From 4e3c4d8c6d60237ead5a1134c426754b9f632298 Mon Sep 17 00:00:00 2001
From: Kurt Zeilenga <kurt@openldap.org>
Date: Sat, 22 Jan 2000 20:55:43 +0000
Subject: [PATCH] Add initial (untested) implementations of ldap_utf8_strtok,
 ldap_utf8_strcspn, ldap_utf8_strtok.

---
 libraries/libldap/ldap-int.h |  4 ++
 libraries/libldap/utf-8.c    | 93 ++++++++++++++++++++++++++++++++++--
 2 files changed, 93 insertions(+), 4 deletions(-)

diff --git a/libraries/libldap/ldap-int.h b/libraries/libldap/ldap-int.h
index 46425fc5d2..e161a6d52b 100644
--- a/libraries/libldap/ldap-int.h
+++ b/libraries/libldap/ldap-int.h
@@ -524,6 +524,10 @@ LIBLDAP_F (int) ldap_utf8_isdigit( const char * );
 LIBLDAP_F (int) ldap_utf8_isxdigit( const char * );
 LIBLDAP_F (int) ldap_utf8_isspace( const char * );
 
+LIBLDAP_F (ber_len_t) ldap_utf8_strcspn( const char* str, const char *set);
+LIBLDAP_F (ber_len_t) ldap_utf8_strspn( const char* str, const char *set);
+LIBLDAP_F (char*) ldap_utf8_strtok( char* sp, const char* sep, char **last);
+
 LDAP_END_DECL
 
 #endif /* _LDAP_INT_H */
diff --git a/libraries/libldap/utf-8.c b/libraries/libldap/utf-8.c
index 152bd8efd5..5eb46ce325 100644
--- a/libraries/libldap/utf-8.c
+++ b/libraries/libldap/utf-8.c
@@ -33,15 +33,15 @@
  */
 ber_len_t ldap_utf8_bytes( const char * p )
 {
-	ber_len_t bytes;
+	ber_len_t bytes = 0;
 
-	if( p == NULL ) return 0;
+	if( p == NULL ) return bytes;
 
-	for( bytes=0; p[bytes] ; bytes++ ) {
+	while( p[bytes++] ) {
 		/* EMPTY */ ;
 	}
 
-	return ++bytes;
+	return bytes;
 }
 
 ber_len_t ldap_utf8_chars( const char * p )
@@ -327,3 +327,88 @@ char* ldap_utf8_fgetc( FILE *s, char *buf )
 
 	return buf;
 }
+
+ber_len_t (ldap_utf8_strcspn)( const char *str, const char *set )
+{
+	int len;
+	const char *cstr;
+
+	for( cstr = str; *cstr != '\0'; cstr += len ) {
+		const char *cset;
+
+		for( cset = set; ; cset += len ) {
+			if( ldap_utf8_to_ucs4( cstr ) == ldap_utf8_to_ucs4( cset ) ) {
+				return cstr - str;
+			} 
+
+			len = ldap_utf8_charlen(cset);
+			if( !len ) break;
+		}
+
+		len = ldap_utf8_charlen(cstr);
+		if( !len ) break;
+	}
+
+	return cstr - str;
+}
+
+ber_len_t (ldap_utf8_strspn)( const char *str, const char *set )
+{
+	int len;
+	const char *cstr;
+
+	for( cstr = str; *cstr != '\0'; cstr += len ) {
+		const char *cset;
+
+		for( cset = set; ; cset += len ) {
+			if( *cset == '\0' ) {
+				return cstr - str;
+			}
+
+			if( ldap_utf8_to_ucs4( cstr ) == ldap_utf8_to_ucs4( cset ) ) {
+				break;
+			} 
+
+			len = ldap_utf8_charlen(cset);
+			if( !len ) break;
+		}
+
+		len = ldap_utf8_charlen(cstr);
+		if( !len ) break;
+	}
+
+	return cstr - str;
+}
+
+char *(ldap_utf8_strtok)(char *str, const char *sep, char **last)
+{
+	char *begin;
+	char *end;
+
+	if( last == NULL ) return NULL;
+
+	begin = str ? str : *last;
+
+	begin += ldap_utf8_strspn( begin, sep );
+
+	if( *begin == '\0' ) {
+		*last = NULL;
+		return NULL;
+	}
+
+	end = &begin[ ldap_utf8_strcpn( begin, sep ) ];
+
+	if( *end != '\0' ) {
+		int len = ldap_utf8_charlen( end );
+		*end = '\0';
+
+		if( len ) {
+			end += len;
+		} else {
+			end = NULL;
+		}
+	}
+
+	*last = end;
+	return begin;
+}
-- 
2.39.5