Fix an edge case in chat::DecodeUtf16::size_hint

There are cases, when data in the buf might or might not be an error.
This commit is contained in:
Maybe Waffle 2022-01-30 15:32:21 +03:00
parent 2c97d1012e
commit 17cd2cd592
2 changed files with 20 additions and 11 deletions

View File

@ -121,23 +121,31 @@ fn next(&mut self) -> Option<Result<char, DecodeUtf16Error>> {
fn size_hint(&self) -> (usize, Option<usize>) {
let (low, high) = self.iter.size_hint();
// If
// - `self.buf` contains a non surrogate (`u < 0xD800 || 0xDFFF < u`), or
// - `high == Some(0)` (and `self.buf` contains a leading surrogate since
// it can never contain a trailing surrogate)
//
// then buf contains an additional character or error that doesn't
// need a pair from `self.iter`, so it's +1 additional element.
let addition_from_buf =
self.buf.map_or(false, |u| u < 0xD800 || 0xDFFF < u || high == Some(0)) as usize;
let (low_buf, high_buf) = match self.buf {
// buf is empty, no additional elements from it.
None => (0, 0),
// `u` is a non surrogate, so it's always an additional character.
Some(u) if u < 0xD800 || 0xDFFF < u => (1, 1),
// `u` is a leading surrogate (it can never be a trailing surrogate and
// it's a surrogate due to the previous branch) and `self.iter` is empty.
//
// `u` can't be paired, since the `self.iter` is empty,
// so it will always become an additional element (error).
Some(_u) if high == Some(0) => (1, 1),
// `u` is a leading surrogate and `iter` may be non-empty.
//
// `u` can either pair with a trailing surrogate, in which case no additional elements
// are produced, or it can become an error, in which case it's an additional character (error).
Some(_u) => (0, 1),
};
// `self.iter` could contain entirely valid surrogates (2 elements per
// char), or entirely non-surrogates (1 element per char).
//
// On odd lower bound, at least one element must stay unpaired
// (with other elements from `self.iter`), so we round up.
let low = low.div_ceil(2) + addition_from_buf;
let high = high.and_then(|h| h.checked_add(addition_from_buf));
let low = low.div_ceil(2) + low_buf;
let high = high.and_then(|h| h.checked_add(high_buf));
(low, high)
}

View File

@ -329,6 +329,7 @@ fn check(s: &[u16]) {
}
check(&[0xD800, 0xD800, 0xDC00]);
check(&[0xD800, 0xD800, 0x0]);
check(&[0xD800, 0x41, 0x42]);
check(&[0xD800, 0]);
check(&[0xD834, 0x006d]);