//! Incremental UTF-8 decoder with Node `StringDecoder` semantics: incomplete trailing
//! sequences are kept for the next chunk, invalid bytes become U+FFFD.

#[derive(Default)]
pub struct Utf8Decoder {
    pending: Vec<u8>,
}

fn sequence_length(first: u8) -> usize {
    match first {
        0x00..=0x7f => 1,
        0xc2..=0xdf => 2,
        0xe0..=0xef => 3,
        0xf0..=0xf4 => 4,
        _ => 0,
    }
}

impl Utf8Decoder {
    pub fn write(&mut self, chunk: &[u8]) -> String {
        let mut bytes = std::mem::take(&mut self.pending);
        bytes.extend_from_slice(chunk);
        // Keep an incomplete (but so far valid) sequence at the end for the next chunk.
        let mut cut = bytes.len();
        for back in 1..=3.min(bytes.len()) {
            let index = bytes.len() - back;
            let byte = bytes[index];
            if byte & 0xc0 == 0x80 {
                continue;
            }
            let needed = sequence_length(byte);
            if needed > back {
                cut = index;
            }
            break;
        }
        self.pending = bytes.split_off(cut);
        String::from_utf8_lossy(&bytes).into_owned()
    }

    pub fn end(&mut self) -> String {
        let rest = std::mem::take(&mut self.pending);
        if rest.is_empty() {
            String::new()
        } else {
            String::from_utf8_lossy(&rest).into_owned()
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn split_multibyte() {
        let text = "Příliš žluťoučký kůň 🐎";
        let bytes = text.as_bytes();
        for split in 0..bytes.len() {
            let mut decoder = Utf8Decoder::default();
            let mut out = decoder.write(&bytes[..split]);
            out.push_str(&decoder.write(&bytes[split..]));
            out.push_str(&decoder.end());
            assert_eq!(out, text, "split at {split}");
        }
        let mut decoder = Utf8Decoder::default();
        assert_eq!(decoder.write(&[0xff, b'a']), "\u{fffd}a");
        assert_eq!(decoder.write(&[0xe2, 0x82]), "");
        assert_eq!(decoder.end(), "\u{fffd}");
    }
}