44
55#[ deny( missing_docs) ]
66#[ deny( clippy:: all) ]
7+ use bit_field:: BitField ;
78
89/// Possible errors returned by the API.
910#[ derive( Debug , Copy , Clone ) ]
9192 Ok ( ( ) )
9293}
9394
95+ /// Decode an input UCS-2 string into a UTF-8 string.
96+ ///
97+ /// The returned `usize` represents the length of the returned buffer,
98+ /// in bytes.
99+ pub fn decode ( input : & [ u16 ] , output : & mut [ u8 ] ) -> Result < usize > {
100+ let buffer_size = output. len ( ) ;
101+ let mut i = 0 ;
102+
103+ for & ch in input. iter ( ) {
104+ /*
105+ * We need to find how many bytes of UTF-8 this UCS-2 code-point needs. Because UCS-2 can only encode
106+ * the Basic Multilingual Plane, a maximum of three bytes are needed.
107+ */
108+ if ( 0x0000 ..0x0080 ) . contains ( & ch) {
109+ // Can be encoded in a single byte
110+ if i >= buffer_size {
111+ return Err ( Error :: BufferOverflow ) ;
112+ }
113+
114+ output[ i] = ch as u8 ;
115+ i += 1 ;
116+ } else if ( 0x0080 ..0x0800 ) . contains ( & ch) {
117+ // Can be encoded as two bytes
118+ if ( i + 1 ) >= buffer_size {
119+ return Err ( Error :: BufferOverflow ) ;
120+ }
121+
122+ output[ i] = 0b11000000 + ch. get_bits ( 6 ..11 ) as u8 ;
123+ output[ i + 1 ] = 0b10000000 + ch. get_bits ( 0 ..6 ) as u8 ;
124+ i += 2 ;
125+ } else {
126+ // Can be encoded as three bytes
127+ if ( i + 2 ) >= buffer_size {
128+ return Err ( Error :: BufferOverflow ) ;
129+ }
130+
131+ output[ i] = 0b11100000 + ch. get_bits ( 12 ..16 ) as u8 ;
132+ output[ i + 1 ] = 0b10000000 + ch. get_bits ( 6 ..12 ) as u8 ;
133+ output[ i + 2 ] = 0b10000000 + ch. get_bits ( 0 ..6 ) as u8 ;
134+ i += 3 ;
135+ }
136+ }
137+
138+ Ok ( i)
139+ }
140+
94141#[ cfg( test) ]
95142mod tests {
96143 use super :: * ;
@@ -105,4 +152,17 @@ mod tests {
105152
106153 assert_eq ! ( buffer[ ..] , [ 0x0151 , 0x044D , 0x254B ] ) ;
107154 }
155+
156+ #[ test]
157+ fn decoding ( ) {
158+ let input = "$¢ह한" ;
159+ let mut u16_buffer = [ 0u16 ; 4 ] ;
160+ let result = encode ( input, & mut u16_buffer) ;
161+ assert_eq ! ( result. unwrap( ) , 4 ) ;
162+
163+ let mut u8_buffer = [ 0u8 ; 9 ] ;
164+ let result = decode ( & u16_buffer, & mut u8_buffer) ;
165+ assert_eq ! ( result. unwrap( ) , 9 ) ;
166+ assert_eq ! ( core:: str :: from_utf8( & u8_buffer[ 0 ..9 ] ) , Ok ( "$¢ह한" ) ) ;
167+ }
108168}
0 commit comments