/* $OpenLDAP$ */
-/*
- * Copyright 1998-2000 The OpenLDAP Foundation, Redwood City, California, USA
+/* This work is part of OpenLDAP Software <http://www.openldap.org/>.
+ *
+ * Copyright 1998-2004 The OpenLDAP Foundation.
* All rights reserved.
*
- * Redistribution and use in source and binary forms are permitted only
- * as authorized by the OpenLDAP Public License. A copy of this
- * license is available at http://www.OpenLDAP.org/license.html or
- * in file LICENSE in the top-level directory of the distribution.
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted only as authorized by the OpenLDAP
+ * Public License.
+ *
+ * A copy of this license is available in file LICENSE in the
+ * top-level directory of the distribution or, alternatively, at
+ * <http://www.OpenLDAP.org/license.html>.
*/
+
/*
* ldap_pvt_uc.h - Header for Unicode functions.
* These are meant to be used by the OpenLDAP distribution only.
+ * These should be named ldap_pvt_....()
*/
#ifndef _LDAP_PVT_UC_H
#define _LDAP_PVT_UC_H 1
-#include <ldap_cdefs.h>
#include <lber.h> /* get ber_slen_t */
-#ifdef _MSC_VER
+#include <ac/bytes.h>
#include "../libraries/liblunicode/ucdata/ucdata.h"
-#else
-#include "../libraries/liblunicode/ucdata.h"
-#endif
LDAP_BEGIN_DECL
-/*
+/*
* UTF-8 (in utf-8.c)
*/
-typedef ber_int_t ldap_ucs4_t;
-#define LDAP_UCS4_INVALID (0x80000000U)
-
typedef short ldap_ucs2_t;
/* UCDATA uses UCS-2 passed in an unsigned long */
typedef unsigned long ldap_unicode_t;
-
-/* conversion routines */
-LDAP_F( ldap_ucs4_t ) ldap_utf8_to_ucs4( const char * p );
-LDAP_F( int ) ldap_ucs4_to_utf8( ldap_ucs4_t c, char *buf );
-
#define ldap_utf8_to_unicode( p ) ldap_utf8_to_ucs4((p))
#define ldap_unicode_to_utf8( c, buf ) ldap_ucs4_to_ucs4((c),(buf))
+/* Convert a string with csize octets per character to UTF-8 */
+LDAP_F( int ) ldap_ucs_to_utf8s LDAP_P((
+ struct berval *ucs, int csize, struct berval *utf8s ));
+
+
/* returns the number of bytes in the UTF-8 string */
LDAP_F (ber_len_t) ldap_utf8_bytes( const char * );
/* returns the number of UTF-8 characters in the string */
LDAP_F (int) ldap_utf8_offset( const char * );
/* returns the length (in bytes) indicated by the UTF-8 character */
LDAP_F (int) ldap_utf8_charlen( const char * );
+
+/* returns the length (in bytes) indicated by the UTF-8 character
+ * also checks that shortest possible encoding was used
+ */
+LDAP_F (int) ldap_utf8_charlen2( const char * );
+
/* copies a UTF-8 character and returning number of bytes copied */
LDAP_F (int) ldap_utf8_copy( char *, const char *);
LDAP_F (char*) ldap_utf8_strtok( char* sp, const char* sep, char **last);
/* Optimizations */
-#define LDAP_UTF8_ISASCII(p) ( * (const unsigned char *) (p) < 0x80 )
+LDAP_V (const char) ldap_utf8_lentab[128];
+LDAP_V (const char) ldap_utf8_mintab[32];
+
+#define LDAP_UTF8_ISASCII(p) ( !(*(const unsigned char *)(p) & 0x80 ) )
#define LDAP_UTF8_CHARLEN(p) ( LDAP_UTF8_ISASCII(p) \
- ? 1 : ldap_utf8_charlen((p)) )
+ ? 1 : ldap_utf8_lentab[*(const unsigned char *)(p) ^ 0x80] )
+
+/* This is like CHARLEN but additionally validates to make sure
+ * the char used the shortest possible encoding.
+ * 'l' is used to temporarily hold the result of CHARLEN.
+ */
+#define LDAP_UTF8_CHARLEN2(p, l) ( ( ( l = LDAP_UTF8_CHARLEN( p )) < 3 || \
+ ( ldap_utf8_mintab[*(const unsigned char *)(p) & 0x1f] & (p)[1] ) ) ? \
+ l : 0 )
+
#define LDAP_UTF8_OFFSET(p) ( LDAP_UTF8_ISASCII(p) \
? 1 : ldap_utf8_offset((p)) )
-#define LDAP_UTF8_COPY(d,s) ( LDAP_UTF8_ISASCII(s) \
+#define LDAP_UTF8_COPY(d,s) ( LDAP_UTF8_ISASCII(s) \
? (*(d) = *(s), 1) : ldap_utf8_copy((d),(s)) )
#define LDAP_UTF8_NEXT(p) ( LDAP_UTF8_ISASCII(p) \
ldap_unicode_t *,
ber_len_t );
-#define UTF8_CASEFOLD 1
-#define UTF8_NOCASEFOLD 0
-
-LDAP_LUNICODE_F(char *) UTF8normalize(
- const char *,
- char );
-
-LDAP_LUNICODE_F(int) UTF8normcmp(
- const char *,
- const char *,
- char );
+#define LDAP_UTF8_NOCASEFOLD 0x0U
+#define LDAP_UTF8_CASEFOLD 0x1U
+#define LDAP_UTF8_ARG1NFC 0x2U
+#define LDAP_UTF8_ARG2NFC 0x4U
+#define LDAP_UTF8_APPROX 0x8U
+
+LDAP_LUNICODE_F(struct berval *) UTF8bvnormalize(
+ struct berval *,
+ struct berval *,
+ unsigned,
+ void *memctx );
+
+LDAP_LUNICODE_F(int) UTF8bvnormcmp(
+ struct berval *,
+ struct berval *,
+ unsigned,
+ void *memctx );
LDAP_END_DECL