Welcome to mirror list, hosted at ThFree Co, Russian Federation.

cygwin.com/git/newlib-cygwin.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorCorinna Vinschen <corinna@vinschen.de>2010-01-16 18:11:56 +0300
committerCorinna Vinschen <corinna@vinschen.de>2010-01-16 18:11:56 +0300
commit011560c87dcf1b338fe60748c4e8330ed994e2fe (patch)
tree9638b0b379ec370bbf76299ec6b9e4ef80b27db4 /winsup/cygwin/libc
parent1b526b480b08c75483b56ff34bab8f0d643081cc (diff)
* libc/fnmatch.c: Replace with multibyte capable version from FreeBSD.
Diffstat (limited to 'winsup/cygwin/libc')
-rw-r--r--winsup/cygwin/libc/fnmatch.c213
1 files changed, 145 insertions, 68 deletions
diff --git a/winsup/cygwin/libc/fnmatch.c b/winsup/cygwin/libc/fnmatch.c
index 32b741c86..295f4c413 100644
--- a/winsup/cygwin/libc/fnmatch.c
+++ b/winsup/cygwin/libc/fnmatch.c
@@ -1,5 +1,3 @@
-/* $OpenBSD: fnmatch.c,v 1.7 2000/03/23 19:13:51 millert Exp $ */
-
/*
* Copyright (c) 1989, 1993, 1994
* The Regents of the University of California. All rights reserved.
@@ -33,57 +31,94 @@
*/
#if defined(LIBC_SCCS) && !defined(lint)
-#if 0
static char sccsid[] = "@(#)fnmatch.c 8.2 (Berkeley) 4/16/94";
-#else
-static char rcsid[] = "$OpenBSD: fnmatch.c,v 1.7 2000/03/23 19:13:51 millert Exp $";
-#endif
#endif /* LIBC_SCCS and not lint */
+#include <sys/cdefs.h>
+#if 0
+__FBSDID("$FreeBSD: src/lib/libc/gen/fnmatch.c,v 1.18 2007/01/09 00:27:53 imp Exp $");
+#endif
/*
* Function fnmatch() as specified in POSIX 1003.2-1992, section B.6.
* Compares a filename or pathname to a pattern.
*/
-/* Just this single line added for Cygwin. */
-#include "winsup.h"
+/*
+ * Some notes on multibyte character support:
+ * 1. Patterns with illegal byte sequences match nothing.
+ * 2. Illegal byte sequences in the "string" argument are handled by treating
+ * them as single-byte characters with a value of the first byte of the
+ * sequence cast to wchar_t.
+ * 3. Multibyte conversion state objects (mbstate_t) are passed around and
+ * used for most, but not all, conversions. Further work will be required
+ * to support state-dependent encodings.
+ */
-#include <ctype.h>
-#include <stdio.h>
-#include <string.h>
#include <fnmatch.h>
+#include <limits.h>
+#include <string.h>
+#include <wchar.h>
+#include <wctype.h>
+
+#include "../posix/collate.h"
#define EOS '\0'
-#define RANGE_MATCH 1
-#define RANGE_NOMATCH 0
-#define RANGE_ERROR (-1)
+#define RANGE_MATCH 1
+#define RANGE_NOMATCH 0
+#define RANGE_ERROR (-1)
-static int rangematch __P((const char *, char, int, char **));
+static int rangematch(const char *, wchar_t, int, char **, mbstate_t *);
+static int fnmatch1(const char *, const char *, int, mbstate_t, mbstate_t);
int
-fnmatch(const char *pattern, const char *string, int flags)
+fnmatch(pattern, string, flags)
+ const char *pattern, *string;
+ int flags;
+{
+ static const mbstate_t initial;
+
+ return (fnmatch1(pattern, string, flags, initial, initial));
+}
+
+static int
+fnmatch1(pattern, string, flags, patmbs, strmbs)
+ const char *pattern, *string;
+ int flags;
+ mbstate_t patmbs, strmbs;
{
const char *stringstart;
char *newp;
- char c, test;
+ char c;
+ wchar_t pc, sc;
+ size_t pclen, sclen;
- for (stringstart = string;;)
- switch (c = *pattern++) {
+ for (stringstart = string;;) {
+ pclen = mbrtowc(&pc, pattern, MB_LEN_MAX, &patmbs);
+ if (pclen == (size_t)-1 || pclen == (size_t)-2)
+ return (FNM_NOMATCH);
+ pattern += pclen;
+ sclen = mbrtowc(&sc, string, MB_LEN_MAX, &strmbs);
+ if (sclen == (size_t)-1 || sclen == (size_t)-2) {
+ sc = (unsigned char)*string;
+ sclen = 1;
+ memset(&strmbs, 0, sizeof(strmbs));
+ }
+ switch (pc) {
case EOS:
- if ((flags & FNM_LEADING_DIR) && *string == '/')
+ if ((flags & FNM_LEADING_DIR) && sc == '/')
return (0);
- return (*string == EOS ? 0 : FNM_NOMATCH);
+ return (sc == EOS ? 0 : FNM_NOMATCH);
case '?':
- if (*string == EOS)
+ if (sc == EOS)
return (FNM_NOMATCH);
- if (*string == '/' && (flags & FNM_PATHNAME))
+ if (sc == '/' && (flags & FNM_PATHNAME))
return (FNM_NOMATCH);
- if (*string == '.' && (flags & FNM_PERIOD) &&
+ if (sc == '.' && (flags & FNM_PERIOD) &&
(string == stringstart ||
((flags & FNM_PATHNAME) && *(string - 1) == '/')))
return (FNM_NOMATCH);
- ++string;
+ string += sclen;
break;
case '*':
c = *pattern;
@@ -91,81 +126,104 @@ fnmatch(const char *pattern, const char *string, int flags)
while (c == '*')
c = *++pattern;
- if (*string == '.' && (flags & FNM_PERIOD) &&
+ if (sc == '.' && (flags & FNM_PERIOD) &&
(string == stringstart ||
((flags & FNM_PATHNAME) && *(string - 1) == '/')))
return (FNM_NOMATCH);
/* Optimize for pattern with * at end or before /. */
- if (c == EOS) {
+ if (c == EOS)
if (flags & FNM_PATHNAME)
return ((flags & FNM_LEADING_DIR) ||
strchr(string, '/') == NULL ?
0 : FNM_NOMATCH);
else
return (0);
- } else if (c == '/' && (flags & FNM_PATHNAME)) {
+ else if (c == '/' && flags & FNM_PATHNAME) {
if ((string = strchr(string, '/')) == NULL)
return (FNM_NOMATCH);
break;
}
/* General case, use recursion. */
- while ((test = *string) != EOS) {
- if (!fnmatch(pattern, string, flags & ~FNM_PERIOD))
+ while (sc != EOS) {
+ if (!fnmatch1(pattern, string,
+ flags & ~FNM_PERIOD, patmbs, strmbs))
return (0);
- if (test == '/' && (flags & FNM_PATHNAME))
+ sclen = mbrtowc(&sc, string, MB_LEN_MAX,
+ &strmbs);
+ if (sclen == (size_t)-1 ||
+ sclen == (size_t)-2) {
+ sc = (unsigned char)*string;
+ sclen = 1;
+ memset(&strmbs, 0, sizeof(strmbs));
+ }
+ if (sc == '/' && flags & FNM_PATHNAME)
break;
- ++string;
+ string += sclen;
}
return (FNM_NOMATCH);
case '[':
- if (*string == EOS)
+ if (sc == EOS)
return (FNM_NOMATCH);
- if (*string == '/' && (flags & FNM_PATHNAME))
+ if (sc == '/' && (flags & FNM_PATHNAME))
return (FNM_NOMATCH);
- if (*string == '.' && (flags & FNM_PERIOD) &&
+ if (sc == '.' && (flags & FNM_PERIOD) &&
(string == stringstart ||
((flags & FNM_PATHNAME) && *(string - 1) == '/')))
return (FNM_NOMATCH);
- switch (rangematch(pattern, *string, flags, &newp)) {
+ switch (rangematch(pattern, sc, flags, &newp,
+ &patmbs)) {
case RANGE_ERROR:
- /* not a good range, treat as normal text */
- goto normal;
+ goto norm;
case RANGE_MATCH:
pattern = newp;
break;
case RANGE_NOMATCH:
return (FNM_NOMATCH);
}
- ++string;
+ string += sclen;
break;
case '\\':
if (!(flags & FNM_NOESCAPE)) {
- if ((c = *pattern++) == EOS) {
- c = '\\';
- --pattern;
- }
+ pclen = mbrtowc(&pc, pattern, MB_LEN_MAX,
+ &patmbs);
+ if (pclen == (size_t)-1 || pclen == (size_t)-2)
+ return (FNM_NOMATCH);
+ if (pclen == 0)
+ pc = '\\';
+ pattern += pclen;
}
/* FALLTHROUGH */
default:
- normal:
- if (c != *string && !((flags & FNM_CASEFOLD) &&
- (tolower((unsigned char)c) ==
- tolower((unsigned char)*string))))
+ norm:
+ if (pc == sc)
+ ;
+ else if ((flags & FNM_CASEFOLD) &&
+ (towlower(pc) == towlower(sc)))
+ ;
+ else
return (FNM_NOMATCH);
- ++string;
+ string += sclen;
break;
}
+ }
/* NOTREACHED */
}
static int
-rangematch(const char *pattern, char test, int flags, char **newp)
+rangematch(pattern, test, flags, newp, patmbs)
+ const char *pattern;
+ wchar_t test;
+ int flags;
+ char **newp;
+ mbstate_t *patmbs;
{
int negate, ok;
- char c, c2;
+ wchar_t c, c2;
+ size_t pclen;
+ const char *origpat;
/*
* A bracket expression starting with an unquoted circumflex
@@ -174,11 +232,11 @@ rangematch(const char *pattern, char test, int flags, char **newp)
* consistency with the regular expression syntax.
* J.T. Conklin (conklin@ngai.kaleida.com)
*/
- if ((negate = (*pattern == '!' || *pattern == '^')))
+ if ( (negate = (*pattern == '!' || *pattern == '^')) )
++pattern;
if (flags & FNM_CASEFOLD)
- test = tolower((unsigned char)test);
+ test = towlower(test);
/*
* A right bracket shall lose its special meaning and represent
@@ -186,30 +244,49 @@ rangematch(const char *pattern, char test, int flags, char **newp)
* -- POSIX.2 2.8.3.2
*/
ok = 0;
- c = *pattern++;
- do {
- if (c == '\\' && !(flags & FNM_NOESCAPE))
- c = *pattern++;
- if (c == EOS)
+ origpat = pattern;
+ for (;;) {
+ if (*pattern == ']' && pattern > origpat) {
+ pattern++;
+ break;
+ } else if (*pattern == '\0') {
return (RANGE_ERROR);
- if (c == '/' && (flags & FNM_PATHNAME))
+ } else if (*pattern == '/' && (flags & FNM_PATHNAME)) {
+ return (RANGE_NOMATCH);
+ } else if (*pattern == '\\' && !(flags & FNM_NOESCAPE))
+ pattern++;
+ pclen = mbrtowc(&c, pattern, MB_LEN_MAX, patmbs);
+ if (pclen == (size_t)-1 || pclen == (size_t)-2)
return (RANGE_NOMATCH);
- if ((flags & FNM_CASEFOLD))
- c = tolower((unsigned char)c);
- if (*pattern == '-'
- && (c2 = *(pattern+1)) != EOS && c2 != ']') {
- pattern += 2;
- if (c2 == '\\' && !(flags & FNM_NOESCAPE))
- c2 = *pattern++;
+ pattern += pclen;
+
+ if (flags & FNM_CASEFOLD)
+ c = towlower(c);
+
+ if (*pattern == '-' && *(pattern + 1) != EOS &&
+ *(pattern + 1) != ']') {
+ if (*++pattern == '\\' && !(flags & FNM_NOESCAPE))
+ if (*pattern != EOS)
+ pattern++;
+ pclen = mbrtowc(&c2, pattern, MB_LEN_MAX, patmbs);
+ if (pclen == (size_t)-1 || pclen == (size_t)-2)
+ return (RANGE_NOMATCH);
+ pattern += pclen;
if (c2 == EOS)
return (RANGE_ERROR);
+
if (flags & FNM_CASEFOLD)
- c2 = tolower((unsigned char)c2);
- if (c <= test && test <= c2)
+ c2 = towlower(c2);
+
+ if (__collate_load_error ?
+ c <= test && test <= c2 :
+ __collate_range_cmp(c, test) <= 0
+ && __collate_range_cmp(test, c2) <= 0
+ )
ok = 1;
} else if (c == test)
ok = 1;
- } while ((c = *pattern++) != ']');
+ }
*newp = (char *)pattern;
return (ok == negate ? RANGE_NOMATCH : RANGE_MATCH);