| 1 | //! The `ByteStr` type and trait implementations. |
| 2 | |
| 3 | mod traits; |
| 4 | |
| 5 | #[unstable (feature = "bstr_internals" , issue = "none" )] |
| 6 | pub use traits::{impl_partial_eq, impl_partial_eq_n, impl_partial_eq_ord}; |
| 7 | |
| 8 | use crate::borrow::{Borrow, BorrowMut}; |
| 9 | use crate::fmt; |
| 10 | use crate::ops::{Deref, DerefMut, DerefPure}; |
| 11 | |
| 12 | /// A wrapper for `&[u8]` representing a human-readable string that's conventionally, but not |
| 13 | /// always, UTF-8. |
| 14 | /// |
| 15 | /// Unlike `&str`, this type permits non-UTF-8 contents, making it suitable for user input, |
| 16 | /// non-native filenames (as `Path` only supports native filenames), and other applications that |
| 17 | /// need to round-trip whatever data the user provides. |
| 18 | /// |
| 19 | /// For an owned, growable byte string buffer, use |
| 20 | /// [`ByteString`](../../std/bstr/struct.ByteString.html). |
| 21 | /// |
| 22 | /// `ByteStr` implements `Deref` to `[u8]`, so all methods available on `[u8]` are available on |
| 23 | /// `ByteStr`. |
| 24 | /// |
| 25 | /// # Representation |
| 26 | /// |
| 27 | /// A `&ByteStr` has the same representation as a `&str`. That is, a `&ByteStr` is a wide pointer |
| 28 | /// which includes a pointer to some bytes and a length. |
| 29 | /// |
| 30 | /// # Trait implementations |
| 31 | /// |
| 32 | /// The `ByteStr` type has a number of trait implementations, and in particular, defines equality |
| 33 | /// and comparisons between `&ByteStr`, `&str`, and `&[u8]`, for convenience. |
| 34 | /// |
| 35 | /// The `Debug` implementation for `ByteStr` shows its bytes as a normal string, with invalid UTF-8 |
| 36 | /// presented as hex escape sequences. |
| 37 | /// |
| 38 | /// The `Display` implementation behaves as if the `ByteStr` were first lossily converted to a |
| 39 | /// `str`, with invalid UTF-8 presented as the Unicode replacement character (�). |
| 40 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 41 | #[repr (transparent)] |
| 42 | #[doc (alias = "BStr" )] |
| 43 | pub struct ByteStr(pub [u8]); |
| 44 | |
| 45 | impl ByteStr { |
| 46 | /// Creates a `ByteStr` slice from anything that can be converted to a byte slice. |
| 47 | /// |
| 48 | /// This is a zero-cost conversion. |
| 49 | /// |
| 50 | /// # Example |
| 51 | /// |
| 52 | /// You can create a `ByteStr` from a byte array, a byte slice or a string slice: |
| 53 | /// |
| 54 | /// ``` |
| 55 | /// # #![feature (bstr)] |
| 56 | /// # use std::bstr::ByteStr; |
| 57 | /// let a = ByteStr::new(b"abc" ); |
| 58 | /// let b = ByteStr::new(&b"abc" [..]); |
| 59 | /// let c = ByteStr::new("abc" ); |
| 60 | /// |
| 61 | /// assert_eq!(a, b); |
| 62 | /// assert_eq!(a, c); |
| 63 | /// ``` |
| 64 | #[inline ] |
| 65 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 66 | pub fn new<B: ?Sized + AsRef<[u8]>>(bytes: &B) -> &Self { |
| 67 | ByteStr::from_bytes(bytes.as_ref()) |
| 68 | } |
| 69 | |
| 70 | #[doc (hidden)] |
| 71 | #[unstable (feature = "bstr_internals" , issue = "none" )] |
| 72 | #[inline ] |
| 73 | pub fn from_bytes(slice: &[u8]) -> &Self { |
| 74 | // SAFETY: `ByteStr` is a transparent wrapper around `[u8]`, so we can turn a reference to |
| 75 | // the wrapped type into a reference to the wrapper type. |
| 76 | unsafe { &*(slice as *const [u8] as *const Self) } |
| 77 | } |
| 78 | |
| 79 | #[doc (hidden)] |
| 80 | #[unstable (feature = "bstr_internals" , issue = "none" )] |
| 81 | #[inline ] |
| 82 | pub fn from_bytes_mut(slice: &mut [u8]) -> &mut Self { |
| 83 | // SAFETY: `ByteStr` is a transparent wrapper around `[u8]`, so we can turn a reference to |
| 84 | // the wrapped type into a reference to the wrapper type. |
| 85 | unsafe { &mut *(slice as *mut [u8] as *mut Self) } |
| 86 | } |
| 87 | |
| 88 | #[doc (hidden)] |
| 89 | #[unstable (feature = "bstr_internals" , issue = "none" )] |
| 90 | #[inline ] |
| 91 | pub fn as_bytes(&self) -> &[u8] { |
| 92 | &self.0 |
| 93 | } |
| 94 | |
| 95 | #[doc (hidden)] |
| 96 | #[unstable (feature = "bstr_internals" , issue = "none" )] |
| 97 | #[inline ] |
| 98 | pub fn as_bytes_mut(&mut self) -> &mut [u8] { |
| 99 | &mut self.0 |
| 100 | } |
| 101 | } |
| 102 | |
| 103 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 104 | impl Deref for ByteStr { |
| 105 | type Target = [u8]; |
| 106 | |
| 107 | #[inline ] |
| 108 | fn deref(&self) -> &[u8] { |
| 109 | &self.0 |
| 110 | } |
| 111 | } |
| 112 | |
| 113 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 114 | impl DerefMut for ByteStr { |
| 115 | #[inline ] |
| 116 | fn deref_mut(&mut self) -> &mut [u8] { |
| 117 | &mut self.0 |
| 118 | } |
| 119 | } |
| 120 | |
| 121 | #[unstable (feature = "deref_pure_trait" , issue = "87121" )] |
| 122 | unsafe impl DerefPure for ByteStr {} |
| 123 | |
| 124 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 125 | impl fmt::Debug for ByteStr { |
| 126 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 127 | write!(f, " \"" )?; |
| 128 | for chunk: Utf8Chunk<'_> in self.utf8_chunks() { |
| 129 | for c: char in chunk.valid().chars() { |
| 130 | match c { |
| 131 | ' \0' => write!(f, " \\0" )?, |
| 132 | ' \x01' ..=' \x7f' => write!(f, " {}" , (c as u8).escape_ascii())?, |
| 133 | _ => write!(f, " {}" , c.escape_debug())?, |
| 134 | } |
| 135 | } |
| 136 | write!(f, " {}" , chunk.invalid().escape_ascii())?; |
| 137 | } |
| 138 | write!(f, " \"" )?; |
| 139 | Ok(()) |
| 140 | } |
| 141 | } |
| 142 | |
| 143 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 144 | impl fmt::Display for ByteStr { |
| 145 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 146 | fn fmt_nopad(this: &ByteStr, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 147 | for chunk in this.utf8_chunks() { |
| 148 | f.write_str(chunk.valid())?; |
| 149 | if !chunk.invalid().is_empty() { |
| 150 | f.write_str(" \u{FFFD}" )?; |
| 151 | } |
| 152 | } |
| 153 | Ok(()) |
| 154 | } |
| 155 | |
| 156 | let Some(align) = f.align() else { |
| 157 | return fmt_nopad(self, f); |
| 158 | }; |
| 159 | let nchars: usize = self |
| 160 | .utf8_chunks() |
| 161 | .map(|chunk| { |
| 162 | chunk.valid().chars().count() + if chunk.invalid().is_empty() { 0 } else { 1 } |
| 163 | }) |
| 164 | .sum(); |
| 165 | let padding = f.width().unwrap_or(0).saturating_sub(nchars); |
| 166 | let fill = f.fill(); |
| 167 | let (lpad, rpad) = match align { |
| 168 | fmt::Alignment::Left => (0, padding), |
| 169 | fmt::Alignment::Right => (padding, 0), |
| 170 | fmt::Alignment::Center => { |
| 171 | let half = padding / 2; |
| 172 | (half, half + padding % 2) |
| 173 | } |
| 174 | }; |
| 175 | for _ in 0..lpad { |
| 176 | write!(f, " {fill}" )?; |
| 177 | } |
| 178 | fmt_nopad(self, f)?; |
| 179 | for _ in 0..rpad { |
| 180 | write!(f, " {fill}" )?; |
| 181 | } |
| 182 | |
| 183 | Ok(()) |
| 184 | } |
| 185 | } |
| 186 | |
| 187 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 188 | impl AsRef<[u8]> for ByteStr { |
| 189 | #[inline ] |
| 190 | fn as_ref(&self) -> &[u8] { |
| 191 | &self.0 |
| 192 | } |
| 193 | } |
| 194 | |
| 195 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 196 | impl AsRef<ByteStr> for ByteStr { |
| 197 | #[inline ] |
| 198 | fn as_ref(&self) -> &ByteStr { |
| 199 | self |
| 200 | } |
| 201 | } |
| 202 | |
| 203 | // `impl AsRef<ByteStr> for [u8]` omitted to avoid widespread inference failures |
| 204 | |
| 205 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 206 | impl AsRef<ByteStr> for str { |
| 207 | #[inline ] |
| 208 | fn as_ref(&self) -> &ByteStr { |
| 209 | ByteStr::new(self) |
| 210 | } |
| 211 | } |
| 212 | |
| 213 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 214 | impl AsMut<[u8]> for ByteStr { |
| 215 | #[inline ] |
| 216 | fn as_mut(&mut self) -> &mut [u8] { |
| 217 | &mut self.0 |
| 218 | } |
| 219 | } |
| 220 | |
| 221 | // `impl AsMut<ByteStr> for [u8]` omitted to avoid widespread inference failures |
| 222 | |
| 223 | // `impl Borrow<ByteStr> for [u8]` omitted to avoid widespread inference failures |
| 224 | |
| 225 | // `impl Borrow<ByteStr> for str` omitted to avoid widespread inference failures |
| 226 | |
| 227 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 228 | impl Borrow<[u8]> for ByteStr { |
| 229 | #[inline ] |
| 230 | fn borrow(&self) -> &[u8] { |
| 231 | &self.0 |
| 232 | } |
| 233 | } |
| 234 | |
| 235 | // `impl BorrowMut<ByteStr> for [u8]` omitted to avoid widespread inference failures |
| 236 | |
| 237 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 238 | impl BorrowMut<[u8]> for ByteStr { |
| 239 | #[inline ] |
| 240 | fn borrow_mut(&mut self) -> &mut [u8] { |
| 241 | &mut self.0 |
| 242 | } |
| 243 | } |
| 244 | |
| 245 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 246 | impl<'a> Default for &'a ByteStr { |
| 247 | fn default() -> Self { |
| 248 | ByteStr::from_bytes(slice:b"" ) |
| 249 | } |
| 250 | } |
| 251 | |
| 252 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 253 | impl<'a> Default for &'a mut ByteStr { |
| 254 | fn default() -> Self { |
| 255 | ByteStr::from_bytes_mut(&mut []) |
| 256 | } |
| 257 | } |
| 258 | |
| 259 | // Omitted due to inference failures |
| 260 | // |
| 261 | // #[unstable(feature = "bstr", issue = "134915")] |
| 262 | // impl<'a, const N: usize> From<&'a [u8; N]> for &'a ByteStr { |
| 263 | // #[inline] |
| 264 | // fn from(s: &'a [u8; N]) -> Self { |
| 265 | // ByteStr::from_bytes(s) |
| 266 | // } |
| 267 | // } |
| 268 | // |
| 269 | // #[unstable(feature = "bstr", issue = "134915")] |
| 270 | // impl<'a> From<&'a [u8]> for &'a ByteStr { |
| 271 | // #[inline] |
| 272 | // fn from(s: &'a [u8]) -> Self { |
| 273 | // ByteStr::from_bytes(s) |
| 274 | // } |
| 275 | // } |
| 276 | |
| 277 | // Omitted due to slice-from-array-issue-113238: |
| 278 | // |
| 279 | // #[unstable(feature = "bstr", issue = "134915")] |
| 280 | // impl<'a> From<&'a ByteStr> for &'a [u8] { |
| 281 | // #[inline] |
| 282 | // fn from(s: &'a ByteStr) -> Self { |
| 283 | // &s.0 |
| 284 | // } |
| 285 | // } |
| 286 | // |
| 287 | // #[unstable(feature = "bstr", issue = "134915")] |
| 288 | // impl<'a> From<&'a mut ByteStr> for &'a mut [u8] { |
| 289 | // #[inline] |
| 290 | // fn from(s: &'a mut ByteStr) -> Self { |
| 291 | // &mut s.0 |
| 292 | // } |
| 293 | // } |
| 294 | |
| 295 | // Omitted due to inference failures |
| 296 | // |
| 297 | // #[unstable(feature = "bstr", issue = "134915")] |
| 298 | // impl<'a> From<&'a str> for &'a ByteStr { |
| 299 | // #[inline] |
| 300 | // fn from(s: &'a str) -> Self { |
| 301 | // ByteStr::from_bytes(s.as_bytes()) |
| 302 | // } |
| 303 | // } |
| 304 | |
| 305 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 306 | impl<'a> TryFrom<&'a ByteStr> for &'a str { |
| 307 | type Error = crate::str::Utf8Error; |
| 308 | |
| 309 | #[inline ] |
| 310 | fn try_from(s: &'a ByteStr) -> Result<Self, Self::Error> { |
| 311 | crate::str::from_utf8(&s.0) |
| 312 | } |
| 313 | } |
| 314 | |
| 315 | #[unstable (feature = "bstr" , issue = "134915" )] |
| 316 | impl<'a> TryFrom<&'a mut ByteStr> for &'a mut str { |
| 317 | type Error = crate::str::Utf8Error; |
| 318 | |
| 319 | #[inline ] |
| 320 | fn try_from(s: &'a mut ByteStr) -> Result<Self, Self::Error> { |
| 321 | crate::str::from_utf8_mut(&mut s.0) |
| 322 | } |
| 323 | } |
| 324 | |