use std::io::IoResult;
static BADCHAR: char = '\uFFFD';
pub struct UTF8Reader<T: Reader>
{
stream: T,
}
fn tochar(codepoint: u32) -> char
{
match ::std::char::from_u32(codepoint)
{
Some(c) => c,
None => BADCHAR,
}
}
impl<T:Reader> UTF8Reader<T>
{
pub fn new(reader: T) -> UTF8Reader<T>
{
UTF8Reader {
stream: reader,
}
}
pub fn getc(&mut self) -> IoResult<char>
{
let ch1 = try!(self.stream.read_byte()) as u32;
if ch1 & 0xC0 == 0x80 {
return Ok( BADCHAR )
}
if ch1 & 0x80 == 0x00
{
Ok( tochar(ch1) )
}
else if ch1 & 0xE0 == 0xC0
{
let ch2 = try!(self.stream.read_byte()) as u32;
if ch2 & 0xC0 != 0x80 {
return Ok( BADCHAR );
}
let ret = (ch1 & 0x1F << 6) | (ch2 & 0x3F << 0);
Ok( tochar(ret) )
}
else if ch1 & 0xF0 == 0xE0
{
let ch2 = try!(self.stream.read_byte()) as u32;
if ch2 & 0xC0 != 0x80 {
return Ok( BADCHAR );
}
let ch3 = try!(self.stream.read_byte()) as u32;
if ch3 & 0xC0 != 0x80 {
return Ok( BADCHAR );
}
let ret = (ch1 & 0x0F << 12) | (ch2 & 0x3F << 6) | (ch3 & 0x3F << 0);
Ok( tochar(ret) )
}
else if ch1 & 0xF8 == 0xF0
{
let ch2 = try!(self.stream.read_byte()) as u32;
if ch2 & 0xC0 != 0x80 {
return Ok( BADCHAR );
}
let ch3 = try!(self.stream.read_byte()) as u32;
if ch3 & 0xC0 != 0x80 {
return Ok( BADCHAR );
}
let ch4 = try!(self.stream.read_byte()) as u32;
if ch4 & 0xC0 != 0x80 {
return Ok( BADCHAR );
}
let ret = (ch1 & 0x07 << 18) | (ch2 & 0x3F << 12) | (ch3 & 0x3F << 6) | (ch4 & 0x3F << 0);
Ok( tochar(ret) )
}
else
{
Ok( BADCHAR )
}
}
}
impl<T:Reader> Iterator<IoResult<char>> for UTF8Reader<T>
{
fn next(&mut self) -> Option<IoResult<char>>
{
match self.getc()
{
Ok(c) => Some( Ok(c) ),
Err(e) => match e.kind {
::std::io::EndOfFile => None,
_ => Some( Err( e ) ),
}
}
}
}
#[test]
fn it_works() {
}