| 1 | //! Character conversions. |
| 2 | |
| 3 | use crate::char::TryFromCharError; |
| 4 | use crate::error::Error; |
| 5 | use crate::fmt; |
| 6 | use crate::mem::transmute; |
| 7 | use crate::str::FromStr; |
| 8 | use crate::ub_checks::assert_unsafe_precondition; |
| 9 | |
| 10 | /// Converts a `u32` to a `char`. See [`char::from_u32`]. |
| 11 | #[must_use ] |
| 12 | #[inline ] |
| 13 | pub(super) const fn from_u32(i: u32) -> Option<char> { |
| 14 | // FIXME(const-hack): once Result::ok is const fn, use it here |
| 15 | match char_try_from_u32(i) { |
| 16 | Ok(c: char) => Some(c), |
| 17 | Err(_) => None, |
| 18 | } |
| 19 | } |
| 20 | |
| 21 | /// Converts a `u32` to a `char`, ignoring validity. See [`char::from_u32_unchecked`]. |
| 22 | #[inline ] |
| 23 | #[must_use ] |
| 24 | #[allow (unnecessary_transmutes)] |
| 25 | #[track_caller ] |
| 26 | pub(super) const unsafe fn from_u32_unchecked(i: u32) -> char { |
| 27 | // SAFETY: the caller must guarantee that `i` is a valid char value. |
| 28 | unsafe { |
| 29 | assert_unsafe_precondition!( |
| 30 | check_language_ub, |
| 31 | "invalid value for `char`" , |
| 32 | (i: u32 = i) => char_try_from_u32(i).is_ok() |
| 33 | ); |
| 34 | transmute(src:i) |
| 35 | } |
| 36 | } |
| 37 | |
| 38 | #[stable (feature = "char_convert" , since = "1.13.0" )] |
| 39 | impl From<char> for u32 { |
| 40 | /// Converts a [`char`] into a [`u32`]. |
| 41 | /// |
| 42 | /// # Examples |
| 43 | /// |
| 44 | /// ``` |
| 45 | /// let c = 'c' ; |
| 46 | /// let u = u32::from(c); |
| 47 | /// assert!(4 == size_of_val(&u)) |
| 48 | /// ``` |
| 49 | #[inline ] |
| 50 | fn from(c: char) -> Self { |
| 51 | c as u32 |
| 52 | } |
| 53 | } |
| 54 | |
| 55 | #[stable (feature = "more_char_conversions" , since = "1.51.0" )] |
| 56 | impl From<char> for u64 { |
| 57 | /// Converts a [`char`] into a [`u64`]. |
| 58 | /// |
| 59 | /// # Examples |
| 60 | /// |
| 61 | /// ``` |
| 62 | /// let c = '👤' ; |
| 63 | /// let u = u64::from(c); |
| 64 | /// assert!(8 == size_of_val(&u)) |
| 65 | /// ``` |
| 66 | #[inline ] |
| 67 | fn from(c: char) -> Self { |
| 68 | // The char is casted to the value of the code point, then zero-extended to 64 bit. |
| 69 | // See [https://doc.rust-lang.org/reference/expressions/operator-expr.html#semantics] |
| 70 | c as u64 |
| 71 | } |
| 72 | } |
| 73 | |
| 74 | #[stable (feature = "more_char_conversions" , since = "1.51.0" )] |
| 75 | impl From<char> for u128 { |
| 76 | /// Converts a [`char`] into a [`u128`]. |
| 77 | /// |
| 78 | /// # Examples |
| 79 | /// |
| 80 | /// ``` |
| 81 | /// let c = 'âš™' ; |
| 82 | /// let u = u128::from(c); |
| 83 | /// assert!(16 == size_of_val(&u)) |
| 84 | /// ``` |
| 85 | #[inline ] |
| 86 | fn from(c: char) -> Self { |
| 87 | // The char is casted to the value of the code point, then zero-extended to 128 bit. |
| 88 | // See [https://doc.rust-lang.org/reference/expressions/operator-expr.html#semantics] |
| 89 | c as u128 |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | /// Maps a `char` with code point in U+0000..=U+00FF to a byte in 0x00..=0xFF with same value, |
| 94 | /// failing if the code point is greater than U+00FF. |
| 95 | /// |
| 96 | /// See [`impl From<u8> for char`](char#impl-From<u8>-for-char) for details on the encoding. |
| 97 | #[stable (feature = "u8_from_char" , since = "1.59.0" )] |
| 98 | impl TryFrom<char> for u8 { |
| 99 | type Error = TryFromCharError; |
| 100 | |
| 101 | /// Tries to convert a [`char`] into a [`u8`]. |
| 102 | /// |
| 103 | /// # Examples |
| 104 | /// |
| 105 | /// ``` |
| 106 | /// let a = 'ÿ' ; // U+00FF |
| 107 | /// let b = 'Ä€' ; // U+0100 |
| 108 | /// assert_eq!(u8::try_from(a), Ok(0xFF_u8)); |
| 109 | /// assert!(u8::try_from(b).is_err()); |
| 110 | /// ``` |
| 111 | #[inline ] |
| 112 | fn try_from(c: char) -> Result<u8, Self::Error> { |
| 113 | u8::try_from(u32::from(c)).map_err(|_| TryFromCharError(())) |
| 114 | } |
| 115 | } |
| 116 | |
| 117 | /// Maps a `char` with code point in U+0000..=U+FFFF to a `u16` in 0x0000..=0xFFFF with same value, |
| 118 | /// failing if the code point is greater than U+FFFF. |
| 119 | /// |
| 120 | /// This corresponds to the UCS-2 encoding, as specified in ISO/IEC 10646:2003. |
| 121 | #[stable (feature = "u16_from_char" , since = "1.74.0" )] |
| 122 | impl TryFrom<char> for u16 { |
| 123 | type Error = TryFromCharError; |
| 124 | |
| 125 | /// Tries to convert a [`char`] into a [`u16`]. |
| 126 | /// |
| 127 | /// # Examples |
| 128 | /// |
| 129 | /// ``` |
| 130 | /// let trans_rights = 'âš§' ; // U+26A7 |
| 131 | /// let ninjas = '🥷' ; // U+1F977 |
| 132 | /// assert_eq!(u16::try_from(trans_rights), Ok(0x26A7_u16)); |
| 133 | /// assert!(u16::try_from(ninjas).is_err()); |
| 134 | /// ``` |
| 135 | #[inline ] |
| 136 | fn try_from(c: char) -> Result<u16, Self::Error> { |
| 137 | u16::try_from(u32::from(c)).map_err(|_| TryFromCharError(())) |
| 138 | } |
| 139 | } |
| 140 | |
| 141 | /// Maps a byte in 0x00..=0xFF to a `char` whose code point has the same value, in U+0000..=U+00FF. |
| 142 | /// |
| 143 | /// Unicode is designed such that this effectively decodes bytes |
| 144 | /// with the character encoding that IANA calls ISO-8859-1. |
| 145 | /// This encoding is compatible with ASCII. |
| 146 | /// |
| 147 | /// Note that this is different from ISO/IEC 8859-1 a.k.a. ISO 8859-1 (with one less hyphen), |
| 148 | /// which leaves some "blanks", byte values that are not assigned to any character. |
| 149 | /// ISO-8859-1 (the IANA one) assigns them to the C0 and C1 control codes. |
| 150 | /// |
| 151 | /// Note that this is *also* different from Windows-1252 a.k.a. code page 1252, |
| 152 | /// which is a superset ISO/IEC 8859-1 that assigns some (not all!) blanks |
| 153 | /// to punctuation and various Latin characters. |
| 154 | /// |
| 155 | /// To confuse things further, [on the Web](https://encoding.spec.whatwg.org/) |
| 156 | /// `ascii`, `iso-8859-1`, and `windows-1252` are all aliases |
| 157 | /// for a superset of Windows-1252 that fills the remaining blanks with corresponding |
| 158 | /// C0 and C1 control codes. |
| 159 | #[stable (feature = "char_convert" , since = "1.13.0" )] |
| 160 | impl From<u8> for char { |
| 161 | /// Converts a [`u8`] into a [`char`]. |
| 162 | /// |
| 163 | /// # Examples |
| 164 | /// |
| 165 | /// ``` |
| 166 | /// let u = 32 as u8; |
| 167 | /// let c = char::from(u); |
| 168 | /// assert!(4 == size_of_val(&c)) |
| 169 | /// ``` |
| 170 | #[inline ] |
| 171 | fn from(i: u8) -> Self { |
| 172 | i as char |
| 173 | } |
| 174 | } |
| 175 | |
| 176 | /// An error which can be returned when parsing a char. |
| 177 | /// |
| 178 | /// This `struct` is created when using the [`char::from_str`] method. |
| 179 | #[stable (feature = "char_from_str" , since = "1.20.0" )] |
| 180 | #[derive (Clone, Debug, PartialEq, Eq)] |
| 181 | pub struct ParseCharError { |
| 182 | kind: CharErrorKind, |
| 183 | } |
| 184 | |
| 185 | #[derive (Copy, Clone, Debug, PartialEq, Eq)] |
| 186 | enum CharErrorKind { |
| 187 | EmptyString, |
| 188 | TooManyChars, |
| 189 | } |
| 190 | |
| 191 | #[stable (feature = "char_from_str" , since = "1.20.0" )] |
| 192 | impl Error for ParseCharError { |
| 193 | #[allow (deprecated)] |
| 194 | fn description(&self) -> &str { |
| 195 | match self.kind { |
| 196 | CharErrorKind::EmptyString => "cannot parse char from empty string" , |
| 197 | CharErrorKind::TooManyChars => "too many characters in string" , |
| 198 | } |
| 199 | } |
| 200 | } |
| 201 | |
| 202 | #[stable (feature = "char_from_str" , since = "1.20.0" )] |
| 203 | impl fmt::Display for ParseCharError { |
| 204 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 205 | #[allow (deprecated)] |
| 206 | self.description().fmt(f) |
| 207 | } |
| 208 | } |
| 209 | |
| 210 | #[stable (feature = "char_from_str" , since = "1.20.0" )] |
| 211 | impl FromStr for char { |
| 212 | type Err = ParseCharError; |
| 213 | |
| 214 | #[inline ] |
| 215 | fn from_str(s: &str) -> Result<Self, Self::Err> { |
| 216 | let mut chars: Chars<'_> = s.chars(); |
| 217 | match (chars.next(), chars.next()) { |
| 218 | (None, _) => Err(ParseCharError { kind: CharErrorKind::EmptyString }), |
| 219 | (Some(c: char), None) => Ok(c), |
| 220 | _ => Err(ParseCharError { kind: CharErrorKind::TooManyChars }), |
| 221 | } |
| 222 | } |
| 223 | } |
| 224 | |
| 225 | #[inline ] |
| 226 | #[allow (unnecessary_transmutes)] |
| 227 | const fn char_try_from_u32(i: u32) -> Result<char, CharTryFromError> { |
| 228 | // This is an optimized version of the check |
| 229 | // (i > MAX as u32) || (i >= 0xD800 && i <= 0xDFFF), |
| 230 | // which can also be written as |
| 231 | // i >= 0x110000 || (i >= 0xD800 && i < 0xE000). |
| 232 | // |
| 233 | // The XOR with 0xD800 permutes the ranges such that 0xD800..0xE000 is |
| 234 | // mapped to 0x0000..0x0800, while keeping all the high bits outside 0xFFFF the same. |
| 235 | // In particular, numbers >= 0x110000 stay in this range. |
| 236 | // |
| 237 | // Subtracting 0x800 causes 0x0000..0x0800 to wrap, meaning that a single |
| 238 | // unsigned comparison against 0x110000 - 0x800 will detect both the wrapped |
| 239 | // surrogate range as well as the numbers originally larger than 0x110000. |
| 240 | // |
| 241 | if (i ^ 0xD800).wrapping_sub(0x800) >= 0x110000 - 0x800 { |
| 242 | Err(CharTryFromError(())) |
| 243 | } else { |
| 244 | // SAFETY: checked that it's a legal unicode value |
| 245 | Ok(unsafe { transmute(src:i) }) |
| 246 | } |
| 247 | } |
| 248 | |
| 249 | #[stable (feature = "try_from" , since = "1.34.0" )] |
| 250 | impl TryFrom<u32> for char { |
| 251 | type Error = CharTryFromError; |
| 252 | |
| 253 | #[inline ] |
| 254 | fn try_from(i: u32) -> Result<Self, Self::Error> { |
| 255 | char_try_from_u32(i) |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | /// The error type returned when a conversion from [`prim@u32`] to [`prim@char`] fails. |
| 260 | /// |
| 261 | /// This `struct` is created by the [`char::try_from<u32>`](char#impl-TryFrom<u32>-for-char) method. |
| 262 | /// See its documentation for more. |
| 263 | #[stable (feature = "try_from" , since = "1.34.0" )] |
| 264 | #[derive (Copy, Clone, Debug, PartialEq, Eq)] |
| 265 | pub struct CharTryFromError(()); |
| 266 | |
| 267 | #[stable (feature = "try_from" , since = "1.34.0" )] |
| 268 | impl fmt::Display for CharTryFromError { |
| 269 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 270 | "converted integer out of range for `char`" .fmt(f) |
| 271 | } |
| 272 | } |
| 273 | |
| 274 | /// Converts a digit in the given radix to a `char`. See [`char::from_digit`]. |
| 275 | #[inline ] |
| 276 | #[must_use ] |
| 277 | pub(super) const fn from_digit(num: u32, radix: u32) -> Option<char> { |
| 278 | if radix > 36 { |
| 279 | panic!("from_digit: radix is too high (maximum 36)" ); |
| 280 | } |
| 281 | if num < radix { |
| 282 | let num: u8 = num as u8; |
| 283 | if num < 10 { Some((b'0' + num) as char) } else { Some((b'a' + num - 10) as char) } |
| 284 | } else { |
| 285 | None |
| 286 | } |
| 287 | } |
| 288 | |