diff --git a/crates/pi-natives/src/utok/bpe.rs b/crates/pi-natives/src/utok/bpe.rs index 21bd02363..3ef8d44df 100644 --- a/crates/pi-natives/src/utok/bpe.rs +++ b/crates/pi-natives/src/utok/bpe.rs @@ -339,14 +339,14 @@ fn piece_bytes<'a, U: Unit>(piece: &'a [U], buf: &'a mut Vec) -> &'a [u8] { // (dominant for code/English u16 input, cf. xutf's ASCII kernels; // the trivial loop autovectorizes). if let Some(b) = piece[i].ascii() { - buf.push(b); - i += 1; - } else { - let (c, n) = U::decode(piece, i); - i += n; - let mut tmp = [0u8; 4]; - buf.extend_from_slice(c.encode_utf8(&mut tmp).as_bytes()); - } + buf.push(b); + i += 1; + } else { + let (c, n) = U::decode(piece, i); + i += n; + let mut tmp = [0u8; 4]; + buf.extend_from_slice(c.encode_utf8(&mut tmp).as_bytes()); + } } buf } diff --git a/crates/pi-natives/src/utok/scan/deepseek.rs b/crates/pi-natives/src/utok/scan/deepseek.rs index e53db3eae..2eb020eb5 100644 --- a/crates/pi-natives/src/utok/scan/deepseek.rs +++ b/crates/pi-natives/src/utok/scan/deepseek.rs @@ -129,9 +129,10 @@ fn main_end(units: &[U], pos: usize) -> Option { if c != '\r' && c != '\n' && !is_ps(c) { // Not L (checked above), not P/S, not CR/LF: prefix-eligible. if let Some((c2, n2)) = decode_at(units, pos + n) - && is_lm(c2) { - return Some(lm_run_end(units, pos + n + n2)); - } + && is_lm(c2) + { + return Some(lm_run_end(units, pos + n + n2)); + } } // ` ?[\p{P}\p{S}]+[\r\n]*`