core/char/
decode.rs

1//! UTF-8 and UTF-16 decoding iterators
2
3use crate::error::Error;
4use crate::fmt;
5use crate::iter::FusedIterator;
6
7/// An iterator that decodes UTF-16 encoded code points from an iterator of `u16`s.
8///
9/// This `struct` is created by the [`decode_utf16`] method on [`char`]. See its
10/// documentation for more.
11///
12/// [`decode_utf16`]: char::decode_utf16
13#[stable(feature = "decode_utf16", since = "1.9.0")]
14#[derive(Clone, Debug)]
15#[ferrocene::prevalidated]
16pub struct DecodeUtf16<I>
17where
18    I: Iterator<Item = u16>,
19{
20    iter: I,
21    buf: Option<u16>,
22}
23
24/// An error that can be returned when decoding UTF-16 code points.
25///
26/// This `struct` is created when using the [`DecodeUtf16`] type.
27#[stable(feature = "decode_utf16", since = "1.9.0")]
28#[derive(Debug, Clone, Eq, PartialEq)]
29#[ferrocene::prevalidated]
30pub struct DecodeUtf16Error {
31    code: u16,
32}
33
34/// Creates an iterator over the UTF-16 encoded code points in `iter`,
35/// returning unpaired surrogates as `Err`s. See [`char::decode_utf16`].
36#[inline]
37#[ferrocene::prevalidated]
38pub(super) fn decode_utf16<I: IntoIterator<Item = u16>>(iter: I) -> DecodeUtf16<I::IntoIter> {
39    DecodeUtf16 { iter: iter.into_iter(), buf: None }
40}
41
42#[stable(feature = "decode_utf16", since = "1.9.0")]
43impl<I: Iterator<Item = u16>> Iterator for DecodeUtf16<I> {
44    type Item = Result<char, DecodeUtf16Error>;
45
46    #[ferrocene::prevalidated]
47    fn next(&mut self) -> Option<Result<char, DecodeUtf16Error>> {
48        let u = match self.buf.take() {
49            Some(buf) => buf,
50            None => self.iter.next()?,
51        };
52
53        if !u.is_utf16_surrogate() {
54            // SAFETY: not a surrogate
55            Some(Ok(unsafe { char::from_u32_unchecked(u as u32) }))
56        } else if u >= 0xDC00 {
57            // a trailing surrogate
58            Some(Err(DecodeUtf16Error { code: u }))
59        } else {
60            let u2 = match self.iter.next() {
61                Some(u2) => u2,
62                // eof
63                None => return Some(Err(DecodeUtf16Error { code: u })),
64            };
65            if u2 < 0xDC00 || u2 > 0xDFFF {
66                // not a trailing surrogate so we're not a valid
67                // surrogate pair, so rewind to redecode u2 next time.
68                self.buf = Some(u2);
69                return Some(Err(DecodeUtf16Error { code: u }));
70            }
71
72            // all ok, so lets decode it.
73            let c = (((u & 0x3ff) as u32) << 10 | (u2 & 0x3ff) as u32) + 0x1_0000;
74            // SAFETY: we checked that it's a legal unicode value
75            Some(Ok(unsafe { char::from_u32_unchecked(c) }))
76        }
77    }
78
79    #[inline]
80    #[ferrocene::prevalidated]
81    fn size_hint(&self) -> (usize, Option<usize>) {
82        let (low, high) = self.iter.size_hint();
83
84        let (low_buf, high_buf) = match self.buf {
85            // buf is empty, no additional elements from it.
86            None => (0, 0),
87            // `u` is a non surrogate, so it's always an additional character.
88            Some(u) if !u.is_utf16_surrogate() => (1, 1),
89            // `u` is a leading surrogate (it can never be a trailing surrogate and
90            // it's a surrogate due to the previous branch) and `self.iter` is empty.
91            //
92            // `u` can't be paired, since the `self.iter` is empty,
93            // so it will always become an additional element (error).
94            Some(_u) if high == Some(0) => (1, 1),
95            // `u` is a leading surrogate and `iter` may be non-empty.
96            //
97            // `u` can either pair with a trailing surrogate, in which case no additional elements
98            // are produced, or it can become an error, in which case it's an additional character (error).
99            Some(_u) => (0, 1),
100        };
101
102        // `self.iter` could contain entirely valid surrogates (2 elements per
103        // char), or entirely non-surrogates (1 element per char).
104        //
105        // On odd lower bound, at least one element must stay unpaired
106        // (with other elements from `self.iter`), so we round up.
107        let low = low.div_ceil(2) + low_buf;
108        let high = high.and_then(|h| h.checked_add(high_buf));
109
110        (low, high)
111    }
112}
113
114#[stable(feature = "decode_utf16_fused_iterator", since = "1.75.0")]
115impl<I: Iterator<Item = u16> + FusedIterator> FusedIterator for DecodeUtf16<I> {}
116
117impl DecodeUtf16Error {
118    /// Returns the unpaired surrogate which caused this error.
119    #[must_use]
120    #[stable(feature = "decode_utf16", since = "1.9.0")]
121    #[ferrocene::prevalidated]
122    pub fn unpaired_surrogate(&self) -> u16 {
123        self.code
124    }
125}
126
127#[stable(feature = "decode_utf16", since = "1.9.0")]
128impl fmt::Display for DecodeUtf16Error {
129    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
130        write!(f, "unpaired surrogate found: {:x}", self.code)
131    }
132}
133
134#[stable(feature = "decode_utf16", since = "1.9.0")]
135impl Error for DecodeUtf16Error {}
core/char/decode.rs

core/char/
decode.rs
