library/core/src/char/decode.rs

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