/* Extended regular expression matching and search library, version
- 0.12. (Implements POSIX draft P10003.2/D11.2, except for
+ 0.12. (Implements POSIX draft P1003.2/D11.2, except for some of the
internationalization features.)
- Copyright (C) 1993,94,95,96,97,98,2000 Free Software Foundation, Inc.
+ Copyright (C) 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001,
+ 2002, 2003, 2004, 2005, 2006 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
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,
+ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301,
USA. */
/* TODO:
- structure the opcode space into opcode+flag.
- - merge with glibc's regex.[ch]
- */
+ - merge with glibc's regex.[ch].
+ - replace (succeed_n + jump_n + set_number_at) with something that doesn't
+ need to modify the compiled regexp so that re_match can be reentrant.
+ - get rid of on_failure_jump_smart by doing the optimization in re_comp
+ rather than at run-time, so that re_match can be reentrant.
+*/
/* AIX requires this to be the first thing in the file. */
-#if defined (_AIX) && !defined (REGEX_MALLOC)
+#if defined _AIX && !defined REGEX_MALLOC
#pragma alloca
#endif
-#undef _GNU_SOURCE
-#define _GNU_SOURCE
+#ifdef HAVE_CONFIG_H
+# include <config.h>
+#endif
-#ifdef emacs
-/* Converts the pointer to the char to BEG-based offset from the start. */
-#define PTR_TO_OFFSET(d) POS_AS_IN_BUFFER (POINTER_TO_OFFSET (d))
-#define POS_AS_IN_BUFFER(p) ((p) + (NILP (re_match_object) || BUFFERP (re_match_object)))
+#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
-#ifdef HAVE_CONFIG_H
-#include <config.h>
+/* Whether to use ISO C Amendment 1 wide char functions.
+ Those should not be used for Emacs since it uses its own. */
+#if defined _LIBC
+#define WIDE_CHAR_SUPPORT 1
+#else
+#define WIDE_CHAR_SUPPORT \
+ (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC && !emacs)
#endif
-/* We need this for `regex.h', and perhaps for the Emacs include files. */
-#include <sys/types.h>
+/* For platform which support the ISO C amendement 1 functionality we
+ support user defined character classes. */
+#if WIDE_CHAR_SUPPORT
+/* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
+# include <wchar.h>
+# include <wctype.h>
+#endif
-/* This is for other GNU distributions with internationalized messages. */
-#if HAVE_LIBINTL_H || defined (_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) \
+ __regerror(errcode, preg, errbuf, errbuf_size)
+# 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) \
+ __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
+# define re_match(bufp, string, size, pos, regs) \
+ __re_match (bufp, string, size, pos, 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) \
+ __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) \
+ __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
+# define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
+
+/* Make sure we call libc's function even if the user overrides them. */
+# define btowc __btowc
+# define iswctype __iswctype
+# define wctype __wctype
+
+# define WEAK_ALIAS(a,b) weak_alias (a, b)
+
+/* We are also using some library internals. */
+# include <locale/localeinfo.h>
+# include <locale/elem-hash.h>
+# include <langinfo.h>
+#else
+# define WEAK_ALIAS(a,b)
+#endif
+
+/* This is for other GNU distributions with internationalized messages. */
+#if HAVE_LIBINTL_H || defined _LIBC
# include <libintl.h>
#else
# define gettext(msgid) (msgid)
#ifndef gettext_noop
/* This define is so xgettext can find the internationalizable
strings. */
-#define gettext_noop(String) String
+# define gettext_noop(String) String
#endif
/* The `emacs' switch turns on certain matching commands
that make sense only in Emacs. */
#ifdef emacs
-#include "lisp.h"
-#include "buffer.h"
+# include "lisp.h"
+# include "buffer.h"
/* Make syntax table lookup grant data in gl_state. */
-#define SYNTAX_ENTRY_VIA_PROPERTY
-
-#include "syntax.h"
-#include "charset.h"
-#include "category.h"
+# define SYNTAX_ENTRY_VIA_PROPERTY
+
+# include "syntax.h"
+# include "charset.h"
+# include "category.h"
+
+# ifdef malloc
+# undef malloc
+# endif
+# define malloc xmalloc
+# ifdef realloc
+# undef realloc
+# endif
+# define realloc xrealloc
+# ifdef free
+# undef free
+# endif
+# define free xfree
-#define malloc xmalloc
-#define realloc xrealloc
-#define free xfree
+/* Converts the pointer to the char to BEG-based offset from the start. */
+# define PTR_TO_OFFSET(d) POS_AS_IN_BUFFER (POINTER_TO_OFFSET (d))
+# define POS_AS_IN_BUFFER(p) ((p) + (NILP (re_match_object) || BUFFERP (re_match_object)))
-#define RE_MULTIBYTE_P(bufp) ((bufp)->multibyte)
-#define RE_STRING_CHAR(p, s) \
+# define RE_MULTIBYTE_P(bufp) ((bufp)->multibyte)
+# define RE_STRING_CHAR(p, s) \
(multibyte ? (STRING_CHAR (p, s)) : (*(p)))
-#define RE_STRING_CHAR_AND_LENGTH(p, s, len) \
+# define RE_STRING_CHAR_AND_LENGTH(p, s, len) \
(multibyte ? (STRING_CHAR_AND_LENGTH (p, s, len)) : ((len) = 1, *(p)))
/* Set C a (possibly multibyte) character before P. P points into a
string which is the virtual concatenation of STR1 (which ends at
END1) or STR2 (which ends at END2). */
-#define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
+# define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
do { \
if (multibyte) \
{ \
- re_char *dtemp = (p) == (str2) ? (end1) : (p); \
- re_char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
- while (dtemp-- > dlimit && !CHAR_HEAD_P (*dtemp)); \
- c = STRING_CHAR (dtemp, (p) - dtemp); \
+ re_char *dtemp = (p) == (str2) ? (end1) : (p); \
+ re_char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
+ re_char *d0 = dtemp; \
+ PREV_CHAR_BOUNDARY (d0, dlimit); \
+ c = STRING_CHAR (d0, dtemp - d0); \
} \
else \
(c = ((p) == (str2) ? (end1) : (p))[-1]); \
/* 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
+# 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
-#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)
-#include <string.h>
-#ifndef bcmp
-#define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
-#endif
-#ifndef bcopy
-#define bcopy(s, d, n) memcpy ((d), (s), (n))
-#endif
-#ifndef bzero
-#define bzero(s, n) memset ((s), 0, (n))
-#endif
-#else
-#include <strings.h>
-#endif
-#endif
+/* When used in Emacs's lib-src, we need xmalloc and xrealloc. */
-/* Define the syntax stuff for \<, \>, etc. */
-
-/* Sword must be nonzero for the wordchar pattern commands in re_match_2. */
-enum syntaxcode { Swhitespace = 0, Sword = 1 };
-
-#ifdef SWITCH_ENUM_BUG
-#define SWITCH_ENUM_CAST(x) ((int)(x))
-#else
-#define SWITCH_ENUM_CAST(x) (x)
-#endif
-
-#ifdef SYNTAX_TABLE
-
-extern char *re_syntax_table;
-
-#else /* not SYNTAX_TABLE */
-
-/* How many characters in the character set. */
-#define CHAR_SET_SIZE 256
-
-static char re_syntax_table[CHAR_SET_SIZE];
-
-static void
-init_syntax_once ()
+void *
+xmalloc (size)
+ size_t size;
{
- 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;
+ register void *val;
+ val = (void *) malloc (size);
+ if (!val && size)
+ {
+ write (2, "virtual memory exhausted\n", 25);
+ exit (1);
+ }
+ return val;
+}
- for (c = 'A'; c <= 'Z'; c++)
- re_syntax_table[c] = Sword;
+void *
+xrealloc (block, size)
+ void *block;
+ size_t size;
+{
+ register void *val;
+ /* We must call malloc explicitly when BLOCK is 0, since some
+ reallocs don't do this. */
+ if (! block)
+ val = (void *) malloc (size);
+ else
+ val = (void *) realloc (block, size);
+ if (!val && size)
+ {
+ write (2, "virtual memory exhausted\n", 25);
+ exit (1);
+ }
+ return val;
+}
- for (c = '0'; c <= '9'; c++)
- re_syntax_table[c] = Sword;
+# ifdef malloc
+# undef malloc
+# endif
+# define malloc xmalloc
+# ifdef realloc
+# undef realloc
+# endif
+# define realloc xrealloc
- re_syntax_table['_'] = Sword;
+/* 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
+# endif
+# endif
+# endif
+
+/* This is the normal way of making sure we have memcpy, memcmp and 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
+# include <string.h>
+# ifndef bzero
+# ifndef _LIBC
+# define bzero(s, n) (memset (s, '\0', n), (s))
+# else
+# 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
- done = 1;
-}
+/* Define the syntax stuff for \<, \>, etc. */
-#endif /* not SYNTAX_TABLE */
+/* Sword must be nonzero for the wordchar pattern commands in re_match_2. */
+enum syntaxcode { Swhitespace = 0, Sword = 1, Ssymbol = 2 };
-#define SYNTAX(c) re_syntax_table[c]
+# ifdef SWITCH_ENUM_BUG
+# define SWITCH_ENUM_CAST(x) ((int)(x))
+# else
+# define SWITCH_ENUM_CAST(x) (x)
+# endif
/* Dummy macros for non-Emacs environments. */
-#define BASE_LEADING_CODE_P(c) (0)
-#define CHAR_CHARSET(c) 0
-#define CHARSET_LEADING_CODE_BASE(c) 0
-#define MAX_MULTIBYTE_LENGTH 1
-#define RE_MULTIBYTE_P(x) 0
-#define WORD_BOUNDARY_P(c1, c2) (0)
-#define CHAR_HEAD_P(p) (1)
-#define SINGLE_BYTE_CHAR_P(c) (1)
-#define SAME_CHARSET_P(c1, c2) (1)
-#define MULTIBYTE_FORM_LENGTH(p, s) (1)
-#define STRING_CHAR(p, s) (*(p))
-#define RE_STRING_CHAR STRING_CHAR
-#define CHAR_STRING(c, s) (*(s) = (c), 1)
-#define STRING_CHAR_AND_LENGTH(p, s, actual_len) ((actual_len) = 1, *(p))
-#define RE_STRING_CHAR_AND_LENGTH STRING_CHAR_AND_LENGTH
-#define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
+# define BASE_LEADING_CODE_P(c) (0)
+# define CHAR_CHARSET(c) 0
+# define CHARSET_LEADING_CODE_BASE(c) 0
+# define MAX_MULTIBYTE_LENGTH 1
+# define RE_MULTIBYTE_P(x) 0
+# define WORD_BOUNDARY_P(c1, c2) (0)
+# define CHAR_HEAD_P(p) (1)
+# define SINGLE_BYTE_CHAR_P(c) (1)
+# define SAME_CHARSET_P(c1, c2) (1)
+# define MULTIBYTE_FORM_LENGTH(p, s) (1)
+# define PREV_CHAR_BOUNDARY(p, limit) ((p)--)
+# define STRING_CHAR(p, s) (*(p))
+# define RE_STRING_CHAR STRING_CHAR
+# define CHAR_STRING(c, s) (*(s) = (c), 1)
+# define STRING_CHAR_AND_LENGTH(p, s, actual_len) ((actual_len) = 1, *(p))
+# define RE_STRING_CHAR_AND_LENGTH STRING_CHAR_AND_LENGTH
+# define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
(c = ((p) == (str2) ? *((end1) - 1) : *((p) - 1)))
+# define MAKE_CHAR(charset, c1, c2) (c1)
#endif /* not emacs */
#ifndef RE_TRANSLATE
-#define RE_TRANSLATE(TBL, C) ((unsigned char)(TBL)[C])
-#define RE_TRANSLATE_P(TBL) (TBL)
+# define RE_TRANSLATE(TBL, C) ((unsigned char)(TBL)[C])
+# define RE_TRANSLATE_P(TBL) (TBL)
#endif
\f
/* Get the interface, including the syntax bits. */
#ifdef emacs
/* 1 if C is an ASCII character. */
-#define IS_REAL_ASCII(c) ((c) < 0200)
+# define IS_REAL_ASCII(c) ((c) < 0200)
/* 1 if C is a unibyte character. */
-#define ISUNIBYTE(c) (SINGLE_BYTE_CHAR_P ((c)))
+# define ISUNIBYTE(c) (SINGLE_BYTE_CHAR_P ((c)))
/* The Emacs definitions should not be directly affected by locales. */
/* In Emacs, these are only used for single-byte characters. */
-#define ISDIGIT(c) ((c) >= '0' && (c) <= '9')
-#define ISCNTRL(c) ((c) < ' ')
-#define ISXDIGIT(c) (((c) >= '0' && (c) <= '9') \
+# define ISDIGIT(c) ((c) >= '0' && (c) <= '9')
+# define ISCNTRL(c) ((c) < ' ')
+# define ISXDIGIT(c) (((c) >= '0' && (c) <= '9') \
|| ((c) >= 'a' && (c) <= 'f') \
|| ((c) >= 'A' && (c) <= 'F'))
/* This is only used for single-byte characters. */
-#define ISBLANK(c) ((c) == ' ' || (c) == '\t')
+# define ISBLANK(c) ((c) == ' ' || (c) == '\t')
/* The rest must handle multibyte characters. */
-#define ISGRAPH(c) (SINGLE_BYTE_CHAR_P (c) \
+# define ISGRAPH(c) (SINGLE_BYTE_CHAR_P (c) \
? (c) > ' ' && !((c) >= 0177 && (c) <= 0237) \
: 1)
-#define ISPRINT(c) (SINGLE_BYTE_CHAR_P (c) \
+# define ISPRINT(c) (SINGLE_BYTE_CHAR_P (c) \
? (c) >= ' ' && !((c) >= 0177 && (c) <= 0237) \
: 1)
-#define ISALNUM(c) (IS_REAL_ASCII (c) \
+# define ISALNUM(c) (IS_REAL_ASCII (c) \
? (((c) >= 'a' && (c) <= 'z') \
|| ((c) >= 'A' && (c) <= 'Z') \
|| ((c) >= '0' && (c) <= '9')) \
: SYNTAX (c) == Sword)
-#define ISALPHA(c) (IS_REAL_ASCII (c) \
+# define ISALPHA(c) (IS_REAL_ASCII (c) \
? (((c) >= 'a' && (c) <= 'z') \
|| ((c) >= 'A' && (c) <= 'Z')) \
: SYNTAX (c) == Sword)
-#define ISLOWER(c) (LOWERCASEP (c))
+# define ISLOWER(c) (LOWERCASEP (c))
-#define ISPUNCT(c) (IS_REAL_ASCII (c) \
+# define ISPUNCT(c) (IS_REAL_ASCII (c) \
? ((c) > ' ' && (c) < 0177 \
&& !(((c) >= 'a' && (c) <= 'z') \
|| ((c) >= 'A' && (c) <= 'Z') \
|| ((c) >= '0' && (c) <= '9'))) \
: SYNTAX (c) != Sword)
-#define ISSPACE(c) (SYNTAX (c) == Swhitespace)
+# define ISSPACE(c) (SYNTAX (c) == Swhitespace)
-#define ISUPPER(c) (UPPERCASEP (c))
+# define ISUPPER(c) (UPPERCASEP (c))
-#define ISWORD(c) (SYNTAX (c) == Sword)
+# define ISWORD(c) (SYNTAX (c) == Sword)
#else /* not emacs */
"... 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
+ 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." */
+ 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 ISASCII(c) 1
-#else
-#define ISASCII(c) isascii(c)
-#endif
+# undef ISASCII
+# if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
+# define ISASCII(c) 1
+# else
+# define ISASCII(c) isascii(c)
+# endif
/* 1 if C is an ASCII character. */
-#define IS_REAL_ASCII(c) ((c) < 0200)
+# define IS_REAL_ASCII(c) ((c) < 0200)
/* This distinction is not meaningful, except in Emacs. */
-#define ISUNIBYTE(c) 1
+# define ISUNIBYTE(c) 1
+
+# 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))
+
+# define ISWORD(c) ISALPHA(c)
+
+# ifdef _tolower
+# define TOLOWER(c) _tolower(c)
+# else
+# define TOLOWER(c) tolower(c)
+# endif
-#define ISDIGIT(c) (ISASCII (c) && isdigit (c))
-#define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
-#define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
+/* How many characters in the character set. */
+# define CHAR_SET_SIZE 256
-#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
+# ifdef SYNTAX_TABLE
+
+extern char *re_syntax_table;
-#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))
+# else /* not SYNTAX_TABLE */
+
+static char re_syntax_table[CHAR_SET_SIZE];
-#define ISWORD(c) ISALPHA(c)
+static void
+init_syntax_once ()
+{
+ register int c;
+ static int done = 0;
+
+ if (done)
+ return;
+
+ bzero (re_syntax_table, sizeof re_syntax_table);
+
+ for (c = 0; c < CHAR_SET_SIZE; ++c)
+ if (ISALNUM (c))
+ re_syntax_table[c] = Sword;
+
+ re_syntax_table['_'] = Ssymbol;
+
+ done = 1;
+}
+
+# endif /* not SYNTAX_TABLE */
+
+# define SYNTAX(c) re_syntax_table[(c)]
#endif /* not emacs */
\f
#ifndef NULL
-#define NULL (void *)0
+# 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.) */
+ (Per Bothner suggested the basic approach.) */
#undef SIGN_EXTEND_CHAR
#if __STDC__
-#define SIGN_EXTEND_CHAR(c) ((signed char) (c))
+# 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)
+# define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
#endif
\f
/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
#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 */
/* 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>
-#else /* not __GNUC__ or HAVE_ALLOCA_H */
-#if 0 /* It is a bad idea to declare alloca. We always cast the result. */
-#ifndef _AIX /* Already did AIX, up at the top. */
-char *alloca ();
-#endif /* not _AIX */
-#endif
-#endif /* not HAVE_ALLOCA_H */
-#endif /* not __GNUC__ */
+# ifdef __GNUC__
+# define alloca __builtin_alloca
+# else /* not __GNUC__ */
+# if HAVE_ALLOCA_H
+# include <alloca.h>
+# endif /* HAVE_ALLOCA_H */
+# endif /* not __GNUC__ */
-#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), \
- bcopy (source, destination, osize), \
- destination)
+ 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 */
/* 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 */
-#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) ((void)0)
+# define REGEX_FREE_STACK(arg) ((void)0)
-#endif /* not REGEX_MALLOC */
+# endif /* not REGEX_MALLOC */
#endif /* not using relocating allocator */
if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), 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 false 0
#define true 1
-static int re_match_2_internal ();
+static int re_match_2_internal _RE_ARGS ((struct re_pattern_buffer *bufp,
+ re_char *string1, int size1,
+ re_char *string2, int size2,
+ int pos,
+ struct re_registers *regs,
+ int stop));
\f
/* These are the command codes that appear in compiled regular
- expressions. Some opcodes are followed by argument bytes. A
+ expressions. Some opcodes are followed by argument bytes. A
command code can specify any interpretation whatsoever for its
arguments. Zero bytes may appear in the compiled regular expression. */
{
no_op = 0,
- /* Succeed right away--no more backtracking. */
+ /* Succeed right away--no more backtracking. */
succeed,
/* Followed by one byte giving n, then by n literal bytes. */
If the length byte has the 0x80 bit set, then that stuff
is followed by a range table:
2 bytes of flags for character sets (low 8 bits, high 8 bits)
- See RANGE_TABLE_WORK_BITS below.
- 2 bytes, the number of pairs that follow
+ See RANGE_TABLE_WORK_BITS below.
+ 2 bytes, the number of pairs that follow (upto 32767)
pairs, each 2 multibyte characters,
- each multibyte character represented as 3 bytes. */
+ each multibyte character represented as 3 bytes. */
charset,
/* Same parameters as charset, but match any character that is
- not one of those specified. */
+ not one of those specified. */
charset_not,
/* Start remembering the text that is matched, for storing in a
stop_memory,
/* Match a duplicate of something remembered. Followed by one
- byte containing the register number. */
+ byte containing the register number. */
duplicate,
/* Fail unless at beginning of line. */
begline,
- /* Fail unless at end of line. */
+ /* Fail unless at end of line. */
endline,
/* Succeeds if at beginning of buffer (if emacs) or at beginning
by a `no_op'. */
on_failure_jump_nastyloop,
- /* A smart `on_failure_jump' used for greedy * and + operators.
+ /* A smart `on_failure_jump' used for greedy * and + operators.
It analyses the loop before which it is put and if the
loop does not require backtracking, it changes itself to
`on_failure_keep_string_jump' and short-circuits the loop,
wordbound, /* Succeeds if at a word boundary. */
notwordbound, /* Succeeds if not at a word boundary. */
+ symbeg, /* Succeeds if at symbol beginning. */
+ symend, /* Succeeds if at symbol end. */
+
/* Matches any character whose syntax is specified. Followed by
a byte which contains a syntax code, e.g., Sword. */
syntaxspec,
} while (0)
#ifdef DEBUG
+static void extract_number _RE_ARGS ((int *dest, re_char *source));
static void
extract_number (dest, source)
int *dest;
- unsigned char *source;
+ re_char *source;
{
int temp = SIGN_EXTEND_CHAR (*(source + 1));
*dest = *source & 0377;
*dest += temp << 8;
}
-#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) extract_number (&dest, src)
+# endif /* not EXTRACT_MACROS */
#endif /* DEBUG */
} while (0)
#ifdef DEBUG
+static void extract_number_and_incr _RE_ARGS ((int *destination,
+ re_char **source));
static void
extract_number_and_incr (destination, source)
int *destination;
- unsigned char **source;
+ re_char **source;
{
extract_number (destination, *source);
*source += 2;
}
-#ifndef EXTRACT_MACROS
-#undef EXTRACT_NUMBER_AND_INCR
-#define EXTRACT_NUMBER_AND_INCR(dest, src) \
+# 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 */
+# endif /* not EXTRACT_MACROS */
#endif /* DEBUG */
\f
#define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \
do \
{ \
- int range_start, range_end; \
- unsigned char *p; \
- unsigned char *range_table_end \
+ re_wchar_t range_start, range_end; \
+ re_char *p; \
+ re_char *range_table_end \
= CHARSET_RANGE_TABLE_END ((range_table), (count)); \
\
for (p = (range_table); p < range_table_end; p += 2 * 3) \
{ \
/* Number of ranges in range table. */ \
int count; \
- unsigned char *range_table = CHARSET_RANGE_TABLE (charset); \
- \
+ re_char *range_table = CHARSET_RANGE_TABLE (charset); \
+ \
EXTRACT_NUMBER_AND_INCR (count, range_table); \
CHARSET_LOOKUP_RANGE_TABLE_RAW ((not), (c), range_table, count); \
} \
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. */
+ the other test files, you can run the already-written tests. */
#ifdef DEBUG
/* 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>
static int debug = -100000;
-#define DEBUG_STATEMENT(e) e
-#define DEBUG_PRINT1(x) if (debug > 0) printf (x)
-#define DEBUG_PRINT2(x1, x2) if (debug > 0) printf (x1, x2)
-#define DEBUG_PRINT3(x1, x2, x3) if (debug > 0) printf (x1, x2, x3)
-#define DEBUG_PRINT4(x1, x2, x3, x4) if (debug > 0) printf (x1, x2, x3, x4)
-#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
+# define DEBUG_STATEMENT(e) e
+# define DEBUG_PRINT1(x) if (debug > 0) printf (x)
+# define DEBUG_PRINT2(x1, x2) if (debug > 0) printf (x1, x2)
+# define DEBUG_PRINT3(x1, x2, x3) if (debug > 0) printf (x1, x2, x3)
+# define DEBUG_PRINT4(x1, x2, x3, x4) if (debug > 0) printf (x1, x2, x3, x4)
+# define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
if (debug > 0) print_partial_compiled_pattern (s, e)
-#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
+# define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
if (debug > 0) print_double_string (w, s1, sz1, s2, sz2)
void
print_partial_compiled_pattern (start, end)
- unsigned char *start;
- unsigned char *end;
+ re_char *start;
+ re_char *end;
{
int mcnt, mcnt2;
- unsigned char *p = start;
- unsigned char *pend = end;
+ re_char *p = start;
+ re_char *pend = end;
if (start == NULL)
{
- printf ("(null)\n");
+ fprintf (stderr, "(null)\n");
return;
}
/* Loop over pattern commands. */
while (p < pend)
{
- printf ("%d:\t", p - start);
+ fprintf (stderr, "%d:\t", p - start);
switch ((re_opcode_t) *p++)
{
case no_op:
- printf ("/no_op");
+ fprintf (stderr, "/no_op");
break;
case succeed:
- printf ("/succeed");
+ fprintf (stderr, "/succeed");
break;
case exactn:
mcnt = *p++;
- printf ("/exactn/%d", mcnt);
+ fprintf (stderr, "/exactn/%d", mcnt);
do
{
- putchar ('/');
- putchar (*p++);
+ fprintf (stderr, "/%c", *p++);
}
while (--mcnt);
break;
case start_memory:
- printf ("/start_memory/%d", *p++);
+ fprintf (stderr, "/start_memory/%d", *p++);
break;
case stop_memory:
- printf ("/stop_memory/%d", *p++);
+ fprintf (stderr, "/stop_memory/%d", *p++);
break;
case duplicate:
- printf ("/duplicate/%d", *p++);
+ fprintf (stderr, "/duplicate/%d", *p++);
break;
case anychar:
- printf ("/anychar");
+ fprintf (stderr, "/anychar");
break;
case charset:
int length = CHARSET_BITMAP_SIZE (p - 1);
int has_range_table = CHARSET_RANGE_TABLE_EXISTS_P (p - 1);
- printf ("/charset [%s",
- (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
+ fprintf (stderr, "/charset [%s",
+ (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
- assert (p + *p < pend);
+ if (p + *p >= pend)
+ fprintf (stderr, " !extends past end of pattern! ");
for (c = 0; c < 256; c++)
if (c / 8 < length
/* Are we starting a range? */
if (last + 1 == c && ! in_range)
{
- putchar ('-');
+ fprintf (stderr, "-");
in_range = 1;
}
/* Have we broken a range? */
else if (last + 1 != c && in_range)
{
- putchar (last);
+ fprintf (stderr, "%c", last);
in_range = 0;
}
if (! in_range)
- putchar (c);
+ fprintf (stderr, "%c", c);
last = c;
}
if (in_range)
- putchar (last);
+ fprintf (stderr, "%c", last);
- putchar (']');
+ fprintf (stderr, "]");
p += 1 + length;
if (has_range_table)
{
int count;
- printf ("has-range-table");
+ fprintf (stderr, "has-range-table");
/* ??? Should print the range table; for now, just skip it. */
p += 2; /* skip range table bits */
break;
case begline:
- printf ("/begline");
+ fprintf (stderr, "/begline");
break;
case endline:
- printf ("/endline");
+ fprintf (stderr, "/endline");
break;
case on_failure_jump:
extract_number_and_incr (&mcnt, &p);
- printf ("/on_failure_jump to %d", p + mcnt - start);
+ fprintf (stderr, "/on_failure_jump to %d", p + mcnt - start);
break;
case on_failure_keep_string_jump:
extract_number_and_incr (&mcnt, &p);
- printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
+ fprintf (stderr, "/on_failure_keep_string_jump to %d", p + mcnt - start);
break;
case on_failure_jump_nastyloop:
extract_number_and_incr (&mcnt, &p);
- printf ("/on_failure_jump_nastyloop to %d", p + mcnt - start);
+ fprintf (stderr, "/on_failure_jump_nastyloop to %d", p + mcnt - start);
break;
case on_failure_jump_loop:
extract_number_and_incr (&mcnt, &p);
- printf ("/on_failure_jump_loop to %d", p + mcnt - start);
+ fprintf (stderr, "/on_failure_jump_loop to %d", p + mcnt - start);
break;
case on_failure_jump_smart:
extract_number_and_incr (&mcnt, &p);
- printf ("/on_failure_jump_smart to %d", p + mcnt - start);
+ fprintf (stderr, "/on_failure_jump_smart to %d", p + mcnt - start);
break;
case jump:
extract_number_and_incr (&mcnt, &p);
- printf ("/jump to %d", p + mcnt - start);
+ fprintf (stderr, "/jump to %d", p + mcnt - start);
break;
case succeed_n:
extract_number_and_incr (&mcnt, &p);
extract_number_and_incr (&mcnt2, &p);
- printf ("/succeed_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
+ fprintf (stderr, "/succeed_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
break;
case jump_n:
extract_number_and_incr (&mcnt, &p);
extract_number_and_incr (&mcnt2, &p);
- printf ("/jump_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
+ fprintf (stderr, "/jump_n to %d, %d times", p - 2 + mcnt - start, mcnt2);
break;
case set_number_at:
extract_number_and_incr (&mcnt, &p);
extract_number_and_incr (&mcnt2, &p);
- printf ("/set_number_at location %d to %d", p - 2 + mcnt - start, mcnt2);
+ fprintf (stderr, "/set_number_at location %d to %d", p - 2 + mcnt - start, mcnt2);
break;
case wordbound:
- printf ("/wordbound");
+ fprintf (stderr, "/wordbound");
break;
case notwordbound:
- printf ("/notwordbound");
+ fprintf (stderr, "/notwordbound");
break;
case wordbeg:
- printf ("/wordbeg");
+ fprintf (stderr, "/wordbeg");
break;
case wordend:
- printf ("/wordend");
+ fprintf (stderr, "/wordend");
+ break;
+
+ case symbeg:
+ fprintf (stderr, "/symbeg");
+ break;
+
+ case symend:
+ fprintf (stderr, "/symend");
+ break;
case syntaxspec:
- printf ("/syntaxspec");
+ fprintf (stderr, "/syntaxspec");
mcnt = *p++;
- printf ("/%d", mcnt);
+ fprintf (stderr, "/%d", mcnt);
break;
case notsyntaxspec:
- printf ("/notsyntaxspec");
+ fprintf (stderr, "/notsyntaxspec");
mcnt = *p++;
- printf ("/%d", mcnt);
+ fprintf (stderr, "/%d", mcnt);
break;
-#ifdef emacs
+# ifdef emacs
case before_dot:
- printf ("/before_dot");
+ fprintf (stderr, "/before_dot");
break;
case at_dot:
- printf ("/at_dot");
+ fprintf (stderr, "/at_dot");
break;
case after_dot:
- printf ("/after_dot");
+ fprintf (stderr, "/after_dot");
break;
case categoryspec:
- printf ("/categoryspec");
+ fprintf (stderr, "/categoryspec");
mcnt = *p++;
- printf ("/%d", mcnt);
+ fprintf (stderr, "/%d", mcnt);
break;
case notcategoryspec:
- printf ("/notcategoryspec");
+ fprintf (stderr, "/notcategoryspec");
mcnt = *p++;
- printf ("/%d", mcnt);
+ fprintf (stderr, "/%d", mcnt);
break;
-#endif /* emacs */
+# endif /* emacs */
case begbuf:
- printf ("/begbuf");
+ fprintf (stderr, "/begbuf");
break;
case endbuf:
- printf ("/endbuf");
+ fprintf (stderr, "/endbuf");
break;
default:
- printf ("?%d", *(p-1));
+ fprintf (stderr, "?%d", *(p-1));
}
- putchar ('\n');
+ fprintf (stderr, "\n");
}
- printf ("%d:\tend of pattern.\n", p - start);
+ fprintf (stderr, "%d:\tend of pattern.\n", p - start);
}
print_compiled_pattern (bufp)
struct re_pattern_buffer *bufp;
{
- unsigned char *buffer = bufp->buffer;
+ re_char *buffer = bufp->buffer;
print_partial_compiled_pattern (buffer, buffer + bufp->used);
- printf ("%ld bytes used/%ld bytes allocated.\n", bufp->used, bufp->allocated);
+ printf ("%ld bytes used/%ld bytes allocated.\n",
+ bufp->used, bufp->allocated);
if (bufp->fastmap_accurate && bufp->fastmap)
{
printf ("re_nsub: %d\t", bufp->re_nsub);
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);
printf ("no_sub: %d\t", bufp->no_sub);
printf ("not_bol: %d\t", bufp->not_bol);
printf ("not_eol: %d\t", bufp->not_eol);
- printf ("syntax: %d\n", bufp->syntax);
+ printf ("syntax: %lx\n", bufp->syntax);
fflush (stdout);
/* Perhaps we should print the translate table? */
}
int size1;
int size2;
{
- unsigned this_char;
+ int this_char;
if (where == NULL)
printf ("(null)");
#else /* not DEBUG */
-#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)
-#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 */
\f
different, incompatible syntaxes.
The argument SYNTAX is a bit mask comprised of the various bits
- defined in regex.h. We return the old syntax. */
+ defined in regex.h. We return the old syntax. */
reg_syntax_t
re_set_syntax (syntax)
- reg_syntax_t syntax;
+ reg_syntax_t syntax;
{
reg_syntax_t ret = re_syntax_options;
re_syntax_options = syntax;
return ret;
}
+WEAK_ALIAS (__re_set_syntax, re_set_syntax)
+
+/* Regexp to use to replace spaces, or NULL meaning don't. */
+static re_char *whitespace_regexp;
+
+void
+re_set_whitespace_regexp (regexp)
+ const char *regexp;
+{
+ whitespace_regexp = (re_char *) regexp;
+}
+WEAK_ALIAS (__re_set_syntax, re_set_syntax)
\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.
+ 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? */
+ but why not be nice? */
static const char *re_error_msgid[] =
{
gettext_noop ("Premature end of regular expression"), /* REG_EEND */
gettext_noop ("Regular expression too big"), /* REG_ESIZE */
gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
+ gettext_noop ("Range striding over charsets") /* REG_ERANGEX */
};
\f
-/* Avoiding alloca during matching, to placate r_alloc. */
+/* Avoiding alloca during matching, to placate r_alloc. */
/* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
searching and matching functions should not call alloca. On some
/* 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
+# 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
+ so REL_ALLOC should not affect this. */
+#if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
+# undef MATCH_MAY_ALLOCATE
#endif
\f
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 20
+# define INIT_FAILURE_ALLOC 20
#endif
/* Roughly the maximum number of failure points on the stack. Would be
exactly that if always used TYPICAL_FAILURE_SIZE items each time we failed.
This is a variable only so users of regex can assign to it; we never
- change it ourselves. */
-#if defined (MATCH_MAY_ALLOCATE)
-/* Note that 4400 is enough to cause a crash on Alpha OSF/1,
+ change it ourselves. We always multiply it by TYPICAL_FAILURE_SIZE
+ before using it, so it should probably be a byte-count instead. */
+# if defined MATCH_MAY_ALLOCATE
+/* Note that 4400 was enough to cause a crash on Alpha OSF/1,
whose default stack limit is 2mb. In order for a larger
value to work reliably, you have to try to make it accord
with the process stack limit. */
-int re_max_failures = 40000;
-#else
-int re_max_failures = 4000;
-#endif
+size_t re_max_failures = 40000;
+# else
+size_t re_max_failures = 4000;
+# endif
union fail_stack_elt
{
- const unsigned char *pointer;
- unsigned int integer;
+ re_char *pointer;
+ /* This should be the biggest `int' that's no bigger than a pointer. */
+ long integer;
};
typedef union fail_stack_elt fail_stack_elt_t;
typedef struct
{
fail_stack_elt_t *stack;
- unsigned size;
- unsigned avail; /* Offset of next open position. */
- unsigned frame; /* Offset of the cur constructed frame. */
+ size_t size;
+ size_t avail; /* Offset of next open position. */
+ size_t frame; /* Offset of the cur constructed frame. */
} fail_stack_type;
-#define PATTERN_STACK_EMPTY() (fail_stack.avail == 0)
#define FAIL_STACK_EMPTY() (fail_stack.frame == 0)
#define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
Do `return -2' if the alloc fails. */
#ifdef MATCH_MAY_ALLOCATE
-#define INIT_FAIL_STACK() \
+# define INIT_FAIL_STACK() \
do { \
fail_stack.stack = (fail_stack_elt_t *) \
REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * TYPICAL_FAILURE_SIZE \
fail_stack.frame = 0; \
} while (0)
-#define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
+# define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
#else
-#define INIT_FAIL_STACK() \
+# define INIT_FAIL_STACK() \
do { \
fail_stack.avail = 0; \
fail_stack.frame = 0; \
} while (0)
-#define RESET_FAIL_STACK() ((void)0)
+# define RESET_FAIL_STACK() ((void)0)
#endif
Return 1 if succeeds, and 0 if either ran out of memory
allocating space for it or it was already too large.
- REGEX_REALLOCATE_STACK requires `destination' be declared. */
+ REGEX_REALLOCATE_STACK requires `destination' be declared. */
/* Factor to increase the failure stack size by
when we increase it.
1)))
-/* 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) \
- ((FAIL_STACK_FULL () \
- && !GROW_FAIL_STACK (FAIL_STACK)) \
- ? 0 \
- : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
- 1))
-#define POP_PATTERN_OP() POP_FAILURE_POINTER ()
-
/* Push a pointer value onto the failure stack.
Assumes the variable `fail_stack'. Probably should only
- be called from within `PUSH_FAILURE_POINT'. */
+ be called from within `PUSH_FAILURE_POINT'. */
#define PUSH_FAILURE_POINTER(item) \
- fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
+ fail_stack.stack[fail_stack.avail++].pointer = (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'. */
+ be called from within `PUSH_FAILURE_POINT'. */
#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'. */
+ be called from within `PUSH_FAILURE_POINT'. */
#define PUSH_FAILURE_ELT(item) \
fail_stack.stack[fail_stack.avail++] = (item)
PUSH_FAILURE_INT (num); \
} while (0)
+/* Change the counter's value to VAL, but make sure that it will
+ be reset when backtracking. */
+#define PUSH_NUMBER(ptr,val) \
+do { \
+ char *destination; \
+ int c; \
+ ENSURE_FAIL_STACK(3); \
+ EXTRACT_NUMBER (c, ptr); \
+ DEBUG_PRINT4 (" Push number %p = %d -> %d\n", ptr, c, val); \
+ PUSH_FAILURE_INT (c); \
+ PUSH_FAILURE_POINTER (ptr); \
+ PUSH_FAILURE_INT (-1); \
+ STORE_NUMBER (ptr, val); \
+} while (0)
+
/* Pop a saved register off the stack. */
-#define POP_FAILURE_REG() \
+#define POP_FAILURE_REG_OR_COUNT() \
do { \
int reg = POP_FAILURE_INT (); \
- regend[reg] = POP_FAILURE_POINTER (); \
- regstart[reg] = POP_FAILURE_POINTER (); \
- DEBUG_PRINT4 (" Pop reg %d (spanning %p -> %p)\n", \
- reg, regstart[reg], regend[reg]); \
+ if (reg == -1) \
+ { \
+ /* It's a counter. */ \
+ /* Here, we discard `const', making re_match non-reentrant. */ \
+ unsigned char *ptr = (unsigned char*) POP_FAILURE_POINTER (); \
+ reg = POP_FAILURE_INT (); \
+ STORE_NUMBER (ptr, reg); \
+ DEBUG_PRINT3 (" Pop counter %p = %d\n", ptr, reg); \
+ } \
+ else \
+ { \
+ regend[reg] = POP_FAILURE_POINTER (); \
+ regstart[reg] = POP_FAILURE_POINTER (); \
+ DEBUG_PRINT4 (" Pop reg %d (spanning %p -> %p)\n", \
+ reg, regstart[reg], regend[reg]); \
+ } \
} while (0)
/* Check that we are not stuck in an infinite loop. */
#define CHECK_INFINITE_LOOP(pat_cur, string_place) \
do { \
- int failure = TOP_FAILURE_HANDLE(); \
+ int failure = TOP_FAILURE_HANDLE (); \
/* Check for infinite matching loops */ \
- while (failure > 0 && \
- (FAILURE_STR (failure) == string_place \
- || FAILURE_STR (failure) == NULL)) \
+ while (failure > 0 \
+ && (FAILURE_STR (failure) == string_place \
+ || FAILURE_STR (failure) == NULL)) \
{ \
assert (FAILURE_PAT (failure) >= bufp->buffer \
&& FAILURE_PAT (failure) <= bufp->buffer + bufp->used); \
if (FAILURE_PAT (failure) == pat_cur) \
- goto fail; \
+ { \
+ cycle = 1; \
+ break; \
+ } \
DEBUG_PRINT2 (" Other pattern: %p\n", FAILURE_PAT (failure)); \
failure = NEXT_FAILURE_HANDLE(failure); \
} \
DEBUG_PRINT2 (" Other string: %p\n", FAILURE_STR (failure)); \
} while (0)
-
+
/* Push the information about the state we will need
if we ever fail back to it.
/* Must be int, so when we don't save any registers, the arithmetic \
of 0 + -1 isn't done as unsigned. */ \
\
- DEBUG_STATEMENT (failure_id++); \
DEBUG_STATEMENT (nfailure_points_pushed++); \
- DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
+ DEBUG_PRINT1 ("\nPUSH_FAILURE_POINT:\n"); \
DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail); \
DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
\
/* Estimate the size of data pushed by a typical failure stack entry.
An estimate is all we need, because all we use this for
is to choose a limit for how big to make the failure stack. */
-
+/* BEWARE, the value `20' is hard-coded in emacs.c:main(). */
#define TYPICAL_FAILURE_SIZE 20
/* How many items can still be added to the stack without overflowing it. */
\
/* Pop the saved registers. */ \
while (fail_stack.frame < fail_stack.avail) \
- POP_FAILURE_REG (); \
+ POP_FAILURE_REG_OR_COUNT (); \
\
- pat = (unsigned char *) POP_FAILURE_POINTER (); \
+ pat = POP_FAILURE_POINTER (); \
DEBUG_PRINT2 (" Popping pattern %p: ", pat); \
DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
\
/* If the saved string location is NULL, it came from an \
on_failure_keep_string_jump opcode, and we want to throw away the \
saved NULL, thus retaining our current position in the string. */ \
- str = (re_char *) POP_FAILURE_POINTER (); \
+ str = POP_FAILURE_POINTER (); \
DEBUG_PRINT2 (" Popping string %p: `", str); \
DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
DEBUG_PRINT1 ("'\n"); \
\f
/* Registers are set to a sentinel when they haven't yet matched. */
-#define REG_UNSET_VALUE NULL
-#define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
+#define REG_UNSET(e) ((e) == NULL)
\f
/* Subroutine declarations and macros for regex_compile. */
-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 unsigned char *pattern,
- const unsigned char *p,
- reg_syntax_t syntax));
-static boolean at_endline_loc_p _RE_ARGS((const unsigned char *p,
- const unsigned char *pend,
- reg_syntax_t syntax));
-static unsigned char *skip_one_char _RE_ARGS((unsigned char *p));
-static int analyse_first _RE_ARGS((unsigned char *p, unsigned char *pend,
- char *fastmap, const int multibyte));
-
-/* 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'). */
-#define PATFETCH(c) \
- do { \
- PATFETCH_RAW (c); \
- c = TRANSLATE (c); \
- } while (0)
+static reg_errcode_t regex_compile _RE_ARGS ((re_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 ((re_char *pattern,
+ re_char *p,
+ reg_syntax_t syntax));
+static boolean at_endline_loc_p _RE_ARGS ((re_char *p,
+ re_char *pend,
+ reg_syntax_t syntax));
+static re_char *skip_one_char _RE_ARGS ((re_char *p));
+static int analyse_first _RE_ARGS ((re_char *p, re_char *pend,
+ char *fastmap, const int multibyte));
/* Fetch the next character in the uncompiled pattern, with no
- translation. */
-#define PATFETCH_RAW(c) \
+ translation. */
+#define PATFETCH(c) \
do { \
int len; \
if (p == pend) return REG_EEND; \
`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) \
+# define TRANSLATE(d) \
(RE_TRANSLATE_P (translate) ? RE_TRANSLATE (translate, (d)) : (d))
#endif
/* If the buffer isn't allocated when it comes in, use this. */
#define INIT_BUF_SIZE 32
-/* Make sure we have at least N more bytes of space in buffer. */
+/* Make sure we have at least N more bytes of space in buffer. */
#define GET_BUFFER_SPACE(n) \
- while (b - bufp->buffer + (n) > bufp->allocated) \
+ while ((size_t) (b - bufp->buffer + (n)) > bufp->allocated) \
EXTEND_BUFFER ()
/* Make sure we have one more byte of buffer space and then add C to it. */
} while (0)
-/* As with BUF_PUSH_2, except for three bytes. */
+/* As with BUF_PUSH_2, except for three bytes. */
#define BUF_PUSH_3(c1, c2, c3) \
do { \
GET_BUFFER_SPACE (3); \
/* 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. */
+ relative address offset by the three bytes the jump itself occupies. */
#define STORE_JUMP(op, loc, to) \
store_op1 (op, loc, (to) - (loc) - 3)
#define STORE_JUMP2(op, loc, to, arg) \
store_op2 (op, loc, (to) - (loc) - 3, arg)
-/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
+/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
#define INSERT_JUMP(op, loc, to) \
insert_op1 (op, loc, (to) - (loc) - 3, 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
+ into the pattern are two bytes long. So if 2^15 bytes turns out to
be too small, many things would have to change. */
-#define MAX_BUF_SIZE (1L << 16)
-
+# define MAX_BUF_SIZE (1L << 15)
+
+#if 0 /* This is when we thought it could be 2^16 bytes. */
+/* Any other compiler which, like MSC, has allocation limit below 2^16
+ bytes will have to use approach similar to what was done below for
+ 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
+/* Microsoft C 16-bit versions limit malloc to approx 65512 bytes. */
+# define MAX_BUF_SIZE 65500L
+#else
+# define MAX_BUF_SIZE (1L << 16)
+#endif
+#endif /* 0 */
/* 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. */
+ being larger than MAX_BUF_SIZE, then flag memory exhausted. */
+#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
#define EXTEND_BUFFER() \
do { \
- unsigned char *old_buffer = bufp->buffer; \
+ re_char *old_buffer = bufp->buffer; \
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 = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
+ RETALLOC (bufp->buffer, bufp->allocated, unsigned char); \
if (bufp->buffer == NULL) \
return REG_ESPACE; \
/* If the buffer moved, move all the pointers into it. */ \
if (old_buffer != bufp->buffer) \
{ \
- b = (b - old_buffer) + bufp->buffer; \
- begalt = (begalt - old_buffer) + bufp->buffer; \
+ int incr = bufp->buffer - old_buffer; \
+ MOVE_BUFFER_POINTER (b); \
+ MOVE_BUFFER_POINTER (begalt); \
if (fixup_alt_jump) \
- fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
+ MOVE_BUFFER_POINTER (fixup_alt_jump); \
if (laststart) \
- laststart = (laststart - old_buffer) + bufp->buffer; \
+ MOVE_BUFFER_POINTER (laststart); \
if (pending_exact) \
- pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
+ MOVE_BUFFER_POINTER (pending_exact); \
} \
+ ELSE_EXTEND_BUFFER_HIGH_BOUND \
} while (0)
/* But patterns can have more than `MAX_REGNUM' registers. We just
ignore the excess. */
-typedef unsigned regnum_t;
+typedef int regnum_t;
/* Macros for the compile stack. */
/* Since offsets can go either forwards or backwards, this type needs to
- be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
-typedef int pattern_offset_t;
+ be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
+/* int may be not enough when sizeof(int) == 2. */
+typedef long pattern_offset_t;
typedef struct
{
#define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
-/* The next available element. */
+/* The next available element. */
#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
-
+/* Explicit quit checking is only used on NTemacs and whenever we
+ use polling to process input events. */
+#if defined emacs && (defined WINDOWSNT || defined SYNC_INPUT) && defined QUIT
+extern int immediate_quit;
+# define IMMEDIATE_QUIT_CHECK \
+ do { \
+ if (immediate_quit) QUIT; \
+ } while (0)
+#else
+# define IMMEDIATE_QUIT_CHECK ((void)0)
+#endif
+\f
/* Structure to manage work area for range table. */
struct range_table_work_area
{
int bits; /* flag to record character classes */
};
-/* Make sure that WORK_AREA can hold more N multibyte characters. */
-#define EXTEND_RANGE_TABLE_WORK_AREA(work_area, n) \
- do { \
- if (((work_area).used + (n)) * sizeof (int) > (work_area).allocated) \
- { \
- (work_area).allocated += 16 * sizeof (int); \
- if ((work_area).table) \
- (work_area).table \
- = (int *) realloc ((work_area).table, (work_area).allocated); \
- else \
- (work_area).table \
- = (int *) malloc ((work_area).allocated); \
- if ((work_area).table == 0) \
- FREE_STACK_RETURN (REG_ESPACE); \
- } \
+/* Make sure that WORK_AREA can hold more N multibyte characters.
+ This is used only in set_image_of_range and set_image_of_range_1.
+ It expects WORK_AREA to be a pointer.
+ If it can't get the space, it returns from the surrounding function. */
+
+#define EXTEND_RANGE_TABLE(work_area, n) \
+ do { \
+ if (((work_area)->used + (n)) * sizeof (int) > (work_area)->allocated) \
+ { \
+ extend_range_table_work_area (work_area); \
+ if ((work_area)->table == 0) \
+ return (REG_ESPACE); \
+ } \
} while (0)
#define SET_RANGE_TABLE_WORK_AREA_BIT(work_area, bit) \
(work_area).bits |= (bit)
-/* These bits represent the various character classes such as [:alnum:]
- in a charset's range table. */
-#define BIT_ALNUM 0x1
-#define BIT_ALPHA 0x2
-#define BIT_WORD 0x4
-#define BIT_ASCII 0x8
-#define BIT_NONASCII 0x10
-#define BIT_GRAPH 0x20
-#define BIT_LOWER 0x40
-#define BIT_PRINT 0x80
-#define BIT_PUNCT 0x100
-#define BIT_SPACE 0x200
-#define BIT_UPPER 0x400
-#define BIT_UNIBYTE 0x800
-#define BIT_MULTIBYTE 0x1000
-
-/* Set a range (RANGE_START, RANGE_END) to WORK_AREA. */
-#define SET_RANGE_TABLE_WORK_AREA(work_area, range_start, range_end) \
+/* Bits used to implement the multibyte-part of the various character classes
+ such as [:alnum:] in a charset's range table. */
+#define BIT_WORD 0x1
+#define BIT_LOWER 0x2
+#define BIT_PUNCT 0x4
+#define BIT_SPACE 0x8
+#define BIT_UPPER 0x10
+#define BIT_MULTIBYTE 0x20
+
+/* Set a range START..END to WORK_AREA.
+ The range is passed through TRANSLATE, so START and END
+ should be untranslated. */
+#define SET_RANGE_TABLE_WORK_AREA(work_area, start, end) \
do { \
- EXTEND_RANGE_TABLE_WORK_AREA ((work_area), 2); \
- (work_area).table[(work_area).used++] = (range_start); \
- (work_area).table[(work_area).used++] = (range_end); \
+ int tem; \
+ tem = set_image_of_range (&work_area, start, end, translate); \
+ if (tem > 0) \
+ FREE_STACK_RETURN (tem); \
} while (0)
/* Free allocated memory for WORK_AREA. */
#define RANGE_TABLE_WORK_USED(work_area) ((work_area).used)
#define RANGE_TABLE_WORK_BITS(work_area) ((work_area).bits)
#define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i])
-
+\f
/* Set the bit for character C in a list. */
-#define SET_LIST_BIT(c) \
- (b[((unsigned char) (c)) / BYTEWIDTH] \
- |= 1 << (((unsigned char) c) % BYTEWIDTH))
+#define SET_LIST_BIT(c) (b[((c)) / BYTEWIDTH] |= 1 << ((c) % BYTEWIDTH))
/* Get the next unsigned number in the uncompiled pattern. */
#define GET_UNSIGNED_NUMBER(num) \
- do { if (p != pend) \
- { \
- PATFETCH (c); \
- while (ISDIGIT (c)) \
- { \
- if (num < 0) \
+ do { \
+ if (p == pend) \
+ FREE_STACK_RETURN (REG_EBRACE); \
+ else \
+ { \
+ PATFETCH (c); \
+ while ('0' <= c && c <= '9') \
+ { \
+ int prev; \
+ if (num < 0) \
num = 0; \
- num = num * 10 + c - '0'; \
- if (p == pend) \
- break; \
- PATFETCH (c); \
- } \
- } \
- } while (0)
+ prev = num; \
+ num = num * 10 + c - '0'; \
+ if (num / 10 != prev) \
+ FREE_STACK_RETURN (REG_BADBR); \
+ if (p == pend) \
+ FREE_STACK_RETURN (REG_EBRACE); \
+ PATFETCH (c); \
+ } \
+ } \
+ } while (0)
+\f
+#if ! WIDE_CHAR_SUPPORT
+
+/* Map a string to the char class it names (if any). */
+re_wctype_t
+re_wctype (str)
+ re_char *str;
+{
+ const char *string = str;
+ if (STREQ (string, "alnum")) return RECC_ALNUM;
+ else if (STREQ (string, "alpha")) return RECC_ALPHA;
+ else if (STREQ (string, "word")) return RECC_WORD;
+ else if (STREQ (string, "ascii")) return RECC_ASCII;
+ else if (STREQ (string, "nonascii")) return RECC_NONASCII;
+ else if (STREQ (string, "graph")) return RECC_GRAPH;
+ else if (STREQ (string, "lower")) return RECC_LOWER;
+ else if (STREQ (string, "print")) return RECC_PRINT;
+ else if (STREQ (string, "punct")) return RECC_PUNCT;
+ else if (STREQ (string, "space")) return RECC_SPACE;
+ else if (STREQ (string, "upper")) return RECC_UPPER;
+ else if (STREQ (string, "unibyte")) return RECC_UNIBYTE;
+ else if (STREQ (string, "multibyte")) return RECC_MULTIBYTE;
+ else if (STREQ (string, "digit")) return RECC_DIGIT;
+ else if (STREQ (string, "xdigit")) return RECC_XDIGIT;
+ else if (STREQ (string, "cntrl")) return RECC_CNTRL;
+ else if (STREQ (string, "blank")) return RECC_BLANK;
+ else return 0;
+}
+
+/* True iff CH is in the char class CC. */
+boolean
+re_iswctype (ch, cc)
+ int ch;
+ re_wctype_t cc;
+{
+ switch (cc)
+ {
+ case RECC_ALNUM: return ISALNUM (ch);
+ case RECC_ALPHA: return ISALPHA (ch);
+ case RECC_BLANK: return ISBLANK (ch);
+ case RECC_CNTRL: return ISCNTRL (ch);
+ case RECC_DIGIT: return ISDIGIT (ch);
+ case RECC_GRAPH: return ISGRAPH (ch);
+ case RECC_LOWER: return ISLOWER (ch);
+ case RECC_PRINT: return ISPRINT (ch);
+ case RECC_PUNCT: return ISPUNCT (ch);
+ case RECC_SPACE: return ISSPACE (ch);
+ case RECC_UPPER: return ISUPPER (ch);
+ case RECC_XDIGIT: return ISXDIGIT (ch);
+ case RECC_ASCII: return IS_REAL_ASCII (ch);
+ case RECC_NONASCII: return !IS_REAL_ASCII (ch);
+ case RECC_UNIBYTE: return ISUNIBYTE (ch);
+ case RECC_MULTIBYTE: return !ISUNIBYTE (ch);
+ case RECC_WORD: return ISWORD (ch);
+ case RECC_ERROR: return false;
+ default:
+ abort();
+ }
+}
+
+/* Return a bit-pattern to use in the range-table bits to match multibyte
+ chars of class CC. */
+static int
+re_wctype_to_bit (cc)
+ re_wctype_t cc;
+{
+ switch (cc)
+ {
+ case RECC_NONASCII: case RECC_PRINT: case RECC_GRAPH:
+ case RECC_MULTIBYTE: return BIT_MULTIBYTE;
+ case RECC_ALPHA: case RECC_ALNUM: case RECC_WORD: return BIT_WORD;
+ case RECC_LOWER: return BIT_LOWER;
+ case RECC_UPPER: return BIT_UPPER;
+ case RECC_PUNCT: return BIT_PUNCT;
+ case RECC_SPACE: return BIT_SPACE;
+ case RECC_ASCII: case RECC_DIGIT: case RECC_XDIGIT: case RECC_CNTRL:
+ case RECC_BLANK: case RECC_UNIBYTE: case RECC_ERROR: return 0;
+ default:
+ abort();
+ }
+}
+#endif
+\f
+/* Filling in the work area of a range. */
+
+/* Actually extend the space in WORK_AREA. */
+
+static void
+extend_range_table_work_area (work_area)
+ struct range_table_work_area *work_area;
+{
+ work_area->allocated += 16 * sizeof (int);
+ if (work_area->table)
+ work_area->table
+ = (int *) realloc (work_area->table, work_area->allocated);
+ else
+ work_area->table
+ = (int *) malloc (work_area->allocated);
+}
+
+#ifdef emacs
+
+/* Carefully find the ranges of codes that are equivalent
+ under case conversion to the range start..end when passed through
+ TRANSLATE. Handle the case where non-letters can come in between
+ two upper-case letters (which happens in Latin-1).
+ Also handle the case of groups of more than 2 case-equivalent chars.
+
+ The basic method is to look at consecutive characters and see
+ if they can form a run that can be handled as one.
+
+ Returns -1 if successful, REG_ESPACE if ran out of space. */
+
+static int
+set_image_of_range_1 (work_area, start, end, translate)
+ RE_TRANSLATE_TYPE translate;
+ struct range_table_work_area *work_area;
+ re_wchar_t start, end;
+{
+ /* `one_case' indicates a character, or a run of characters,
+ each of which is an isolate (no case-equivalents).
+ This includes all ASCII non-letters.
+
+ `two_case' indicates a character, or a run of characters,
+ each of which has two case-equivalent forms.
+ This includes all ASCII letters.
+
+ `strange' indicates a character that has more than one
+ case-equivalent. */
+
+ enum case_type {one_case, two_case, strange};
+
+ /* Describe the run that is in progress,
+ which the next character can try to extend.
+ If run_type is strange, that means there really is no run.
+ If run_type is one_case, then run_start...run_end is the run.
+ If run_type is two_case, then the run is run_start...run_end,
+ and the case-equivalents end at run_eqv_end. */
+
+ enum case_type run_type = strange;
+ int run_start, run_end, run_eqv_end;
+
+ Lisp_Object eqv_table;
+
+ if (!RE_TRANSLATE_P (translate))
+ {
+ EXTEND_RANGE_TABLE (work_area, 2);
+ work_area->table[work_area->used++] = (start);
+ work_area->table[work_area->used++] = (end);
+ return -1;
+ }
+
+ eqv_table = XCHAR_TABLE (translate)->extras[2];
+
+ for (; start <= end; start++)
+ {
+ enum case_type this_type;
+ int eqv = RE_TRANSLATE (eqv_table, start);
+ int minchar, maxchar;
+
+ /* Classify this character */
+ if (eqv == start)
+ this_type = one_case;
+ else if (RE_TRANSLATE (eqv_table, eqv) == start)
+ this_type = two_case;
+ else
+ this_type = strange;
+
+ if (start < eqv)
+ minchar = start, maxchar = eqv;
+ else
+ minchar = eqv, maxchar = start;
+
+ /* Can this character extend the run in progress? */
+ if (this_type == strange || this_type != run_type
+ || !(minchar == run_end + 1
+ && (run_type == two_case
+ ? maxchar == run_eqv_end + 1 : 1)))
+ {
+ /* No, end the run.
+ Record each of its equivalent ranges. */
+ if (run_type == one_case)
+ {
+ EXTEND_RANGE_TABLE (work_area, 2);
+ work_area->table[work_area->used++] = run_start;
+ work_area->table[work_area->used++] = run_end;
+ }
+ else if (run_type == two_case)
+ {
+ EXTEND_RANGE_TABLE (work_area, 4);
+ work_area->table[work_area->used++] = run_start;
+ work_area->table[work_area->used++] = run_end;
+ work_area->table[work_area->used++]
+ = RE_TRANSLATE (eqv_table, run_start);
+ work_area->table[work_area->used++]
+ = RE_TRANSLATE (eqv_table, run_end);
+ }
+ run_type = strange;
+ }
-#define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
-
-#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") \
- || STREQ (string, "word") \
- || STREQ (string, "ascii") || STREQ (string, "nonascii") \
- || STREQ (string, "unibyte") || STREQ (string, "multibyte"))
-
-/* QUIT is only used on NTemacs. */
-#if !defined (WINDOWSNT) || !defined (emacs)
-#undef QUIT
-#define QUIT
+ if (this_type == strange)
+ {
+ /* For a strange character, add each of its equivalents, one
+ by one. Don't start a range. */
+ do
+ {
+ EXTEND_RANGE_TABLE (work_area, 2);
+ work_area->table[work_area->used++] = eqv;
+ work_area->table[work_area->used++] = eqv;
+ eqv = RE_TRANSLATE (eqv_table, eqv);
+ }
+ while (eqv != start);
+ }
+
+ /* Add this char to the run, or start a new run. */
+ else if (run_type == strange)
+ {
+ /* Initialize a new range. */
+ run_type = this_type;
+ run_start = start;
+ run_end = start;
+ run_eqv_end = RE_TRANSLATE (eqv_table, run_end);
+ }
+ else
+ {
+ /* Extend a running range. */
+ run_end = minchar;
+ run_eqv_end = RE_TRANSLATE (eqv_table, run_end);
+ }
+ }
+
+ /* If a run is still in progress at the end, finish it now
+ by recording its equivalent ranges. */
+ if (run_type == one_case)
+ {
+ EXTEND_RANGE_TABLE (work_area, 2);
+ work_area->table[work_area->used++] = run_start;
+ work_area->table[work_area->used++] = run_end;
+ }
+ else if (run_type == two_case)
+ {
+ EXTEND_RANGE_TABLE (work_area, 4);
+ work_area->table[work_area->used++] = run_start;
+ work_area->table[work_area->used++] = run_end;
+ work_area->table[work_area->used++]
+ = RE_TRANSLATE (eqv_table, run_start);
+ work_area->table[work_area->used++]
+ = RE_TRANSLATE (eqv_table, run_end);
+ }
+
+ return -1;
+}
+
+#endif /* emacs */
+
+/* Record the the image of the range start..end when passed through
+ TRANSLATE. This is not necessarily TRANSLATE(start)..TRANSLATE(end)
+ and is not even necessarily contiguous.
+ Normally we approximate it with the smallest contiguous range that contains
+ all the chars we need. However, for Latin-1 we go to extra effort
+ to do a better job.
+
+ This function is not called for ASCII ranges.
+
+ Returns -1 if successful, REG_ESPACE if ran out of space. */
+
+static int
+set_image_of_range (work_area, start, end, translate)
+ RE_TRANSLATE_TYPE translate;
+ struct range_table_work_area *work_area;
+ re_wchar_t start, end;
+{
+ re_wchar_t cmin, cmax;
+
+#ifdef emacs
+ /* For Latin-1 ranges, use set_image_of_range_1
+ to get proper handling of ranges that include letters and nonletters.
+ For a range that includes the whole of Latin-1, this is not necessary.
+ For other character sets, we don't bother to get this right. */
+ if (RE_TRANSLATE_P (translate) && start < 04400
+ && !(start < 04200 && end >= 04377))
+ {
+ int newend;
+ int tem;
+ newend = end;
+ if (newend > 04377)
+ newend = 04377;
+ tem = set_image_of_range_1 (work_area, start, newend, translate);
+ if (tem > 0)
+ return tem;
+
+ start = 04400;
+ if (end < 04400)
+ return -1;
+ }
#endif
+
+ EXTEND_RANGE_TABLE (work_area, 2);
+ work_area->table[work_area->used++] = (start);
+ work_area->table[work_area->used++] = (end);
+
+ cmin = -1, cmax = -1;
+
+ if (RE_TRANSLATE_P (translate))
+ {
+ int ch;
+
+ for (ch = start; ch <= end; ch++)
+ {
+ re_wchar_t c = TRANSLATE (ch);
+ if (! (start <= c && c <= end))
+ {
+ if (cmin == -1)
+ cmin = c, cmax = c;
+ else
+ {
+ cmin = MIN (cmin, c);
+ cmax = MAX (cmax, c);
+ }
+ }
+ }
+
+ if (cmin != -1)
+ {
+ EXTEND_RANGE_TABLE (work_area, 2);
+ work_area->table[work_area->used++] = (cmin);
+ work_area->table[work_area->used++] = (cmax);
+ }
+ }
+
+ return -1;
+}
\f
#ifndef MATCH_MAY_ALLOCATE
/* Size with which the following vectors are currently allocated.
That is so we can make them bigger as needed,
- but never make them smaller. */
+ but never make them smaller. */
static int regs_allocated_size;
static re_char ** regstart, ** regend;
static re_char **best_regstart, **best_regend;
/* Make the register vectors big enough for NUM_REGS registers,
- but don't make them smaller. */
+ but don't make them smaller. */
static
regex_grow_registers (num_regs)
`re_nsub' is the number of subexpressions in PATTERN;
`not_bol' and `not_eol' are zero;
- The `fastmap' and `newline_anchor' fields are neither
- examined nor set. */
+ The `fastmap' field is neither examined nor set. */
/* Insert the `jump' from the end of last alternative to "here".
The space for the jump has already been allocated. */
static reg_errcode_t
regex_compile (pattern, size, syntax, bufp)
re_char *pattern;
- int size;
+ size_t 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 int c, c1;
+ /* We fetch characters from PATTERN here. */
+ register re_wchar_t c, c1;
/* A random temporary spot in PATTERN. */
re_char *p1;
/* If the object matched can contain multibyte characters. */
const boolean multibyte = RE_MULTIBYTE_P (bufp);
+ /* Nonzero if we have pushed down into a subpattern. */
+ int in_subpattern = 0;
+
+ /* These hold the values of p, pattern, and pend from the main
+ pattern when we have pushed into a subpattern. */
+ re_char *main_p;
+ re_char *main_pattern;
+ re_char *main_pend;
+
#ifdef DEBUG
debug++;
DEBUG_PRINT1 ("\nCompiling pattern: ");
/* Always count groups, whether or not bufp->no_sub is set. */
bufp->re_nsub = 0;
-#if !defined (emacs) && !defined (SYNTAX_TABLE)
+#if !defined emacs && !defined SYNTAX_TABLE
/* Initialize the syntax table. */
init_syntax_once ();
#endif
begalt = b = bufp->buffer;
/* Loop through the uncompiled pattern until we're at the end. */
- while (p != pend)
+ while (1)
{
+ if (p == pend)
+ {
+ /* If this is the end of an included regexp,
+ pop back to the main regexp and try again. */
+ if (in_subpattern)
+ {
+ in_subpattern = 0;
+ pattern = main_pattern;
+ p = main_p;
+ pend = main_pend;
+ continue;
+ }
+ /* If this is the end of the main regexp, we are done. */
+ break;
+ }
+
PATFETCH (c);
- switch (c)
- {
+ switch (c)
+ {
+ case ' ':
+ {
+ re_char *p1 = p;
+
+ /* If there's no special whitespace regexp, treat
+ spaces normally. And don't try to do this recursively. */
+ if (!whitespace_regexp || in_subpattern)
+ goto normal_char;
+
+ /* Peek past following spaces. */
+ while (p1 != pend)
+ {
+ if (*p1 != ' ')
+ break;
+ p1++;
+ }
+ /* If the spaces are followed by a repetition op,
+ treat them normally. */
+ if (p1 != pend
+ && (*p1 == '*' || *p1 == '+' || *p1 == '?'
+ || (*p1 == '\\' && p1 + 1 != pend && p1[1] == '{')))
+ goto normal_char;
+
+ /* Replace the spaces with the whitespace regexp. */
+ in_subpattern = 1;
+ main_p = p1;
+ main_pend = pend;
+ main_pattern = pattern;
+ p = pattern = whitespace_regexp;
+ pend = p + strlen (p);
+ break;
+ }
+
case '^':
{
if ( /* If at start of pattern, it's an operator. */
|| syntax & RE_CONTEXT_INDEP_ANCHORS
/* Otherwise, depends on what's come before. */
|| at_begline_loc_p (pattern, p, syntax))
- BUF_PUSH (begline);
+ BUF_PUSH ((syntax & RE_NO_NEWLINE_ANCHOR) ? begbuf : begline);
else
goto normal_char;
}
|| syntax & RE_CONTEXT_INDEP_ANCHORS
/* Otherwise, depends on what's next. */
|| at_endline_loc_p (p, pend, syntax))
- BUF_PUSH (endline);
+ BUF_PUSH ((syntax & RE_NO_NEWLINE_ANCHOR) ? endbuf : endline);
else
goto normal_char;
}
for (;;)
{
- if (!(syntax & RE_ALL_GREEDY)
+ if ((syntax & RE_FRUGAL)
&& c == '?' && (zero_times_ok || many_times_ok))
greedy = 0;
else
boolean simple = skip_one_char (laststart) == b;
unsigned int startoffset = 0;
re_opcode_t ofj =
- (simple || !analyse_first (laststart, b, NULL, 0)) ?
- on_failure_jump : on_failure_jump_loop;
+ /* Check if the loop can match the empty string. */
+ (simple || !analyse_first (laststart, b, NULL, 0))
+ ? on_failure_jump : on_failure_jump_loop;
assert (skip_one_char (laststart) <= b);
-
+
if (!zero_times_ok && simple)
{ /* Since simple * loops can be made faster by using
on_failure_keep_string_jump, we turn simple P+
{
boolean emptyp = analyse_first (laststart, b, NULL, 0);
- /* The non-greedy multiple match looks like a repeat..until:
- we only need a conditional jump at the end of the loop */
+ /* The non-greedy multiple match looks like
+ a repeat..until: we only need a conditional jump
+ at the end of the loop. */
if (emptyp) BUF_PUSH (no_op);
STORE_JUMP (emptyp ? on_failure_jump_nastyloop
: on_failure_jump, b, laststart);
{
/* The repeat...until naturally matches one or more.
To also match zero times, we need to first jump to
- the end of the loop (its conditional jump). */
+ the end of the loop (its conditional jump). */
INSERT_JUMP (jump, laststart, b);
b += 3;
}
if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+ /* Don't translate yet. The range TRANSLATE(X..Y) cannot
+ always be determined from TRANSLATE(X) and TRANSLATE(Y)
+ So the translation is done later in a loop. Example:
+ (let ((case-fold-search t)) (string-match "[A-_]" "A")) */
PATFETCH (c);
/* \ might escape characters inside [...] and [^...]. */
syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
{
/* Leave room for the null. */
- char str[CHAR_CLASS_MAX_LENGTH + 1];
+ unsigned char str[CHAR_CLASS_MAX_LENGTH + 1];
const unsigned char *class_beg;
PATFETCH (c);
for (;;)
{
- PATFETCH (c);
- if (c == ':' || c == ']' || p == pend
- || c1 == CHAR_CLASS_MAX_LENGTH)
- break;
- str[c1++] = c;
+ 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';
them). */
if (c == ':' && *p == ']')
{
- int ch;
- boolean is_alnum = STREQ (str, "alnum");
- boolean is_alpha = STREQ (str, "alpha");
- boolean is_ascii = STREQ (str, "ascii");
- boolean is_blank = STREQ (str, "blank");
- boolean is_cntrl = STREQ (str, "cntrl");
- boolean is_digit = STREQ (str, "digit");
- boolean is_graph = STREQ (str, "graph");
- boolean is_lower = STREQ (str, "lower");
- boolean is_multibyte = STREQ (str, "multibyte");
- boolean is_nonascii = STREQ (str, "nonascii");
- boolean is_print = STREQ (str, "print");
- boolean is_punct = STREQ (str, "punct");
- boolean is_space = STREQ (str, "space");
- boolean is_unibyte = STREQ (str, "unibyte");
- boolean is_upper = STREQ (str, "upper");
- boolean is_word = STREQ (str, "word");
- boolean is_xdigit = STREQ (str, "xdigit");
-
- if (!IS_CHAR_CLASS (str))
+ re_wchar_t ch;
+ re_wctype_t cc;
+
+ cc = re_wctype (str);
+
+ if (cc == 0)
FREE_STACK_RETURN (REG_ECTYPE);
- /* Throw away the ] at the end of the character
- class. */
- PATFETCH (c);
+ /* Throw away the ] at the end of the character
+ class. */
+ PATFETCH (c);
- if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
+ if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
/* Most character classes in a multibyte match
just set a flag. Exceptions are is_blank,
is_digit, is_cntrl, and is_xdigit, since
they can only match ASCII characters. We
- don't need to handle them for multibyte. */
+ don't need to handle them for multibyte.
+ They are distinguished by a negative wctype. */
if (multibyte)
- {
- int bit = 0;
-
- if (is_alnum) bit = BIT_ALNUM;
- if (is_alpha) bit = BIT_ALPHA;
- if (is_ascii) bit = BIT_ASCII;
- if (is_graph) bit = BIT_GRAPH;
- if (is_lower) bit = BIT_LOWER;
- if (is_multibyte) bit = BIT_MULTIBYTE;
- if (is_nonascii) bit = BIT_NONASCII;
- if (is_print) bit = BIT_PRINT;
- if (is_punct) bit = BIT_PUNCT;
- if (is_space) bit = BIT_SPACE;
- if (is_unibyte) bit = BIT_UNIBYTE;
- if (is_upper) bit = BIT_UPPER;
- if (is_word) bit = BIT_WORD;
- if (bit)
- SET_RANGE_TABLE_WORK_AREA_BIT (range_table_work,
- bit);
- }
+ SET_RANGE_TABLE_WORK_AREA_BIT (range_table_work,
+ re_wctype_to_bit (cc));
- /* Handle character classes for ASCII characters. */
- for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
+ for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
{
int translated = TRANSLATE (ch);
- /* This was split into 3 if's to
- avoid an arbitrary limit in some compiler. */
- if ( (is_alnum && ISALNUM (ch))
- || (is_alpha && ISALPHA (ch))
- || (is_blank && ISBLANK (ch))
- || (is_cntrl && ISCNTRL (ch)))
- SET_LIST_BIT (translated);
- if ( (is_digit && ISDIGIT (ch))
- || (is_graph && ISGRAPH (ch))
- || (is_lower && ISLOWER (ch))
- || (is_print && ISPRINT (ch)))
- SET_LIST_BIT (translated);
- if ( (is_punct && ISPUNCT (ch))
- || (is_space && ISSPACE (ch))
- || (is_upper && ISUPPER (ch))
- || (is_xdigit && ISXDIGIT (ch)))
- SET_LIST_BIT (translated);
- if ( (is_ascii && IS_REAL_ASCII (ch))
- || (is_nonascii && !IS_REAL_ASCII (ch))
- || (is_unibyte && ISUNIBYTE (ch))
- || (is_multibyte && !ISUNIBYTE (ch)))
- SET_LIST_BIT (translated);
-
- if ( (is_word && ISWORD (ch)))
+ if (re_iswctype (btowc (ch), cc))
SET_LIST_BIT (translated);
}
/* Fetch the character which ends the range. */
PATFETCH (c1);
- if (SINGLE_BYTE_CHAR_P (c)
- && ! SINGLE_BYTE_CHAR_P (c1))
+ if (SINGLE_BYTE_CHAR_P (c))
{
- /* Handle a range such as \177-\377 in multibyte mode.
- Split that into two ranges,,
- the low one ending at 0237, and the high one
- starting at ...040. */
- /* Unless I'm missing something,
- this line is useless. -sm
- int c1_base = (c1 & ~0177) | 040; */
- SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1);
- c1 = 0237;
+ if (! SINGLE_BYTE_CHAR_P (c1))
+ {
+ /* Handle a range starting with a
+ character of less than 256, and ending
+ with a character of not less than 256.
+ Split that into two ranges, the low one
+ ending at 0377, and the high one
+ starting at the smallest character in
+ the charset of C1 and ending at C1. */
+ int charset = CHAR_CHARSET (c1);
+ re_wchar_t c2 = MAKE_CHAR (charset, 0, 0);
+
+ SET_RANGE_TABLE_WORK_AREA (range_table_work,
+ c2, c1);
+ c1 = 0377;
+ }
}
else if (!SAME_CHARSET_P (c, c1))
- FREE_STACK_RETURN (REG_ERANGE);
+ FREE_STACK_RETURN (REG_ERANGEX);
}
else
/* Range from C to C. */
if (SINGLE_BYTE_CHAR_P (c))
/* ... into bitmap. */
{
- unsigned this_char;
- int range_start = c, range_end = c1;
+ re_wchar_t this_char;
+ re_wchar_t range_start = c, range_end = c1;
/* If the start is after the end, the range is empty. */
if (range_start > range_end)
/* Do not translate the character after the \, so that we can
distinguish, e.g., \B from \b, even if we normally would
translate, e.g., B to b. */
- PATFETCH_RAW (c);
+ PATFETCH (c);
switch (c)
{
if (!(syntax & RE_INTERVALS)
/* If we're at `\{' and it's not the open-interval
operator. */
- || (syntax & RE_NO_BK_BRACES)
- /* What is that? -sm */
- /* || (p - 2 == pattern && p == pend) */)
+ || (syntax & RE_NO_BK_BRACES))
goto normal_backslash;
handle_interval:
beg_interval = p;
- if (p == pend)
- {
- if (syntax & RE_NO_BK_BRACES)
- goto unfetch_interval;
- else
- FREE_STACK_RETURN (REG_EBRACE);
- }
-
GET_UNSIGNED_NUMBER (lower_bound);
if (c == ',')
if (lower_bound < 0 || upper_bound > RE_DUP_MAX
|| (upper_bound >= 0 && lower_bound > upper_bound))
- {
- if (syntax & RE_NO_BK_BRACES)
- goto unfetch_interval;
- else
- FREE_STACK_RETURN (REG_BADBR);
- }
+ FREE_STACK_RETURN (REG_BADBR);
if (!(syntax & RE_NO_BK_BRACES))
{
- if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
-
+ if (c != '\\')
+ FREE_STACK_RETURN (REG_BADBR);
+ if (p == pend)
+ FREE_STACK_RETURN (REG_EESCAPE);
PATFETCH (c);
}
if (c != '}')
- {
- if (syntax & RE_NO_BK_BRACES)
- goto unfetch_interval;
- else
- FREE_STACK_RETURN (REG_BADBR);
- }
+ FREE_STACK_RETURN (REG_BADBR);
/* We just parsed a valid interval. */
goto unfetch_interval;
}
- if (upper_bound == 0)
- /* If the upper bound is zero, just drop the sub pattern
- altogether. */
- b = laststart;
- else if (lower_bound == 1 && upper_bound == 1)
- /* Just match it once: nothing to do here. */
- ;
-
- /* Otherwise, we have a nontrivial interval. When
- we're all done, the pattern will look like:
- set_number_at <jump count> <upper bound>
- set_number_at <succeed_n count> <lower bound>
-
succeed_n <after jump addr> <succeed_n count>
- <body of loop>
- jump_n <succeed_n addr> <jump count>
- (The upper bound and `jump_n' are omitted if
- `upper_bound' is 1, though.) */
- else
- { /* If the upper bound is > 1, we need to insert
- more at the end of the loop. */
- unsigned int nbytes = (upper_bound < 0 ? 3
- : upper_bound > 1 ? 5 : 0);
- unsigned int startoffset = 0;
-
- GET_BUFFER_SPACE (20); /* We might use less. */
-
- if (lower_bound == 0)
- {
- /* A succeed_n that starts with 0 is really a
- a simple on_failure_jump_loop. */
- INSERT_JUMP (on_failure_jump_loop, laststart,
- b + 3 + nbytes);
- b += 3;
- }
- else
- {
- /* Initialize lower bound of the `succeed_n', even
- though it will be set during matching by its
- attendant `set_number_at' (inserted next),
- because `re_compile_fastmap' needs to know.
- Jump to the `jump_n' we might insert below. */
- INSERT_JUMP2 (succeed_n, laststart,
- b + 5 + nbytes,
- lower_bound);
- b += 5;
-
- /* 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;
- startoffset += 5;
- }
-
- if (upper_bound < 0)
- {
- /* A negative upper bound stands for infinity,
- in which case it degenerates to a plain jump. */
- STORE_JUMP (jump, b, laststart + startoffset);
- b += 3;
- }
- else if (upper_bound > 1)
- { /* More than one repetition is allowed, so
- append a backward jump to the `succeed_n'
- that starts this interval.
-
- 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 + startoffset,
- upper_bound - 1);
- b += 5;
-
- /* The location we want to set is the second
- parameter of the `jump_n'; that is `b-2' as
- an absolute address. `laststart' will be
- the `set_number_at' we're about to insert;
- `laststart+3' the number to set, the source
- for the relative address. But we are
- inserting into the middle of the pattern --
- so everything is getting moved up by 5.
- Conclusion: (b - 2) - (laststart + 3) + 5,
- i.e., b - laststart.
-
- 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;
- }
- }
+ if (upper_bound == 0)
+ /* If the upper bound is zero, just drop the sub pattern
+ altogether. */
+ b = laststart;
+ else if (lower_bound == 1 && upper_bound == 1)
+ /* Just match it once: nothing to do here. */
+ ;
+
+ /* Otherwise, we have a nontrivial interval. When
+ we're all done, the pattern will look like:
+ set_number_at <jump count> <upper bound>
+ set_number_at <succeed_n count> <lower bound>
+ succeed_n <after jump addr> <succeed_n count>
+ <body of loop>
+ jump_n <succeed_n addr> <jump count>
+ (The upper bound and `jump_n' are omitted if
+ `upper_bound' is 1, though.) */
+ else
+ { /* If the upper bound is > 1, we need to insert
+ more at the end of the loop. */
+ unsigned int nbytes = (upper_bound < 0 ? 3
+ : upper_bound > 1 ? 5 : 0);
+ unsigned int startoffset = 0;
+
+ GET_BUFFER_SPACE (20); /* We might use less. */
+
+ if (lower_bound == 0)
+ {
+ /* A succeed_n that starts with 0 is really a
+ a simple on_failure_jump_loop. */
+ INSERT_JUMP (on_failure_jump_loop, laststart,
+ b + 3 + nbytes);
+ b += 3;
+ }
+ else
+ {
+ /* Initialize lower bound of the `succeed_n', even
+ though it will be set during matching by its
+ attendant `set_number_at' (inserted next),
+ because `re_compile_fastmap' needs to know.
+ Jump to the `jump_n' we might insert below. */
+ INSERT_JUMP2 (succeed_n, laststart,
+ b + 5 + nbytes,
+ lower_bound);
+ b += 5;
+
+ /* 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;
+ startoffset += 5;
+ }
+
+ if (upper_bound < 0)
+ {
+ /* A negative upper bound stands for infinity,
+ in which case it degenerates to a plain jump. */
+ STORE_JUMP (jump, b, laststart + startoffset);
+ b += 3;
+ }
+ else if (upper_bound > 1)
+ { /* More than one repetition is allowed, so
+ append a backward jump to the `succeed_n'
+ that starts this interval.
+
+ 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 + startoffset,
+ upper_bound - 1);
+ b += 5;
+
+ /* The location we want to set is the second
+ parameter of the `jump_n'; that is `b-2' as
+ an absolute address. `laststart' will be
+ the `set_number_at' we're about to insert;
+ `laststart+3' the number to set, the source
+ for the relative address. But we are
+ inserting into the middle of the pattern --
+ so everything is getting moved up by 5.
+ Conclusion: (b - 2) - (laststart + 3) + 5,
+ i.e., b - laststart.
+
+ 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;
+ }
+ }
pending_exact = 0;
beg_interval = NULL;
}
case 'c':
laststart = b;
- PATFETCH_RAW (c);
+ PATFETCH (c);
BUF_PUSH_2 (categoryspec, c);
break;
case 'C':
laststart = b;
- PATFETCH_RAW (c);
+ PATFETCH (c);
BUF_PUSH_2 (notcategoryspec, c);
break;
#endif /* emacs */
case 'w':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
laststart = b;
BUF_PUSH_2 (syntaxspec, Sword);
break;
case 'W':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
laststart = b;
BUF_PUSH_2 (notsyntaxspec, Sword);
break;
case '<':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
BUF_PUSH (wordbeg);
break;
case '>':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
BUF_PUSH (wordend);
break;
+ case '_':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
+ laststart = b;
+ PATFETCH (c);
+ if (c == '<')
+ BUF_PUSH (symbeg);
+ else if (c == '>')
+ BUF_PUSH (symend);
+ else
+ FREE_STACK_RETURN (REG_BADPAT);
+ break;
+
case 'b':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
BUF_PUSH (wordbound);
break;
case 'B':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
BUF_PUSH (notwordbound);
break;
case '`':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
BUF_PUSH (begbuf);
break;
case '\'':
+ if (syntax & RE_NO_GNU_OPS)
+ goto normal_char;
BUF_PUSH (endbuf);
break;
case '1': case '2': case '3': case '4': case '5':
case '6': case '7': case '8': case '9':
- if (syntax & RE_NO_BK_REFS)
- goto normal_char;
+ {
+ regnum_t reg;
- c1 = c - '0';
+ if (syntax & RE_NO_BK_REFS)
+ goto normal_backslash;
- if (c1 > regnum)
- FREE_STACK_RETURN (REG_ESUBREG);
+ reg = c - '0';
- /* Can't back reference to a subexpression if inside of it. */
- if (group_in_compile_stack (compile_stack, c1))
- goto normal_char;
+ /* Can't back reference to a subexpression before its end. */
+ if (reg > regnum || group_in_compile_stack (compile_stack, reg))
+ FREE_STACK_RETURN (REG_ESUBREG);
- laststart = b;
- BUF_PUSH_2 (duplicate, c1);
+ laststart = b;
+ BUF_PUSH_2 (duplicate, reg);
+ }
break;
normal_backslash:
/* You might think it would be useful for \ to mean
not to translate; but if we don't translate it
- it will never match anything. */
- c = TRANSLATE (c);
+ it will never match anything. */
goto normal_char;
}
break;
default:
/* Expects the character in `c'. */
normal_char:
- /* If no exactn currently being built. */
+ /* If no exactn currently being built. */
if (!pending_exact
/* If last exactn not at current position. */
GET_BUFFER_SPACE (MAX_MULTIBYTE_LENGTH);
{
- int len = CHAR_STRING (c, b);
+ int len;
+
+ c = TRANSLATE (c);
+ if (multibyte)
+ len = CHAR_STRING (c, b);
+ else
+ *b = c, len = 1;
b += len;
(*pending_exact) += len;
}
if (syntax & RE_NO_POSIX_BACKTRACKING)
BUF_PUSH (succeed);
- free (compile_stack.stack);
-
/* We have succeeded; set the length of the buffer. */
bufp->used = b - bufp->buffer;
}
#endif /* not MATCH_MAY_ALLOCATE */
- return REG_NOERROR;
+ FREE_STACK_RETURN (REG_NOERROR);
} /* regex_compile */
\f
/* Subroutines for `regex_compile'. */
static boolean
at_begline_loc_p (pattern, p, syntax)
- const unsigned char *pattern, *p;
+ re_char *pattern, *p;
reg_syntax_t syntax;
{
- const unsigned char *prev = p - 2;
+ re_char *prev = p - 2;
boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
return
static boolean
at_endline_loc_p (p, pend, syntax)
- const unsigned char *p, *pend;
+ re_char *p, *pend;
reg_syntax_t syntax;
{
- const unsigned char *next = p;
+ re_char *next = p;
boolean next_backslash = *next == '\\';
- const unsigned char *next_next = p + 1 < pend ? p + 1 : 0;
+ re_char *next_next = p + 1 < pend ? p + 1 : 0;
return
/* Before a subexpression? */
Return 1 if p..pend might match the empty string.
Return 0 if p..pend matches at least one char.
- Return -1 if p..pend matches at least one char, but fastmap was not
- updated accurately.
- Return -2 if an error occurred. */
+ Return -1 if fastmap was not updated accurately. */
static int
analyse_first (p, pend, fastmap, multibyte)
- unsigned char *p, *pend;
+ re_char *p, *pend;
char *fastmap;
const int multibyte;
{
int j, k;
boolean not;
-#ifdef MATCH_MAY_ALLOCATE
- fail_stack_type fail_stack;
-#endif
-#ifndef REGEX_MALLOC
- char *destination;
-#endif
-
-#if defined (REL_ALLOC) && defined (REGEX_MALLOC)
- /* This holds the pointer to the failure stack, when
- it is allocated relocatably. */
- fail_stack_elt_t *failure_stack_ptr;
-#endif
-
- /* Assume that each path through the pattern can be null until
- proven otherwise. We set this false at the bottom of switch
- statement, to which we get only if a particular path doesn't
- match the empty string. */
- boolean path_can_be_null = true;
/* If all elements for base leading-codes in fastmap is set, this
flag is set true. */
assert (p);
- INIT_FAIL_STACK ();
-
/* The loop below works as follows:
- It has a working-list kept in the PATTERN_STACK and which basically
starts by only containing a pointer to the first operation.
so that `p' is monotonically increasing. More to the point, we
never set `p' (or push) anything `<= p1'. */
- /* If can_be_null is set, then the fastmap will not be used anyway. */
- while (1)
+ while (p < pend)
{
/* `p1' is used as a marker of how far back a `on_failure_jump'
can go without being ignored. It is normally equal to `p'
3..9: <body>
10: on_failure_jump 3
as used for the *? operator. */
- unsigned char *p1 = p;
-
- if (p >= pend)
- {
- if (path_can_be_null)
- return (RESET_FAIL_STACK (), 1);
-
- /* We have reached the (effective) end of pattern. */
- if (PATTERN_STACK_EMPTY ())
- return (RESET_FAIL_STACK (), 0);
-
- p = (unsigned char*) POP_PATTERN_OP ();
- path_can_be_null = true;
- continue;
- }
-
- /* We should never be about to go beyond the end of the pattern. */
- assert (p < pend);
+ re_char *p1 = p;
switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
{
case succeed:
- p = pend;
+ return 1;
continue;
case duplicate:
with `break'. */
case exactn:
- if (fastmap) fastmap[p[1]] = 1;
+ if (fastmap)
+ {
+ int c = RE_STRING_CHAR (p + 1, pend - p);
+
+ if (SINGLE_BYTE_CHAR_P (c))
+ fastmap[c] = 1;
+ else
+ fastmap[p[1]] = 1;
+ }
break;
/* We could put all the chars except for \n (and maybe \0)
but we don't bother since it is generally not worth it. */
if (!fastmap) break;
- return (RESET_FAIL_STACK (), -1);
+ return -1;
case charset_not:
int c, count;
/* Make P points the range table. `+ 2' is to skip flag
- bits for a character class. */
+ bits for a character class. */
p += CHARSET_BITMAP_SIZE (&p[-2]) + 2;
/* Extract the number of ranges in range table into COUNT. */
#else /* emacs */
/* This match depends on text properties. These end with
aborting optimizations. */
- return (RESET_FAIL_STACK (), -1);
+ return -1;
case categoryspec:
case notcategoryspec:
case notwordbound:
case wordbeg:
case wordend:
+ case symbeg:
+ case symend:
continue;
EXTRACT_NUMBER_AND_INCR (j, p);
if (p + j <= p1)
; /* Backward jump to be ignored. */
- else if (!PUSH_PATTERN_OP (p + j, fail_stack))
- return (RESET_FAIL_STACK (), -2);
+ else
+ { /* We have to look down both arms.
+ We first go down the "straight" path so as to minimize
+ stack usage when going through alternatives. */
+ int r = analyse_first (p, pend, fastmap, multibyte);
+ if (r) return r;
+ p += j;
+ }
continue;
case has already been handled, so we only need to look at the
fallthrough case. */
continue;
-
+
case succeed_n:
/* If N == 0, it should be an on_failure_jump_loop instead. */
DEBUG_STATEMENT (EXTRACT_NUMBER (j, p + 2); assert (j > 0));
/* Getting here means we have found the possible starting
characters for one path of the pattern -- and that the empty
- string does not match. We need not follow this path further.
- Instead, look at the next alternative (remembered on the
- stack), or quit if no more. The test at the top of the loop
- does these things. */
- path_can_be_null = false;
- p = pend;
+ string does not match. We need not follow this path further. */
+ return 0;
} /* while p */
- return (RESET_FAIL_STACK (), 0);
+ /* We reached the end without matching anything. */
+ return 1;
+
} /* analyse_first */
\f
/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
analysis = analyse_first (bufp->buffer, bufp->buffer + bufp->used,
fastmap, RE_MULTIBYTE_P (bufp));
- if (analysis < -1)
- return analysis;
bufp->can_be_null = (analysis != 0);
return 0;
} /* re_compile_fastmap */
regs->start = regs->end = (regoff_t *) 0;
}
}
+WEAK_ALIAS (__re_set_registers, re_set_registers)
\f
/* Searching routines. */
return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
regs, size);
}
+WEAK_ALIAS (__re_search, re_search)
+
+/* Head address of virtual concatenation of string. */
+#define HEAD_ADDR_VSTRING(P) \
+ (((P) >= size1 ? string2 : string1))
/* End address of virtual concatenation of string. */
#define STOP_ADDR_VSTRING(P) \
register RE_TRANSLATE_TYPE translate = bufp->translate;
int total_size = size1 + size2;
int endpos = startpos + range;
- int anchored_start = 0;
+ boolean anchored_start;
/* Nonzero if we have to concern multibyte character. */
const boolean multibyte = RE_MULTIBYTE_P (bufp);
/* Update the fastmap now if not correct already. */
if (fastmap && !bufp->fastmap_accurate)
- if (re_compile_fastmap (bufp) == -2)
- return -2;
+ re_compile_fastmap (bufp);
/* See whether the pattern is anchored. */
- if (bufp->buffer[0] == begline)
- anchored_start = 1;
+ anchored_start = (bufp->buffer[0] == begline);
#ifdef emacs
gl_state.object = re_match_object;
because that case doesn't repeat. */
if (anchored_start && startpos > 0)
{
- if (! (bufp->newline_anchor
- && ((startpos <= size1 ? string1[startpos - 1]
- : string2[startpos - size1 - 1])
- == '\n')))
+ if (! ((startpos <= size1 ? string1[startpos - 1]
+ : string2[startpos - size1 - 1])
+ == '\n'))
goto advance;
}
if (fastmap && startpos < total_size && !bufp->can_be_null)
{
register re_char *d;
- register unsigned int buf_ch;
+ register re_wchar_t buf_ch;
d = POS_ADDR_VSTRING (startpos);
d += buf_charlen;
}
else
- while (range > lim
- && !fastmap[RE_TRANSLATE (translate, *d)])
- {
- d++;
- range--;
- }
+ {
+ /* Convert *d to integer to shut up GCC's
+ whining about comparison that is always
+ true. */
+ int di = *d;
+
+ while (range > lim
+ && !fastmap[RE_TRANSLATE (translate, di)])
+ {
+ di = *(++d);
+ range--;
+ }
+ }
}
else
while (range > lim && !fastmap[*d])
val = re_match_2_internal (bufp, string1, size1, string2, size2,
startpos, regs, stop);
#ifndef REGEX_MALLOC
-#ifdef C_ALLOCA
+# ifdef C_ALLOCA
alloca (0);
-#endif
+# endif
#endif
if (val >= 0)
/* Update STARTPOS to the previous character boundary. */
if (multibyte)
{
- re_char *p = POS_ADDR_VSTRING (startpos);
- int len = 0;
+ re_char *p = POS_ADDR_VSTRING (startpos) + 1;
+ re_char *p0 = p;
+ re_char *phead = HEAD_ADDR_VSTRING (startpos);
/* Find the head of multibyte form. */
- while (!CHAR_HEAD_P (*p))
- p--, len++;
-
- /* Adjust it. */
-#if 0 /* XXX */
- if (MULTIBYTE_FORM_LENGTH (p, len + 1) != (len + 1))
- ;
- else
-#endif
- {
- range += len;
- if (range > 0)
- break;
+ PREV_CHAR_BOUNDARY (p, phead);
+ range += p0 - 1 - p;
+ if (range > 0)
+ break;
- startpos -= len;
- }
+ startpos -= p0 - 1 - p;
}
}
}
return -1;
} /* re_search_2 */
+WEAK_ALIAS (__re_search_2, re_search_2)
\f
/* Declarations and macros for re_match_2. */
/* End of string2 => fail. */ \
if (dend == end_match_2) \
goto fail; \
- /* End of string1 => advance to string2. */ \
+ /* End of string1 => advance to string2. */ \
d = string2; \
dend = end_match_2; \
}
/* Free everything we malloc. */
#ifdef MATCH_MAY_ALLOCATE
-#define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
-#define FREE_VARIABLES() \
+# define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
+# define FREE_VARIABLES() \
do { \
REGEX_FREE_STACK (fail_stack.stack); \
FREE_VAR (regstart); \
FREE_VAR (best_regend); \
} while (0)
#else
-#define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
+# define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
#endif /* not MATCH_MAY_ALLOCATE */
\f
/* If the operation is a match against one or more chars,
return a pointer to the next operation, else return NULL. */
-static unsigned char *
+static re_char *
skip_one_char (p)
- unsigned char *p;
+ re_char *p;
{
switch (SWITCH_ENUM_CAST (*p++))
{
case anychar:
break;
-
+
case exactn:
p += *p + 1;
break;
else
p += 1 + CHARSET_BITMAP_SIZE (p - 1);
break;
-
+
case syntaxspec:
case notsyntaxspec:
#ifdef emacs
/* Jump over non-matching operations. */
-static unsigned char *
+static re_char *
skip_noops (p, pend)
- unsigned char *p, *pend;
+ re_char *p, *pend;
{
int mcnt;
while (p < pend)
static int
mutually_exclusive_p (bufp, p1, p2)
struct re_pattern_buffer *bufp;
- unsigned char *p1, *p2;
+ re_char *p1, *p2;
{
re_opcode_t op2;
const boolean multibyte = RE_MULTIBYTE_P (bufp);
return 1;
}
break;
-
+
case endline:
- if (!bufp->newline_anchor)
- break;
- /* Fallthrough */
case exactn:
{
- register unsigned int c
+ register re_wchar_t c
= (re_opcode_t) *p2 == endline ? '\n'
- : RE_STRING_CHAR(p2 + 2, pend - p2 - 2);
+ : RE_STRING_CHAR (p2 + 2, pend - p2 - 2);
if ((re_opcode_t) *p1 == exactn)
{
break;
case charset:
- case charset_not:
{
if ((re_opcode_t) *p1 == exactn)
/* Reuse the code above. */
return mutually_exclusive_p (bufp, p2, p1);
-
/* It is hard to list up all the character in charset
P2 if it includes multibyte character. Give up in
such case. */
P2 is ASCII, it is enough to test only bitmap
table of P1. */
- if (*p1 == *p2)
+ if ((re_opcode_t) *p1 == charset)
{
int idx;
/* We win if the charset inside the loop
return 1;
}
}
- else if ((re_opcode_t) *p1 == charset
- || (re_opcode_t) *p1 == charset_not)
+ else if ((re_opcode_t) *p1 == charset_not)
{
int idx;
/* We win if the charset_not inside the loop lists
}
}
}
-
+ break;
+
+ case charset_not:
+ switch (SWITCH_ENUM_CAST (*p1))
+ {
+ case exactn:
+ case charset:
+ /* Reuse the code above. */
+ return mutually_exclusive_p (bufp, p2, p1);
+ case charset_not:
+ /* When we have two charset_not, it's very unlikely that
+ they don't overlap. The union of the two sets of excluded
+ chars should cover all possible chars, which, as a matter of
+ fact, is virtually impossible in multibyte buffers. */
+ break;
+ }
+ break;
+
case wordend:
- case notsyntaxspec:
+ return ((re_opcode_t) *p1 == syntaxspec && p1[1] == Sword);
+ case symend:
return ((re_opcode_t) *p1 == syntaxspec
- && p1[1] == (op2 == wordend ? Sword : p2[1]));
+ && (p1[1] == Ssymbol || p1[1] == Sword));
+ case notsyntaxspec:
+ return ((re_opcode_t) *p1 == syntaxspec && p1[1] == p2[1]);
case wordbeg:
- case syntaxspec:
+ return ((re_opcode_t) *p1 == notsyntaxspec && p1[1] == Sword);
+ case symbeg:
return ((re_opcode_t) *p1 == notsyntaxspec
- && p1[1] == (op2 == wordend ? Sword : p2[1]));
+ && (p1[1] == Ssymbol || p1[1] == Sword));
+ case syntaxspec:
+ return ((re_opcode_t) *p1 == notsyntaxspec && p1[1] == p2[1]);
case wordbound:
return (((re_opcode_t) *p1 == notsyntaxspec
int size, pos;
struct re_registers *regs;
{
- int result = re_match_2_internal (bufp, NULL, 0, string, size,
+ int result = re_match_2_internal (bufp, NULL, 0, (re_char*) string, size,
pos, regs, size);
+# if defined C_ALLOCA && !defined REGEX_MALLOC
alloca (0);
+# endif
return result;
}
+WEAK_ALIAS (__re_match, re_match)
#endif /* not emacs */
#ifdef emacs
SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object, charpos, 1);
#endif
- result = re_match_2_internal (bufp, string1, size1, string2, size2,
+ result = re_match_2_internal (bufp, (re_char*) string1, size1,
+ (re_char*) string2, size2,
pos, regs, stop);
+#if defined C_ALLOCA && !defined REGEX_MALLOC
alloca (0);
+#endif
return result;
}
+WEAK_ALIAS (__re_match_2, re_match_2)
/* This is a separate function so that we can force an alloca cleanup
afterwards. */
{
/* General temporaries. */
int mcnt;
+ size_t reg;
boolean not;
- unsigned char *p1;
/* Just past the end of the corresponding string. */
re_char *end1, *end2;
re_char *dfail;
/* Where we are in the pattern, and the end of the pattern. */
- unsigned char *p = bufp->buffer;
- register unsigned char *pend = p + bufp->used;
+ re_char *p = bufp->buffer;
+ re_char *pend = p + bufp->used;
/* We use this to map every character in the string. */
RE_TRANSLATE_TYPE translate = bufp->translate;
fail_stack_type fail_stack;
#endif
#ifdef DEBUG
- static unsigned failure_id = 0;
unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
#endif
-#if defined (REL_ALLOC) && defined (REGEX_MALLOC)
+#if defined REL_ALLOC && defined REGEX_MALLOC
/* This holds the pointer to the failure stack, when
it is allocated relocatably. */
fail_stack_elt_t *failure_stack_ptr;
/* We fill all the registers internally, independent of what we
return, for use in backreferences. The number here includes
an element for register zero. */
- unsigned num_regs = bufp->re_nsub + 1;
+ size_t num_regs = bufp->re_nsub + 1;
/* Information on the contents of registers. These are pointers into
the input strings; they record just what was matched (on this
/* 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. */
- for (mcnt = 1; mcnt < num_regs; mcnt++)
- regstart[mcnt] = regend[mcnt] = REG_UNSET_VALUE;
+ for (reg = 1; reg < num_regs; reg++)
+ regstart[reg] = regend[reg] = NULL;
/* We move `string1' into `string2' if the latter's empty -- but not if
`string1' is null. */
DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
- for (mcnt = 1; mcnt < num_regs; mcnt++)
+ for (reg = 1; reg < num_regs; reg++)
{
- best_regstart[mcnt] = regstart[mcnt];
- best_regend[mcnt] = regend[mcnt];
+ best_regstart[reg] = regstart[reg];
+ best_regend[reg] = regend[reg];
}
}
goto fail;
dend = ((d >= string1 && d <= end1)
? end_match_1 : end_match_2);
- for (mcnt = 1; mcnt < num_regs; mcnt++)
+ for (reg = 1; reg < num_regs; reg++)
{
- regstart[mcnt] = best_regstart[mcnt];
- regend[mcnt] = best_regend[mcnt];
+ regstart[reg] = best_regstart[reg];
+ regend[reg] = best_regend[reg];
}
}
} /* d != end_match_2 */
/* Go through the first `min (num_regs, regs->num_regs)'
registers, since that is all we initialized. */
- for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
+ for (reg = 1; reg < MIN (num_regs, regs->num_regs); reg++)
{
- if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
- regs->start[mcnt] = regs->end[mcnt] = -1;
+ if (REG_UNSET (regstart[reg]) || REG_UNSET (regend[reg]))
+ regs->start[reg] = regs->end[reg] = -1;
else
{
- regs->start[mcnt]
- = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
- regs->end[mcnt]
- = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
+ regs->start[reg]
+ = (regoff_t) POINTER_TO_OFFSET (regstart[reg]);
+ regs->end[reg]
+ = (regoff_t) POINTER_TO_OFFSET (regend[reg]);
}
}
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; mcnt < regs->num_regs; mcnt++)
- regs->start[mcnt] = regs->end[mcnt] = -1;
+ for (reg = num_regs; reg < regs->num_regs; reg++)
+ regs->start[reg] = regs->end[reg] = -1;
} /* regs && !bufp->no_sub */
DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
else
do
{
+ /* Avoid compiler whining about comparison being
+ always true. */
+ int di;
+
PREFETCH ();
- if (RE_TRANSLATE (translate, *d) != *p++)
+ di = *d;
+ if (RE_TRANSLATE (translate, di) != *p++)
{
d = dfail;
goto fail;
case anychar:
{
int buf_charlen;
- unsigned int buf_ch;
+ re_wchar_t buf_ch;
DEBUG_PRINT1 ("EXECUTING anychar.\n");
/* Start of actual range_table, or end of bitmap if there is no
range table. */
- unsigned char *range_table;
+ re_char *range_table;
/* Nonzero if there is a range table. */
int range_table_exists;
{
int class_bits = CHARSET_RANGE_TABLE_BITS (&p[-1]);
- if ( (class_bits & BIT_ALNUM && ISALNUM (c))
- | (class_bits & BIT_ALPHA && ISALPHA (c))
- | (class_bits & BIT_ASCII && IS_REAL_ASCII (c))
- | (class_bits & BIT_GRAPH && ISGRAPH (c))
- | (class_bits & BIT_LOWER && ISLOWER (c))
- | (class_bits & BIT_MULTIBYTE && !ISUNIBYTE (c))
- | (class_bits & BIT_NONASCII && !IS_REAL_ASCII (c))
- | (class_bits & BIT_PRINT && ISPRINT (c))
+ if ( (class_bits & BIT_LOWER && ISLOWER (c))
+ | (class_bits & BIT_MULTIBYTE)
| (class_bits & BIT_PUNCT && ISPUNCT (c))
| (class_bits & BIT_SPACE && ISSPACE (c))
- | (class_bits & BIT_UNIBYTE && ISUNIBYTE (c))
| (class_bits & BIT_UPPER && ISUPPER (c))
| (class_bits & BIT_WORD && ISWORD (c)))
not = !not;
PUSH_FAILURE_REG ((unsigned int)*p);
regstart[*p] = d;
- regend[*p] = REG_UNSET_VALUE; /* probably unnecessary. -sm */
+ regend[*p] = NULL; /* probably unnecessary. -sm */
DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
/* Move past the register number and inner group count. */
assert (!REG_UNSET (regstart[*p]));
/* Strictly speaking, there should be code such as:
-
- assert (REG_UNSET (regend[*p]));
+
+ assert (REG_UNSET (regend[*p]));
PUSH_FAILURE_REGSTOP ((unsigned int)*p);
But the only info to be pushed is regend[*p] and it is known to
past them. */
if (RE_TRANSLATE_P (translate)
? bcmp_translate (d, d2, mcnt, translate, multibyte)
- : bcmp (d, d2, mcnt))
+ : memcmp (d, d2, mcnt))
{
d = dfail;
goto fail;
/* 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. */
+ (unless `not_bol' is set in `bufp'), and after newlines. */
case begline:
DEBUG_PRINT1 ("EXECUTING begline.\n");
{
unsigned char c;
GET_CHAR_BEFORE_2 (c, d, string1, end1, string2, end2);
- if (c == '\n' && bufp->newline_anchor)
+ if (c == '\n')
break;
}
/* In all other cases, we fail. */
else
{
PREFETCH_NOLIMIT ();
- if (*d == '\n' && bufp->newline_anchor)
+ if (*d == '\n')
break;
}
goto fail;
cycle detection cannot work. Worse yet, such a detection
can not only fail to detect a cycle, but it can also wrongly
detect a cycle (between different instantiations of the same
- loop.
+ loop).
So the method used for those nasty loops is a little different:
We use a special cycle-detection-stack-frame which is pushed
when the on_failure_jump_nastyloop failure-point is *popped*.
mcnt, p + mcnt);
assert ((re_opcode_t)p[-4] == no_op);
- CHECK_INFINITE_LOOP (p - 4, d);
- PUSH_FAILURE_POINT (p - 3, d);
+ {
+ int cycle = 0;
+ CHECK_INFINITE_LOOP (p - 4, d);
+ if (!cycle)
+ /* If there's a cycle, just continue without pushing
+ this failure point. The failure point is the "try again"
+ option, which shouldn't be tried.
+ We want (x?)*?y\1z to match both xxyz and xxyxz. */
+ PUSH_FAILURE_POINT (p - 3, d);
+ }
break;
-
/* Simple loop detecting on_failure_jump: just check on the
failure stack if the same spot was already hit earlier. */
case on_failure_jump_loop:
EXTRACT_NUMBER_AND_INCR (mcnt, p);
DEBUG_PRINT3 ("EXECUTING on_failure_jump_loop %d (to %p):\n",
mcnt, p + mcnt);
-
- CHECK_INFINITE_LOOP (p - 3, d);
- PUSH_FAILURE_POINT (p - 3, d);
+ {
+ int cycle = 0;
+ CHECK_INFINITE_LOOP (p - 3, d);
+ if (cycle)
+ /* If there's a cycle, get out of the loop, as if the matching
+ had failed. We used to just `goto fail' here, but that was
+ aborting the search a bit too early: we want to keep the
+ empty-loop-match and keep matching after the loop.
+ We want (x?)*y\1z to match both xxyz and xxyxz. */
+ p += mcnt;
+ else
+ PUSH_FAILURE_POINT (p - 3, d);
+ }
break;
the repetition text and either the following jump or
pop_failure_jump back to this on_failure_jump. */
case on_failure_jump:
- QUIT;
EXTRACT_NUMBER_AND_INCR (mcnt, p);
DEBUG_PRINT3 ("EXECUTING on_failure_jump %d (to %p):\n",
mcnt, p + mcnt);
then we can use a non-backtracking loop based on
on_failure_keep_string_jump instead of on_failure_jump. */
case on_failure_jump_smart:
- QUIT;
EXTRACT_NUMBER_AND_INCR (mcnt, p);
DEBUG_PRINT3 ("EXECUTING on_failure_jump_smart %d (to %p).\n",
mcnt, p + mcnt);
{
- unsigned char *p1 = p; /* Next operation. */
- unsigned char *p2 = p + mcnt; /* Destination of the jump. */
+ re_char *p1 = p; /* Next operation. */
+ /* Here, we discard `const', making re_match non-reentrant. */
+ unsigned char *p2 = (unsigned char*) p + mcnt; /* Jump dest. */
+ unsigned char *p3 = (unsigned char*) p - 3; /* opcode location. */
p -= 3; /* Reset so that we will re-execute the
instruction once it's been changed. */
{
/* Use a fast `on_failure_keep_string_jump' loop. */
DEBUG_PRINT1 (" smart exclusive => fast loop.\n");
- *p = (unsigned char) on_failure_keep_string_jump;
+ *p3 = (unsigned char) on_failure_keep_string_jump;
STORE_NUMBER (p2 - 2, mcnt + 3);
}
else
{
/* Default to a safe `on_failure_jump' loop. */
DEBUG_PRINT1 (" smart default => slow loop.\n");
- *p = (unsigned char) on_failure_jump;
+ *p3 = (unsigned char) on_failure_jump;
}
DEBUG_STATEMENT (debug -= 2);
}
/* Unconditionally jump (without popping any failure points). */
case jump:
unconditional_jump:
- QUIT;
+ IMMEDIATE_QUIT_CHECK;
EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
p += mcnt; /* Do the jump. */
/* Have to succeed matching what follows at least n times.
After that, handle like `on_failure_jump'. */
case succeed_n:
+ /* Signedness doesn't matter since we only compare MCNT to 0. */
EXTRACT_NUMBER (mcnt, p + 2);
DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
- assert (mcnt >= 0);
- /* Originally, this is how many times we HAVE to succeed. */
- if (mcnt > 0)
- {
- mcnt--;
- p += 2;
- STORE_NUMBER_AND_INCR (p, mcnt);
- DEBUG_PRINT3 (" Setting %p to %d.\n", p, mcnt);
- }
- else if (mcnt == 0)
+ /* Originally, mcnt is how many times we HAVE to succeed. */
+ if (mcnt != 0)
{
- DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
- p[2] = (unsigned char) no_op;
- p[3] = (unsigned char) no_op;
- goto on_failure;
+ /* Here, we discard `const', making re_match non-reentrant. */
+ unsigned char *p2 = (unsigned char*) p + 2; /* counter loc. */
+ mcnt--;
+ p += 4;
+ PUSH_NUMBER (p2, mcnt);
}
+ else
+ /* The two bytes encoding mcnt == 0 are two no_op opcodes. */
+ goto on_failure;
break;
case jump_n:
+ /* Signedness doesn't matter since we only compare MCNT to 0. */
EXTRACT_NUMBER (mcnt, p + 2);
DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
/* Originally, this is how many times we CAN jump. */
- if (mcnt)
+ if (mcnt != 0)
{
- mcnt--;
- STORE_NUMBER (p + 2, mcnt);
- goto unconditional_jump;
+ /* Here, we discard `const', making re_match non-reentrant. */
+ unsigned char *p2 = (unsigned char*) p + 2; /* counter loc. */
+ mcnt--;
+ PUSH_NUMBER (p2, mcnt);
+ goto unconditional_jump;
}
/* If don't have to jump any more, skip over the rest of command. */
else
case set_number_at:
{
+ unsigned char *p2; /* Location of the counter. */
DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
EXTRACT_NUMBER_AND_INCR (mcnt, p);
- p1 = p + mcnt;
+ /* Here, we discard `const', making re_match non-reentrant. */
+ p2 = (unsigned char*) p + mcnt;
+ /* Signedness doesn't matter since we only copy MCNT's bits . */
EXTRACT_NUMBER_AND_INCR (mcnt, p);
- DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
- STORE_NUMBER (p1, mcnt);
+ DEBUG_PRINT3 (" Setting %p to %d.\n", p2, mcnt);
+ PUSH_NUMBER (p2, mcnt);
break;
}
{
/* C1 is the character before D, S1 is the syntax of C1, C2
is the character at D, and S2 is the syntax of C2. */
- int c1, c2, s1, s2;
+ re_wchar_t c1, c2;
+ int s1, s2;
#ifdef emacs
int offset = PTR_TO_OFFSET (d - 1);
int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
{
/* C1 is the character before D, S1 is the syntax of C1, C2
is the character at D, and S2 is the syntax of C2. */
- int c1, c2, s1, s2;
+ re_wchar_t c1, c2;
+ int s1, s2;
#ifdef emacs
int offset = PTR_TO_OFFSET (d);
int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
PREFETCH ();
c2 = RE_STRING_CHAR (d, dend - d);
s2 = SYNTAX (c2);
-
+
/* Case 2: S2 is not Sword. */
if (s2 != Sword)
goto fail;
{
/* C1 is the character before D, S1 is the syntax of C1, C2
is the character at D, and S2 is the syntax of C2. */
- int c1, c2, s1, s2;
+ re_wchar_t c1, c2;
+ int s1, s2;
#ifdef emacs
int offset = PTR_TO_OFFSET (d) - 1;
int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
PREFETCH_NOLIMIT ();
c2 = RE_STRING_CHAR (d, dend - d);
#ifdef emacs
- UPDATE_SYNTAX_TABLE_FORWARD (charpos);
+ UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1);
#endif
s2 = SYNTAX (c2);
}
break;
+ case symbeg:
+ DEBUG_PRINT1 ("EXECUTING symbeg.\n");
+
+ /* We FAIL in one of the following cases: */
+
+ /* Case 1: D is at the end of string. */
+ if (AT_STRINGS_END (d))
+ goto fail;
+ else
+ {
+ /* C1 is the character before D, S1 is the syntax of C1, C2
+ is the character at D, and S2 is the syntax of C2. */
+ re_wchar_t c1, c2;
+ int s1, s2;
+#ifdef emacs
+ int offset = PTR_TO_OFFSET (d);
+ int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
+ UPDATE_SYNTAX_TABLE (charpos);
+#endif
+ PREFETCH ();
+ c2 = RE_STRING_CHAR (d, dend - d);
+ s2 = SYNTAX (c2);
+
+ /* Case 2: S2 is neither Sword nor Ssymbol. */
+ if (s2 != Sword && s2 != Ssymbol)
+ goto fail;
+
+ /* Case 3: D is not at the beginning of string ... */
+ if (!AT_STRINGS_BEG (d))
+ {
+ GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
+#ifdef emacs
+ UPDATE_SYNTAX_TABLE_BACKWARD (charpos - 1);
+#endif
+ s1 = SYNTAX (c1);
+
+ /* ... and S1 is Sword or Ssymbol. */
+ if (s1 == Sword || s1 == Ssymbol)
+ goto fail;
+ }
+ }
+ break;
+
+ case symend:
+ DEBUG_PRINT1 ("EXECUTING symend.\n");
+
+ /* We FAIL in one of the following cases: */
+
+ /* Case 1: D is at the beginning of string. */
+ if (AT_STRINGS_BEG (d))
+ goto fail;
+ else
+ {
+ /* C1 is the character before D, S1 is the syntax of C1, C2
+ is the character at D, and S2 is the syntax of C2. */
+ re_wchar_t c1, c2;
+ int s1, s2;
+#ifdef emacs
+ int offset = PTR_TO_OFFSET (d) - 1;
+ int charpos = SYNTAX_TABLE_BYTE_TO_CHAR (offset);
+ UPDATE_SYNTAX_TABLE (charpos);
+#endif
+ GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
+ s1 = SYNTAX (c1);
+
+ /* Case 2: S1 is neither Ssymbol nor Sword. */
+ if (s1 != Sword && s1 != Ssymbol)
+ goto fail;
+
+ /* Case 3: D is not at the end of string ... */
+ if (!AT_STRINGS_END (d))
+ {
+ PREFETCH_NOLIMIT ();
+ c2 = RE_STRING_CHAR (d, dend - d);
+#ifdef emacs
+ UPDATE_SYNTAX_TABLE_FORWARD (charpos + 1);
+#endif
+ s2 = SYNTAX (c2);
+
+ /* ... and S2 is Sword or Ssymbol. */
+ if (s2 == Sword || s2 == Ssymbol)
+ goto fail;
+ }
+ }
+ break;
+
case syntaxspec:
case notsyntaxspec:
not = (re_opcode_t) *(p - 1) == notsyntaxspec;
}
#endif
{
- int c, len;
+ int len;
+ re_wchar_t c;
c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
DEBUG_PRINT3 ("EXECUTING %scategoryspec %d.\n", not?"not":"", mcnt);
PREFETCH ();
{
- int c, len;
+ int len;
+ re_wchar_t c;
+
c = RE_STRING_CHAR_AND_LENGTH (d, dend - d, len);
if ((!CHAR_HAS_CATEGORY (c, mcnt)) ^ not)
#endif /* emacs */
- default:
- abort ();
+ default:
+ abort ();
}
continue; /* Successfully executed one pattern command; keep going. */
/* We goto here if a matching operation fails. */
fail:
- QUIT;
+ IMMEDIATE_QUIT_CHECK;
if (!FAIL_STACK_EMPTY ())
{
- re_char *str;
- unsigned char *pat;
+ re_char *str, *pat;
/* A restart point is known. Restore to that state. */
- DEBUG_PRINT1 ("\nFAIL:\n");
- POP_FAILURE_POINT (str, pat);
+ DEBUG_PRINT1 ("\nFAIL:\n");
+ POP_FAILURE_POINT (str, pat);
switch (SWITCH_ENUM_CAST ((re_opcode_t) *pat++))
{
case on_failure_keep_string_jump:
assert (p >= bufp->buffer && p <= pend);
- if (d >= string1 && d <= end1)
+ if (d >= string1 && d <= end1)
dend = end_match_1;
- }
+ }
else
- break; /* Matching at this starting point really fails. */
+ break; /* Matching at this starting point really fails. */
} /* for (;;) */
if (best_regs_set)
re_char *p1_end = s1 + len;
re_char *p2_end = s2 + len;
- while (p1 != p1_end && p2 != p2_end)
+ /* FIXME: Checking both p1 and p2 presumes that the two strings might have
+ different lengths, but relying on a single `len' would break this. -sm */
+ while (p1 < p1_end && p2 < p2_end)
{
int p1_charlen, p2_charlen;
- int p1_ch, p2_ch;
+ re_wchar_t p1_ch, p2_ch;
p1_ch = RE_STRING_CHAR_AND_LENGTH (p1, p1_end - p1, p1_charlen);
p2_ch = RE_STRING_CHAR_AND_LENGTH (p2, p2_end - p2, p2_charlen);
const char *
re_compile_pattern (pattern, length, bufp)
const char *pattern;
- int length;
+ size_t length;
struct re_pattern_buffer *bufp;
{
reg_errcode_t ret;
setting no_sub. */
bufp->no_sub = 0;
- /* Match anchors at newline. */
- bufp->newline_anchor = 1;
-
- ret = regex_compile (pattern, length, re_syntax_options, bufp);
+ ret = regex_compile ((re_char*) pattern, length, re_syntax_options, bufp);
if (!ret)
return NULL;
return gettext (re_error_msgid[(int) ret]);
}
+WEAK_ALIAS (__re_compile_pattern, re_compile_pattern)
\f
/* Entry points compatible with 4.2 BSD regex library. We don't define
them unless specifically requested. */
-#if defined (_REGEX_RE_COMP) || defined (_LIBC)
+#if defined _REGEX_RE_COMP || defined _LIBC
/* BSD has one and only one pattern buffer. */
static struct re_pattern_buffer re_comp_buf;
char *
-#ifdef _LIBC
+# ifdef _LIBC
/* Make these definitions weak in libc, so POSIX programs can redefine
these names if they don't use our functions, and still use
regcomp/regexec below without link errors. */
weak_function
-#endif
+# endif
re_comp (s)
const char *s;
{
if (!s)
{
if (!re_comp_buf.buffer)
- return gettext ("No previous regular expression");
+ /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
+ return (char *) gettext ("No previous regular expression");
return 0;
}
{
re_comp_buf.buffer = (unsigned char *) malloc (200);
if (re_comp_buf.buffer == NULL)
- return gettext (re_error_msgid[(int) REG_ESPACE]);
+ /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
+ return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
re_comp_buf.allocated = 200;
re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
if (re_comp_buf.fastmap == NULL)
- return gettext (re_error_msgid[(int) REG_ESPACE]);
+ /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
+ return (char *) gettext (re_error_msgid[(int) REG_ESPACE]);
}
/* Since `re_exec' always passes NULL for the `regs' argument, we
don't need to initialize the pattern buffer fields which affect it. */
- /* Match anchors at newlines. */
- re_comp_buf.newline_anchor = 1;
-
ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
if (!ret)
int
-#ifdef _LIBC
+# ifdef _LIBC
weak_function
-#endif
+# endif
re_exec (s)
const char *s;
{
`syntax' to RE_SYNTAX_POSIX_EXTENDED if the
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.
int
regcomp (preg, pattern, cflags)
- regex_t *preg;
- const char *pattern;
+ regex_t *__restrict preg;
+ const char *__restrict pattern;
int cflags;
{
reg_errcode_t ret;
- unsigned syntax
+ reg_syntax_t syntax
= (cflags & REG_EXTENDED) ?
RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
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)
{
= (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
* sizeof (*(RE_TRANSLATE_TYPE)0));
if (preg->translate == NULL)
- return (int) REG_ESPACE;
+ 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;
{ /* REG_NEWLINE implies neither . nor [^...] match newline. */
syntax &= ~RE_DOT_NEWLINE;
syntax |= RE_HAT_LISTS_NOT_NEWLINE;
- /* It also changes the matching behavior. */
- preg->newline_anchor = 1;
}
else
- preg->newline_anchor = 0;
+ syntax |= RE_NO_NEWLINE_ANCHOR;
preg->no_sub = !!(cflags & REG_NOSUB);
/* 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);
+ ret = regex_compile ((re_char*) 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_ERPAREN)
+ ret = REG_EPAREN;
+
+ if (ret == REG_NOERROR && preg->fastmap)
+ { /* Compute the fastmap now, since regexec cannot modify the pattern
+ buffer. */
+ re_compile_fastmap (preg);
+ if (preg->can_be_null)
+ { /* The fastmap can't be used anyway. */
+ free (preg->fastmap);
+ preg->fastmap = NULL;
+ }
+ }
return (int) ret;
}
+WEAK_ALIAS (__regcomp, regcomp)
/* regexec searches for a given pattern, specified by PREG, in the
int
regexec (preg, string, nmatch, pmatch, eflags)
- const regex_t *preg;
- const char *string;
+ const regex_t *__restrict preg;
+ const char *__restrict string;
size_t nmatch;
- regmatch_t pmatch[];
+ regmatch_t pmatch[__restrict_arr];
int eflags;
{
int ret;
struct re_registers regs;
regex_t private_preg;
int len = strlen (string);
- boolean want_reg_info = !preg->no_sub && nmatch > 0;
+ boolean want_reg_info = !preg->no_sub && nmatch > 0 && pmatch;
private_preg = *preg;
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)
- return (int) REG_NOMATCH;
+ regs.start = TALLOC (nmatch * 2, regoff_t);
+ if (regs.start == NULL)
+ return (int) REG_NOMATCH;
+ regs.end = regs.start + nmatch;
}
+ /* Instead of using not_eol to implement REG_NOTEOL, we could simply
+ pass (&private_preg, string, len + 1, 0, len, ...) pretending the string
+ was a little bit longer but still only matching the real part.
+ This works because the `endline' will check for a '\n' and will find a
+ '\0', correctly deciding that this is not the end of a line.
+ But it doesn't work out so nicely for REG_NOTBOL, since we don't have
+ a convenient '\0' there. For all we know, the string could be preceded
+ by '\n' which would throw things off. */
+
/* Perform the searching operation. */
ret = re_search (&private_preg, string, len,
- /* start: */ 0, /* range: */ len,
- want_reg_info ? ®s : (struct re_registers *) 0);
+ /* start: */ 0, /* range: */ len,
+ want_reg_info ? ®s : (struct re_registers *) 0);
/* Copy the register information to the POSIX structure. */
if (want_reg_info)
{
if (ret >= 0)
- {
- unsigned r;
+ {
+ unsigned r;
- for (r = 0; r < nmatch; r++)
- {
- pmatch[r].rm_so = regs.start[r];
- pmatch[r].rm_eo = regs.end[r];
- }
- }
+ for (r = 0; r < nmatch; r++)
+ {
+ pmatch[r].rm_so = regs.start[r];
+ pmatch[r].rm_eo = regs.end[r];
+ }
+ }
/* 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'. */
return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
}
+WEAK_ALIAS (__regexec, regexec)
/* Returns a message corresponding to an error code, ERRCODE, returned
if (errbuf_size != 0)
{
if (msg_size > errbuf_size)
- {
- strncpy (errbuf, msg, errbuf_size - 1);
- errbuf[errbuf_size - 1] = 0;
- }
+ {
+ strncpy (errbuf, msg, errbuf_size - 1);
+ errbuf[errbuf_size - 1] = 0;
+ }
else
- strcpy (errbuf, msg);
+ strcpy (errbuf, msg);
}
return msg_size;
}
+WEAK_ALIAS (__regerror, regerror)
/* Free dynamically allocated space used by PREG. */
free (preg->translate);
preg->translate = NULL;
}
+WEAK_ALIAS (__regfree, regfree)
#endif /* not emacs */
+
+/* arch-tag: 4ffd68ba-2a9e-435b-a21a-018990f9eeb2
+ (do not change this comment) */