| 1 | // Copyright Mozilla Foundation |
| 2 | // |
| 3 | // Licensed under the Apache License (Version 2.0), or the MIT license, |
| 4 | // (the "Licenses") at your option. You may not use this file except in |
| 5 | // compliance with one of the Licenses. You may obtain copies of the |
| 6 | // Licenses at: |
| 7 | // |
| 8 | // https://www.apache.org/licenses/LICENSE-2.0 |
| 9 | // https://opensource.org/licenses/MIT |
| 10 | // |
| 11 | // Unless required by applicable law or agreed to in writing, software |
| 12 | // distributed under the Licenses is distributed on an "AS IS" BASIS, |
| 13 | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 14 | // See the Licenses for the specific language governing permissions and |
| 15 | // limitations under the Licenses. |
| 16 | |
| 17 | #![no_std ] |
| 18 | |
| 19 | //! Provides iteration by `char` over `&[u8]` containing potentially-invalid |
| 20 | //! UTF-8 such that errors are handled according to the [WHATWG Encoding |
| 21 | //! Standard](https://encoding.spec.whatwg.org/#utf-8-decoder) (i.e. the same |
| 22 | //! way as in `String::from_utf8_lossy`). |
| 23 | //! |
| 24 | //! The trait `Utf8CharsEx` provides the convenience method `chars()` on |
| 25 | //! byte slices themselves instead of having to use the more verbose |
| 26 | //! `Utf8Chars::new(slice)`. |
| 27 | //! |
| 28 | //! ```rust |
| 29 | //! use utf8_iter::Utf8CharsEx; |
| 30 | //! let data = b" \xFF\xC2\xE2\xE2\x98\xF0\xF0\x9F\xF0\x9F\x92\xE2\x98\x83" ; |
| 31 | //! let from_iter: String = data.chars().collect(); |
| 32 | //! let from_std = String::from_utf8_lossy(data); |
| 33 | //! assert_eq!(from_iter, from_std); |
| 34 | //! ``` |
| 35 | |
| 36 | mod indices; |
| 37 | mod report; |
| 38 | |
| 39 | pub use crate::indices::Utf8CharIndices; |
| 40 | pub use crate::report::ErrorReportingUtf8Chars; |
| 41 | pub use crate::report::Utf8CharsError; |
| 42 | use core::iter::FusedIterator; |
| 43 | |
| 44 | #[repr (align(64))] // Align to cache lines |
| 45 | struct Utf8Data { |
| 46 | pub table: [u8; 384], |
| 47 | } |
| 48 | |
| 49 | // This is generated code copied and pasted from utf_8.rs of encoding_rs. |
| 50 | // Please don't edit by hand but instead regenerate as instructed in that |
| 51 | // file. |
| 52 | |
| 53 | static UTF8_DATA: Utf8Data = Utf8Data { |
| 54 | table: [ |
| 55 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 56 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 57 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 58 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 59 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 60 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 61 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 62 | 252, 252, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 84, 148, 148, 148, |
| 63 | 148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 148, 164, 164, 164, 164, 164, |
| 64 | 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, 164, |
| 65 | 164, 164, 164, 164, 164, 164, 164, 164, 164, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 66 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 67 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 68 | 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, 252, |
| 69 | 252, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, |
| 70 | 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, |
| 71 | 4, 4, 4, 4, 4, 4, 4, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, |
| 72 | 8, 8, 8, 8, 8, 8, 8, 16, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 32, 8, 8, 64, 8, 8, 8, 128, 4, |
| 73 | 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, |
| 74 | ], |
| 75 | }; |
| 76 | |
| 77 | // End manually copypasted generated code. |
| 78 | |
| 79 | #[inline (always)] |
| 80 | fn in_inclusive_range8(i: u8, start: u8, end: u8) -> bool { |
| 81 | i.wrapping_sub(start) <= (end - start) |
| 82 | } |
| 83 | |
| 84 | /// Iterator by `char` over `&[u8]` that contains |
| 85 | /// potentially-invalid UTF-8. See the crate documentation. |
| 86 | #[derive (Debug, Clone)] |
| 87 | pub struct Utf8Chars<'a> { |
| 88 | remaining: &'a [u8], |
| 89 | } |
| 90 | |
| 91 | impl<'a> Utf8Chars<'a> { |
| 92 | #[inline (always)] |
| 93 | /// Creates the iterator from a byte slice. |
| 94 | pub fn new(bytes: &'a [u8]) -> Self { |
| 95 | Utf8Chars::<'a> { remaining: bytes } |
| 96 | } |
| 97 | |
| 98 | /// Views the current remaining data in the iterator as a subslice |
| 99 | /// of the original slice. |
| 100 | #[inline (always)] |
| 101 | pub fn as_slice(&self) -> &'a [u8] { |
| 102 | self.remaining |
| 103 | } |
| 104 | |
| 105 | #[inline (never)] |
| 106 | fn next_fallback(&mut self) -> Option<char> { |
| 107 | if self.remaining.is_empty() { |
| 108 | return None; |
| 109 | } |
| 110 | let first = self.remaining[0]; |
| 111 | if first < 0x80 { |
| 112 | self.remaining = &self.remaining[1..]; |
| 113 | return Some(char::from(first)); |
| 114 | } |
| 115 | if !in_inclusive_range8(first, 0xC2, 0xF4) || self.remaining.len() == 1 { |
| 116 | self.remaining = &self.remaining[1..]; |
| 117 | return Some(' \u{FFFD}' ); |
| 118 | } |
| 119 | let second = self.remaining[1]; |
| 120 | let (lower_bound, upper_bound) = match first { |
| 121 | 0xE0 => (0xA0, 0xBF), |
| 122 | 0xED => (0x80, 0x9F), |
| 123 | 0xF0 => (0x90, 0xBF), |
| 124 | 0xF4 => (0x80, 0x8F), |
| 125 | _ => (0x80, 0xBF), |
| 126 | }; |
| 127 | if !in_inclusive_range8(second, lower_bound, upper_bound) { |
| 128 | self.remaining = &self.remaining[1..]; |
| 129 | return Some(' \u{FFFD}' ); |
| 130 | } |
| 131 | if first < 0xE0 { |
| 132 | self.remaining = &self.remaining[2..]; |
| 133 | let point = ((u32::from(first) & 0x1F) << 6) | (u32::from(second) & 0x3F); |
| 134 | return Some(unsafe { char::from_u32_unchecked(point) }); |
| 135 | } |
| 136 | if self.remaining.len() == 2 { |
| 137 | self.remaining = &self.remaining[2..]; |
| 138 | return Some(' \u{FFFD}' ); |
| 139 | } |
| 140 | let third = self.remaining[2]; |
| 141 | if !in_inclusive_range8(third, 0x80, 0xBF) { |
| 142 | self.remaining = &self.remaining[2..]; |
| 143 | return Some(' \u{FFFD}' ); |
| 144 | } |
| 145 | if first < 0xF0 { |
| 146 | self.remaining = &self.remaining[3..]; |
| 147 | let point = ((u32::from(first) & 0xF) << 12) |
| 148 | | ((u32::from(second) & 0x3F) << 6) |
| 149 | | (u32::from(third) & 0x3F); |
| 150 | return Some(unsafe { char::from_u32_unchecked(point) }); |
| 151 | } |
| 152 | // At this point, we have a valid 3-byte prefix of a |
| 153 | // four-byte sequence that has to be incomplete, because |
| 154 | // otherwise `next()` would have succeeded. |
| 155 | self.remaining = &self.remaining[3..]; |
| 156 | Some(' \u{FFFD}' ) |
| 157 | } |
| 158 | } |
| 159 | |
| 160 | impl<'a> Iterator for Utf8Chars<'a> { |
| 161 | type Item = char; |
| 162 | |
| 163 | #[inline ] |
| 164 | fn next(&mut self) -> Option<char> { |
| 165 | // Not delegating directly to `ErrorReportingUtf8Chars` to avoid |
| 166 | // an extra branch in the common case based on a cursory inspection |
| 167 | // of generated code in a similar case. Be sure to inspect the |
| 168 | // generated code as inlined into an actual usage site carefully |
| 169 | // if attempting to consolidate the source code here. |
| 170 | |
| 171 | // This loop is only broken out of as goto forward |
| 172 | #[allow (clippy::never_loop)] |
| 173 | loop { |
| 174 | if self.remaining.len() < 4 { |
| 175 | break; |
| 176 | } |
| 177 | let first = self.remaining[0]; |
| 178 | if first < 0x80 { |
| 179 | self.remaining = &self.remaining[1..]; |
| 180 | return Some(char::from(first)); |
| 181 | } |
| 182 | let second = self.remaining[1]; |
| 183 | if in_inclusive_range8(first, 0xC2, 0xDF) { |
| 184 | if !in_inclusive_range8(second, 0x80, 0xBF) { |
| 185 | break; |
| 186 | } |
| 187 | let point = ((u32::from(first) & 0x1F) << 6) | (u32::from(second) & 0x3F); |
| 188 | self.remaining = &self.remaining[2..]; |
| 189 | return Some(unsafe { char::from_u32_unchecked(point) }); |
| 190 | } |
| 191 | // This table-based formulation was benchmark-based in encoding_rs, |
| 192 | // but it hasn't been re-benchmarked in this iterator context. |
| 193 | let third = self.remaining[2]; |
| 194 | if first < 0xF0 { |
| 195 | if ((UTF8_DATA.table[usize::from(second)] |
| 196 | & UTF8_DATA.table[usize::from(first) + 0x80]) |
| 197 | | (third >> 6)) |
| 198 | != 2 |
| 199 | { |
| 200 | break; |
| 201 | } |
| 202 | let point = ((u32::from(first) & 0xF) << 12) |
| 203 | | ((u32::from(second) & 0x3F) << 6) |
| 204 | | (u32::from(third) & 0x3F); |
| 205 | self.remaining = &self.remaining[3..]; |
| 206 | return Some(unsafe { char::from_u32_unchecked(point) }); |
| 207 | } |
| 208 | let fourth = self.remaining[3]; |
| 209 | if (u16::from( |
| 210 | UTF8_DATA.table[usize::from(second)] & UTF8_DATA.table[usize::from(first) + 0x80], |
| 211 | ) | u16::from(third >> 6) |
| 212 | | (u16::from(fourth & 0xC0) << 2)) |
| 213 | != 0x202 |
| 214 | { |
| 215 | break; |
| 216 | } |
| 217 | let point = ((u32::from(first) & 0x7) << 18) |
| 218 | | ((u32::from(second) & 0x3F) << 12) |
| 219 | | ((u32::from(third) & 0x3F) << 6) |
| 220 | | (u32::from(fourth) & 0x3F); |
| 221 | self.remaining = &self.remaining[4..]; |
| 222 | return Some(unsafe { char::from_u32_unchecked(point) }); |
| 223 | } |
| 224 | self.next_fallback() |
| 225 | } |
| 226 | } |
| 227 | |
| 228 | impl<'a> DoubleEndedIterator for Utf8Chars<'a> { |
| 229 | #[inline ] |
| 230 | fn next_back(&mut self) -> Option<char> { |
| 231 | if self.remaining.is_empty() { |
| 232 | return None; |
| 233 | } |
| 234 | let mut attempt = 1; |
| 235 | for b in self.remaining.iter().rev() { |
| 236 | if b & 0xC0 != 0x80 { |
| 237 | let (head, tail) = self.remaining.split_at(self.remaining.len() - attempt); |
| 238 | let mut inner = Utf8Chars::new(tail); |
| 239 | let candidate = inner.next(); |
| 240 | if inner.as_slice().is_empty() { |
| 241 | self.remaining = head; |
| 242 | return candidate; |
| 243 | } |
| 244 | break; |
| 245 | } |
| 246 | if attempt == 4 { |
| 247 | break; |
| 248 | } |
| 249 | attempt += 1; |
| 250 | } |
| 251 | |
| 252 | self.remaining = &self.remaining[..self.remaining.len() - 1]; |
| 253 | Some(' \u{FFFD}' ) |
| 254 | } |
| 255 | } |
| 256 | |
| 257 | impl FusedIterator for Utf8Chars<'_> {} |
| 258 | |
| 259 | /// Convenience trait that adds `chars()` and `char_indices()` methods |
| 260 | /// similar to the ones on string slices to byte slices. |
| 261 | pub trait Utf8CharsEx { |
| 262 | fn chars(&self) -> Utf8Chars<'_>; |
| 263 | fn char_indices(&self) -> Utf8CharIndices<'_>; |
| 264 | } |
| 265 | |
| 266 | impl Utf8CharsEx for [u8] { |
| 267 | /// Convenience method for creating an UTF-8 iterator |
| 268 | /// for the slice. |
| 269 | #[inline ] |
| 270 | fn chars(&self) -> Utf8Chars<'_> { |
| 271 | Utf8Chars::new(self) |
| 272 | } |
| 273 | /// Convenience method for creating a byte index and |
| 274 | /// UTF-8 iterator for the slice. |
| 275 | #[inline ] |
| 276 | fn char_indices(&self) -> Utf8CharIndices<'_> { |
| 277 | Utf8CharIndices::new(self) |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | // No manually-written tests for forward-iteration, because the code passed multiple |
| 282 | // days of fuzzing comparing with known-good behavior. |
| 283 | |