BYTE */
+# define TRANSLATE(d) \
(translate ? (char) translate[(unsigned char) (d)] : (d))
-#endif
+# endif /* WCHAR */
+# endif
/* Macros for outputting the compiled pattern into `buffer'. */
/* If the buffer isn't allocated when it comes in, use this. */
-#define INIT_BUF_SIZE 32
+# define INIT_BUF_SIZE (32 * sizeof(UCHAR_T))
/* Make sure we have at least N more bytes of space in buffer. */
-#define GET_BUFFER_SPACE(n) \
+# ifdef WCHAR
+# define GET_BUFFER_SPACE(n) \
+ while (((unsigned long)b - (unsigned long)COMPILED_BUFFER_VAR \
+ + (n)*sizeof(CHAR_T)) > bufp->allocated) \
+ EXTEND_BUFFER ()
+# else /* BYTE */
+# define GET_BUFFER_SPACE(n) \
while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
EXTEND_BUFFER ()
+# endif /* WCHAR */
/* Make sure we have one more byte of buffer space and then add C to it. */
-#define BUF_PUSH(c) \
+# define BUF_PUSH(c) \
do { \
GET_BUFFER_SPACE (1); \
- *b++ = (unsigned char) (c); \
+ *b++ = (UCHAR_T) (c); \
} while (0)
/* Ensure we have two more bytes of buffer space and then append C1 and C2. */
-#define BUF_PUSH_2(c1, c2) \
+# define BUF_PUSH_2(c1, c2) \
do { \
GET_BUFFER_SPACE (2); \
- *b++ = (unsigned char) (c1); \
- *b++ = (unsigned char) (c2); \
+ *b++ = (UCHAR_T) (c1); \
+ *b++ = (UCHAR_T) (c2); \
} while (0)
/* As with BUF_PUSH_2, except for three bytes. */
-#define BUF_PUSH_3(c1, c2, c3) \
+# define BUF_PUSH_3(c1, c2, c3) \
do { \
GET_BUFFER_SPACE (3); \
- *b++ = (unsigned char) (c1); \
- *b++ = (unsigned char) (c2); \
- *b++ = (unsigned char) (c3); \
+ *b++ = (UCHAR_T) (c1); \
+ *b++ = (UCHAR_T) (c2); \
+ *b++ = (UCHAR_T) (c3); \
} while (0)
-
/* Store a jump with opcode OP at LOC to location TO. We store a
relative address offset by the three bytes the jump itself occupies. */
-#define STORE_JUMP(op, loc, to) \
- store_op1 (op, loc, (int) ((to) - (loc) - 3))
+# define STORE_JUMP(op, loc, to) \
+ PREFIX(store_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)))
/* Likewise, for a two-argument jump. */
-#define STORE_JUMP2(op, loc, to, arg) \
- store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
+# define STORE_JUMP2(op, loc, to, arg) \
+ PREFIX(store_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), arg)
/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
-#define INSERT_JUMP(op, loc, to) \
- insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
+# define INSERT_JUMP(op, loc, to) \
+ PREFIX(insert_op1) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)), b)
/* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
-#define INSERT_JUMP2(op, loc, to, arg) \
- insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
-
+# define INSERT_JUMP2(op, loc, to, arg) \
+ PREFIX(insert_op2) (op, loc, (int) ((to) - (loc) - (1 + OFFSET_ADDRESS_SIZE)),\
+ arg, b)
/* This is not an arbitrary limit: the arguments which represent offsets
into the pattern are two bytes long. So if 2^16 bytes turns out to
MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
reallocating to 0 bytes. Such thing is not going to work too well.
You have been warned!! */
-#if defined _MSC_VER && !defined WIN32
+# ifndef DEFINED_ONCE
+# if defined _MSC_VER && !defined WIN32
/* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
The REALLOC define eliminates a flurry of conversion warnings,
but is not required. */
-# define MAX_BUF_SIZE 65500L
-# define REALLOC(p,s) realloc ((p), (size_t) (s))
-#else
-# define MAX_BUF_SIZE (1L << 16)
-# define REALLOC(p,s) realloc ((p), (s))
-#endif
+# define MAX_BUF_SIZE 65500L
+# define REALLOC(p,s) realloc ((p), (size_t) (s))
+# else
+# define MAX_BUF_SIZE (1L << 16)
+# define REALLOC(p,s) realloc ((p), (s))
+# endif
/* Extend the buffer by twice its current size via realloc and
reset the pointers that pointed into the old block to point to the
correct places in the new one. If extending the buffer results in it
being larger than MAX_BUF_SIZE, then flag memory exhausted. */
-#define EXTEND_BUFFER() \
- do { \
- unsigned char *old_buffer = bufp->buffer; \
- if (bufp->allocated == MAX_BUF_SIZE) \
+# if __BOUNDED_POINTERS__
+# define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
+# define MOVE_BUFFER_POINTER(P) \
+ (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
+# define ELSE_EXTEND_BUFFER_HIGH_BOUND \
+ else \
+ { \
+ SET_HIGH_BOUND (b); \
+ SET_HIGH_BOUND (begalt); \
+ if (fixup_alt_jump) \
+ SET_HIGH_BOUND (fixup_alt_jump); \
+ if (laststart) \
+ SET_HIGH_BOUND (laststart); \
+ if (pending_exact) \
+ SET_HIGH_BOUND (pending_exact); \
+ }
+# else
+# define MOVE_BUFFER_POINTER(P) (P) += incr
+# define ELSE_EXTEND_BUFFER_HIGH_BOUND
+# endif
+# endif /* not DEFINED_ONCE */
+
+# ifdef WCHAR
+# define EXTEND_BUFFER() \
+ do { \
+ UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
+ int wchar_count; \
+ if (bufp->allocated + sizeof(UCHAR_T) > MAX_BUF_SIZE) \
return REG_ESIZE; \
bufp->allocated <<= 1; \
if (bufp->allocated > MAX_BUF_SIZE) \
- bufp->allocated = MAX_BUF_SIZE; \
- bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
- if (bufp->buffer == NULL) \
+ bufp->allocated = MAX_BUF_SIZE; \
+ /* How many characters the new buffer can have? */ \
+ wchar_count = bufp->allocated / sizeof(UCHAR_T); \
+ if (wchar_count == 0) wchar_count = 1; \
+ /* Truncate the buffer to CHAR_T align. */ \
+ bufp->allocated = wchar_count * sizeof(UCHAR_T); \
+ RETALLOC (COMPILED_BUFFER_VAR, wchar_count, UCHAR_T); \
+ bufp->buffer = (char*)COMPILED_BUFFER_VAR; \
+ if (COMPILED_BUFFER_VAR == NULL) \
return REG_ESPACE; \
/* If the buffer moved, move all the pointers into it. */ \
- if (old_buffer != bufp->buffer) \
+ if (old_buffer != COMPILED_BUFFER_VAR) \
{ \
- b = (b - old_buffer) + bufp->buffer; \
- begalt = (begalt - old_buffer) + bufp->buffer; \
- if (fixup_alt_jump) \
- fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
- if (laststart) \
- laststart = (laststart - old_buffer) + bufp->buffer; \
- if (pending_exact) \
- pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
+ int incr = COMPILED_BUFFER_VAR - old_buffer; \
+ MOVE_BUFFER_POINTER (b); \
+ MOVE_BUFFER_POINTER (begalt); \
+ if (fixup_alt_jump) \
+ MOVE_BUFFER_POINTER (fixup_alt_jump); \
+ if (laststart) \
+ MOVE_BUFFER_POINTER (laststart); \
+ if (pending_exact) \
+ MOVE_BUFFER_POINTER (pending_exact); \
} \
+ ELSE_EXTEND_BUFFER_HIGH_BOUND \
} while (0)
+# else /* BYTE */
+# define EXTEND_BUFFER() \
+ do { \
+ UCHAR_T *old_buffer = COMPILED_BUFFER_VAR; \
+ if (bufp->allocated == MAX_BUF_SIZE) \
+ return REG_ESIZE; \
+ bufp->allocated <<= 1; \
+ if (bufp->allocated > MAX_BUF_SIZE) \
+ bufp->allocated = MAX_BUF_SIZE; \
+ bufp->buffer \
+ = (UCHAR_T *) REALLOC (COMPILED_BUFFER_VAR, bufp->allocated); \
+ if (COMPILED_BUFFER_VAR == NULL) \
+ return REG_ESPACE; \
+ /* If the buffer moved, move all the pointers into it. */ \
+ if (old_buffer != COMPILED_BUFFER_VAR) \
+ { \
+ int incr = COMPILED_BUFFER_VAR - old_buffer; \
+ MOVE_BUFFER_POINTER (b); \
+ MOVE_BUFFER_POINTER (begalt); \
+ if (fixup_alt_jump) \
+ MOVE_BUFFER_POINTER (fixup_alt_jump); \
+ if (laststart) \
+ MOVE_BUFFER_POINTER (laststart); \
+ if (pending_exact) \
+ MOVE_BUFFER_POINTER (pending_exact); \
+ } \
+ ELSE_EXTEND_BUFFER_HIGH_BOUND \
+ } while (0)
+# endif /* WCHAR */
-
+# ifndef DEFINED_ONCE
/* Since we have one byte reserved for the register number argument to
{start,stop}_memory, the maximum number of groups we can report
things about is what fits in that byte. */
-#define MAX_REGNUM 255
+# define MAX_REGNUM 255
/* But patterns can have more than `MAX_REGNUM' registers. We just
ignore the excess. */
} compile_stack_type;
-#define INIT_COMPILE_STACK_SIZE 32
+# define INIT_COMPILE_STACK_SIZE 32
-#define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
-#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
+# define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
+# define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
/* The next available element. */
-#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
+# define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
+# endif /* not DEFINED_ONCE */
/* Set the bit for character C in a list. */
-#define SET_LIST_BIT(c) \
+# ifndef DEFINED_ONCE
+# define SET_LIST_BIT(c) \
(b[((unsigned char) (c)) / BYTEWIDTH] \
|= 1 << (((unsigned char) c) % BYTEWIDTH))
-
+# endif /* DEFINED_ONCE */
/* Get the next unsigned number in the uncompiled pattern. */
-#define GET_UNSIGNED_NUMBER(num) \
- { if (p != pend) \
- { \
- PATFETCH (c); \
- while (ISDIGIT (c)) \
- { \
- if (num < 0) \
- num = 0; \
- num = num * 10 + c - '0'; \
- if (p == pend) \
- break; \
- PATFETCH (c); \
- } \
- } \
- }
+# define GET_UNSIGNED_NUMBER(num) \
+ { \
+ while (p != pend) \
+ { \
+ PATFETCH (c); \
+ if (c < '0' || c > '9') \
+ break; \
+ if (num <= RE_DUP_MAX) \
+ { \
+ if (num < 0) \
+ num = 0; \
+ num = num * 10 + c - '0'; \
+ } \
+ } \
+ }
-#if WIDE_CHAR_SUPPORT
+# ifndef DEFINED_ONCE
+# if defined _LIBC || WIDE_CHAR_SUPPORT
/* The GNU C library provides support for user-defined character classes
and the functions from ISO C amendement 1. */
-# ifdef CHARCLASS_NAME_MAX
-# define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
-# else
+# ifdef CHARCLASS_NAME_MAX
+# define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
+# else
/* This shouldn't happen but some implementation might still have this
problem. Use a reasonable default value. */
-# define CHAR_CLASS_MAX_LENGTH 256
-# endif
+# define CHAR_CLASS_MAX_LENGTH 256
+# endif
-# ifdef _LIBC
-# define IS_CHAR_CLASS(string) __wctype (string)
-# else
-# define IS_CHAR_CLASS(string) wctype (string)
-# endif
-#else
-# define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
+# ifdef _LIBC
+# define IS_CHAR_CLASS(string) __wctype (string)
+# else
+# define IS_CHAR_CLASS(string) wctype (string)
+# endif
+# else
+# define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
-# define IS_CHAR_CLASS(string) \
+# define IS_CHAR_CLASS(string) \
(STREQ (string, "alpha") || STREQ (string, "upper") \
|| STREQ (string, "lower") || STREQ (string, "digit") \
|| STREQ (string, "alnum") || STREQ (string, "xdigit") \
|| STREQ (string, "space") || STREQ (string, "print") \
|| STREQ (string, "punct") || STREQ (string, "graph") \
|| STREQ (string, "cntrl") || STREQ (string, "blank"))
-#endif
+# endif
+# endif /* DEFINED_ONCE */
\f
-#ifndef MATCH_MAY_ALLOCATE
+# ifndef MATCH_MAY_ALLOCATE
/* If we cannot allocate large objects within re_match_2_internal,
we make the fail stack and register vectors global.
The register vectors, we adjust in size each time we
compile a regexp, according to the number of registers it needs. */
-static fail_stack_type fail_stack;
+static PREFIX(fail_stack_type) fail_stack;
/* Size with which the following vectors are currently allocated.
That is so we can make them bigger as needed,
but never make them smaller. */
+# ifdef DEFINED_ONCE
static int regs_allocated_size;
static const char ** regstart, ** regend;
static const char ** old_regstart, ** old_regend;
static const char **best_regstart, **best_regend;
-static register_info_type *reg_info;
static const char **reg_dummy;
-static register_info_type *reg_info_dummy;
+# endif /* DEFINED_ONCE */
+
+static PREFIX(register_info_type) *PREFIX(reg_info);
+static PREFIX(register_info_type) *PREFIX(reg_info_dummy);
/* Make the register vectors big enough for NUM_REGS registers,
but don't make them smaller. */
-static
-regex_grow_registers (num_regs)
- int num_regs;
+static void
+PREFIX(regex_grow_registers) (int num_regs)
{
if (num_regs > regs_allocated_size)
{
RETALLOC_IF (old_regend, num_regs, const char *);
RETALLOC_IF (best_regstart, num_regs, const char *);
RETALLOC_IF (best_regend, num_regs, const char *);
- RETALLOC_IF (reg_info, num_regs, register_info_type);
+ RETALLOC_IF (PREFIX(reg_info), num_regs, PREFIX(register_info_type));
RETALLOC_IF (reg_dummy, num_regs, const char *);
- RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
+ RETALLOC_IF (PREFIX(reg_info_dummy), num_regs, PREFIX(register_info_type));
regs_allocated_size = num_regs;
}
}
-#endif /* not MATCH_MAY_ALLOCATE */
+# endif /* not MATCH_MAY_ALLOCATE */
\f
-static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
- compile_stack,
- regnum_t regnum));
+# ifndef DEFINED_ONCE
+static boolean group_in_compile_stack (compile_stack_type
+ compile_stack,
+ regnum_t regnum);
+# endif /* not DEFINED_ONCE */
/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
Returns one of error codes defined in `regex.h', or zero for success.
examined nor set. */
/* Return, freeing storage we allocated. */
-#define FREE_STACK_RETURN(value) \
+# ifdef WCHAR
+# define FREE_STACK_RETURN(value) \
+ return (free(pattern), free(mbs_offset), free(is_binary), free (compile_stack.stack), value)
+# else
+# define FREE_STACK_RETURN(value) \
return (free (compile_stack.stack), value)
+# endif /* WCHAR */
static reg_errcode_t
-regex_compile (pattern, size, syntax, bufp)
- const char *pattern;
- size_t size;
- reg_syntax_t syntax;
- struct re_pattern_buffer *bufp;
+PREFIX(regex_compile) (const char *ARG_PREFIX(pattern),
+ size_t ARG_PREFIX(size),
+ reg_syntax_t syntax,
+ struct re_pattern_buffer *bufp)
{
/* We fetch characters from PATTERN here. Even though PATTERN is
`char *' (i.e., signed), we declare these variables as unsigned, so
they can be reliably used as array indices. */
- register unsigned char c, c1;
+ register UCHAR_T c, c1;
+
+#ifdef WCHAR
+ /* A temporary space to keep wchar_t pattern and compiled pattern. */
+ CHAR_T *pattern, *COMPILED_BUFFER_VAR;
+ size_t size;
+ /* offset buffer for optimization. See convert_mbs_to_wc. */
+ int *mbs_offset = NULL;
+ /* It hold whether each wchar_t is binary data or not. */
+ char *is_binary = NULL;
+ /* A flag whether exactn is handling binary data or not. */
+ char is_exactn_bin = FALSE;
+#endif /* WCHAR */
/* A random temporary spot in PATTERN. */
- const char *p1;
+ const CHAR_T *p1;
/* Points to the end of the buffer, where we should append. */
- register unsigned char *b;
+ register UCHAR_T *b;
/* Keeps track of unclosed groups. */
compile_stack_type compile_stack;
/* Points to the current (ending) position in the pattern. */
- const char *p = pattern;
- const char *pend = pattern + size;
+#ifdef WCHAR
+ const CHAR_T *p;
+ const CHAR_T *pend;
+#else /* BYTE */
+ const CHAR_T *p = pattern;
+ const CHAR_T *pend = pattern + size;
+#endif /* WCHAR */
/* How to translate the characters in the pattern. */
RE_TRANSLATE_TYPE translate = bufp->translate;
command. This makes it possible to tell if a new exact-match
character can be added to that command or if the character requires
a new `exactn' command. */
- unsigned char *pending_exact = 0;
+ UCHAR_T *pending_exact = 0;
/* Address of start of the most recently finished expression.
This tells, e.g., postfix * where to find the start of its
operand. Reset at the beginning of groups and alternatives. */
- unsigned char *laststart = 0;
+ UCHAR_T *laststart = 0;
/* Address of beginning of regexp, or inside of last group. */
- unsigned char *begalt;
-
- /* Place in the uncompiled pattern (i.e., the {) to
- which to go back if the interval is invalid. */
- const char *beg_interval;
+ UCHAR_T *begalt;
/* Address of the place where a forward jump should go to the end of
the containing expression. Each alternative of an `or' -- except the
last -- ends with a forward jump of this sort. */
- unsigned char *fixup_alt_jump = 0;
+ UCHAR_T *fixup_alt_jump = 0;
/* Counts open-groups as they are encountered. Remembered for the
matching close-group on the compile stack, so the same register
number is put in the stop_memory as the start_memory. */
regnum_t regnum = 0;
+#ifdef WCHAR
+ /* Initialize the wchar_t PATTERN and offset_buffer. */
+ p = pend = pattern = TALLOC(csize + 1, CHAR_T);
+ mbs_offset = TALLOC(csize + 1, int);
+ is_binary = TALLOC(csize + 1, char);
+ if (pattern == NULL || mbs_offset == NULL || is_binary == NULL)
+ {
+ free(pattern);
+ free(mbs_offset);
+ free(is_binary);
+ return REG_ESPACE;
+ }
+ pattern[csize] = L'\0'; /* sentinel */
+ size = convert_mbs_to_wcs(pattern, cpattern, csize, mbs_offset, is_binary);
+ pend = p + size;
+ if (size < 0)
+ {
+ free(pattern);
+ free(mbs_offset);
+ free(is_binary);
+ return REG_BADPAT;
+ }
+#endif
+
#ifdef DEBUG
DEBUG_PRINT1 ("\nCompiling pattern: ");
if (debug)
unsigned debug_count;
for (debug_count = 0; debug_count < size; debug_count++)
- putchar (pattern[debug_count]);
+ PUT_CHAR (pattern[debug_count]);
putchar ('\n');
}
#endif /* DEBUG */
/* Initialize the compile stack. */
compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
if (compile_stack.stack == NULL)
- return REG_ESPACE;
+ {
+#ifdef WCHAR
+ free(pattern);
+ free(mbs_offset);
+ free(is_binary);
+#endif
+ return REG_ESPACE;
+ }
compile_stack.size = INIT_COMPILE_STACK_SIZE;
compile_stack.avail = 0;
{ /* If zero allocated, but buffer is non-null, try to realloc
enough space. This loses if buffer's address is bogus, but
that is the user's responsibility. */
- RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
+#ifdef WCHAR
+ /* Free bufp->buffer and allocate an array for wchar_t pattern
+ buffer. */
+ free(bufp->buffer);
+ COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE/sizeof(UCHAR_T),
+ UCHAR_T);
+#else
+ RETALLOC (COMPILED_BUFFER_VAR, INIT_BUF_SIZE, UCHAR_T);
+#endif /* WCHAR */
}
else
{ /* Caller did not allocate a buffer. Do it for them. */
- bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
+ COMPILED_BUFFER_VAR = TALLOC (INIT_BUF_SIZE / sizeof(UCHAR_T),
+ UCHAR_T);
}
- if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
+ if (!COMPILED_BUFFER_VAR) FREE_STACK_RETURN (REG_ESPACE);
+#ifdef WCHAR
+ bufp->buffer = (char*)COMPILED_BUFFER_VAR;
+#endif /* WCHAR */
bufp->allocated = INIT_BUF_SIZE;
}
+#ifdef WCHAR
+ else
+ COMPILED_BUFFER_VAR = (UCHAR_T*) bufp->buffer;
+#endif
- begalt = b = bufp->buffer;
+ begalt = b = COMPILED_BUFFER_VAR;
/* Loop through the uncompiled pattern until we're at the end. */
while (p != pend)
/* If context independent, it's an operator. */
|| syntax & RE_CONTEXT_INDEP_ANCHORS
/* Otherwise, depends on what's come before. */
- || at_begline_loc_p (pattern, p, syntax))
+ || PREFIX(at_begline_loc_p) (pattern, p, syntax))
BUF_PUSH (begline);
else
goto normal_char;
/* If context independent, it's an operator. */
|| syntax & RE_CONTEXT_INDEP_ANCHORS
/* Otherwise, depends on what's next. */
- || at_endline_loc_p (p, pend, syntax))
+ || PREFIX(at_endline_loc_p) (p, pend, syntax))
BUF_PUSH (endline);
else
goto normal_char;
assert (p - 1 > pattern);
/* Allocate the space for the jump. */
- GET_BUFFER_SPACE (3);
+ GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
/* We know we are not at the first character of the pattern,
because laststart was nonzero. And we've already
}
else
/* Anything else. */
- STORE_JUMP (maybe_pop_jump, b, laststart - 3);
+ STORE_JUMP (maybe_pop_jump, b, laststart -
+ (1 + OFFSET_ADDRESS_SIZE));
/* We've added more stuff to the buffer. */
- b += 3;
+ b += 1 + OFFSET_ADDRESS_SIZE;
}
/* On failure, jump from laststart to b + 3, which will be the
end of the buffer after this jump is inserted. */
- GET_BUFFER_SPACE (3);
+ /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE' instead of
+ 'b + 3'. */
+ GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
: on_failure_jump,
- laststart, b + 3);
+ laststart, b + 1 + OFFSET_ADDRESS_SIZE);
pending_exact = 0;
- b += 3;
+ b += 1 + OFFSET_ADDRESS_SIZE;
if (!zero_times_ok)
{
`on_failure_jump' instruction of the loop. This
effects a skip over that instruction the first time
we hit that loop. */
- GET_BUFFER_SPACE (3);
- INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
- b += 3;
+ GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
+ INSERT_JUMP (dummy_failure_jump, laststart, laststart +
+ 2 + 2 * OFFSET_ADDRESS_SIZE);
+ b += 1 + OFFSET_ADDRESS_SIZE;
}
}
break;
case '[':
{
boolean had_char_class = false;
-
+#ifdef WCHAR
+ CHAR_T range_start = 0xffffffff;
+#else
+ unsigned int range_start = 0xffffffff;
+#endif
if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+#ifdef WCHAR
+ /* We assume a charset(_not) structure as a wchar_t array.
+ charset[0] = (re_opcode_t) charset(_not)
+ charset[1] = l (= length of char_classes)
+ charset[2] = m (= length of collating_symbols)
+ charset[3] = n (= length of equivalence_classes)
+ charset[4] = o (= length of char_ranges)
+ charset[5] = p (= length of chars)
+
+ charset[6] = char_class (wctype_t)
+ charset[6+CHAR_CLASS_SIZE] = char_class (wctype_t)
+ ...
+ charset[l+5] = char_class (wctype_t)
+
+ charset[l+6] = collating_symbol (wchar_t)
+ ...
+ charset[l+m+5] = collating_symbol (wchar_t)
+ ifdef _LIBC we use the index if
+ _NL_COLLATE_SYMB_EXTRAMB instead of
+ wchar_t string.
+
+ charset[l+m+6] = equivalence_classes (wchar_t)
+ ...
+ charset[l+m+n+5] = equivalence_classes (wchar_t)
+ ifdef _LIBC we use the index in
+ _NL_COLLATE_WEIGHT instead of
+ wchar_t string.
+
+ charset[l+m+n+6] = range_start
+ charset[l+m+n+7] = range_end
+ ...
+ charset[l+m+n+2o+4] = range_start
+ charset[l+m+n+2o+5] = range_end
+ ifdef _LIBC we use the value looked up
+ in _NL_COLLATE_COLLSEQ instead of
+ wchar_t character.
+
+ charset[l+m+n+2o+6] = char
+ ...
+ charset[l+m+n+2o+p+5] = char
+
+ */
+
+ /* We need at least 6 spaces: the opcode, the length of
+ char_classes, the length of collating_symbols, the length of
+ equivalence_classes, the length of char_ranges, the length of
+ chars. */
+ GET_BUFFER_SPACE (6);
+
+ /* Save b as laststart. And We use laststart as the pointer
+ to the first element of the charset here.
+ In other words, laststart[i] indicates charset[i]. */
+ laststart = b;
+
+ /* We test `*p == '^' twice, instead of using an if
+ statement, so we only need one BUF_PUSH. */
+ BUF_PUSH (*p == '^' ? charset_not : charset);
+ if (*p == '^')
+ p++;
+
+ /* Push the length of char_classes, the length of
+ collating_symbols, the length of equivalence_classes, the
+ length of char_ranges and the length of chars. */
+ BUF_PUSH_3 (0, 0, 0);
+ BUF_PUSH_2 (0, 0);
+
+ /* Remember the first position in the bracket expression. */
+ p1 = p;
+
+ /* charset_not matches newline according to a syntax bit. */
+ if ((re_opcode_t) b[-6] == charset_not
+ && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
+ {
+ BUF_PUSH('\n');
+ laststart[5]++; /* Update the length of characters */
+ }
+
+ /* Read in characters and ranges, setting map bits. */
+ for (;;)
+ {
+ if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+
+ PATFETCH (c);
+
+ /* \ might escape characters inside [...] and [^...]. */
+ if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
+ {
+ if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
+
+ PATFETCH (c1);
+ BUF_PUSH(c1);
+ laststart[5]++; /* Update the length of chars */
+ range_start = c1;
+ continue;
+ }
+
+ /* Could be the end of the bracket expression. If it's
+ not (i.e., when the bracket expression is `[]' so
+ far), the ']' character bit gets set way below. */
+ if (c == ']' && p != p1 + 1)
+ break;
+
+ /* Look ahead to see if it's a range when the last thing
+ was a character class. */
+ if (had_char_class && c == '-' && *p != ']')
+ FREE_STACK_RETURN (REG_ERANGE);
+
+ /* Look ahead to see if it's a range when the last thing
+ was a character: if this is a hyphen not at the
+ beginning or the end of a list, then it's the range
+ operator. */
+ if (c == '-'
+ && !(p - 2 >= pattern && p[-2] == '[')
+ && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
+ && *p != ']')
+ {
+ reg_errcode_t ret;
+ /* Allocate the space for range_start and range_end. */
+ GET_BUFFER_SPACE (2);
+ /* Update the pointer to indicate end of buffer. */
+ b += 2;
+ ret = wcs_compile_range (range_start, &p, pend, translate,
+ syntax, b, laststart);
+ if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
+ range_start = 0xffffffff;
+ }
+ else if (p[0] == '-' && p[1] != ']')
+ { /* This handles ranges made up of characters only. */
+ reg_errcode_t ret;
+
+ /* Move past the `-'. */
+ PATFETCH (c1);
+ /* Allocate the space for range_start and range_end. */
+ GET_BUFFER_SPACE (2);
+ /* Update the pointer to indicate end of buffer. */
+ b += 2;
+ ret = wcs_compile_range (c, &p, pend, translate, syntax, b,
+ laststart);
+ if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
+ range_start = 0xffffffff;
+ }
+
+ /* See if we're at the beginning of a possible character
+ class. */
+ else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
+ { /* Leave room for the null. */
+ char str[CHAR_CLASS_MAX_LENGTH + 1];
+
+ PATFETCH (c);
+ c1 = 0;
+
+ /* If pattern is `[[:'. */
+ if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+
+ for (;;)
+ {
+ PATFETCH (c);
+ if ((c == ':' && *p == ']') || p == pend)
+ break;
+ if (c1 < CHAR_CLASS_MAX_LENGTH)
+ str[c1++] = c;
+ else
+ /* This is in any case an invalid class name. */
+ str[0] = '\0';
+ }
+ str[c1] = '\0';
+
+ /* If isn't a word bracketed by `[:' and `:]':
+ undo the ending character, the letters, and leave
+ the leading `:' and `[' (but store them as character). */
+ if (c == ':' && *p == ']')
+ {
+ wctype_t wt;
+ uintptr_t alignedp;
+
+ /* Query the character class as wctype_t. */
+ wt = IS_CHAR_CLASS (str);
+ if (wt == 0)
+ FREE_STACK_RETURN (REG_ECTYPE);
+
+ /* Throw away the ] at the end of the character
+ class. */
+ PATFETCH (c);
+
+ if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+
+ /* Allocate the space for character class. */
+ GET_BUFFER_SPACE(CHAR_CLASS_SIZE);
+ /* Update the pointer to indicate end of buffer. */
+ b += CHAR_CLASS_SIZE;
+ /* Move data which follow character classes
+ not to violate the data. */
+ insert_space(CHAR_CLASS_SIZE,
+ laststart + 6 + laststart[1],
+ b - 1);
+ alignedp = ((uintptr_t)(laststart + 6 + laststart[1])
+ + __alignof__(wctype_t) - 1)
+ & ~(uintptr_t)(__alignof__(wctype_t) - 1);
+ /* Store the character class. */
+ *((wctype_t*)alignedp) = wt;
+ /* Update length of char_classes */
+ laststart[1] += CHAR_CLASS_SIZE;
+
+ had_char_class = true;
+ }
+ else
+ {
+ c1++;
+ while (c1--)
+ PATUNFETCH;
+ BUF_PUSH ('[');
+ BUF_PUSH (':');
+ laststart[5] += 2; /* Update the length of characters */
+ range_start = ':';
+ had_char_class = false;
+ }
+ }
+ else if (syntax & RE_CHAR_CLASSES && c == '[' && (*p == '='
+ || *p == '.'))
+ {
+ CHAR_T str[128]; /* Should be large enough. */
+ CHAR_T delim = *p; /* '=' or '.' */
+# ifdef _LIBC
+ uint32_t nrules =
+ _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
+# endif
+ PATFETCH (c);
+ c1 = 0;
+
+ /* If pattern is `[[=' or '[[.'. */
+ if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+
+ for (;;)
+ {
+ PATFETCH (c);
+ if ((c == delim && *p == ']') || p == pend)
+ break;
+ if (c1 < sizeof (str) - 1)
+ str[c1++] = c;
+ else
+ /* This is in any case an invalid class name. */
+ str[0] = '\0';
+ }
+ str[c1] = '\0';
+
+ if (c == delim && *p == ']' && str[0] != '\0')
+ {
+ unsigned int i, offset;
+ /* If we have no collation data we use the default
+ collation in which each character is in a class
+ by itself. It also means that ASCII is the
+ character set and therefore we cannot have character
+ with more than one byte in the multibyte
+ representation. */
+
+ /* If not defined _LIBC, we push the name and
+ `\0' for the sake of matching performance. */
+ int datasize = c1 + 1;
+
+# ifdef _LIBC
+ int32_t idx = 0;
+ if (nrules == 0)
+# endif
+ {
+ if (c1 != 1)
+ FREE_STACK_RETURN (REG_ECOLLATE);
+ }
+# ifdef _LIBC
+ else
+ {
+ const int32_t *table;
+ const int32_t *weights;
+ const int32_t *extra;
+ const int32_t *indirect;
+ wint_t *cp;
+
+ /* This #include defines a local function! */
+# include <locale/weightwc.h>
+
+ if(delim == '=')
+ {
+ /* We push the index for equivalence class. */
+ cp = (wint_t*)str;
+
+ table = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_TABLEWC);
+ weights = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_WEIGHTWC);
+ extra = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_EXTRAWC);
+ indirect = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_INDIRECTWC);
+
+ idx = findidx ((const wint_t**)&cp);
+ if (idx == 0 || cp < (wint_t*) str + c1)
+ /* This is no valid character. */
+ FREE_STACK_RETURN (REG_ECOLLATE);
+
+ str[0] = (wchar_t)idx;
+ }
+ else /* delim == '.' */
+ {
+ /* We push collation sequence value
+ for collating symbol. */
+ int32_t table_size;
+ const int32_t *symb_table;
+ const unsigned char *extra;
+ int32_t idx;
+ int32_t elem;
+ int32_t second;
+ int32_t hash;
+ char char_str[c1];
+
+ /* We have to convert the name to a single-byte
+ string. This is possible since the names
+ consist of ASCII characters and the internal
+ representation is UCS4. */
+ for (i = 0; i < c1; ++i)
+ char_str[i] = str[i];
+
+ table_size =
+ _NL_CURRENT_WORD (LC_COLLATE,
+ _NL_COLLATE_SYMB_HASH_SIZEMB);
+ symb_table = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_SYMB_TABLEMB);
+ extra = (const unsigned char *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_SYMB_EXTRAMB);
+
+ /* Locate the character in the hashing table. */
+ hash = elem_hash (char_str, c1);
+
+ idx = 0;
+ elem = hash % table_size;
+ second = hash % (table_size - 2);
+ while (symb_table[2 * elem] != 0)
+ {
+ /* First compare the hashing value. */
+ if (symb_table[2 * elem] == hash
+ && c1 == extra[symb_table[2 * elem + 1]]
+ && memcmp (char_str,
+ &extra[symb_table[2 * elem + 1]
+ + 1], c1) == 0)
+ {
+ /* Yep, this is the entry. */
+ idx = symb_table[2 * elem + 1];
+ idx += 1 + extra[idx];
+ break;
+ }
+
+ /* Next entry. */
+ elem += second;
+ }
+
+ if (symb_table[2 * elem] != 0)
+ {
+ /* Compute the index of the byte sequence
+ in the table. */
+ idx += 1 + extra[idx];
+ /* Adjust for the alignment. */
+ idx = (idx + 3) & ~3;
+
+ str[0] = (wchar_t) idx + 4;
+ }
+ else if (symb_table[2 * elem] == 0 && c1 == 1)
+ {
+ /* No valid character. Match it as a
+ single byte character. */
+ had_char_class = false;
+ BUF_PUSH(str[0]);
+ /* Update the length of characters */
+ laststart[5]++;
+ range_start = str[0];
+
+ /* Throw away the ] at the end of the
+ collating symbol. */
+ PATFETCH (c);
+ /* exit from the switch block. */
+ continue;
+ }
+ else
+ FREE_STACK_RETURN (REG_ECOLLATE);
+ }
+ datasize = 1;
+ }
+# endif
+ /* Throw away the ] at the end of the equivalence
+ class (or collating symbol). */
+ PATFETCH (c);
+
+ /* Allocate the space for the equivalence class
+ (or collating symbol) (and '\0' if needed). */
+ GET_BUFFER_SPACE(datasize);
+ /* Update the pointer to indicate end of buffer. */
+ b += datasize;
+
+ if (delim == '=')
+ { /* equivalence class */
+ /* Calculate the offset of char_ranges,
+ which is next to equivalence_classes. */
+ offset = laststart[1] + laststart[2]
+ + laststart[3] +6;
+ /* Insert space. */
+ insert_space(datasize, laststart + offset, b - 1);
+
+ /* Write the equivalence_class and \0. */
+ for (i = 0 ; i < datasize ; i++)
+ laststart[offset + i] = str[i];
+
+ /* Update the length of equivalence_classes. */
+ laststart[3] += datasize;
+ had_char_class = true;
+ }
+ else /* delim == '.' */
+ { /* collating symbol */
+ /* Calculate the offset of the equivalence_classes,
+ which is next to collating_symbols. */
+ offset = laststart[1] + laststart[2] + 6;
+ /* Insert space and write the collationg_symbol
+ and \0. */
+ insert_space(datasize, laststart + offset, b-1);
+ for (i = 0 ; i < datasize ; i++)
+ laststart[offset + i] = str[i];
+
+ /* In re_match_2_internal if range_start < -1, we
+ assume -range_start is the offset of the
+ collating symbol which is specified as
+ the character of the range start. So we assign
+ -(laststart[1] + laststart[2] + 6) to
+ range_start. */
+ range_start = -(laststart[1] + laststart[2] + 6);
+ /* Update the length of collating_symbol. */
+ laststart[2] += datasize;
+ had_char_class = false;
+ }
+ }
+ else
+ {
+ c1++;
+ while (c1--)
+ PATUNFETCH;
+ BUF_PUSH ('[');
+ BUF_PUSH (delim);
+ laststart[5] += 2; /* Update the length of characters */
+ range_start = delim;
+ had_char_class = false;
+ }
+ }
+ else
+ {
+ had_char_class = false;
+ BUF_PUSH(c);
+ laststart[5]++; /* Update the length of characters */
+ range_start = c;
+ }
+ }
+
+#else /* BYTE */
/* Ensure that we have enough space to push a charset: the
opcode, the length count, and the bitset; 34 bytes in all. */
GET_BUFFER_SPACE (34);
PATFETCH (c1);
SET_LIST_BIT (c1);
+ range_start = c1;
continue;
}
&& *p != ']')
{
reg_errcode_t ret
- = compile_range (&p, pend, translate, syntax, b);
+ = byte_compile_range (range_start, &p, pend, translate,
+ syntax, b);
if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
+ range_start = 0xffffffff;
}
else if (p[0] == '-' && p[1] != ']')
/* Move past the `-'. */
PATFETCH (c1);
- ret = compile_range (&p, pend, translate, syntax, b);
+ ret = byte_compile_range (c, &p, pend, translate, syntax, b);
if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
+ range_start = 0xffffffff;
}
/* See if we're at the beginning of a possible character
for (;;)
{
PATFETCH (c);
- if ((c == ':' && *p == ']') || p == pend
- || c1 == CHAR_CLASS_MAX_LENGTH)
+ if ((c == ':' && *p == ']') || p == pend)
break;
- str[c1++] = c;
+ if (c1 < CHAR_CLASS_MAX_LENGTH)
+ str[c1++] = c;
+ else
+ /* This is in any case an invalid class name. */
+ str[0] = '\0';
}
str[c1] = '\0';
the leading `:' and `[' (but set bits for them). */
if (c == ':' && *p == ']')
{
-#if WIDE_CHAR_SUPPORT
+# if defined _LIBC || WIDE_CHAR_SUPPORT
boolean is_lower = STREQ (str, "lower");
boolean is_upper = STREQ (str, "upper");
wctype_t wt;
for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
{
-# ifdef _LIBC
- if (__iswctype (__btowc (ch), wt))
- SET_LIST_BIT (ch);
-#else
if (iswctype (btowc (ch), wt))
SET_LIST_BIT (ch);
-#endif
if (translate && (is_upper || is_lower)
&& (ISUPPER (ch) || ISLOWER (ch)))
}
had_char_class = true;
-#else
+# else
int ch;
boolean is_alnum = STREQ (str, "alnum");
boolean is_alpha = STREQ (str, "alpha");
SET_LIST_BIT (ch);
}
had_char_class = true;
-#endif /* libc || wctype.h */
+# endif /* libc || wctype.h */
+ }
+ else
+ {
+ c1++;
+ while (c1--)
+ PATUNFETCH;
+ SET_LIST_BIT ('[');
+ SET_LIST_BIT (':');
+ range_start = ':';
+ had_char_class = false;
+ }
+ }
+ else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
+ {
+ unsigned char str[MB_LEN_MAX + 1];
+# ifdef _LIBC
+ uint32_t nrules =
+ _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
+# endif
+
+ PATFETCH (c);
+ c1 = 0;
+
+ /* If pattern is `[[='. */
+ if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+
+ for (;;)
+ {
+ PATFETCH (c);
+ if ((c == '=' && *p == ']') || p == pend)
+ break;
+ if (c1 < MB_LEN_MAX)
+ str[c1++] = c;
+ else
+ /* This is in any case an invalid class name. */
+ str[0] = '\0';
+ }
+ str[c1] = '\0';
+
+ if (c == '=' && *p == ']' && str[0] != '\0')
+ {
+ /* If we have no collation data we use the default
+ collation in which each character is in a class
+ by itself. It also means that ASCII is the
+ character set and therefore we cannot have character
+ with more than one byte in the multibyte
+ representation. */
+# ifdef _LIBC
+ if (nrules == 0)
+# endif
+ {
+ if (c1 != 1)
+ FREE_STACK_RETURN (REG_ECOLLATE);
+
+ /* Throw away the ] at the end of the equivalence
+ class. */
+ PATFETCH (c);
+
+ /* Set the bit for the character. */
+ SET_LIST_BIT (str[0]);
+ }
+# ifdef _LIBC
+ else
+ {
+ /* Try to match the byte sequence in `str' against
+ those known to the collate implementation.
+ First find out whether the bytes in `str' are
+ actually from exactly one character. */
+ const int32_t *table;
+ const unsigned char *weights;
+ const unsigned char *extra;
+ const int32_t *indirect;
+ int32_t idx;
+ const unsigned char *cp = str;
+ int ch;
+
+ /* This #include defines a local function! */
+# include <locale/weight.h>
+
+ table = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
+ weights = (const unsigned char *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
+ extra = (const unsigned char *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
+ indirect = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
+
+ idx = findidx (&cp);
+ if (idx == 0 || cp < str + c1)
+ /* This is no valid character. */
+ FREE_STACK_RETURN (REG_ECOLLATE);
+
+ /* Throw away the ] at the end of the equivalence
+ class. */
+ PATFETCH (c);
+
+ /* Now we have to go throught the whole table
+ and find all characters which have the same
+ first level weight.
+
+ XXX Note that this is not entirely correct.
+ we would have to match multibyte sequences
+ but this is not possible with the current
+ implementation. */
+ for (ch = 1; ch < 256; ++ch)
+ /* XXX This test would have to be changed if we
+ would allow matching multibyte sequences. */
+ if (table[ch] > 0)
+ {
+ int32_t idx2 = table[ch];
+ size_t len = weights[idx2];
+
+ /* Test whether the lenghts match. */
+ if (weights[idx] == len)
+ {
+ /* They do. New compare the bytes of
+ the weight. */
+ size_t cnt = 0;
+
+ while (cnt < len
+ && (weights[idx + 1 + cnt]
+ == weights[idx2 + 1 + cnt]))
+ ++cnt;
+
+ if (cnt == len)
+ /* They match. Mark the character as
+ acceptable. */
+ SET_LIST_BIT (ch);
+ }
+ }
+ }
+# endif
+ had_char_class = true;
+ }
+ else
+ {
+ c1++;
+ while (c1--)
+ PATUNFETCH;
+ SET_LIST_BIT ('[');
+ SET_LIST_BIT ('=');
+ range_start = '=';
+ had_char_class = false;
+ }
+ }
+ else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
+ {
+ unsigned char str[128]; /* Should be large enough. */
+# ifdef _LIBC
+ uint32_t nrules =
+ _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
+# endif
+
+ PATFETCH (c);
+ c1 = 0;
+
+ /* If pattern is `[[.'. */
+ if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+
+ for (;;)
+ {
+ PATFETCH (c);
+ if ((c == '.' && *p == ']') || p == pend)
+ break;
+ if (c1 < sizeof (str))
+ str[c1++] = c;
+ else
+ /* This is in any case an invalid class name. */
+ str[0] = '\0';
}
+ str[c1] = '\0';
+
+ if (c == '.' && *p == ']' && str[0] != '\0')
+ {
+ /* If we have no collation data we use the default
+ collation in which each character is the name
+ for its own class which contains only the one
+ character. It also means that ASCII is the
+ character set and therefore we cannot have character
+ with more than one byte in the multibyte
+ representation. */
+# ifdef _LIBC
+ if (nrules == 0)
+# endif
+ {
+ if (c1 != 1)
+ FREE_STACK_RETURN (REG_ECOLLATE);
+
+ /* Throw away the ] at the end of the equivalence
+ class. */
+ PATFETCH (c);
+
+ /* Set the bit for the character. */
+ SET_LIST_BIT (str[0]);
+ range_start = ((const unsigned char *) str)[0];
+ }
+# ifdef _LIBC
+ else
+ {
+ /* Try to match the byte sequence in `str' against
+ those known to the collate implementation.
+ First find out whether the bytes in `str' are
+ actually from exactly one character. */
+ int32_t table_size;
+ const int32_t *symb_table;
+ const unsigned char *extra;
+ int32_t idx;
+ int32_t elem;
+ int32_t second;
+ int32_t hash;
+
+ table_size =
+ _NL_CURRENT_WORD (LC_COLLATE,
+ _NL_COLLATE_SYMB_HASH_SIZEMB);
+ symb_table = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_SYMB_TABLEMB);
+ extra = (const unsigned char *)
+ _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_SYMB_EXTRAMB);
+
+ /* Locate the character in the hashing table. */
+ hash = elem_hash (str, c1);
+
+ idx = 0;
+ elem = hash % table_size;
+ second = hash % (table_size - 2);
+ while (symb_table[2 * elem] != 0)
+ {
+ /* First compare the hashing value. */
+ if (symb_table[2 * elem] == hash
+ && c1 == extra[symb_table[2 * elem + 1]]
+ && memcmp (str,
+ &extra[symb_table[2 * elem + 1]
+ + 1],
+ c1) == 0)
+ {
+ /* Yep, this is the entry. */
+ idx = symb_table[2 * elem + 1];
+ idx += 1 + extra[idx];
+ break;
+ }
+
+ /* Next entry. */
+ elem += second;
+ }
+
+ if (symb_table[2 * elem] == 0)
+ /* This is no valid character. */
+ FREE_STACK_RETURN (REG_ECOLLATE);
+
+ /* Throw away the ] at the end of the equivalence
+ class. */
+ PATFETCH (c);
+
+ /* Now add the multibyte character(s) we found
+ to the accept list.
+
+ XXX Note that this is not entirely correct.
+ we would have to match multibyte sequences
+ but this is not possible with the current
+ implementation. Also, we have to match
+ collating symbols, which expand to more than
+ one file, as a whole and not allow the
+ individual bytes. */
+ c1 = extra[idx++];
+ if (c1 == 1)
+ range_start = extra[idx];
+ while (c1-- > 0)
+ {
+ SET_LIST_BIT (extra[idx]);
+ ++idx;
+ }
+ }
+# endif
+ had_char_class = false;
+ }
else
{
c1++;
while (c1--)
PATUNFETCH;
SET_LIST_BIT ('[');
- SET_LIST_BIT (':');
+ SET_LIST_BIT ('.');
+ range_start = '.';
had_char_class = false;
}
- }
+ }
else
{
had_char_class = false;
SET_LIST_BIT (c);
+ range_start = c;
}
}
while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
b[-1]--;
b += b[-1];
+#endif /* WCHAR */
}
break;
group. They are all relative offsets, so that if the
whole pattern moves because of realloc, they will still
be valid. */
- COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
+ COMPILE_STACK_TOP.begalt_offset = begalt - COMPILED_BUFFER_VAR;
COMPILE_STACK_TOP.fixup_alt_jump
- = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
- COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
+ = fixup_alt_jump ? fixup_alt_jump - COMPILED_BUFFER_VAR + 1 : 0;
+ COMPILE_STACK_TOP.laststart_offset = b - COMPILED_BUFFER_VAR;
COMPILE_STACK_TOP.regnum = regnum;
/* We will eventually replace the 0 with the number of
represent in the compiled pattern. */
if (regnum <= MAX_REGNUM)
{
- COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
+ COMPILE_STACK_TOP.inner_group_offset = b
+ - COMPILED_BUFFER_VAR + 2;
BUF_PUSH_3 (start_memory, regnum, 0);
}
regnum_t this_group_regnum;
compile_stack.avail--;
- begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
+ begalt = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.begalt_offset;
fixup_alt_jump
= COMPILE_STACK_TOP.fixup_alt_jump
- ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
+ ? COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.fixup_alt_jump - 1
: 0;
- laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
+ laststart = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.laststart_offset;
this_group_regnum = COMPILE_STACK_TOP.regnum;
/* If we've reached MAX_REGNUM groups, then this open
won't actually generate any code, so we'll have to
groups were inside this one. */
if (this_group_regnum <= MAX_REGNUM)
{
- unsigned char *inner_group_loc
- = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
+ UCHAR_T *inner_group_loc
+ = COMPILED_BUFFER_VAR + COMPILE_STACK_TOP.inner_group_offset;
*inner_group_loc = regnum - this_group_regnum;
BUF_PUSH_3 (stop_memory, this_group_regnum,
/* Insert before the previous alternative a jump which
jumps to this alternative if the former fails. */
- GET_BUFFER_SPACE (3);
- INSERT_JUMP (on_failure_jump, begalt, b + 6);
+ GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
+ INSERT_JUMP (on_failure_jump, begalt,
+ b + 2 + 2 * OFFSET_ADDRESS_SIZE);
pending_exact = 0;
- b += 3;
+ b += 1 + OFFSET_ADDRESS_SIZE;
/* The alternative before this one has a jump after it
which gets executed if it gets matched. Adjust that
to be filled in later either by next alternative or
when know we're at the end of a series of alternatives. */
fixup_alt_jump = b;
- GET_BUFFER_SPACE (3);
- b += 3;
+ GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
+ b += 1 + OFFSET_ADDRESS_SIZE;
laststart = 0;
begalt = b;
if (!(syntax & RE_INTERVALS)
/* If we're at `\{' and it's not the open-interval
operator. */
- || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
- || (p - 2 == pattern && p == pend))
+ || (syntax & RE_NO_BK_BRACES))
goto normal_backslash;
handle_interval:
/* At least (most) this many matches must be made. */
int lower_bound = -1, upper_bound = -1;
- beg_interval = p - 1;
+ /* Place in the uncompiled pattern (i.e., just after
+ the '{') to go back to if the interval is invalid. */
+ const CHAR_T *beg_interval = p;
if (p == pend)
- {
- if (syntax & RE_NO_BK_BRACES)
- goto unfetch_interval;
- else
- FREE_STACK_RETURN (REG_EBRACE);
- }
+ goto invalid_interval;
GET_UNSIGNED_NUMBER (lower_bound);
if (c == ',')
{
GET_UNSIGNED_NUMBER (upper_bound);
- if (upper_bound < 0) upper_bound = RE_DUP_MAX;
+ if (upper_bound < 0)
+ upper_bound = RE_DUP_MAX;
}
else
/* Interval such as `{1}' => match exactly once. */
upper_bound = lower_bound;
- if (lower_bound < 0 || upper_bound > RE_DUP_MAX
- || lower_bound > upper_bound)
- {
- if (syntax & RE_NO_BK_BRACES)
- goto unfetch_interval;
- else
- FREE_STACK_RETURN (REG_BADBR);
- }
+ if (! (0 <= lower_bound && lower_bound <= upper_bound))
+ goto invalid_interval;
if (!(syntax & RE_NO_BK_BRACES))
{
- if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
-
+ if (c != '\\' || p == pend)
+ goto invalid_interval;
PATFETCH (c);
}
if (c != '}')
- {
- if (syntax & RE_NO_BK_BRACES)
- goto unfetch_interval;
- else
- FREE_STACK_RETURN (REG_BADBR);
- }
-
- /* We just parsed a valid interval. */
+ goto invalid_interval;
/* If it's invalid to have no preceding re. */
if (!laststart)
{
- if (syntax & RE_CONTEXT_INVALID_OPS)
+ if (syntax & RE_CONTEXT_INVALID_OPS
+ && !(syntax & RE_INVALID_INTERVAL_ORD))
FREE_STACK_RETURN (REG_BADRPT);
else if (syntax & RE_CONTEXT_INDEP_OPS)
laststart = b;
goto unfetch_interval;
}
+ /* We just parsed a valid interval. */
+
+ if (RE_DUP_MAX < upper_bound)
+ FREE_STACK_RETURN (REG_BADBR);
+
/* If the upper bound is zero, don't want to succeed at
all; jump from `laststart' to `b + 3', which will be
- the end of the buffer after we insert the jump. */
+ the end of the buffer after we insert the jump. */
+ /* ifdef WCHAR, 'b + 1 + OFFSET_ADDRESS_SIZE'
+ instead of 'b + 3'. */
if (upper_bound == 0)
{
- GET_BUFFER_SPACE (3);
- INSERT_JUMP (jump, laststart, b + 3);
- b += 3;
+ GET_BUFFER_SPACE (1 + OFFSET_ADDRESS_SIZE);
+ INSERT_JUMP (jump, laststart, b + 1
+ + OFFSET_ADDRESS_SIZE);
+ b += 1 + OFFSET_ADDRESS_SIZE;
}
/* Otherwise, we have a nontrivial interval. When
else
{ /* If the upper bound is > 1, we need to insert
more at the end of the loop. */
- unsigned nbytes = 10 + (upper_bound > 1) * 10;
+ unsigned nbytes = 2 + 4 * OFFSET_ADDRESS_SIZE +
+ (upper_bound > 1) * (2 + 4 * OFFSET_ADDRESS_SIZE);
GET_BUFFER_SPACE (nbytes);
because `re_compile_fastmap' needs to know.
Jump to the `jump_n' we might insert below. */
INSERT_JUMP2 (succeed_n, laststart,
- b + 5 + (upper_bound > 1) * 5,
- lower_bound);
- b += 5;
+ b + 1 + 2 * OFFSET_ADDRESS_SIZE
+ + (upper_bound > 1) * (1 + 2 * OFFSET_ADDRESS_SIZE)
+ , lower_bound);
+ b += 1 + 2 * OFFSET_ADDRESS_SIZE;
/* Code to initialize the lower bound. Insert
before the `succeed_n'. The `5' is the last two
bytes of this `set_number_at', plus 3 bytes of
the following `succeed_n'. */
- insert_op2 (set_number_at, laststart, 5, lower_bound, b);
- b += 5;
+ /* ifdef WCHAR, The '1+2*OFFSET_ADDRESS_SIZE'
+ is the 'set_number_at', plus '1+OFFSET_ADDRESS_SIZE'
+ of the following `succeed_n'. */
+ PREFIX(insert_op2) (set_number_at, laststart, 1
+ + 2 * OFFSET_ADDRESS_SIZE, lower_bound, b);
+ b += 1 + 2 * OFFSET_ADDRESS_SIZE;
if (upper_bound > 1)
{ /* More than one repetition is allowed, so
When we've reached this during matching,
we'll have matched the interval once, so
jump back only `upper_bound - 1' times. */
- STORE_JUMP2 (jump_n, b, laststart + 5,
+ STORE_JUMP2 (jump_n, b, laststart
+ + 2 * OFFSET_ADDRESS_SIZE + 1,
upper_bound - 1);
- b += 5;
+ b += 1 + 2 * OFFSET_ADDRESS_SIZE;
/* The location we want to set is the second
parameter of the `jump_n'; that is `b-2' as
We insert this at the beginning of the loop
so that if we fail during matching, we'll
reinitialize the bounds. */
- insert_op2 (set_number_at, laststart, b - laststart,
- upper_bound - 1, b);
- b += 5;
+ PREFIX(insert_op2) (set_number_at, laststart,
+ b - laststart,
+ upper_bound - 1, b);
+ b += 1 + 2 * OFFSET_ADDRESS_SIZE;
}
}
pending_exact = 0;
- beg_interval = NULL;
- }
- break;
-
- unfetch_interval:
- /* If an invalid interval, match the characters as literals. */
- assert (beg_interval);
- p = beg_interval;
- beg_interval = NULL;
-
- /* normal_char and normal_backslash need `c'. */
- PATFETCH (c);
-
- if (!(syntax & RE_NO_BK_BRACES))
- {
- if (p > pattern && p[-1] == '\\')
- goto normal_backslash;
- }
- goto normal_char;
+ break;
+
+ invalid_interval:
+ if (!(syntax & RE_INVALID_INTERVAL_ORD))
+ FREE_STACK_RETURN (p == pend ? REG_EBRACE : REG_BADBR);
+ unfetch_interval:
+ /* Match the characters as literals. */
+ p = beg_interval;
+ c = '{';
+ if (syntax & RE_NO_BK_BRACES)
+ goto normal_char;
+ else
+ goto normal_backslash;
+ }
#ifdef emacs
/* There is no way to specify the before_dot and after_dot
normal_char:
/* If no exactn currently being built. */
if (!pending_exact
+#ifdef WCHAR
+ /* If last exactn handle binary(or character) and
+ new exactn handle character(or binary). */
+ || is_exactn_bin != is_binary[p - 1 - pattern]
+#endif /* WCHAR */
/* If last exactn not at current position. */
|| pending_exact + *pending_exact + 1 != b
laststart = b;
+#ifdef WCHAR
+ /* Is this exactn binary data or character? */
+ is_exactn_bin = is_binary[p - 1 - pattern];
+ if (is_exactn_bin)
+ BUF_PUSH_2 (exactn_bin, 0);
+ else
+ BUF_PUSH_2 (exactn, 0);
+#else
BUF_PUSH_2 (exactn, 0);
+#endif /* WCHAR */
pending_exact = b - 1;
}
if (syntax & RE_NO_POSIX_BACKTRACKING)
BUF_PUSH (succeed);
+#ifdef WCHAR
+ free (pattern);
+ free (mbs_offset);
+ free (is_binary);
+#endif
free (compile_stack.stack);
/* We have succeeded; set the length of the buffer. */
+#ifdef WCHAR
+ bufp->used = (uintptr_t) b - (uintptr_t) COMPILED_BUFFER_VAR;
+#else
bufp->used = b - bufp->buffer;
+#endif
#ifdef DEBUG
if (debug)
{
DEBUG_PRINT1 ("\nCompiled pattern: \n");
- print_compiled_pattern (bufp);
+ PREFIX(print_compiled_pattern) (bufp);
}
#endif /* DEBUG */
# ifdef emacs
if (! fail_stack.stack)
fail_stack.stack
- = (fail_stack_elt_t *) xmalloc (fail_stack.size
- * sizeof (fail_stack_elt_t));
+ = (PREFIX(fail_stack_elt_t) *) xmalloc (fail_stack.size
+ * sizeof (PREFIX(fail_stack_elt_t)));
else
fail_stack.stack
- = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
- (fail_stack.size
- * sizeof (fail_stack_elt_t)));
+ = (PREFIX(fail_stack_elt_t) *) xrealloc (fail_stack.stack,
+ (fail_stack.size
+ * sizeof (PREFIX(fail_stack_elt_t))));
# else /* not emacs */
if (! fail_stack.stack)
fail_stack.stack
- = (fail_stack_elt_t *) malloc (fail_stack.size
- * sizeof (fail_stack_elt_t));
+ = malloc (fail_stack.size * sizeof (PREFIX(fail_stack_elt_t)));
else
fail_stack.stack
- = (fail_stack_elt_t *) realloc (fail_stack.stack,
- (fail_stack.size
- * sizeof (fail_stack_elt_t)));
+ = realloc (fail_stack.stack,
+ fail_stack.size * sizeof (PREFIX(fail_stack_elt_t)));
# endif /* not emacs */
}
- regex_grow_registers (num_regs);
+ PREFIX(regex_grow_registers) (num_regs);
}
#endif /* not MATCH_MAY_ALLOCATE */
return REG_NOERROR;
} /* regex_compile */
-\f
+
/* Subroutines for `regex_compile'. */
/* Store OP at LOC followed by two-byte integer parameter ARG. */
+/* ifdef WCHAR, integer parameter is 1 wchar_t. */
static void
-store_op1 (op, loc, arg)
- re_opcode_t op;
- unsigned char *loc;
- int arg;
+PREFIX(store_op1) (re_opcode_t op, UCHAR_T *loc, int arg)
{
- *loc = (unsigned char) op;
+ *loc = (UCHAR_T) op;
STORE_NUMBER (loc + 1, arg);
}
/* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
+/* ifdef WCHAR, integer parameter is 1 wchar_t. */
static void
-store_op2 (op, loc, arg1, arg2)
- re_opcode_t op;
- unsigned char *loc;
- int arg1, arg2;
+PREFIX(store_op2) (re_opcode_t op, UCHAR_T *loc, int arg1, int arg2)
{
- *loc = (unsigned char) op;
+ *loc = (UCHAR_T) op;
STORE_NUMBER (loc + 1, arg1);
- STORE_NUMBER (loc + 3, arg2);
+ STORE_NUMBER (loc + 1 + OFFSET_ADDRESS_SIZE, arg2);
}
/* Copy the bytes from LOC to END to open up three bytes of space at LOC
for OP followed by two-byte integer parameter ARG. */
+/* ifdef WCHAR, integer parameter is 1 wchar_t. */
static void
-insert_op1 (op, loc, arg, end)
- re_opcode_t op;
- unsigned char *loc;
- int arg;
- unsigned char *end;
+PREFIX(insert_op1) (re_opcode_t op, UCHAR_T *loc, int arg, UCHAR_T *end)
{
- register unsigned char *pfrom = end;
- register unsigned char *pto = end + 3;
+ register UCHAR_T *pfrom = end;
+ register UCHAR_T *pto = end + 1 + OFFSET_ADDRESS_SIZE;
while (pfrom != loc)
*--pto = *--pfrom;
- store_op1 (op, loc, arg);
+ PREFIX(store_op1) (op, loc, arg);
}
/* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
+/* ifdef WCHAR, integer parameter is 1 wchar_t. */
static void
-insert_op2 (op, loc, arg1, arg2, end)
- re_opcode_t op;
- unsigned char *loc;
- int arg1, arg2;
- unsigned char *end;
+PREFIX(insert_op2) (re_opcode_t op, UCHAR_T *loc, int arg1, int arg2,
+ UCHAR_T *end)
{
- register unsigned char *pfrom = end;
- register unsigned char *pto = end + 5;
+ register UCHAR_T *pfrom = end;
+ register UCHAR_T *pto = end + 1 + 2 * OFFSET_ADDRESS_SIZE;
while (pfrom != loc)
*--pto = *--pfrom;
- store_op2 (op, loc, arg1, arg2);
+ PREFIX(store_op2) (op, loc, arg1, arg2);
}
least one character before the ^. */
static boolean
-at_begline_loc_p (pattern, p, syntax)
- const char *pattern, *p;
- reg_syntax_t syntax;
+PREFIX(at_begline_loc_p) (const CHAR_T *pattern, const CHAR_T *p,
+ reg_syntax_t syntax)
{
- const char *prev = p - 2;
+ const CHAR_T *prev = p - 2;
boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
return
at least one character after the $, i.e., `P < PEND'. */
static boolean
-at_endline_loc_p (p, pend, syntax)
- const char *p, *pend;
- reg_syntax_t syntax;
+PREFIX(at_endline_loc_p) (const CHAR_T *p, const CHAR_T *pend,
+ reg_syntax_t syntax)
{
- const char *next = p;
+ const CHAR_T *next = p;
boolean next_backslash = *next == '\\';
- const char *next_next = p + 1 < pend ? p + 1 : 0;
+ const CHAR_T *next_next = p + 1 < pend ? p + 1 : 0;
return
/* Before a subexpression? */
: next_backslash && next_next && *next_next == '|');
}
+#else /* not INSIDE_RECURSION */
/* Returns true if REGNUM is in one of COMPILE_STACK's elements and
false if it's not. */
static boolean
-group_in_compile_stack (compile_stack, regnum)
- compile_stack_type compile_stack;
- regnum_t regnum;
+group_in_compile_stack (compile_stack_type compile_stack,
+ regnum_t regnum)
{
int this_element;
return false;
}
+#endif /* not INSIDE_RECURSION */
+
+#ifdef INSIDE_RECURSION
+
+#ifdef WCHAR
+/* This insert space, which size is "num", into the pattern at "loc".
+ "end" must point the end of the allocated buffer. */
+static void
+insert_space (int num, CHAR_T *loc, CHAR_T *end)
+{
+ register CHAR_T *pto = end;
+ register CHAR_T *pfrom = end - num;
+
+ while (pfrom >= loc)
+ *pto-- = *pfrom--;
+}
+#endif /* WCHAR */
+
+#ifdef WCHAR
+static reg_errcode_t
+wcs_compile_range (CHAR_T range_start_char,
+ const CHAR_T **p_ptr, const CHAR_T *pend,
+ RE_TRANSLATE_TYPE translate, reg_syntax_t syntax,
+ CHAR_T *b, CHAR_T *char_set)
+{
+ const CHAR_T *p = *p_ptr;
+ CHAR_T range_start, range_end;
+ reg_errcode_t ret;
+# ifdef _LIBC
+ uint32_t nrules;
+ uint32_t start_val, end_val;
+# endif
+ if (p == pend)
+ return REG_ERANGE;
+
+# ifdef _LIBC
+ nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
+ if (nrules != 0)
+ {
+ const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE,
+ _NL_COLLATE_COLLSEQWC);
+ const unsigned char *extra = (const unsigned char *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
+
+ if (range_start_char < -1)
+ {
+ /* range_start is a collating symbol. */
+ int32_t *wextra;
+ /* Retreive the index and get collation sequence value. */
+ wextra = (int32_t*)(extra + char_set[-range_start_char]);
+ start_val = wextra[1 + *wextra];
+ }
+ else
+ start_val = collseq_table_lookup(collseq, TRANSLATE(range_start_char));
+
+ end_val = collseq_table_lookup (collseq, TRANSLATE (p[0]));
+ /* Report an error if the range is empty and the syntax prohibits
+ this. */
+ ret = ((syntax & RE_NO_EMPTY_RANGES)
+ && (start_val > end_val))? REG_ERANGE : REG_NOERROR;
+
+ /* Insert space to the end of the char_ranges. */
+ insert_space(2, b - char_set[5] - 2, b - 1);
+ *(b - char_set[5] - 2) = (wchar_t)start_val;
+ *(b - char_set[5] - 1) = (wchar_t)end_val;
+ char_set[4]++; /* ranges_index */
+ }
+ else
+# endif
+ {
+ range_start = (range_start_char >= 0)? TRANSLATE (range_start_char):
+ range_start_char;
+ range_end = TRANSLATE (p[0]);
+ /* Report an error if the range is empty and the syntax prohibits
+ this. */
+ ret = ((syntax & RE_NO_EMPTY_RANGES)
+ && (range_start > range_end))? REG_ERANGE : REG_NOERROR;
+
+ /* Insert space to the end of the char_ranges. */
+ insert_space(2, b - char_set[5] - 2, b - 1);
+ *(b - char_set[5] - 2) = range_start;
+ *(b - char_set[5] - 1) = range_end;
+ char_set[4]++; /* ranges_index */
+ }
+ /* Have to increment the pointer into the pattern string, so the
+ caller isn't still at the ending character. */
+ (*p_ptr)++;
+ return ret;
+}
+#else /* BYTE */
/* Read the ending character of a range (in a bracket expression) from the
uncompiled pattern *P_PTR (which ends at PEND). We assume the
starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
`regex_compile' itself. */
static reg_errcode_t
-compile_range (p_ptr, pend, translate, syntax, b)
- const char **p_ptr, *pend;
- RE_TRANSLATE_TYPE translate;
- reg_syntax_t syntax;
- unsigned char *b;
+byte_compile_range (unsigned int range_start_char,
+ const char **p_ptr, const char *pend,
+ RE_TRANSLATE_TYPE translate, reg_syntax_t syntax,
+ unsigned char *b)
{
unsigned this_char;
-
const char *p = *p_ptr;
- unsigned int range_start, range_end;
+ reg_errcode_t ret;
+# if _LIBC
+ const unsigned char *collseq;
+ unsigned int start_colseq;
+ unsigned int end_colseq;
+# else
+ unsigned end_char;
+# endif
if (p == pend)
return REG_ERANGE;
- /* Even though the pattern is a signed `char *', we need to fetch
- with unsigned char *'s; if the high bit of the pattern character
- is set, the range endpoints will be negative if we fetch using a
- signed char *.
-
- We also want to fetch the endpoints without translating them; the
- appropriate translation is done in the bit-setting loop below. */
- /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
- range_start = ((const unsigned char *) p)[-2];
- range_end = ((const unsigned char *) p)[0];
-
/* Have to increment the pointer into the pattern string, so the
caller isn't still at the ending character. */
(*p_ptr)++;
- /* If the start is after the end, the range is empty. */
- if (range_start > range_end)
- return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
+ /* Report an error if the range is empty and the syntax prohibits this. */
+ ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
+
+# if _LIBC
+ collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
+ _NL_COLLATE_COLLSEQMB);
+
+ start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
+ end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
+ for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
+ {
+ unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
+ if (start_colseq <= this_colseq && this_colseq <= end_colseq)
+ {
+ SET_LIST_BIT (TRANSLATE (this_char));
+ ret = REG_NOERROR;
+ }
+ }
+# else
/* Here we see why `this_char' has to be larger than an `unsigned
- char' -- the range is inclusive, so if `range_end' == 0xff
- (assuming 8-bit characters), we would otherwise go into an infinite
- loop, since all characters <= 0xff. */
- for (this_char = range_start; this_char <= range_end; this_char++)
+ char' -- we would otherwise go into an infinite loop, since all
+ characters <= 0xff. */
+ range_start_char = TRANSLATE (range_start_char);
+ /* TRANSLATE(p[0]) is casted to char (not unsigned char) in TRANSLATE,
+ and some compilers cast it to int implicitly, so following for_loop
+ may fall to (almost) infinite loop.
+ e.g. If translate[p[0]] = 0xff, end_char may equals to 0xffffffff.
+ To avoid this, we cast p[0] to unsigned int and truncate it. */
+ end_char = ((unsigned)TRANSLATE(p[0]) & ((1 << BYTEWIDTH) - 1));
+
+ for (this_char = range_start_char; this_char <= end_char; ++this_char)
{
SET_LIST_BIT (TRANSLATE (this_char));
+ ret = REG_NOERROR;
}
+# endif
- return REG_NOERROR;
+ return ret;
}
+#endif /* WCHAR */
\f
/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
Returns 0 if we succeed, -2 if an internal error. */
-int
-re_compile_fastmap (bufp)
- struct re_pattern_buffer *bufp;
+#ifdef WCHAR
+/* local function for re_compile_fastmap.
+ truncate wchar_t character to char. */
+
+static unsigned char
+truncate_wchar (CHAR_T c)
+{
+ unsigned char buf[MB_CUR_MAX];
+ mbstate_t state;
+ int retval;
+ memset (&state, '\0', sizeof (state));
+ retval = wcrtomb (buf, c, &state);
+ return retval > 0 ? buf[0] : (unsigned char) c;
+}
+#endif /* WCHAR */
+
+static int
+PREFIX(re_compile_fastmap) (struct re_pattern_buffer *bufp)
{
int j, k;
#ifdef MATCH_MAY_ALLOCATE
- fail_stack_type fail_stack;
+ PREFIX(fail_stack_type) fail_stack;
#endif
#ifndef REGEX_MALLOC
char *destination;
#endif
register char *fastmap = bufp->fastmap;
- unsigned char *pattern = bufp->buffer;
- unsigned char *p = pattern;
- register unsigned char *pend = pattern + bufp->used;
+
+#ifdef WCHAR
+ /* We need to cast pattern to (wchar_t*), because we casted this compiled
+ pattern to (char*) in regex_compile. */
+ UCHAR_T *pattern = (UCHAR_T*)bufp->buffer;
+ register UCHAR_T *pend = (UCHAR_T*) (bufp->buffer + bufp->used);
+#else /* BYTE */
+ UCHAR_T *pattern = bufp->buffer;
+ register UCHAR_T *pend = pattern + bufp->used;
+#endif /* WCHAR */
+ UCHAR_T *p = pattern;
#ifdef REL_ALLOC
/* This holds the pointer to the failure stack, when
/* Following are the cases which match a character. These end
with `break'. */
+#ifdef WCHAR
+ case exactn:
+ fastmap[truncate_wchar(p[1])] = 1;
+ break;
+#else /* BYTE */
case exactn:
fastmap[p[1]] = 1;
break;
+#endif /* WCHAR */
+#ifdef MBS_SUPPORT
+ case exactn_bin:
+ fastmap[p[1]] = 1;
+ break;
+#endif
-
+#ifdef WCHAR
+ /* It is hard to distinguish fastmap from (multi byte) characters
+ which depends on current locale. */
+ case charset:
+ case charset_not:
+ case wordchar:
+ case notwordchar:
+ bufp->can_be_null = 1;
+ goto done;
+#else /* BYTE */
case charset:
for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
if (SYNTAX (j) != Sword)
fastmap[j] = 1;
break;
-
+#endif /* WCHAR */
case anychar:
{
case succeed_n:
/* Get to the number of times to succeed. */
- p += 2;
+ p += OFFSET_ADDRESS_SIZE;
/* Increment p past the n for when k != 0. */
EXTRACT_NUMBER_AND_INCR (k, p);
if (k == 0)
{
- p -= 4;
+ p -= 2 * OFFSET_ADDRESS_SIZE;
succeed_n_p = true; /* Spaghetti code alert. */
goto handle_on_failure_jump;
}
case set_number_at:
- p += 4;
+ p += 2 * OFFSET_ADDRESS_SIZE;
continue;
done:
RESET_FAIL_STACK ();
return 0;
+}
+
+#else /* not INSIDE_RECURSION */
+
+int
+re_compile_fastmap (struct re_pattern_buffer *bufp)
+{
+# ifdef MBS_SUPPORT
+ if (MB_CUR_MAX != 1)
+ return wcs_re_compile_fastmap(bufp);
+ else
+# endif
+ return byte_re_compile_fastmap(bufp);
} /* re_compile_fastmap */
#ifdef _LIBC
weak_alias (__re_compile_fastmap, re_compile_fastmap)
#endif
\f
+
/* Set REGS to hold NUM_REGS registers, storing them in STARTS and
ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
this memory for recording register information. STARTS and ENDS
freeing the old data. */
void
-re_set_registers (bufp, regs, num_regs, starts, ends)
- struct re_pattern_buffer *bufp;
- struct re_registers *regs;
- unsigned num_regs;
- regoff_t *starts, *ends;
+re_set_registers (struct re_pattern_buffer *bufp,
+ struct re_registers *regs,
+ unsigned int num_regs,
+ regoff_t *starts, regoff_t *ends)
{
if (num_regs)
{
/* Searching routines. */
/* Like re_search_2, below, but only one string is specified, and
- doesn't let you say where to stop matching. */
+ doesn't let you say where to stop matching. */
int
-re_search (bufp, string, size, startpos, range, regs)
- struct re_pattern_buffer *bufp;
- const char *string;
- int size, startpos, range;
- struct re_registers *regs;
+re_search (struct re_pattern_buffer *bufp,
+ const char *string,
+ int size, int startpos, int range,
+ struct re_registers *regs)
{
return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
regs, size);
stack overflow). */
int
-re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
- struct re_pattern_buffer *bufp;
- const char *string1, *string2;
- int size1, size2;
- int startpos;
- int range;
- struct re_registers *regs;
- int stop;
+re_search_2 (struct re_pattern_buffer *bufp,
+ const char *string1, int size1,
+ const char *string2, int size2,
+ int startpos, int range,
+ struct re_registers *regs,
+ int stop)
+{
+# ifdef MBS_SUPPORT
+ if (MB_CUR_MAX != 1)
+ return wcs_re_search_2 (bufp, string1, size1, string2, size2, startpos,
+ range, regs, stop);
+ else
+# endif
+ return byte_re_search_2 (bufp, string1, size1, string2, size2, startpos,
+ range, regs, stop);
+} /* re_search_2 */
+#ifdef _LIBC
+weak_alias (__re_search_2, re_search_2)
+#endif
+
+#endif /* not INSIDE_RECURSION */
+
+#ifdef INSIDE_RECURSION
+
+#ifdef MATCH_MAY_ALLOCATE
+# define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
+#else
+# define FREE_VAR(var) if (var) free (var); var = NULL
+#endif
+
+#ifdef WCHAR
+# define MAX_ALLOCA_SIZE 2000
+
+# define FREE_WCS_BUFFERS() \
+ do { \
+ if (size1 > MAX_ALLOCA_SIZE) \
+ { \
+ free (wcs_string1); \
+ free (mbs_offset1); \
+ } \
+ else \
+ { \
+ FREE_VAR (wcs_string1); \
+ FREE_VAR (mbs_offset1); \
+ } \
+ if (size2 > MAX_ALLOCA_SIZE) \
+ { \
+ free (wcs_string2); \
+ free (mbs_offset2); \
+ } \
+ else \
+ { \
+ FREE_VAR (wcs_string2); \
+ FREE_VAR (mbs_offset2); \
+ } \
+ } while (0)
+
+#endif
+
+
+static int
+PREFIX(re_search_2) (struct re_pattern_buffer *bufp,
+ const char *string1, int size1,
+ const char *string2, int size2,
+ int startpos, int range,
+ struct re_registers *regs,
+ int stop)
{
int val;
register char *fastmap = bufp->fastmap;
register RE_TRANSLATE_TYPE translate = bufp->translate;
int total_size = size1 + size2;
int endpos = startpos + range;
+#ifdef WCHAR
+ /* We need wchar_t* buffers correspond to cstring1, cstring2. */
+ wchar_t *wcs_string1 = NULL, *wcs_string2 = NULL;
+ /* We need the size of wchar_t buffers correspond to csize1, csize2. */
+ int wcs_size1 = 0, wcs_size2 = 0;
+ /* offset buffer for optimizatoin. See convert_mbs_to_wc. */
+ int *mbs_offset1 = NULL, *mbs_offset2 = NULL;
+ /* They hold whether each wchar_t is binary data or not. */
+ char *is_binary = NULL;
+#endif /* WCHAR */
/* Check for out-of-range STARTPOS. */
if (startpos < 0 || startpos > total_size)
if (re_compile_fastmap (bufp) == -2)
return -2;
+#ifdef WCHAR
+ /* Allocate wchar_t array for wcs_string1 and wcs_string2 and
+ fill them with converted string. */
+ if (size1 != 0)
+ {
+ if (size1 > MAX_ALLOCA_SIZE)
+ {
+ wcs_string1 = TALLOC (size1 + 1, CHAR_T);
+ mbs_offset1 = TALLOC (size1 + 1, int);
+ is_binary = TALLOC (size1 + 1, char);
+ }
+ else
+ {
+ wcs_string1 = REGEX_TALLOC (size1 + 1, CHAR_T);
+ mbs_offset1 = REGEX_TALLOC (size1 + 1, int);
+ is_binary = REGEX_TALLOC (size1 + 1, char);
+ }
+ if (!wcs_string1 || !mbs_offset1 || !is_binary)
+ {
+ if (size1 > MAX_ALLOCA_SIZE)
+ {
+ free (wcs_string1);
+ free (mbs_offset1);
+ free (is_binary);
+ }
+ else
+ {
+ FREE_VAR (wcs_string1);
+ FREE_VAR (mbs_offset1);
+ FREE_VAR (is_binary);
+ }
+ return -2;
+ }
+ wcs_size1 = convert_mbs_to_wcs(wcs_string1, string1, size1,
+ mbs_offset1, is_binary);
+ wcs_string1[wcs_size1] = L'\0'; /* for a sentinel */
+ if (size1 > MAX_ALLOCA_SIZE)
+ free (is_binary);
+ else
+ FREE_VAR (is_binary);
+ }
+ if (size2 != 0)
+ {
+ if (size2 > MAX_ALLOCA_SIZE)
+ {
+ wcs_string2 = TALLOC (size2 + 1, CHAR_T);
+ mbs_offset2 = TALLOC (size2 + 1, int);
+ is_binary = TALLOC (size2 + 1, char);
+ }
+ else
+ {
+ wcs_string2 = REGEX_TALLOC (size2 + 1, CHAR_T);
+ mbs_offset2 = REGEX_TALLOC (size2 + 1, int);
+ is_binary = REGEX_TALLOC (size2 + 1, char);
+ }
+ if (!wcs_string2 || !mbs_offset2 || !is_binary)
+ {
+ FREE_WCS_BUFFERS ();
+ if (size2 > MAX_ALLOCA_SIZE)
+ free (is_binary);
+ else
+ FREE_VAR (is_binary);
+ return -2;
+ }
+ wcs_size2 = convert_mbs_to_wcs(wcs_string2, string2, size2,
+ mbs_offset2, is_binary);
+ wcs_string2[wcs_size2] = L'\0'; /* for a sentinel */
+ if (size2 > MAX_ALLOCA_SIZE)
+ free (is_binary);
+ else
+ FREE_VAR (is_binary);
+ }
+#endif /* WCHAR */
+
+
/* Loop through the string, looking for a place to start matching. */
for (;;)
{
}
else /* Searching backwards. */
{
- register char c = (size1 == 0 || startpos >= size1
- ? string2[startpos - size1]
- : string1[startpos]);
+ register CHAR_T c = (size1 == 0 || startpos >= size1
+ ? string2[startpos - size1]
+ : string1[startpos]);
if (!fastmap[(unsigned char) TRANSLATE (c)])
goto advance;
/* If can't match the null string, and that's all we have left, fail. */
if (range >= 0 && startpos == total_size && fastmap
&& !bufp->can_be_null)
- return -1;
+ {
+#ifdef WCHAR
+ FREE_WCS_BUFFERS ();
+#endif
+ return -1;
+ }
+
+#ifdef WCHAR
+ val = wcs_re_match_2_internal (bufp, string1, size1, string2,
+ size2, startpos, regs, stop,
+ wcs_string1, wcs_size1,
+ wcs_string2, wcs_size2,
+ mbs_offset1, mbs_offset2);
+#else /* BYTE */
+ val = byte_re_match_2_internal (bufp, string1, size1, string2,
+ size2, startpos, regs, stop);
+#endif /* BYTE */
- val = re_match_2_internal (bufp, string1, size1, string2, size2,
- startpos, regs, stop);
#ifndef REGEX_MALLOC
# ifdef C_ALLOCA
alloca (0);
#endif
if (val >= 0)
- return startpos;
+ {
+#ifdef WCHAR
+ FREE_WCS_BUFFERS ();
+#endif
+ return startpos;
+ }
if (val == -2)
- return -2;
+ {
+#ifdef WCHAR
+ FREE_WCS_BUFFERS ();
+#endif
+ return -2;
+ }
advance:
if (!range)
startpos--;
}
}
- return -1;
-} /* re_search_2 */
-#ifdef _LIBC
-weak_alias (__re_search_2, re_search_2)
+#ifdef WCHAR
+ FREE_WCS_BUFFERS ();
#endif
-\f
+ return -1;
+}
+
+#ifdef WCHAR
+/* This converts PTR, a pointer into one of the search wchar_t strings
+ `string1' and `string2' into an multibyte string offset from the
+ beginning of that string. We use mbs_offset to optimize.
+ See convert_mbs_to_wcs. */
+# define POINTER_TO_OFFSET(ptr) \
+ (FIRST_STRING_P (ptr) \
+ ? ((regoff_t)(mbs_offset1 != NULL? mbs_offset1[(ptr)-string1] : 0)) \
+ : ((regoff_t)((mbs_offset2 != NULL? mbs_offset2[(ptr)-string2] : 0) \
+ + csize1)))
+#else /* BYTE */
/* This converts PTR, a pointer into one of the search strings `string1'
and `string2' into an offset from the beginning of that string. */
-#define POINTER_TO_OFFSET(ptr) \
+# define POINTER_TO_OFFSET(ptr) \
(FIRST_STRING_P (ptr) \
? ((regoff_t) ((ptr) - string1)) \
: ((regoff_t) ((ptr) - string2 + size1)))
+#endif /* WCHAR */
/* Macros for dealing with the split strings in re_match_2. */
dend = end_match_2; \
}
-
/* Test if at very beginning or at very end of the virtual concatenation
of `string1' and `string2'. If only one string, it's `string2'. */
#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
two special cases to check for: if past the end of string1, look at
the first character in string2; and if before the beginning of
string2, look at the last character in string1. */
-#define WORDCHAR_P(d) \
+#ifdef WCHAR
+/* Use internationalized API instead of SYNTAX. */
+# define WORDCHAR_P(d) \
+ (iswalnum ((wint_t)((d) == end1 ? *string2 \
+ : (d) == string2 - 1 ? *(end1 - 1) : *(d))) != 0 \
+ || ((d) == end1 ? *string2 \
+ : (d) == string2 - 1 ? *(end1 - 1) : *(d)) == L'_')
+#else /* BYTE */
+# define WORDCHAR_P(d) \
(SYNTAX ((d) == end1 ? *string2 \
: (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
== Sword)
+#endif /* WCHAR */
/* Disabled due to a compiler bug -- see comment at case wordbound */
#if 0
/* Free everything we malloc. */
#ifdef MATCH_MAY_ALLOCATE
-# define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
-# define FREE_VARIABLES() \
+# ifdef WCHAR
+# define FREE_VARIABLES() \
+ do { \
+ REGEX_FREE_STACK (fail_stack.stack); \
+ FREE_VAR (regstart); \
+ FREE_VAR (regend); \
+ FREE_VAR (old_regstart); \
+ FREE_VAR (old_regend); \
+ FREE_VAR (best_regstart); \
+ FREE_VAR (best_regend); \
+ FREE_VAR (reg_info); \
+ FREE_VAR (reg_dummy); \
+ FREE_VAR (reg_info_dummy); \
+ if (!cant_free_wcs_buf) \
+ { \
+ FREE_VAR (string1); \
+ FREE_VAR (string2); \
+ FREE_VAR (mbs_offset1); \
+ FREE_VAR (mbs_offset2); \
+ } \
+ } while (0)
+# else /* BYTE */
+# define FREE_VARIABLES() \
do { \
REGEX_FREE_STACK (fail_stack.stack); \
FREE_VAR (regstart); \
FREE_VAR (reg_dummy); \
FREE_VAR (reg_info_dummy); \
} while (0)
+# endif /* WCHAR */
#else
-# define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
+# ifdef WCHAR
+# define FREE_VARIABLES() \
+ do { \
+ if (!cant_free_wcs_buf) \
+ { \
+ FREE_VAR (string1); \
+ FREE_VAR (string2); \
+ FREE_VAR (mbs_offset1); \
+ FREE_VAR (mbs_offset2); \
+ } \
+ } while (0)
+# else /* BYTE */
+# define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
+# endif /* WCHAR */
#endif /* not MATCH_MAY_ALLOCATE */
/* These values must meet several constraints. They must not be valid
#define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
#define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
\f
+#else /* not INSIDE_RECURSION */
/* Matching routines. */
#ifndef emacs /* Emacs never uses this. */
/* re_match is like re_match_2 except it takes only a single string. */
int
-re_match (bufp, string, size, pos, regs)
- struct re_pattern_buffer *bufp;
- const char *string;
- int size, pos;
- struct re_registers *regs;
+re_match (struct re_pattern_buffer *bufp,
+ const char *string,
+ int size, int pos,
+ struct re_registers *regs)
{
- int result = re_match_2_internal (bufp, NULL, 0, string, size,
- pos, regs, size);
+ int result;
+# ifdef MBS_SUPPORT
+ if (MB_CUR_MAX != 1)
+ result = wcs_re_match_2_internal (bufp, NULL, 0, string, size,
+ pos, regs, size,
+ NULL, 0, NULL, 0, NULL, NULL);
+ else
+# endif
+ result = byte_re_match_2_internal (bufp, NULL, 0, string, size,
+ pos, regs, size);
# ifndef REGEX_MALLOC
# ifdef C_ALLOCA
alloca (0);
# endif
#endif /* not emacs */
-static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
- unsigned char *end,
- register_info_type *reg_info));
-static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
- unsigned char *end,
- register_info_type *reg_info));
-static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
- unsigned char *end,
- register_info_type *reg_info));
-static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
- int len, char *translate));
+#endif /* not INSIDE_RECURSION */
+
+#ifdef INSIDE_RECURSION
+static boolean PREFIX(group_match_null_string_p) (UCHAR_T **p,
+ UCHAR_T *end,
+ PREFIX(register_info_type) *reg_info);
+static boolean PREFIX(alt_match_null_string_p) (UCHAR_T *p,
+ UCHAR_T *end,
+ PREFIX(register_info_type) *reg_info);
+static boolean PREFIX(common_op_match_null_string_p) (UCHAR_T **p,
+ UCHAR_T *end,
+ PREFIX(register_info_type) *reg_info);
+static int PREFIX(bcmp_translate) (const CHAR_T *s1, const CHAR_T *s2,
+ int len, char *translate);
+#else /* not INSIDE_RECURSION */
/* re_match_2 matches the compiled pattern in BUFP against the
the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
matched substring. */
int
-re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
- struct re_pattern_buffer *bufp;
- const char *string1, *string2;
- int size1, size2;
- int pos;
- struct re_registers *regs;
- int stop;
+re_match_2 (struct re_pattern_buffer *bufp,
+ const char *string1, int size1,
+ const char *string2, int size2,
+ int pos, struct re_registers *regs,
+ int stop)
{
- int result = re_match_2_internal (bufp, string1, size1, string2, size2,
- pos, regs, stop);
+ int result;
+# ifdef MBS_SUPPORT
+ if (MB_CUR_MAX != 1)
+ result = wcs_re_match_2_internal (bufp, string1, size1, string2, size2,
+ pos, regs, stop,
+ NULL, 0, NULL, 0, NULL, NULL);
+ else
+# endif
+ result = byte_re_match_2_internal (bufp, string1, size1, string2, size2,
+ pos, regs, stop);
+
#ifndef REGEX_MALLOC
# ifdef C_ALLOCA
alloca (0);
weak_alias (__re_match_2, re_match_2)
#endif
+#endif /* not INSIDE_RECURSION */
+
+#ifdef INSIDE_RECURSION
+
+#ifdef WCHAR
+
+/* This check the substring (from 0, to length) of the multibyte string,
+ to which offset_buffer correspond. And count how many wchar_t_characters
+ the substring occupy. We use offset_buffer to optimization.
+ See convert_mbs_to_wcs. */
+
+static int
+count_mbs_length (int *offset_buffer, int length)
+{
+ int upper, lower;
+
+ /* Check whether the size is valid. */
+ if (length < 0)
+ return -1;
+
+ if (offset_buffer == NULL)
+ return 0;
+
+ /* If there are no multibyte character, offset_buffer[i] == i.
+ Optmize for this case. */
+ if (offset_buffer[length] == length)
+ return length;
+
+ /* Set up upper with length. (because for all i, offset_buffer[i] >= i) */
+ upper = length;
+ lower = 0;
+
+ while (true)
+ {
+ int middle = (lower + upper) / 2;
+ if (middle == lower || middle == upper)
+ break;
+ if (offset_buffer[middle] > length)
+ upper = middle;
+ else if (offset_buffer[middle] < length)
+ lower = middle;
+ else
+ return middle;
+ }
+
+ return -1;
+}
+#endif /* WCHAR */
+
/* This is a separate function so that we can force an alloca cleanup
afterwards. */
+#ifdef WCHAR
+static int
+wcs_re_match_2_internal (struct re_pattern_buffer *bufp,
+ const char *cstring1, int csize1,
+ const char *cstring2, int csize2,
+ int pos,
+ struct re_registers *regs,
+ int stop,
+ /* string1 == string2 == NULL means
+ string1/2, size1/2 and mbs_offset1/2 need
+ setting up in this function. */
+ /* We need wchar_t * buffers corresponding to
+ cstring1, cstring2. */
+ wchar_t *string1, int size1,
+ wchar_t *string2, int size2,
+ /* Offset buffer for optimization. See
+ convert_mbs_to_wc. */
+ int *mbs_offset1,
+ int *mbs_offset2)
+#else /* BYTE */
static int
-re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
- struct re_pattern_buffer *bufp;
- const char *string1, *string2;
- int size1, size2;
- int pos;
- struct re_registers *regs;
- int stop;
+byte_re_match_2_internal (struct re_pattern_buffer *bufp,
+ const char *string1, int size1,
+ const char *string2, int size2,
+ int pos,
+ struct re_registers *regs,
+ int stop)
+#endif /* BYTE */
{
/* General temporaries. */
int mcnt;
- unsigned char *p1;
+ UCHAR_T *p1;
+#ifdef WCHAR
+ /* They hold whether each wchar_t is binary data or not. */
+ char *is_binary = NULL;
+ /* If true, we can't free string1/2, mbs_offset1/2. */
+ int cant_free_wcs_buf = 1;
+#endif /* WCHAR */
/* Just past the end of the corresponding string. */
- const char *end1, *end2;
+ const CHAR_T *end1, *end2;
/* Pointers into string1 and string2, just past the last characters in
each to consider matching. */
- const char *end_match_1, *end_match_2;
+ const CHAR_T *end_match_1, *end_match_2;
/* Where we are in the data, and the end of the current string. */
- const char *d, *dend;
+ const CHAR_T *d, *dend;
/* Where we are in the pattern, and the end of the pattern. */
- unsigned char *p = bufp->buffer;
- register unsigned char *pend = p + bufp->used;
+#ifdef WCHAR
+ UCHAR_T *pattern, *p;
+ register UCHAR_T *pend;
+#else /* BYTE */
+ UCHAR_T *p = bufp->buffer;
+ register UCHAR_T *pend = p + bufp->used;
+#endif /* WCHAR */
/* Mark the opcode just after a start_memory, so we can test for an
empty subpattern when we get to the stop_memory. */
- unsigned char *just_past_start_mem = 0;
+ UCHAR_T *just_past_start_mem = 0;
/* We use this to map every character in the string. */
RE_TRANSLATE_TYPE translate = bufp->translate;
a ``dummy''; if a failure happens and the failure point is a dummy,
it gets discarded and the next next one is tried. */
#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
- fail_stack_type fail_stack;
+ PREFIX(fail_stack_type) fail_stack;
#endif
#ifdef DEBUG
- static unsigned failure_id = 0;
+ static unsigned failure_id;
unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
#endif
stopped matching the regnum-th subexpression. (The zeroth register
keeps track of what the whole pattern matches.) */
#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
- const char **regstart, **regend;
+ const CHAR_T **regstart, **regend;
#endif
/* If a group that's operated upon by a repetition operator fails to
are when we last see its open-group operator. Similarly for a
register's end. */
#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
- const char **old_regstart, **old_regend;
+ const CHAR_T **old_regstart, **old_regend;
#endif
/* The is_active field of reg_info helps us keep track of which (possibly
subexpression. These two fields get reset each time through any
loop their register is in. */
#ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
- register_info_type *reg_info;
+ PREFIX(register_info_type) *reg_info;
#endif
/* The following record the register info as found in the above
turn happens only if we have not yet matched the entire string. */
unsigned best_regs_set = false;
#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
- const char **best_regstart, **best_regend;
+ const CHAR_T **best_regstart, **best_regend;
#endif
/* Logically, this is `best_regend[0]'. But we don't want to have to
the end of the best match so far in a separate variable. We
initialize this to NULL so that when we backtrack the first time
and need to test it, it's not garbage. */
- const char *match_end = NULL;
+ const CHAR_T *match_end = NULL;
/* This helps SET_REGS_MATCHED avoid doing redundant work. */
int set_regs_matched_done = 0;
/* Used when we pop values we don't care about. */
#ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
- const char **reg_dummy;
- register_info_type *reg_info_dummy;
+ const CHAR_T **reg_dummy;
+ PREFIX(register_info_type) *reg_info_dummy;
#endif
#ifdef DEBUG
unsigned num_regs_pushed = 0;
#endif
+ /* Definitions for state transitions. More efficiently for gcc. */
+#ifdef __GNUC__
+# if defined HAVE_SUBTRACT_LOCAL_LABELS && defined SHARED
+# define NEXT \
+ do \
+ { \
+ int offset; \
+ const void *__unbounded ptr; \
+ offset = (p == pend \
+ ? 0 : jmptable[SWITCH_ENUM_CAST ((re_opcode_t) *p++)]); \
+ ptr = &&end_of_pattern + offset; \
+ goto *ptr; \
+ } \
+ while (0)
+# define REF(x) \
+ &&label_##x - &&end_of_pattern
+# define JUMP_TABLE_TYPE const int
+# else
+# define NEXT \
+ do \
+ { \
+ const void *__unbounded ptr; \
+ ptr = (p == pend ? &&end_of_pattern \
+ : jmptable[SWITCH_ENUM_CAST ((re_opcode_t) *p++)]); \
+ goto *ptr; \
+ } \
+ while (0)
+# define REF(x) \
+ &&label_##x
+# define JUMP_TABLE_TYPE const void *const
+# endif
+# define CASE(x) label_##x
+ static JUMP_TABLE_TYPE jmptable[] =
+ {
+ REF (no_op),
+ REF (succeed),
+ REF (exactn),
+# ifdef MBS_SUPPORT
+ REF (exactn_bin),
+# endif
+ REF (anychar),
+ REF (charset),
+ REF (charset_not),
+ REF (start_memory),
+ REF (stop_memory),
+ REF (duplicate),
+ REF (begline),
+ REF (endline),
+ REF (begbuf),
+ REF (endbuf),
+ REF (jump),
+ REF (jump_past_alt),
+ REF (on_failure_jump),
+ REF (on_failure_keep_string_jump),
+ REF (pop_failure_jump),
+ REF (maybe_pop_jump),
+ REF (dummy_failure_jump),
+ REF (push_dummy_failure),
+ REF (succeed_n),
+ REF (jump_n),
+ REF (set_number_at),
+ REF (wordchar),
+ REF (notwordchar),
+ REF (wordbeg),
+ REF (wordend),
+ REF (wordbound),
+ REF (notwordbound)
+# ifdef emacs
+ ,REF (before_dot),
+ REF (at_dot),
+ REF (after_dot),
+ REF (syntaxspec),
+ REF (notsyntaxspec)
+# endif
+ };
+#else
+# define NEXT \
+ break
+# define CASE(x) \
+ case x
+#endif
+
DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
INIT_FAIL_STACK ();
array indexing. We should fix this. */
if (bufp->re_nsub)
{
- regstart = REGEX_TALLOC (num_regs, const char *);
- regend = REGEX_TALLOC (num_regs, const char *);
- old_regstart = REGEX_TALLOC (num_regs, const char *);
- old_regend = REGEX_TALLOC (num_regs, const char *);
- best_regstart = REGEX_TALLOC (num_regs, const char *);
- best_regend = REGEX_TALLOC (num_regs, const char *);
- reg_info = REGEX_TALLOC (num_regs, register_info_type);
- reg_dummy = REGEX_TALLOC (num_regs, const char *);
- reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
+ regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
+ regend = REGEX_TALLOC (num_regs, const CHAR_T *);
+ old_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
+ old_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
+ best_regstart = REGEX_TALLOC (num_regs, const CHAR_T *);
+ best_regend = REGEX_TALLOC (num_regs, const CHAR_T *);
+ reg_info = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
+ reg_dummy = REGEX_TALLOC (num_regs, const CHAR_T *);
+ reg_info_dummy = REGEX_TALLOC (num_regs, PREFIX(register_info_type));
if (!(regstart && regend && old_regstart && old_regend && reg_info
&& best_regstart && best_regend && reg_dummy && reg_info_dummy))
`FREE_VARIABLES' doesn't try to free them. */
regstart = regend = old_regstart = old_regend = best_regstart
= best_regend = reg_dummy = NULL;
- reg_info = reg_info_dummy = (register_info_type *) NULL;
+ reg_info = reg_info_dummy = (PREFIX(register_info_type) *) NULL;
}
#endif /* MATCH_MAY_ALLOCATE */
/* The starting position is bogus. */
+#ifdef WCHAR
+ if (pos < 0 || pos > csize1 + csize2)
+#else /* BYTE */
if (pos < 0 || pos > size1 + size2)
+#endif
{
FREE_VARIABLES ();
return -1;
}
+#ifdef WCHAR
+ /* Allocate wchar_t array for string1 and string2 and
+ fill them with converted string. */
+ if (string1 == NULL && string2 == NULL)
+ {
+ /* We need seting up buffers here. */
+
+ /* We must free wcs buffers in this function. */
+ cant_free_wcs_buf = 0;
+
+ if (csize1 != 0)
+ {
+ string1 = REGEX_TALLOC (csize1 + 1, CHAR_T);
+ mbs_offset1 = REGEX_TALLOC (csize1 + 1, int);
+ is_binary = REGEX_TALLOC (csize1 + 1, char);
+ if (!string1 || !mbs_offset1 || !is_binary)
+ {
+ FREE_VAR (string1);
+ FREE_VAR (mbs_offset1);
+ FREE_VAR (is_binary);
+ return -2;
+ }
+ }
+ if (csize2 != 0)
+ {
+ string2 = REGEX_TALLOC (csize2 + 1, CHAR_T);
+ mbs_offset2 = REGEX_TALLOC (csize2 + 1, int);
+ is_binary = REGEX_TALLOC (csize2 + 1, char);
+ if (!string2 || !mbs_offset2 || !is_binary)
+ {
+ FREE_VAR (string1);
+ FREE_VAR (mbs_offset1);
+ FREE_VAR (string2);
+ FREE_VAR (mbs_offset2);
+ FREE_VAR (is_binary);
+ return -2;
+ }
+ size2 = convert_mbs_to_wcs(string2, cstring2, csize2,
+ mbs_offset2, is_binary);
+ string2[size2] = L'\0'; /* for a sentinel */
+ FREE_VAR (is_binary);
+ }
+ }
+
+ /* We need to cast pattern to (wchar_t*), because we casted this compiled
+ pattern to (char*) in regex_compile. */
+ p = pattern = (CHAR_T*)bufp->buffer;
+ pend = (CHAR_T*)(bufp->buffer + bufp->used);
+
+#endif /* WCHAR */
+
/* Initialize subexpression text positions to -1 to mark ones that no
start_memory/stop_memory has been seen for. Also initialize the
register information struct. */
size2 = size1;
string1 = 0;
size1 = 0;
+#ifdef WCHAR
+ mbs_offset2 = mbs_offset1;
+ csize2 = csize1;
+ mbs_offset1 = NULL;
+ csize1 = 0;
+#endif
}
end1 = string1 + size1;
end2 = string2 + size2;
/* Compute where to stop matching, within the two strings. */
+#ifdef WCHAR
+ if (stop <= csize1)
+ {
+ mcnt = count_mbs_length(mbs_offset1, stop);
+ end_match_1 = string1 + mcnt;
+ end_match_2 = string2;
+ }
+ else
+ {
+ if (stop > csize1 + csize2)
+ stop = csize1 + csize2;
+ end_match_1 = end1;
+ mcnt = count_mbs_length(mbs_offset2, stop-csize1);
+ end_match_2 = string2 + mcnt;
+ }
+ if (mcnt < 0)
+ { /* count_mbs_length return error. */
+ FREE_VARIABLES ();
+ return -1;
+ }
+#else
if (stop <= size1)
{
end_match_1 = string1 + stop;
end_match_1 = end1;
end_match_2 = string2 + stop - size1;
}
+#endif /* WCHAR */
/* `p' scans through the pattern as `d' scans through the data.
`dend' is the end of the input string that `d' points within. `d'
this happens before fetching; therefore, at the beginning of the
loop, `d' can be pointing at the end of a string, but it cannot
equal `string2'. */
+#ifdef WCHAR
+ if (size1 > 0 && pos <= csize1)
+ {
+ mcnt = count_mbs_length(mbs_offset1, pos);
+ d = string1 + mcnt;
+ dend = end_match_1;
+ }
+ else
+ {
+ mcnt = count_mbs_length(mbs_offset2, pos-csize1);
+ d = string2 + mcnt;
+ dend = end_match_2;
+ }
+
+ if (mcnt < 0)
+ { /* count_mbs_length return error. */
+ FREE_VARIABLES ();
+ return -1;
+ }
+#else
if (size1 > 0 && pos <= size1)
{
d = string1 + pos;
d = string2 + pos - size1;
dend = end_match_2;
}
+#endif /* WCHAR */
DEBUG_PRINT1 ("The compiled pattern is:\n");
DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
DEBUG_PRINT2 ("\n0x%x: ", p);
#endif
+#ifdef __GNUC__
+ NEXT;
+#else
if (p == pend)
- { /* End of pattern means we might have succeeded. */
- DEBUG_PRINT1 ("end of pattern ... ");
+#endif
+ {
+#ifdef __GNUC__
+ end_of_pattern:
+#endif
+ /* End of pattern means we might have succeeded. */
+ DEBUG_PRINT1 ("end of pattern ... ");
/* If we haven't matched the entire string, and we want the
- longest match, try backtracking. */
- if (d != end_match_2)
+ longest match, try backtracking. */
+ if (d != end_match_2)
{
- /* 1 if this match ends in the same string (string1 or string2)
- as the best previous match. */
- boolean same_str_p = (FIRST_STRING_P (match_end)
- == MATCHING_IN_FIRST_STRING);
/* 1 if this match is the best seen so far. */
boolean best_match_p;
- /* AIX compiler got confused when this was combined
- with the previous declaration. */
- if (same_str_p)
- best_match_p = d > match_end;
- else
- best_match_p = !MATCHING_IN_FIRST_STRING;
+ {
+ /* 1 if this match ends in the same string (string1 or string2)
+ as the best previous match. */
+ boolean same_str_p = (FIRST_STRING_P (match_end)
+ == MATCHING_IN_FIRST_STRING);
+
+ /* AIX compiler got confused when this was combined
+ with the previous declaration. */
+ if (same_str_p)
+ best_match_p = d > match_end;
+ else
+ best_match_p = !MATCHING_IN_FIRST_STRING;
+ }
- DEBUG_PRINT1 ("backtracking.\n");
+ DEBUG_PRINT1 ("backtracking.\n");
- if (!FAIL_STACK_EMPTY ())
- { /* More failure points to try. */
+ if (!FAIL_STACK_EMPTY ())
+ { /* More failure points to try. */
- /* If exceeds best match so far, save it. */
- if (!best_regs_set || best_match_p)
- {
- best_regs_set = true;
- match_end = d;
+ /* If exceeds best match so far, save it. */
+ if (!best_regs_set || best_match_p)
+ {
+ best_regs_set = true;
+ match_end = d;
- DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
+ DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
- for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
- {
- best_regstart[mcnt] = regstart[mcnt];
- best_regend[mcnt] = regend[mcnt];
- }
- }
- goto fail;
- }
+ for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
+ {
+ best_regstart[mcnt] = regstart[mcnt];
+ best_regend[mcnt] = regend[mcnt];
+ }
+ }
+ goto fail;
+ }
- /* If no failure points, don't restore garbage. And if
- last match is real best match, don't restore second
- best one. */
- else if (best_regs_set && !best_match_p)
- {
- restore_best_regs:
- /* Restore best match. It may happen that `dend ==
- end_match_1' while the restored d is in string2.
- For example, the pattern `x.*y.*z' against the
- strings `x-' and `y-z-', if the two strings are
- not consecutive in memory. */
- DEBUG_PRINT1 ("Restoring best registers.\n");
-
- d = match_end;
- dend = ((d >= string1 && d <= end1)
- ? end_match_1 : end_match_2);
+ /* If no failure points, don't restore garbage. And if
+ last match is real best match, don't restore second
+ best one. */
+ else if (best_regs_set && !best_match_p)
+ {
+ restore_best_regs:
+ /* Restore best match. It may happen that `dend ==
+ end_match_1' while the restored d is in string2.
+ For example, the pattern `x.*y.*z' against the
+ strings `x-' and `y-z-', if the two strings are
+ not consecutive in memory. */
+ DEBUG_PRINT1 ("Restoring best registers.\n");
+
+ d = match_end;
+ dend = ((d >= string1 && d <= end1)
+ ? end_match_1 : end_match_2);
for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
{
regstart[mcnt] = best_regstart[mcnt];
regend[mcnt] = best_regend[mcnt];
}
- }
- } /* d != end_match_2 */
+ }
+ } /* d != end_match_2 */
succeed_label:
- DEBUG_PRINT1 ("Accepting match.\n");
-
- /* If caller wants register contents data back, do it. */
- if (regs && !bufp->no_sub)
+ DEBUG_PRINT1 ("Accepting match.\n");
+ /* If caller wants register contents data back, do it. */
+ if (regs && !bufp->no_sub)
{
- /* Have the register data arrays been allocated? */
- if (bufp->regs_allocated == REGS_UNALLOCATED)
- { /* No. So allocate them with malloc. We need one
- extra element beyond `num_regs' for the `-1' marker
- GNU code uses. */
- regs->num_regs = MAX (RE_NREGS, num_regs + 1);
- regs->start = TALLOC (regs->num_regs, regoff_t);
- regs->end = TALLOC (regs->num_regs, regoff_t);
- if (regs->start == NULL || regs->end == NULL)
+ /* Have the register data arrays been allocated? */
+ if (bufp->regs_allocated == REGS_UNALLOCATED)
+ { /* No. So allocate them with malloc. We need one
+ extra element beyond `num_regs' for the `-1' marker
+ GNU code uses. */
+ regs->num_regs = MAX (RE_NREGS, num_regs + 1);
+ regs->start = TALLOC (regs->num_regs, regoff_t);
+ regs->end = TALLOC (regs->num_regs, regoff_t);
+ if (regs->start == NULL || regs->end == NULL)
{
FREE_VARIABLES ();
return -2;
}
- bufp->regs_allocated = REGS_REALLOCATE;
- }
- else if (bufp->regs_allocated == REGS_REALLOCATE)
- { /* Yes. If we need more elements than were already
- allocated, reallocate them. If we need fewer, just
- leave it alone. */
- if (regs->num_regs < num_regs + 1)
- {
- regs->num_regs = num_regs + 1;
- RETALLOC (regs->start, regs->num_regs, regoff_t);
- RETALLOC (regs->end, regs->num_regs, regoff_t);
- if (regs->start == NULL || regs->end == NULL)
+ bufp->regs_allocated = REGS_REALLOCATE;
+ }
+ else if (bufp->regs_allocated == REGS_REALLOCATE)
+ { /* Yes. If we need more elements than were already
+ allocated, reallocate them. If we need fewer, just
+ leave it alone. */
+ if (regs->num_regs < num_regs + 1)
+ {
+ regs->num_regs = num_regs + 1;
+ RETALLOC (regs->start, regs->num_regs, regoff_t);
+ RETALLOC (regs->end, regs->num_regs, regoff_t);
+ if (regs->start == NULL || regs->end == NULL)
{
FREE_VARIABLES ();
return -2;
}
- }
- }
- else
+ }
+ }
+ else
{
/* These braces fend off a "empty body in an else-statement"
warning under GCC when assert expands to nothing. */
assert (bufp->regs_allocated == REGS_FIXED);
}
- /* Convert the pointer data in `regstart' and `regend' to
- indices. Register zero has to be set differently,
- since we haven't kept track of any info for it. */
- if (regs->num_regs > 0)
- {
- regs->start[0] = pos;
- regs->end[0] = (MATCHING_IN_FIRST_STRING
+ /* Convert the pointer data in `regstart' and `regend' to
+ indices. Register zero has to be set differently,
+ since we haven't kept track of any info for it. */
+ if (regs->num_regs > 0)
+ {
+ regs->start[0] = pos;
+#ifdef WCHAR
+ if (MATCHING_IN_FIRST_STRING)
+ regs->end[0] = (mbs_offset1 != NULL ?
+ mbs_offset1[d-string1] : 0);
+ else
+ regs->end[0] = csize1 + (mbs_offset2 != NULL
+ ? mbs_offset2[d-string2] : 0);
+#else
+ regs->end[0] = (MATCHING_IN_FIRST_STRING
? ((regoff_t) (d - string1))
- : ((regoff_t) (d - string2 + size1)));
- }
+ : ((regoff_t) (d - string2 + size1)));
+#endif /* WCHAR */
+ }
- /* Go through the first `min (num_regs, regs->num_regs)'
- registers, since that is all we initialized. */
+ /* Go through the first `min (num_regs, regs->num_regs)'
+ registers, since that is all we initialized. */
for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
mcnt++)
{
- if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
- regs->start[mcnt] = regs->end[mcnt] = -1;
- else
- {
+ if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
+ regs->start[mcnt] = regs->end[mcnt] = -1;
+ else
+ {
regs->start[mcnt]
= (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
- regs->end[mcnt]
+ regs->end[mcnt]
= (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
- }
+ }
}
- /* If the regs structure we return has more elements than
- were in the pattern, set the extra elements to -1. If
- we (re)allocated the registers, this is the case,
- because we always allocate enough to have at least one
- -1 at the end. */
- for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
- regs->start[mcnt] = regs->end[mcnt] = -1;
+ /* If the regs structure we return has more elements than
+ were in the pattern, set the extra elements to -1. If
+ we (re)allocated the registers, this is the case,
+ because we always allocate enough to have at least one
+ -1 at the end. */
+ for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
+ regs->start[mcnt] = regs->end[mcnt] = -1;
} /* regs && !bufp->no_sub */
- DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
- nfailure_points_pushed, nfailure_points_popped,
- nfailure_points_pushed - nfailure_points_popped);
- DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
+ DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
+ nfailure_points_pushed, nfailure_points_popped,
+ nfailure_points_pushed - nfailure_points_popped);
+ DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
- mcnt = d - pos - (MATCHING_IN_FIRST_STRING
- ? string1
- : string2 - size1);
+#ifdef WCHAR
+ if (MATCHING_IN_FIRST_STRING)
+ mcnt = mbs_offset1 != NULL ? mbs_offset1[d-string1] : 0;
+ else
+ mcnt = (mbs_offset2 != NULL ? mbs_offset2[d-string2] : 0) +
+ csize1;
+ mcnt -= pos;
+#else
+ mcnt = d - pos - (MATCHING_IN_FIRST_STRING
+ ? string1 : string2 - size1);
+#endif /* WCHAR */
- DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
+ DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
- FREE_VARIABLES ();
- return mcnt;
- }
+ FREE_VARIABLES ();
+ return mcnt;
+ }
+#ifndef __GNUC__
/* Otherwise match next pattern command. */
switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
{
+#endif
/* Ignore these. Used to ignore the n of succeed_n's which
currently have n == 0. */
- case no_op:
+ CASE (no_op):
DEBUG_PRINT1 ("EXECUTING no_op.\n");
- break;
+ NEXT;
- case succeed:
+ CASE (succeed):
DEBUG_PRINT1 ("EXECUTING succeed.\n");
goto succeed_label;
/* Match the next n pattern characters exactly. The following
byte in the pattern defines n, and the n bytes after that
are the characters to match. */
- case exactn:
+ CASE (exactn):
+#ifdef MBS_SUPPORT
+ CASE (exactn_bin):
+#endif
mcnt = *p++;
DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
do
{
PREFETCH ();
- if ((unsigned char) translate[(unsigned char) *d++]
- != (unsigned char) *p++)
+#ifdef WCHAR
+ if (*d <= 0xff)
+ {
+ if ((UCHAR_T) translate[(unsigned char) *d++]
+ != (UCHAR_T) *p++)
+ goto fail;
+ }
+ else
+ {
+ if (*d++ != (CHAR_T) *p++)
+ goto fail;
+ }
+#else
+ if ((UCHAR_T) translate[(unsigned char) *d++]
+ != (UCHAR_T) *p++)
goto fail;
+#endif /* WCHAR */
}
while (--mcnt);
}
do
{
PREFETCH ();
- if (*d++ != (char) *p++) goto fail;
+ if (*d++ != (CHAR_T) *p++) goto fail;
}
while (--mcnt);
}
SET_REGS_MATCHED ();
- break;
+ NEXT;
/* Match any character except possibly a newline or a null. */
- case anychar:
+ CASE (anychar):
DEBUG_PRINT1 ("EXECUTING anychar.\n");
PREFETCH ();
goto fail;
SET_REGS_MATCHED ();
- DEBUG_PRINT2 (" Matched `%d'.\n", *d);
+ DEBUG_PRINT2 (" Matched `%ld'.\n", (long int) *d);
d++;
- break;
+ NEXT;
- case charset:
- case charset_not:
+ CASE (charset):
+ CASE (charset_not):
{
- register unsigned char c;
- boolean not = (re_opcode_t) *(p - 1) == charset_not;
-
- DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
+ register UCHAR_T c;
+#ifdef WCHAR
+ unsigned int i, char_class_length, coll_symbol_length,
+ equiv_class_length, ranges_length, chars_length, length;
+ CHAR_T *workp, *workp2, *charset_top;
+#define WORK_BUFFER_SIZE 128
+ CHAR_T str_buf[WORK_BUFFER_SIZE];
+# ifdef _LIBC
+ uint32_t nrules;
+# endif /* _LIBC */
+#endif /* WCHAR */
+ boolean negate = (re_opcode_t) *(p - 1) == charset_not;
+ DEBUG_PRINT2 ("EXECUTING charset%s.\n", negate ? "_not" : "");
PREFETCH ();
c = TRANSLATE (*d); /* The character to match. */
+#ifdef WCHAR
+# ifdef _LIBC
+ nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
+# endif /* _LIBC */
+ charset_top = p - 1;
+ char_class_length = *p++;
+ coll_symbol_length = *p++;
+ equiv_class_length = *p++;
+ ranges_length = *p++;
+ chars_length = *p++;
+ /* p points charset[6], so the address of the next instruction
+ (charset[l+m+n+2o+k+p']) equals p[l+m+n+2*o+p'],
+ where l=length of char_classes, m=length of collating_symbol,
+ n=equivalence_class, o=length of char_range,
+ p'=length of character. */
+ workp = p;
+ /* Update p to indicate the next instruction. */
+ p += char_class_length + coll_symbol_length+ equiv_class_length +
+ 2*ranges_length + chars_length;
+
+ /* match with char_class? */
+ for (i = 0; i < char_class_length ; i += CHAR_CLASS_SIZE)
+ {
+ wctype_t wctype;
+ uintptr_t alignedp = ((uintptr_t)workp
+ + __alignof__(wctype_t) - 1)
+ & ~(uintptr_t)(__alignof__(wctype_t) - 1);
+ wctype = *((wctype_t*)alignedp);
+ workp += CHAR_CLASS_SIZE;
+ if (iswctype((wint_t)c, wctype))
+ goto char_set_matched;
+ }
+
+ /* match with collating_symbol? */
+# ifdef _LIBC
+ if (nrules != 0)
+ {
+ const unsigned char *extra = (const unsigned char *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB);
+
+ for (workp2 = workp + coll_symbol_length ; workp < workp2 ;
+ workp++)
+ {
+ int32_t *wextra;
+ wextra = (int32_t*)(extra + *workp++);
+ for (i = 0; i < *wextra; ++i)
+ if (TRANSLATE(d[i]) != wextra[1 + i])
+ break;
+
+ if (i == *wextra)
+ {
+ /* Update d, however d will be incremented at
+ char_set_matched:, we decrement d here. */
+ d += i - 1;
+ goto char_set_matched;
+ }
+ }
+ }
+ else /* (nrules == 0) */
+# endif
+ /* If we can't look up collation data, we use wcscoll
+ instead. */
+ {
+ for (workp2 = workp + coll_symbol_length ; workp < workp2 ;)
+ {
+ const CHAR_T *backup_d = d, *backup_dend = dend;
+ length = wcslen (workp);
+
+ /* If wcscoll(the collating symbol, whole string) > 0,
+ any substring of the string never match with the
+ collating symbol. */
+ if (wcscoll (workp, d) > 0)
+ {
+ workp += length + 1;
+ continue;
+ }
+
+ /* First, we compare the collating symbol with
+ the first character of the string.
+ If it don't match, we add the next character to
+ the compare buffer in turn. */
+ for (i = 0 ; i < WORK_BUFFER_SIZE-1 ; i++, d++)
+ {
+ int match;
+ if (d == dend)
+ {
+ if (dend == end_match_2)
+ break;
+ d = string2;
+ dend = end_match_2;
+ }
+
+ /* add next character to the compare buffer. */
+ str_buf[i] = TRANSLATE(*d);
+ str_buf[i+1] = '\0';
+
+ match = wcscoll (workp, str_buf);
+ if (match == 0)
+ goto char_set_matched;
+
+ if (match < 0)
+ /* (str_buf > workp) indicate (str_buf + X > workp),
+ because for all X (str_buf + X > str_buf).
+ So we don't need continue this loop. */
+ break;
+
+ /* Otherwise(str_buf < workp),
+ (str_buf+next_character) may equals (workp).
+ So we continue this loop. */
+ }
+ /* not matched */
+ d = backup_d;
+ dend = backup_dend;
+ workp += length + 1;
+ }
+ }
+ /* match with equivalence_class? */
+# ifdef _LIBC
+ if (nrules != 0)
+ {
+ const CHAR_T *backup_d = d, *backup_dend = dend;
+ /* Try to match the equivalence class against
+ those known to the collate implementation. */
+ const int32_t *table;
+ const int32_t *weights;
+ const int32_t *extra;
+ const int32_t *indirect;
+ int32_t idx, idx2;
+ wint_t *cp;
+ size_t len;
+
+ /* This #include defines a local function! */
+# include <locale/weightwc.h>
+
+ table = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC);
+ weights = (const wint_t *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC);
+ extra = (const wint_t *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC);
+ indirect = (const int32_t *)
+ _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC);
+
+ /* Write 1 collating element to str_buf, and
+ get its index. */
+ idx2 = 0;
+
+ for (i = 0 ; idx2 == 0 && i < WORK_BUFFER_SIZE - 1; i++)
+ {
+ cp = (wint_t*)str_buf;
+ if (d == dend)
+ {
+ if (dend == end_match_2)
+ break;
+ d = string2;
+ dend = end_match_2;
+ }
+ str_buf[i] = TRANSLATE(*(d+i));
+ str_buf[i+1] = '\0'; /* sentinel */
+ idx2 = findidx ((const wint_t**)&cp);
+ }
+
+ /* Update d, however d will be incremented at
+ char_set_matched:, we decrement d here. */
+ d = backup_d + ((wchar_t*)cp - (wchar_t*)str_buf - 1);
+ if (d >= dend)
+ {
+ if (dend == end_match_2)
+ d = dend;
+ else
+ {
+ d = string2;
+ dend = end_match_2;
+ }
+ }
+
+ len = weights[idx2];
+
+ for (workp2 = workp + equiv_class_length ; workp < workp2 ;
+ workp++)
+ {
+ idx = (int32_t)*workp;
+ /* We already checked idx != 0 in regex_compile. */
+
+ if (idx2 != 0 && len == weights[idx])
+ {
+ int cnt = 0;
+ while (cnt < len && (weights[idx + 1 + cnt]
+ == weights[idx2 + 1 + cnt]))
+ ++cnt;
+
+ if (cnt == len)
+ goto char_set_matched;
+ }
+ }
+ /* not matched */
+ d = backup_d;
+ dend = backup_dend;
+ }
+ else /* (nrules == 0) */
+# endif
+ /* If we can't look up collation data, we use wcscoll
+ instead. */
+ {
+ for (workp2 = workp + equiv_class_length ; workp < workp2 ;)
+ {
+ const CHAR_T *backup_d = d, *backup_dend = dend;
+ length = wcslen (workp);
+
+ /* If wcscoll(the collating symbol, whole string) > 0,
+ any substring of the string never match with the
+ collating symbol. */
+ if (wcscoll (workp, d) > 0)
+ {
+ workp += length + 1;
+ break;
+ }
+
+ /* First, we compare the equivalence class with
+ the first character of the string.
+ If it don't match, we add the next character to
+ the compare buffer in turn. */
+ for (i = 0 ; i < WORK_BUFFER_SIZE - 1 ; i++, d++)
+ {
+ int match;
+ if (d == dend)
+ {
+ if (dend == end_match_2)
+ break;
+ d = string2;
+ dend = end_match_2;
+ }
+
+ /* add next character to the compare buffer. */
+ str_buf[i] = TRANSLATE(*d);
+ str_buf[i+1] = '\0';
+
+ match = wcscoll (workp, str_buf);
+
+ if (match == 0)
+ goto char_set_matched;
+
+ if (match < 0)
+ /* (str_buf > workp) indicate (str_buf + X > workp),
+ because for all X (str_buf + X > str_buf).
+ So we don't need continue this loop. */
+ break;
+
+ /* Otherwise(str_buf < workp),
+ (str_buf+next_character) may equals (workp).
+ So we continue this loop. */
+ }
+ /* not matched */
+ d = backup_d;
+ dend = backup_dend;
+ workp += length + 1;
+ }
+ }
+
+ /* match with char_range? */
+# ifdef _LIBC
+ if (nrules != 0)
+ {
+ uint32_t collseqval;
+ const char *collseq = (const char *)
+ _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC);
+
+ collseqval = collseq_table_lookup (collseq, c);
+
+ for (; workp < p - chars_length ;)
+ {
+ uint32_t start_val, end_val;
+ /* We already compute the collation sequence value
+ of the characters (or collating symbols). */
+ start_val = (uint32_t) *workp++; /* range_start */
+ end_val = (uint32_t) *workp++; /* range_end */
+
+ if (start_val <= collseqval && collseqval <= end_val)
+ goto char_set_matched;
+ }
+ }
+ else
+# endif
+ {
+ /* We set range_start_char at str_buf[0], range_end_char
+ at str_buf[4], and compared char at str_buf[2]. */
+ str_buf[1] = 0;
+ str_buf[2] = c;
+ str_buf[3] = 0;
+ str_buf[5] = 0;
+ for (; workp < p - chars_length ;)
+ {
+ wchar_t *range_start_char, *range_end_char;
+
+ /* match if (range_start_char <= c <= range_end_char). */
+
+ /* If range_start(or end) < 0, we assume -range_start(end)
+ is the offset of the collating symbol which is specified
+ as the character of the range start(end). */
+
+ /* range_start */
+ if (*workp < 0)
+ range_start_char = charset_top - (*workp++);
+ else
+ {
+ str_buf[0] = *workp++;
+ range_start_char = str_buf;
+ }
+
+ /* range_end */
+ if (*workp < 0)
+ range_end_char = charset_top - (*workp++);
+ else
+ {
+ str_buf[4] = *workp++;
+ range_end_char = str_buf + 4;
+ }
+
+ if (wcscoll (range_start_char, str_buf+2) <= 0
+ && wcscoll (str_buf+2, range_end_char) <= 0)
+ goto char_set_matched;
+ }
+ }
+
+ /* match with char? */
+ for (; workp < p ; workp++)
+ if (c == *workp)
+ goto char_set_matched;
+
+ negate = !negate;
+
+ char_set_matched:
+ if (negate) goto fail;
+#else
/* Cast to `unsigned' instead of `unsigned char' in case the
bit list is a full 32 bytes long. */
if (c < (unsigned) (*p * BYTEWIDTH)
&& p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
- not = !not;
+ negate = !negate;
p += 1 + *p;
- if (!not) goto fail;
-
+ if (!negate) goto fail;
+#undef WORK_BUFFER_SIZE
+#endif /* WCHAR */
SET_REGS_MATCHED ();
d++;
- break;
+ NEXT;
}
number of groups inner to this one in the next. The text
matched within the group is recorded (in the internal
registers data structure) under the register number. */
- case start_memory:
- DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
+ CASE (start_memory):
+ DEBUG_PRINT3 ("EXECUTING start_memory %ld (%ld):\n",
+ (long int) *p, (long int) p[1]);
/* Find out if this group can match the empty string. */
p1 = p; /* To send to group_match_null_string_p. */
if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
REG_MATCH_NULL_STRING_P (reg_info[*p])
- = group_match_null_string_p (&p1, pend, reg_info);
+ = PREFIX(group_match_null_string_p) (&p1, pend, reg_info);
/* Save the position in the string where we were the last time
we were at this open-group operator in case the group is
p += 2;
just_past_start_mem = p;
- break;
+ NEXT;
/* The stop_memory opcode represents the end of a group. Its
arguments are the same as start_memory's: the register
number, and the number of inner groups. */
- case stop_memory:
- DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
+ CASE (stop_memory):
+ DEBUG_PRINT3 ("EXECUTING stop_memory %ld (%ld):\n",
+ (long int) *p, (long int) p[1]);
/* We need to save the string position the last time we were at
this close-group operator in case the group is operated
it isn't necessarily one less than now: consider
(a(b)c(d(e)f)g). When group 3 ends, after the f), the
new highest active register is 1. */
- unsigned char r = *p - 1;
+ UCHAR_T r = *p - 1;
while (r > 0 && !IS_ACTIVE (reg_info[r]))
r--;
case dummy_failure_jump:
EXTRACT_NUMBER_AND_INCR (mcnt, p1);
if (is_a_jump_n)
- p1 += 2;
+ p1 += OFFSET_ADDRESS_SIZE;
break;
default:
by forcing a failure after pushing on the stack the
on_failure_jump's jump in the pattern, and d. */
if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
- && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
+ && (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == start_memory
+ && p1[2+OFFSET_ADDRESS_SIZE] == *p)
{
/* If this group ever matched anything, then restore
what its registers were before trying this last
/* Move past the register number and the inner group count. */
p += 2;
- break;
+ NEXT;
/* \<digit> has been turned into a `duplicate' command which is
followed by the numeric value of <digit> as the register number. */
- case duplicate:
+ CASE (duplicate):
{
- register const char *d2, *dend2;
+ register const CHAR_T *d2, *dend2;
int regno = *p++; /* Get which register to match against. */
DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
/* Compare that many; failure if mismatch, else move
past them. */
if (translate
- ? bcmp_translate (d, d2, mcnt, translate)
- : memcmp (d, d2, mcnt))
+ ? PREFIX(bcmp_translate) (d, d2, mcnt, translate)
+ : memcmp (d, d2, mcnt*sizeof(UCHAR_T)))
goto fail;
d += mcnt, d2 += mcnt;
SET_REGS_MATCHED ();
}
}
- break;
+ NEXT;
/* begline matches the empty string at the beginning of the string
(unless `not_bol' is set in `bufp'), and, if
`newline_anchor' is set, after newlines. */
- case begline:
+ CASE (begline):
DEBUG_PRINT1 ("EXECUTING begline.\n");
if (AT_STRINGS_BEG (d))
{
- if (!bufp->not_bol) break;
+ if (!bufp->not_bol)
+ {
+ NEXT;
+ }
}
else if (d[-1] == '\n' && bufp->newline_anchor)
{
- break;
+ NEXT;
}
/* In all other cases, we fail. */
goto fail;
/* endline is the dual of begline. */
- case endline:
+ CASE (endline):
DEBUG_PRINT1 ("EXECUTING endline.\n");
if (AT_STRINGS_END (d))
{
- if (!bufp->not_eol) break;
+ if (!bufp->not_eol)
+ {
+ NEXT;
+ }
}
/* We have to ``prefetch'' the next character. */
else if ((d == end1 ? *string2 : *d) == '\n'
&& bufp->newline_anchor)
{
- break;
+ NEXT;
}
goto fail;
/* Match at the very beginning of the data. */
- case begbuf:
+ CASE (begbuf):
DEBUG_PRINT1 ("EXECUTING begbuf.\n");
if (AT_STRINGS_BEG (d))
- break;
+ {
+ NEXT;
+ }
goto fail;
/* Match at the very end of the data. */
- case endbuf:
+ CASE (endbuf):
DEBUG_PRINT1 ("EXECUTING endbuf.\n");
if (AT_STRINGS_END (d))
- break;
+ {
+ NEXT;
+ }
goto fail;
stack at all is that otherwise we would have to change
`anychar's code to do something besides goto fail in this
case; that seems worse than this. */
- case on_failure_keep_string_jump:
+ CASE (on_failure_keep_string_jump):
DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
EXTRACT_NUMBER_AND_INCR (mcnt, p);
#endif
PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
- break;
+ NEXT;
/* Uses of on_failure_jump:
Repeats start with an on_failure_jump that points past both
the repetition text and either the following jump or
pop_failure_jump back to this on_failure_jump. */
- case on_failure_jump:
+ CASE (on_failure_jump):
on_failure:
DEBUG_PRINT1 ("EXECUTING on_failure_jump");
DEBUG_PRINT1 (":\n");
PUSH_FAILURE_POINT (p + mcnt, d, -2);
- break;
+ NEXT;
/* A smart repeat ends with `maybe_pop_jump'.
We change it to either `pop_failure_jump' or `jump'. */
- case maybe_pop_jump:
+ CASE (maybe_pop_jump):
EXTRACT_NUMBER_AND_INCR (mcnt, p);
DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
{
- register unsigned char *p2 = p;
+ register UCHAR_T *p2 = p;
/* Compare the beginning of the repeat with what in the
pattern follows its end. If we can establish that there
&& ((re_opcode_t) *p2 == stop_memory
|| (re_opcode_t) *p2 == start_memory))
p2 += 3;
- else if (p2 + 6 < pend
+ else if (p2 + 2 + 2 * OFFSET_ADDRESS_SIZE < pend
&& (re_opcode_t) *p2 == dummy_failure_jump)
- p2 += 6;
+ p2 += 2 + 2 * OFFSET_ADDRESS_SIZE;
else
break;
}
/* Consider what happens when matching ":\(.*\)"
against ":/". I don't really understand this code
yet. */
- p[-3] = (unsigned char) pop_failure_jump;
+ p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
+ pop_failure_jump;
DEBUG_PRINT1
(" End of pattern: change to `pop_failure_jump'.\n");
}
else if ((re_opcode_t) *p2 == exactn
+#ifdef MBS_SUPPORT
+ || (re_opcode_t) *p2 == exactn_bin
+#endif
|| (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
{
- register unsigned char c
- = *p2 == (unsigned char) endline ? '\n' : p2[2];
+ register UCHAR_T c
+ = *p2 == (UCHAR_T) endline ? '\n' : p2[2];
- if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
+ if (((re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn
+#ifdef MBS_SUPPORT
+ || (re_opcode_t) p1[1+OFFSET_ADDRESS_SIZE] == exactn_bin
+#endif
+ ) && p1[3+OFFSET_ADDRESS_SIZE] != c)
{
- p[-3] = (unsigned char) pop_failure_jump;
- DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
- c, p1[5]);
+ p[-(1+OFFSET_ADDRESS_SIZE)] = (UCHAR_T)
+ pop_failure_jump;
+#ifdef WCHAR
+ DEBUG_PRINT3 (" %C != %C => pop_failure_jump.\n",
+ (wint_t) c,
+ (wint_t) p1[3+OFFSET_ADDRESS_SIZE]);
+#else
+ DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
+ (char) c,
+ (char) p1[3+OFFSET_ADDRESS_SIZE]);
+#endif
}
+#ifndef WCHAR
else if ((re_opcode_t) p1[3] == charset
|| (re_opcode_t) p1[3] == charset_not)
{
- int not = (re_opcode_t) p1[3] == charset_not;
+ int negate = (re_opcode_t) p1[3] == charset_not;
- if (c < (unsigned char) (p1[4] * BYTEWIDTH)
+ if (c < (unsigned) (p1[4] * BYTEWIDTH)
&& p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
- not = !not;
+ negate = !negate;
- /* `not' is equal to 1 if c would match, which means
+ /* `negate' is equal to 1 if c would match, which means
that we can't change to pop_failure_jump. */
- if (!not)
+ if (!negate)
{
p[-3] = (unsigned char) pop_failure_jump;
DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
}
}
+#endif /* not WCHAR */
}
+#ifndef WCHAR
else if ((re_opcode_t) *p2 == charset)
{
-#ifdef DEBUG
- register unsigned char c
- = *p2 == (unsigned char) endline ? '\n' : p2[2];
-#endif
-
-#if 0
+ /* We win if the first character of the loop is not part
+ of the charset. */
if ((re_opcode_t) p1[3] == exactn
- && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
- && (p2[2 + p1[5] / BYTEWIDTH]
- & (1 << (p1[5] % BYTEWIDTH)))))
-#else
- if ((re_opcode_t) p1[3] == exactn
- && ! ((int) p2[1] * BYTEWIDTH > (int) p1[4]
- && (p2[2 + p1[4] / BYTEWIDTH]
- & (1 << (p1[4] % BYTEWIDTH)))))
-#endif
- {
- p[-3] = (unsigned char) pop_failure_jump;
- DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
- c, p1[5]);
+ && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
+ && (p2[2 + p1[5] / BYTEWIDTH]
+ & (1 << (p1[5] % BYTEWIDTH)))))
+ {
+ p[-3] = (unsigned char) pop_failure_jump;
+ DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
}
else if ((re_opcode_t) p1[3] == charset_not)
}
}
}
+#endif /* not WCHAR */
}
- p -= 2; /* Point at relative address again. */
+ p -= OFFSET_ADDRESS_SIZE; /* Point at relative address again. */
if ((re_opcode_t) p[-1] != pop_failure_jump)
{
- p[-1] = (unsigned char) jump;
+ p[-1] = (UCHAR_T) jump;
DEBUG_PRINT1 (" Match => jump.\n");
goto unconditional_jump;
}
points put on by this pop_failure_jump's matching
on_failure_jump; we got through the pattern to here from the
matching on_failure_jump, so didn't fail. */
- case pop_failure_jump:
+ CASE (pop_failure_jump):
{
/* We need to pass separate storage for the lowest and
highest registers, even though we don't care about the
register from the stack, since lowest will == highest in
`pop_failure_point'. */
active_reg_t dummy_low_reg, dummy_high_reg;
- unsigned char *pdummy;
- const char *sdummy;
+ UCHAR_T *pdummy = NULL;
+ const CHAR_T *sdummy = NULL;
DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
POP_FAILURE_POINT (sdummy, pdummy,
/* Note fall through. */
/* Unconditionally jump (without popping any failure points). */
- case jump:
+ CASE (jump):
EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
p += mcnt; /* Do the jump. */
#else
DEBUG_PRINT2 ("(to 0x%x).\n", p);
#endif
- break;
+ NEXT;
/* We need this opcode so we can detect where alternatives end
in `group_match_null_string_p' et al. */
- case jump_past_alt:
+ CASE (jump_past_alt):
DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
goto unconditional_jump;
pop_failure_jump, also, and with a pattern of, say, `a+', we
are skipping over the on_failure_jump, so we have to push
something meaningless for pop_failure_jump to pop. */
- case dummy_failure_jump:
+ CASE (dummy_failure_jump):
DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
/* It doesn't matter what we push for the string here. What
the code at `fail' tests is the value for the pattern. */
we don't want the failure point for the alternative to be
popped. For example, matching `(a|ab)*' against `aab'
requires that we match the `ab' alternative. */
- case push_dummy_failure:
+ CASE (push_dummy_failure):
DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
/* See comments just above at `dummy_failure_jump' about the
two zeroes. */
PUSH_FAILURE_POINT (NULL, NULL, -2);
- break;
+ NEXT;
/* Have to succeed matching what follows at least n times.
After that, handle like `on_failure_jump'. */
- case succeed_n:
- EXTRACT_NUMBER (mcnt, p + 2);
+ CASE (succeed_n):
+ EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
assert (mcnt >= 0);
if (mcnt > 0)
{
mcnt--;
- p += 2;
+ p += OFFSET_ADDRESS_SIZE;
STORE_NUMBER_AND_INCR (p, mcnt);
#ifdef _LIBC
- DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
+ DEBUG_PRINT3 (" Setting %p to %d.\n", p - OFFSET_ADDRESS_SIZE
+ , mcnt);
#else
- DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
+ DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - OFFSET_ADDRESS_SIZE
+ , mcnt);
#endif
}
else if (mcnt == 0)
{
#ifdef _LIBC
- DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
+ DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n",
+ p + OFFSET_ADDRESS_SIZE);
#else
- DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
-#endif
- p[2] = (unsigned char) no_op;
- p[3] = (unsigned char) no_op;
+ DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n",
+ p + OFFSET_ADDRESS_SIZE);
+#endif /* _LIBC */
+
+#ifdef WCHAR
+ p[1] = (UCHAR_T) no_op;
+#else
+ p[2] = (UCHAR_T) no_op;
+ p[3] = (UCHAR_T) no_op;
+#endif /* WCHAR */
goto on_failure;
}
- break;
+ NEXT;
- case jump_n:
- EXTRACT_NUMBER (mcnt, p + 2);
+ CASE (jump_n):
+ EXTRACT_NUMBER (mcnt, p + OFFSET_ADDRESS_SIZE);
DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
/* Originally, this is how many times we CAN jump. */
if (mcnt)
{
mcnt--;
- STORE_NUMBER (p + 2, mcnt);
+ STORE_NUMBER (p + OFFSET_ADDRESS_SIZE, mcnt);
+
#ifdef _LIBC
- DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
+ DEBUG_PRINT3 (" Setting %p to %d.\n", p + OFFSET_ADDRESS_SIZE,
+ mcnt);
#else
- DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
-#endif
+ DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + OFFSET_ADDRESS_SIZE,
+ mcnt);
+#endif /* _LIBC */
goto unconditional_jump;
}
/* If don't have to jump any more, skip over the rest of command. */
else
- p += 4;
- break;
+ p += 2 * OFFSET_ADDRESS_SIZE;
+ NEXT;
- case set_number_at:
+ CASE (set_number_at):
{
DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
#endif
STORE_NUMBER (p1, mcnt);
- break;
+ NEXT;
}
#if 0
AT_WORD_BOUNDARY, so this code is disabled. Expanding the
macro and introducing temporary variables works around the bug. */
- case wordbound:
+ CASE (wordbound):
DEBUG_PRINT1 ("EXECUTING wordbound.\n");
if (AT_WORD_BOUNDARY (d))
- break;
+ {
+ NEXT;
+ }
goto fail;
- case notwordbound:
+ CASE (notwordbound):
DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
if (AT_WORD_BOUNDARY (d))
goto fail;
- break;
+ NEXT;
#else
- case wordbound:
+ CASE (wordbound):
{
boolean prevchar, thischar;
DEBUG_PRINT1 ("EXECUTING wordbound.\n");
if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
- break;
+ {
+ NEXT;
+ }
prevchar = WORDCHAR_P (d - 1);
thischar = WORDCHAR_P (d);
if (prevchar != thischar)
- break;
+ {
+ NEXT;
+ }
goto fail;
}
- case notwordbound:
+ CASE (notwordbound):
{
boolean prevchar, thischar;
thischar = WORDCHAR_P (d);
if (prevchar != thischar)
goto fail;
- break;
+ NEXT;
}
#endif
- case wordbeg:
+ CASE (wordbeg):
DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
- if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
- break;
+ if (!AT_STRINGS_END (d) && WORDCHAR_P (d)
+ && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
+ {
+ NEXT;
+ }
goto fail;
- case wordend:
+ CASE (wordend):
DEBUG_PRINT1 ("EXECUTING wordend.\n");
if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
- && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
- break;
+ && (AT_STRINGS_END (d) || !WORDCHAR_P (d)))
+ {
+ NEXT;
+ }
goto fail;
#ifdef emacs
- case before_dot:
+ CASE (before_dot):
DEBUG_PRINT1 ("EXECUTING before_dot.\n");
if (PTR_CHAR_POS ((unsigned char *) d) >= point)
goto fail;
- break;
+ NEXT;
- case at_dot:
+ CASE (at_dot):
DEBUG_PRINT1 ("EXECUTING at_dot.\n");
if (PTR_CHAR_POS ((unsigned char *) d) != point)
goto fail;
- break;
+ NEXT;
- case after_dot:
+ CASE (after_dot):
DEBUG_PRINT1 ("EXECUTING after_dot.\n");
if (PTR_CHAR_POS ((unsigned char *) d) <= point)
goto fail;
- break;
+ NEXT;
- case syntaxspec:
+ CASE (syntaxspec):
DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
mcnt = *p++;
goto matchsyntax;
- case wordchar:
+ CASE (wordchar):
DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
mcnt = (int) Sword;
matchsyntax:
if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
goto fail;
SET_REGS_MATCHED ();
- break;
+ NEXT;
- case notsyntaxspec:
+ CASE (notsyntaxspec):
DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
mcnt = *p++;
goto matchnotsyntax;
- case notwordchar:
+ CASE (notwordchar):
DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
mcnt = (int) Sword;
matchnotsyntax:
if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
goto fail;
SET_REGS_MATCHED ();
- break;
+ NEXT;
#else /* not emacs */
- case wordchar:
+ CASE (wordchar):
DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
PREFETCH ();
if (!WORDCHAR_P (d))
goto fail;
SET_REGS_MATCHED ();
d++;
- break;
+ NEXT;
- case notwordchar:
+ CASE (notwordchar):
DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
PREFETCH ();
if (WORDCHAR_P (d))
goto fail;
SET_REGS_MATCHED ();
d++;
- break;
+ NEXT;
#endif /* not emacs */
+#ifndef __GNUC__
default:
abort ();
}
continue; /* Successfully executed one pattern command; keep going. */
+#endif
/* We goto here if a matching operation fails. */
We don't handle duplicates properly (yet). */
static boolean
-group_match_null_string_p (p, end, reg_info)
- unsigned char **p, *end;
- register_info_type *reg_info;
+PREFIX(group_match_null_string_p) (UCHAR_T **p, UCHAR_T *end,
+ PREFIX(register_info_type) *reg_info)
{
int mcnt;
/* Point to after the args to the start_memory. */
- unsigned char *p1 = *p + 2;
+ UCHAR_T *p1 = *p + 2;
while (p1 < end)
{
with an on_failure_jump (see above) that jumps to right
past a jump_past_alt. */
- while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
+ while ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] ==
+ jump_past_alt)
{
/* `mcnt' holds how many bytes long the alternative
is, including the ending `jump_past_alt' and
its number. */
- if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
- reg_info))
+ if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt -
+ (1 + OFFSET_ADDRESS_SIZE),
+ reg_info))
return false;
/* Move to right after this alternative, including the
alternative that starts with an on_failure_jump. */
p1++;
EXTRACT_NUMBER_AND_INCR (mcnt, p1);
- if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
+ if ((re_opcode_t) p1[mcnt-(1+OFFSET_ADDRESS_SIZE)] !=
+ jump_past_alt)
{
/* Get to the beginning of the n-th alternative. */
- p1 -= 3;
+ p1 -= 1 + OFFSET_ADDRESS_SIZE;
break;
}
}
/* Deal with the last alternative: go back and get number
of the `jump_past_alt' just before it. `mcnt' contains
the length of the alternative. */
- EXTRACT_NUMBER (mcnt, p1 - 2);
+ EXTRACT_NUMBER (mcnt, p1 - OFFSET_ADDRESS_SIZE);
- if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
+ if (!PREFIX(alt_match_null_string_p) (p1, p1 + mcnt, reg_info))
return false;
p1 += mcnt; /* Get past the n-th alternative. */
default:
- if (!common_op_match_null_string_p (&p1, end, reg_info))
+ if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
return false;
}
} /* while p1 < end */
byte past the last. The alternative can contain groups. */
static boolean
-alt_match_null_string_p (p, end, reg_info)
- unsigned char *p, *end;
- register_info_type *reg_info;
+PREFIX(alt_match_null_string_p) (UCHAR_T *p, UCHAR_T *end,
+ PREFIX(register_info_type) *reg_info)
{
int mcnt;
- unsigned char *p1 = p;
+ UCHAR_T *p1 = p;
while (p1 < end)
{
break;
default:
- if (!common_op_match_null_string_p (&p1, end, reg_info))
+ if (!PREFIX(common_op_match_null_string_p) (&p1, end, reg_info))
return false;
}
} /* while p1 < end */
Sets P to one after the op and its arguments, if any. */
static boolean
-common_op_match_null_string_p (p, end, reg_info)
- unsigned char **p, *end;
- register_info_type *reg_info;
+PREFIX(common_op_match_null_string_p) (UCHAR_T **p, UCHAR_T *end,
+ PREFIX(register_info_type) *reg_info)
{
int mcnt;
boolean ret;
int reg_no;
- unsigned char *p1 = *p;
+ UCHAR_T *p1 = *p;
switch ((re_opcode_t) *p1++)
{
case start_memory:
reg_no = *p1;
assert (reg_no > 0 && reg_no <= MAX_REGNUM);
- ret = group_match_null_string_p (&p1, end, reg_info);
+ ret = PREFIX(group_match_null_string_p) (&p1, end, reg_info);
/* Have to set this here in case we're checking a group which
contains a group and a back reference to it. */
case succeed_n:
/* Get to the number of times to succeed. */
- p1 += 2;
+ p1 += OFFSET_ADDRESS_SIZE;
EXTRACT_NUMBER_AND_INCR (mcnt, p1);
if (mcnt == 0)
{
- p1 -= 4;
+ p1 -= 2 * OFFSET_ADDRESS_SIZE;
EXTRACT_NUMBER_AND_INCR (mcnt, p1);
p1 += mcnt;
}
break;
case set_number_at:
- p1 += 4;
+ p1 += 2 * OFFSET_ADDRESS_SIZE;
default:
/* All other opcodes mean we cannot match the empty string. */
bytes; nonzero otherwise. */
static int
-bcmp_translate (s1, s2, len, translate)
- const char *s1, *s2;
- register int len;
- RE_TRANSLATE_TYPE translate;
+PREFIX(bcmp_translate) (const CHAR_T *s1, const CHAR_T *s2,
+ register int len,
+ RE_TRANSLATE_TYPE translate)
{
- register const unsigned char *p1 = (const unsigned char *) s1;
- register const unsigned char *p2 = (const unsigned char *) s2;
+ register const UCHAR_T *p1 = (const UCHAR_T *) s1;
+ register const UCHAR_T *p2 = (const UCHAR_T *) s2;
while (len)
{
+#ifdef WCHAR
+ if (((*p1<=0xff)?translate[*p1++]:*p1++)
+ != ((*p2<=0xff)?translate[*p2++]:*p2++))
+ return 1;
+#else /* BYTE */
if (translate[*p1++] != translate[*p2++]) return 1;
+#endif /* WCHAR */
len--;
}
return 0;
}
\f
+
+#else /* not INSIDE_RECURSION */
+
/* Entry points for GNU code. */
/* re_compile_pattern is the GNU regular expression compiler: it
We call regex_compile to do the actual compilation. */
const char *
-re_compile_pattern (pattern, length, bufp)
- const char *pattern;
- size_t length;
- struct re_pattern_buffer *bufp;
+re_compile_pattern (const char *pattern,
+ size_t length,
+ struct re_pattern_buffer *bufp)
{
reg_errcode_t ret;
/* Match anchors at newline. */
bufp->newline_anchor = 1;
- ret = regex_compile (pattern, length, re_syntax_options, bufp);
+# ifdef MBS_SUPPORT
+ if (MB_CUR_MAX != 1)
+ ret = wcs_regex_compile (pattern, length, re_syntax_options, bufp);
+ else
+# endif
+ ret = byte_regex_compile (pattern, length, re_syntax_options, bufp);
if (!ret)
return NULL;
- return gettext (re_error_msgid[(int) ret]);
+ return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
}
#ifdef _LIBC
weak_alias (__re_compile_pattern, re_compile_pattern)
regcomp/regexec below without link errors. */
weak_function
#endif
-re_comp (s)
- const char *s;
+re_comp (const char *s)
{
reg_errcode_t ret;
if (!s)
{
if (!re_comp_buf.buffer)
- return gettext ("No previous regular expression");
+ return (char *) gettext ("No previous regular expression");
return 0;
}
if (!re_comp_buf.buffer)
{
- re_comp_buf.buffer = (unsigned char *) malloc (200);
+ re_comp_buf.buffer = malloc (200);
if (re_comp_buf.buffer == NULL)
- return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
+ return (char *) gettext (re_error_msgid
+ + re_error_msgid_idx[(int) REG_ESPACE]);
re_comp_buf.allocated = 200;
- re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
+ re_comp_buf.fastmap = malloc (1 << BYTEWIDTH);
if (re_comp_buf.fastmap == NULL)
- return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
+ return (char *) gettext (re_error_msgid
+ + re_error_msgid_idx[(int) REG_ESPACE]);
}
/* Since `re_exec' always passes NULL for the `regs' argument, we
/* Match anchors at newlines. */
re_comp_buf.newline_anchor = 1;
- ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
+# ifdef MBS_SUPPORT
+ if (MB_CUR_MAX != 1)
+ ret = wcs_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
+ else
+# endif
+ ret = byte_regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
if (!ret)
return NULL;
/* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
- return (char *) gettext (re_error_msgid[(int) ret]);
+ return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
}
#ifdef _LIBC
weak_function
#endif
-re_exec (s)
- const char *s;
+re_exec (const char *s)
{
const int len = strlen (s);
return
- 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
+ 0 <= re_search (&re_comp_buf, s, len, 0, len, 0);
}
#endif /* _REGEX_RE_COMP */
REG_EXTENDED bit in CFLAGS is set; otherwise, to
RE_SYNTAX_POSIX_BASIC;
`newline_anchor' to REG_NEWLINE being set in CFLAGS;
- `fastmap' and `fastmap_accurate' to zero;
+ `fastmap' to an allocated space for the fastmap;
+ `fastmap_accurate' to zero;
`re_nsub' to the number of subexpressions in PATTERN.
PATTERN is the address of the pattern string.
the return codes and their meanings.) */
int
-regcomp (preg, pattern, cflags)
- regex_t *preg;
- const char *pattern;
- int cflags;
+regcomp (regex_t *preg, const char *pattern, int cflags)
{
reg_errcode_t ret;
reg_syntax_t syntax
preg->allocated = 0;
preg->used = 0;
- /* Don't bother to use a fastmap when searching. This simplifies the
- REG_NEWLINE case: if we used a fastmap, we'd have to put all the
- characters after newlines into the fastmap. This way, we just try
- every character. */
- preg->fastmap = 0;
+ /* Try to allocate space for the fastmap. */
+ preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
if (cflags & REG_ICASE)
{
unsigned i;
- preg->translate
- = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
- * sizeof (*(RE_TRANSLATE_TYPE)0));
+ preg->translate =
+ (RE_TRANSLATE_TYPE)
+ malloc (CHAR_SET_SIZE * sizeof (*(RE_TRANSLATE_TYPE)0));
if (preg->translate == NULL)
return (int) REG_ESPACE;
/* Map uppercase characters to corresponding lowercase ones. */
for (i = 0; i < CHAR_SET_SIZE; i++)
- preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
+ preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
}
else
preg->translate = NULL;
/* POSIX says a null character in the pattern terminates it, so we
can use strlen here in compiling the pattern. */
- ret = regex_compile (pattern, strlen (pattern), syntax, preg);
+# ifdef MBS_SUPPORT
+ if (MB_CUR_MAX != 1)
+ ret = wcs_regex_compile (pattern, strlen (pattern), syntax, preg);
+ else
+# endif
+ ret = byte_regex_compile (pattern, strlen (pattern), syntax, preg);
/* POSIX doesn't distinguish between an unmatched open-group and an
unmatched close-group: both are REG_EPAREN. */
if (ret == REG_ERPAREN) ret = REG_EPAREN;
+ if (ret == REG_NOERROR && preg->fastmap)
+ {
+ /* Compute the fastmap now, since regexec cannot modify the pattern
+ buffer. */
+ if (re_compile_fastmap (preg) == -2)
+ {
+ /* Some error occurred while computing the fastmap, just forget
+ about it. */
+ free (preg->fastmap);
+ preg->fastmap = NULL;
+ }
+ }
+
return (int) ret;
}
#ifdef _LIBC
We return 0 if we find a match and REG_NOMATCH if not. */
int
-regexec (preg, string, nmatch, pmatch, eflags)
- const regex_t *preg;
- const char *string;
- size_t nmatch;
- regmatch_t pmatch[];
- int eflags;
+regexec (const regex_t *preg, const char *string,
+ size_t nmatch, regmatch_t pmatch[], int eflags)
{
int ret;
struct re_registers regs;
if (want_reg_info)
{
regs.num_regs = nmatch;
- regs.start = TALLOC (nmatch, regoff_t);
- regs.end = TALLOC (nmatch, regoff_t);
- if (regs.start == NULL || regs.end == NULL)
+ regs.start = TALLOC (nmatch * 2, regoff_t);
+ if (regs.start == NULL)
return (int) REG_NOMATCH;
+ regs.end = regs.start + nmatch;
}
/* Perform the searching operation. */
ret = re_search (&private_preg, string, len,
/* start: */ 0, /* range: */ len,
- want_reg_info ? ®s : (struct re_registers *) 0);
+ want_reg_info ? ®s : 0);
/* Copy the register information to the POSIX structure. */
if (want_reg_info)
/* If we needed the temporary register info, free the space now. */
free (regs.start);
- free (regs.end);
}
/* We want zero return to mean success, unlike `re_search'. */
from either regcomp or regexec. We don't use PREG here. */
size_t
-regerror (errcode, preg, errbuf, errbuf_size)
- int errcode;
- const regex_t *preg;
- char *errbuf;
- size_t errbuf_size;
+regerror (int errcode, const regex_t *preg, char *errbuf, size_t errbuf_size)
{
const char *msg;
size_t msg_size;
if (errcode < 0
- || errcode >= (int) (sizeof (re_error_msgid)
- / sizeof (re_error_msgid[0])))
+ || errcode >= (int) (sizeof (re_error_msgid_idx)
+ / sizeof (re_error_msgid_idx[0])))
/* Only error codes returned by the rest of the code should be passed
to this routine. If we are given anything else, or if other regex
code generates an invalid error code, then the program has a bug.
Dump core so we can fix it. */
abort ();
- msg = gettext (re_error_msgid[errcode]);
+ msg = gettext (re_error_msgid + re_error_msgid_idx[errcode]);
msg_size = strlen (msg) + 1; /* Includes the null. */
/* Free dynamically allocated space used by PREG. */
void
-regfree (preg)
- regex_t *preg;
+regfree (regex_t *preg)
{
if (preg->buffer != NULL)
free (preg->buffer);
#endif
#endif /* not emacs */
+
+#endif /* not INSIDE_RECURSION */
+
+\f
+#undef STORE_NUMBER
+#undef STORE_NUMBER_AND_INCR
+#undef EXTRACT_NUMBER
+#undef EXTRACT_NUMBER_AND_INCR
+
+#undef DEBUG_PRINT_COMPILED_PATTERN
+#undef DEBUG_PRINT_DOUBLE_STRING
+
+#undef INIT_FAIL_STACK
+#undef RESET_FAIL_STACK
+#undef DOUBLE_FAIL_STACK
+#undef PUSH_PATTERN_OP
+#undef PUSH_FAILURE_POINTER
+#undef PUSH_FAILURE_INT
+#undef PUSH_FAILURE_ELT
+#undef POP_FAILURE_POINTER
+#undef POP_FAILURE_INT
+#undef POP_FAILURE_ELT
+#undef DEBUG_PUSH
+#undef DEBUG_POP
+#undef PUSH_FAILURE_POINT
+#undef POP_FAILURE_POINT
+
+#undef REG_UNSET_VALUE
+#undef REG_UNSET
+
+#undef PATFETCH
+#undef PATFETCH_RAW
+#undef PATUNFETCH
+#undef TRANSLATE
+
+#undef INIT_BUF_SIZE
+#undef GET_BUFFER_SPACE
+#undef BUF_PUSH
+#undef BUF_PUSH_2
+#undef BUF_PUSH_3
+#undef STORE_JUMP
+#undef STORE_JUMP2
+#undef INSERT_JUMP
+#undef INSERT_JUMP2
+#undef EXTEND_BUFFER
+#undef GET_UNSIGNED_NUMBER
+#undef FREE_STACK_RETURN
+
+# undef POINTER_TO_OFFSET
+# undef MATCHING_IN_FRST_STRING
+# undef PREFETCH
+# undef AT_STRINGS_BEG
+# undef AT_STRINGS_END
+# undef WORDCHAR_P
+# undef FREE_VAR
+# undef FREE_VARIABLES
+# undef NO_HIGHEST_ACTIVE_REG
+# undef NO_LOWEST_ACTIVE_REG
+
+# undef CHAR_T
+# undef UCHAR_T
+# undef COMPILED_BUFFER_VAR
+# undef OFFSET_ADDRESS_SIZE
+# undef CHAR_CLASS_SIZE
+# undef PREFIX
+# undef ARG_PREFIX
+# undef PUT_CHAR
+# undef BYTE
+# undef WCHAR
+
+# define DEFINED_ONCE