src/libcore/str/lossy.rs

   1 use crate::char;
   2 use crate::str as core_str;
   3 use crate::fmt::{self, Write};
   4 use crate::mem;
   5
   6 /// Lossy UTF-8 string.
   7 #[unstable(feature = "str_internals", issue = "0")]
   8 pub struct Utf8Lossy {
   9     bytes: [u8]
  10 }
  11
  12 impl Utf8Lossy {
  13     pub fn from_str(s: &str) -> &Utf8Lossy {
  14         Utf8Lossy::from_bytes(s.as_bytes())
  15     }
  16
  17     pub fn from_bytes(bytes: &[u8]) -> &Utf8Lossy {
  18         unsafe { mem::transmute(bytes) }
  19     }
  20
  21     pub fn chunks(&self) -> Utf8LossyChunksIter<'_> {
  22         Utf8LossyChunksIter { source: &self.bytes }
  23     }
  24 }
  25
  26
  27 /// Iterator over lossy UTF-8 string
  28 #[unstable(feature = "str_internals", issue = "0")]
  29 #[allow(missing_debug_implementations)]
  30 pub struct Utf8LossyChunksIter<'a> {
  31     source: &'a [u8],
  32 }
  33
  34 #[unstable(feature = "str_internals", issue = "0")]
  35 #[derive(PartialEq, Eq, Debug)]
  36 pub struct Utf8LossyChunk<'a> {
  37     /// Sequence of valid chars.
  38     /// Can be empty between broken UTF-8 chars.
  39     pub valid: &'a str,
  40     /// Single broken char, empty if none.
  41     /// Empty iff iterator item is last.
  42     pub broken: &'a [u8],
  43 }
  44
  45 impl<'a> Iterator for Utf8LossyChunksIter<'a> {
  46     type Item = Utf8LossyChunk<'a>;
  47
  48     fn next(&mut self) -> Option<Utf8LossyChunk<'a>> {
  49         if self.source.is_empty() {
  50             return None;
  51         }
  52
  53         const TAG_CONT_U8: u8 = 128;
  54         fn safe_get(xs: &[u8], i: usize) -> u8 {
  55             *xs.get(i).unwrap_or(&0)
  56         }
  57
  58         let mut i = 0;
  59         while i < self.source.len() {
  60             let i_ = i;
  61
  62             let byte = unsafe { *self.source.get_unchecked(i) };
  63             i += 1;
  64
  65             if byte < 128 {
  66
  67             } else {
  68                 let w = core_str::utf8_char_width(byte);
  69
  70                 macro_rules! error { () => ({
  71                     unsafe {
  72                         let r = Utf8LossyChunk {
  73                             valid: core_str::from_utf8_unchecked(&self.source[0..i_]),
  74                             broken: &self.source[i_..i],
  75                         };
  76                         self.source = &self.source[i..];
  77                         return Some(r);
  78                     }
  79                 })}
  80
  81                 match w {
  82                     2 => {
  83                         if safe_get(self.source, i) & 192 != TAG_CONT_U8 {
  84                             error!();
  85                         }
  86                         i += 1;
  87                     }
  88                     3 => {
  89                         match (byte, safe_get(self.source, i)) {
  90                             (0xE0, 0xA0 ..= 0xBF) => (),
  91                             (0xE1 ..= 0xEC, 0x80 ..= 0xBF) => (),
  92                             (0xED, 0x80 ..= 0x9F) => (),
  93                             (0xEE ..= 0xEF, 0x80 ..= 0xBF) => (),
  94                             _ => {
  95                                 error!();
  96                             }
  97                         }
  98                         i += 1;
  99                         if safe_get(self.source, i) & 192 != TAG_CONT_U8 {
 100                             error!();
 101                         }
 102                         i += 1;
 103                     }
 104                     4 => {
 105                         match (byte, safe_get(self.source, i)) {
 106                             (0xF0, 0x90 ..= 0xBF) => (),
 107                             (0xF1 ..= 0xF3, 0x80 ..= 0xBF) => (),
 108                             (0xF4, 0x80 ..= 0x8F) => (),
 109                             _ => {
 110                                 error!();
 111                             }
 112                         }
 113                         i += 1;
 114                         if safe_get(self.source, i) & 192 != TAG_CONT_U8 {
 115                             error!();
 116                         }
 117                         i += 1;
 118                         if safe_get(self.source, i) & 192 != TAG_CONT_U8 {
 119                             error!();
 120                         }
 121                         i += 1;
 122                     }
 123                     _ => {
 124                         error!();
 125                     }
 126                 }
 127             }
 128         }
 129
 130         let r = Utf8LossyChunk {
 131             valid: unsafe { core_str::from_utf8_unchecked(self.source) },
 132             broken: &[],
 133         };
 134         self.source = &[];
 135         Some(r)
 136     }
 137 }
 138
 139
 140 impl fmt::Display for Utf8Lossy {
 141     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 142         // If we're the empty string then our iterator won't actually yield
 143         // anything, so perform the formatting manually
 144         if self.bytes.is_empty() {
 145             return "".fmt(f)
 146         }
 147
 148         for Utf8LossyChunk { valid, broken } in self.chunks() {
 149             // If we successfully decoded the whole chunk as a valid string then
 150             // we can return a direct formatting of the string which will also
 151             // respect various formatting flags if possible.
 152             if valid.len() == self.bytes.len() {
 153                 assert!(broken.is_empty());
 154                 return valid.fmt(f)
 155             }
 156
 157             f.write_str(valid)?;
 158             if !broken.is_empty() {
 159                 f.write_char(char::REPLACEMENT_CHARACTER)?;
 160             }
 161         }
 162         Ok(())
 163     }
 164 }
 165
 166 impl fmt::Debug for Utf8Lossy {
 167     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
 168         f.write_char('"')?;
 169
 170         for Utf8LossyChunk { valid, broken } in self.chunks() {
 171
 172             // Valid part.
 173             // Here we partially parse UTF-8 again which is suboptimal.
 174             {
 175                 let mut from = 0;
 176                 for (i, c) in valid.char_indices() {
 177                     let esc = c.escape_debug();
 178                     // If char needs escaping, flush backlog so far and write, else skip
 179                     if esc.len() != 1 {
 180                         f.write_str(&valid[from..i])?;
 181                         for c in esc {
 182                             f.write_char(c)?;
 183                         }
 184                         from = i + c.len_utf8();
 185                     }
 186                 }
 187                 f.write_str(&valid[from..])?;
 188             }
 189
 190             // Broken parts of string as hex escape.
 191             for &b in broken {
 192                 write!(f, "\\x{:02x}", b)?;
 193             }
 194         }
 195
 196         f.write_char('"')
 197     }
 198 }