| 1 | // This file is part of ICU4X. For terms of use, please see the file |
| 2 | // called LICENSE at the top level of the ICU4X source tree |
| 3 | // (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ). |
| 4 | |
| 5 | use crate::error::{DataError, DataErrorKind}; |
| 6 | |
| 7 | use crate::fallback::{LocaleFallbackConfig, LocaleFallbackPriority, LocaleFallbackSupplement}; |
| 8 | use alloc::borrow::Cow; |
| 9 | use core::fmt; |
| 10 | use core::fmt::Write; |
| 11 | use core::ops::Deref; |
| 12 | use writeable::{LengthHint, Writeable}; |
| 13 | use zerovec::ule::*; |
| 14 | |
| 15 | #[doc (hidden)] |
| 16 | #[macro_export ] |
| 17 | macro_rules! leading_tag { |
| 18 | () => { |
| 19 | " \nicu4x_key_tag" |
| 20 | }; |
| 21 | } |
| 22 | |
| 23 | #[doc (hidden)] |
| 24 | #[macro_export ] |
| 25 | macro_rules! trailing_tag { |
| 26 | () => { |
| 27 | " \n" |
| 28 | }; |
| 29 | } |
| 30 | |
| 31 | #[doc (hidden)] |
| 32 | #[macro_export ] |
| 33 | macro_rules! tagged { |
| 34 | ($without_tags:expr) => { |
| 35 | concat!( |
| 36 | $crate::leading_tag!(), |
| 37 | $without_tags, |
| 38 | $crate::trailing_tag!() |
| 39 | ) |
| 40 | }; |
| 41 | } |
| 42 | |
| 43 | /// A compact hash of a [`DataKey`]. Useful for keys in maps. |
| 44 | /// |
| 45 | /// The hash will be stable over time within major releases. |
| 46 | #[derive (Debug, PartialEq, Eq, PartialOrd, Ord, Copy, Clone, Hash, ULE)] |
| 47 | #[cfg_attr (feature = "serde" , derive(serde::Serialize, serde::Deserialize))] |
| 48 | #[repr (transparent)] |
| 49 | pub struct DataKeyHash([u8; 4]); |
| 50 | |
| 51 | impl DataKeyHash { |
| 52 | const fn compute_from_path(path: DataKeyPath) -> Self { |
| 53 | let hash: u32 = fxhash_32( |
| 54 | path.tagged.as_bytes(), |
| 55 | ignore_leading:leading_tag!().len(), |
| 56 | ignore_trailing:trailing_tag!().len(), |
| 57 | ); |
| 58 | Self(hash.to_le_bytes()) |
| 59 | } |
| 60 | |
| 61 | /// Gets the hash value as a byte array. |
| 62 | pub const fn to_bytes(self) -> [u8; 4] { |
| 63 | self.0 |
| 64 | } |
| 65 | } |
| 66 | |
| 67 | /// Const function to compute the FxHash of a byte array. |
| 68 | /// |
| 69 | /// FxHash is a speedy hash algorithm used within rustc. The algorithm is satisfactory for our |
| 70 | /// use case since the strings being hashed originate from a trusted source (the ICU4X |
| 71 | /// components), and the hashes are computed at compile time, so we can check for collisions. |
| 72 | /// |
| 73 | /// We could have considered a SHA or other cryptographic hash function. However, we are using |
| 74 | /// FxHash because: |
| 75 | /// |
| 76 | /// 1. There is precedent for this algorithm in Rust |
| 77 | /// 2. The algorithm is easy to implement as a const function |
| 78 | /// 3. The amount of code is small enough that we can reasonably keep the algorithm in-tree |
| 79 | /// 4. FxHash is designed to output 32-bit or 64-bit values, whereas SHA outputs more bits, |
| 80 | /// such that truncation would be required in order to fit into a u32, partially reducing |
| 81 | /// the benefit of a cryptographically secure algorithm |
| 82 | // The indexing operations in this function have been reviewed in detail and won't panic. |
| 83 | #[allow (clippy::indexing_slicing)] |
| 84 | const fn fxhash_32(bytes: &[u8], ignore_leading: usize, ignore_trailing: usize) -> u32 { |
| 85 | // This code is adapted from https://github.com/rust-lang/rustc-hash, |
| 86 | // whose license text is reproduced below. |
| 87 | // |
| 88 | // Copyright 2015 The Rust Project Developers. See the COPYRIGHT |
| 89 | // file at the top-level directory of this distribution and at |
| 90 | // http://rust-lang.org/COPYRIGHT. |
| 91 | // |
| 92 | // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
| 93 | // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
| 94 | // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your |
| 95 | // option. This file may not be copied, modified, or distributed |
| 96 | // except according to those terms. |
| 97 | |
| 98 | if ignore_leading + ignore_trailing >= bytes.len() { |
| 99 | return 0; |
| 100 | } |
| 101 | |
| 102 | #[inline ] |
| 103 | const fn hash_word_32(mut hash: u32, word: u32) -> u32 { |
| 104 | const ROTATE: u32 = 5; |
| 105 | const SEED32: u32 = 0x9e_37_79_b9; |
| 106 | hash = hash.rotate_left(ROTATE); |
| 107 | hash ^= word; |
| 108 | hash = hash.wrapping_mul(SEED32); |
| 109 | hash |
| 110 | } |
| 111 | |
| 112 | let mut cursor = ignore_leading; |
| 113 | let end = bytes.len() - ignore_trailing; |
| 114 | let mut hash = 0; |
| 115 | |
| 116 | while end - cursor >= 4 { |
| 117 | let word = u32::from_le_bytes([ |
| 118 | bytes[cursor], |
| 119 | bytes[cursor + 1], |
| 120 | bytes[cursor + 2], |
| 121 | bytes[cursor + 3], |
| 122 | ]); |
| 123 | hash = hash_word_32(hash, word); |
| 124 | cursor += 4; |
| 125 | } |
| 126 | |
| 127 | if end - cursor >= 2 { |
| 128 | let word = u16::from_le_bytes([bytes[cursor], bytes[cursor + 1]]); |
| 129 | hash = hash_word_32(hash, word as u32); |
| 130 | cursor += 2; |
| 131 | } |
| 132 | |
| 133 | if end - cursor >= 1 { |
| 134 | hash = hash_word_32(hash, bytes[cursor] as u32); |
| 135 | } |
| 136 | |
| 137 | hash |
| 138 | } |
| 139 | |
| 140 | impl<'a> zerovec::maps::ZeroMapKV<'a> for DataKeyHash { |
| 141 | type Container = zerovec::ZeroVec<'a, DataKeyHash>; |
| 142 | type Slice = zerovec::ZeroSlice<DataKeyHash>; |
| 143 | type GetType = <DataKeyHash as AsULE>::ULE; |
| 144 | type OwnedType = DataKeyHash; |
| 145 | } |
| 146 | |
| 147 | impl AsULE for DataKeyHash { |
| 148 | type ULE = Self; |
| 149 | #[inline ] |
| 150 | fn to_unaligned(self) -> Self::ULE { |
| 151 | self |
| 152 | } |
| 153 | #[inline ] |
| 154 | fn from_unaligned(unaligned: Self::ULE) -> Self { |
| 155 | unaligned |
| 156 | } |
| 157 | } |
| 158 | |
| 159 | // Safe since the ULE type is `self`. |
| 160 | unsafe impl EqULE for DataKeyHash {} |
| 161 | |
| 162 | /// The string path of a data key. For example, "foo@1" |
| 163 | #[derive (Debug, Copy, Clone, PartialEq, Eq, PartialOrd, Ord)] |
| 164 | pub struct DataKeyPath { |
| 165 | // This string literal is wrapped in leading_tag!() and trailing_tag!() to make it detectable |
| 166 | // in a compiled binary. |
| 167 | tagged: &'static str, |
| 168 | } |
| 169 | |
| 170 | impl DataKeyPath { |
| 171 | /// Gets the path as a static string slice. |
| 172 | #[inline ] |
| 173 | pub const fn get(self) -> &'static str { |
| 174 | unsafe { |
| 175 | // Safe due to invariant that self.path is tagged correctly |
| 176 | core::str::from_utf8_unchecked(core::slice::from_raw_parts( |
| 177 | self.tagged.as_ptr().add(leading_tag!().len()), |
| 178 | self.tagged.len() - trailing_tag!().len() - leading_tag!().len(), |
| 179 | )) |
| 180 | } |
| 181 | } |
| 182 | } |
| 183 | |
| 184 | impl Deref for DataKeyPath { |
| 185 | type Target = str; |
| 186 | #[inline ] |
| 187 | fn deref(&self) -> &Self::Target { |
| 188 | self.get() |
| 189 | } |
| 190 | } |
| 191 | |
| 192 | /// Metadata statically associated with a particular [`DataKey`]. |
| 193 | #[derive (Debug, PartialEq, Eq, Copy, Clone, PartialOrd, Ord)] |
| 194 | #[non_exhaustive ] |
| 195 | pub struct DataKeyMetadata { |
| 196 | /// What to prioritize when fallbacking on this [`DataKey`]. |
| 197 | pub fallback_priority: LocaleFallbackPriority, |
| 198 | /// A Unicode extension keyword to consider when loading data for this [`DataKey`]. |
| 199 | pub extension_key: Option<icu_locid::extensions::unicode::Key>, |
| 200 | /// Optional choice for additional fallbacking data required for loading this marker. |
| 201 | /// |
| 202 | /// For more information, see `LocaleFallbackConfig::fallback_supplement`. |
| 203 | pub fallback_supplement: Option<LocaleFallbackSupplement>, |
| 204 | /// Whether the key has a singleton value, as opposed to per-locale values. Singleton |
| 205 | /// keys behave differently, e.g. they never perform fallback, and can be optimized |
| 206 | /// in data providers. |
| 207 | pub singleton: bool, |
| 208 | } |
| 209 | |
| 210 | impl DataKeyMetadata { |
| 211 | /// Const-friendly version of [`Default::default`]. |
| 212 | pub const fn const_default() -> Self { |
| 213 | Self { |
| 214 | fallback_priority: LocaleFallbackPriority::const_default(), |
| 215 | extension_key: None, |
| 216 | fallback_supplement: None, |
| 217 | singleton: false, |
| 218 | } |
| 219 | } |
| 220 | |
| 221 | #[doc (hidden)] |
| 222 | pub const fn construct_internal( |
| 223 | fallback_priority: LocaleFallbackPriority, |
| 224 | extension_key: Option<icu_locid::extensions::unicode::Key>, |
| 225 | fallback_supplement: Option<LocaleFallbackSupplement>, |
| 226 | singleton: bool, |
| 227 | ) -> Self { |
| 228 | Self { |
| 229 | fallback_priority, |
| 230 | extension_key, |
| 231 | fallback_supplement, |
| 232 | singleton, |
| 233 | } |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | impl Default for DataKeyMetadata { |
| 238 | #[inline ] |
| 239 | fn default() -> Self { |
| 240 | Self::const_default() |
| 241 | } |
| 242 | } |
| 243 | |
| 244 | /// Used for loading data from an ICU4X data provider. |
| 245 | /// |
| 246 | /// A resource key is tightly coupled with the code that uses it to load data at runtime. |
| 247 | /// Executables can be searched for `DataKey` instances to produce optimized data files. |
| 248 | /// Therefore, users should not generally create DataKey instances; they should instead use |
| 249 | /// the ones exported by a component. |
| 250 | /// |
| 251 | /// `DataKey`s are created with the [`data_key!`](crate::data_key) macro: |
| 252 | /// |
| 253 | /// ``` |
| 254 | /// # use icu_provider::DataKey; |
| 255 | /// const K: DataKey = icu_provider::data_key!("foo/bar@1" ); |
| 256 | /// ``` |
| 257 | /// |
| 258 | /// The human-readable path string ends with `@` followed by one or more digits (the version |
| 259 | /// number). Paths do not contain characters other than ASCII letters and digits, `_`, `/`. |
| 260 | /// |
| 261 | /// Invalid paths are compile-time errors (as [`data_key!`](crate::data_key) uses `const`). |
| 262 | /// |
| 263 | /// ```compile_fail,E0080 |
| 264 | /// # use icu_provider::DataKey; |
| 265 | /// const K: DataKey = icu_provider::data_key!("foo/../bar@1" ); |
| 266 | /// ``` |
| 267 | #[derive (Copy, Clone)] |
| 268 | pub struct DataKey { |
| 269 | path: DataKeyPath, |
| 270 | hash: DataKeyHash, |
| 271 | metadata: DataKeyMetadata, |
| 272 | } |
| 273 | |
| 274 | impl PartialEq for DataKey { |
| 275 | #[inline ] |
| 276 | fn eq(&self, other: &Self) -> bool { |
| 277 | self.hash == other.hash && self.path == other.path && self.metadata == other.metadata |
| 278 | } |
| 279 | } |
| 280 | |
| 281 | impl Eq for DataKey {} |
| 282 | |
| 283 | impl Ord for DataKey { |
| 284 | fn cmp(&self, other: &Self) -> core::cmp::Ordering { |
| 285 | self.path |
| 286 | .cmp(&other.path) |
| 287 | .then_with(|| self.metadata.cmp(&other.metadata)) |
| 288 | } |
| 289 | } |
| 290 | |
| 291 | impl PartialOrd for DataKey { |
| 292 | #[inline ] |
| 293 | fn partial_cmp(&self, other: &Self) -> Option<core::cmp::Ordering> { |
| 294 | Some(self.cmp(other)) |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | impl core::hash::Hash for DataKey { |
| 299 | #[inline ] |
| 300 | fn hash<H: core::hash::Hasher>(&self, state: &mut H) { |
| 301 | self.hash.hash(state) |
| 302 | } |
| 303 | } |
| 304 | |
| 305 | impl DataKey { |
| 306 | /// Gets a human-readable representation of a [`DataKey`]. |
| 307 | /// |
| 308 | /// The human-readable path string ends with `@` followed by one or more digits (the version |
| 309 | /// number). Paths do not contain characters other than ASCII letters and digits, `_`, `/`. |
| 310 | /// |
| 311 | /// Useful for reading and writing data to a file system. |
| 312 | #[inline ] |
| 313 | pub const fn path(self) -> DataKeyPath { |
| 314 | self.path |
| 315 | } |
| 316 | |
| 317 | /// Gets a platform-independent hash of a [`DataKey`]. |
| 318 | /// |
| 319 | /// The hash is 4 bytes and allows for fast key comparison. |
| 320 | /// |
| 321 | /// # Example |
| 322 | /// |
| 323 | /// ``` |
| 324 | /// use icu_provider::DataKey; |
| 325 | /// use icu_provider::DataKeyHash; |
| 326 | /// |
| 327 | /// const KEY: DataKey = icu_provider::data_key!("foo@1" ); |
| 328 | /// const KEY_HASH: DataKeyHash = KEY.hashed(); |
| 329 | /// |
| 330 | /// assert_eq!(KEY_HASH.to_bytes(), [0xe2, 0xb6, 0x17, 0x71]); |
| 331 | /// ``` |
| 332 | #[inline ] |
| 333 | pub const fn hashed(self) -> DataKeyHash { |
| 334 | self.hash |
| 335 | } |
| 336 | |
| 337 | /// Gets the metadata associated with this [`DataKey`]. |
| 338 | #[inline ] |
| 339 | pub const fn metadata(self) -> DataKeyMetadata { |
| 340 | self.metadata |
| 341 | } |
| 342 | |
| 343 | /// Returns the [`LocaleFallbackConfig`] for this [`DataKey`]. |
| 344 | #[inline ] |
| 345 | pub const fn fallback_config(self) -> LocaleFallbackConfig { |
| 346 | let mut config = LocaleFallbackConfig::const_default(); |
| 347 | config.priority = self.metadata.fallback_priority; |
| 348 | config.extension_key = self.metadata.extension_key; |
| 349 | config.fallback_supplement = self.metadata.fallback_supplement; |
| 350 | config |
| 351 | } |
| 352 | |
| 353 | /// Constructs a [`DataKey`] from a path and metadata. |
| 354 | /// |
| 355 | /// # Examples |
| 356 | /// |
| 357 | /// ``` |
| 358 | /// use icu_provider::data_key; |
| 359 | /// use icu_provider::DataKey; |
| 360 | /// |
| 361 | /// const CONST_KEY: DataKey = data_key!("foo@1" ); |
| 362 | /// |
| 363 | /// let runtime_key = |
| 364 | /// DataKey::from_path_and_metadata(CONST_KEY.path(), CONST_KEY.metadata()); |
| 365 | /// |
| 366 | /// assert_eq!(CONST_KEY, runtime_key); |
| 367 | /// ``` |
| 368 | #[inline ] |
| 369 | pub const fn from_path_and_metadata(path: DataKeyPath, metadata: DataKeyMetadata) -> Self { |
| 370 | Self { |
| 371 | path, |
| 372 | hash: DataKeyHash::compute_from_path(path), |
| 373 | metadata, |
| 374 | } |
| 375 | } |
| 376 | |
| 377 | #[doc (hidden)] |
| 378 | // Error is a str of the expected character class and the index where it wasn't encountered |
| 379 | // The indexing operations in this function have been reviewed in detail and won't panic. |
| 380 | #[allow (clippy::indexing_slicing)] |
| 381 | pub const fn construct_internal( |
| 382 | path: &'static str, |
| 383 | metadata: DataKeyMetadata, |
| 384 | ) -> Result<Self, (&'static str, usize)> { |
| 385 | if path.len() < leading_tag!().len() + trailing_tag!().len() { |
| 386 | return Err(("tag" , 0)); |
| 387 | } |
| 388 | // Start and end of the untagged part |
| 389 | let start = leading_tag!().len(); |
| 390 | let end = path.len() - trailing_tag!().len(); |
| 391 | |
| 392 | // Check tags |
| 393 | let mut i = 0; |
| 394 | while i < leading_tag!().len() { |
| 395 | if path.as_bytes()[i] != leading_tag!().as_bytes()[i] { |
| 396 | return Err(("tag" , 0)); |
| 397 | } |
| 398 | i += 1; |
| 399 | } |
| 400 | i = 0; |
| 401 | while i < trailing_tag!().len() { |
| 402 | if path.as_bytes()[end + i] != trailing_tag!().as_bytes()[i] { |
| 403 | return Err(("tag" , end + 1)); |
| 404 | } |
| 405 | i += 1; |
| 406 | } |
| 407 | |
| 408 | match Self::validate_path_manual_slice(path, start, end) { |
| 409 | Ok(()) => (), |
| 410 | Err(e) => return Err(e), |
| 411 | }; |
| 412 | |
| 413 | let path = DataKeyPath { tagged: path }; |
| 414 | |
| 415 | Ok(Self { |
| 416 | path, |
| 417 | hash: DataKeyHash::compute_from_path(path), |
| 418 | metadata, |
| 419 | }) |
| 420 | } |
| 421 | |
| 422 | const fn validate_path_manual_slice( |
| 423 | path: &'static str, |
| 424 | start: usize, |
| 425 | end: usize, |
| 426 | ) -> Result<(), (&'static str, usize)> { |
| 427 | debug_assert!(start <= end); |
| 428 | debug_assert!(end <= path.len()); |
| 429 | // Regex: [a-zA-Z0-9_][a-zA-Z0-9_/]*@[0-9]+ |
| 430 | enum State { |
| 431 | Empty, |
| 432 | Body, |
| 433 | At, |
| 434 | Version, |
| 435 | } |
| 436 | use State::*; |
| 437 | let mut i = start; |
| 438 | let mut state = Empty; |
| 439 | loop { |
| 440 | let byte = if i < end { |
| 441 | #[allow (clippy::indexing_slicing)] // protected by debug assertion |
| 442 | Some(path.as_bytes()[i]) |
| 443 | } else { |
| 444 | None |
| 445 | }; |
| 446 | state = match (state, byte) { |
| 447 | (Empty | Body, Some(b'a' ..=b'z' | b'A' ..=b'Z' | b'0' ..=b'9' | b'_' )) => Body, |
| 448 | (Body, Some(b'/' )) => Body, |
| 449 | (Body, Some(b'@' )) => At, |
| 450 | (At | Version, Some(b'0' ..=b'9' )) => Version, |
| 451 | // One of these cases will be hit at the latest when i == end, so the loop converges. |
| 452 | (Version, None) => { |
| 453 | return Ok(()); |
| 454 | } |
| 455 | |
| 456 | (Empty, _) => return Err(("[a-zA-Z0-9_]" , i)), |
| 457 | (Body, _) => return Err(("[a-zA-z0-9_/@]" , i)), |
| 458 | (At, _) => return Err(("[0-9]" , i)), |
| 459 | (Version, _) => return Err(("[0-9]" , i)), |
| 460 | }; |
| 461 | i += 1; |
| 462 | } |
| 463 | } |
| 464 | |
| 465 | /// Returns [`Ok`] if this data key matches the argument, or the appropriate error. |
| 466 | /// |
| 467 | /// Convenience method for data providers that support a single [`DataKey`]. |
| 468 | /// |
| 469 | /// # Examples |
| 470 | /// |
| 471 | /// ``` |
| 472 | /// use icu_provider::prelude::*; |
| 473 | /// |
| 474 | /// const FOO_BAR: DataKey = icu_provider::data_key!("foo/bar@1" ); |
| 475 | /// const FOO_BAZ: DataKey = icu_provider::data_key!("foo/baz@1" ); |
| 476 | /// const BAR_BAZ: DataKey = icu_provider::data_key!("bar/baz@1" ); |
| 477 | /// |
| 478 | /// assert!(matches!(FOO_BAR.match_key(FOO_BAR), Ok(()))); |
| 479 | /// assert!(matches!( |
| 480 | /// FOO_BAR.match_key(FOO_BAZ), |
| 481 | /// Err(DataError { |
| 482 | /// kind: DataErrorKind::MissingDataKey, |
| 483 | /// .. |
| 484 | /// }) |
| 485 | /// )); |
| 486 | /// assert!(matches!( |
| 487 | /// FOO_BAR.match_key(BAR_BAZ), |
| 488 | /// Err(DataError { |
| 489 | /// kind: DataErrorKind::MissingDataKey, |
| 490 | /// .. |
| 491 | /// }) |
| 492 | /// )); |
| 493 | /// |
| 494 | /// // The error context contains the argument: |
| 495 | /// assert_eq!(FOO_BAR.match_key(BAR_BAZ).unwrap_err().key, Some(BAR_BAZ)); |
| 496 | /// ``` |
| 497 | pub fn match_key(self, key: Self) -> Result<(), DataError> { |
| 498 | if self == key { |
| 499 | Ok(()) |
| 500 | } else { |
| 501 | Err(DataErrorKind::MissingDataKey.with_key(key)) |
| 502 | } |
| 503 | } |
| 504 | } |
| 505 | |
| 506 | /// See [`DataKey`]. |
| 507 | #[macro_export ] |
| 508 | macro_rules! data_key { |
| 509 | ($path:expr) => {{ |
| 510 | $crate::data_key!($path, $crate::DataKeyMetadata::const_default()) |
| 511 | }}; |
| 512 | ($path:expr, $metadata:expr) => {{ |
| 513 | // Force the DataKey into a const context |
| 514 | const RESOURCE_KEY_MACRO_CONST: $crate::DataKey = { |
| 515 | match $crate::DataKey::construct_internal($crate::tagged!($path), $metadata) { |
| 516 | Ok(v) => v, |
| 517 | #[allow(clippy::panic)] // Const context |
| 518 | Err(_) => panic!(concat!("Invalid resource key: " , $path)), |
| 519 | // TODO Once formatting is const: |
| 520 | // Err((expected, index)) => panic!( |
| 521 | // "Invalid resource key {:?}: expected {:?}, found {:?} ", |
| 522 | // $path, |
| 523 | // expected, |
| 524 | // $crate::tagged!($path).get(index..)) |
| 525 | // ); |
| 526 | } |
| 527 | }; |
| 528 | RESOURCE_KEY_MACRO_CONST |
| 529 | }}; |
| 530 | } |
| 531 | |
| 532 | impl fmt::Debug for DataKey { |
| 533 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 534 | f.write_str(data:"DataKey{" )?; |
| 535 | fmt::Display::fmt(self, f)?; |
| 536 | f.write_char('}' )?; |
| 537 | Ok(()) |
| 538 | } |
| 539 | } |
| 540 | |
| 541 | impl Writeable for DataKey { |
| 542 | fn write_to<W: core::fmt::Write + ?Sized>(&self, sink: &mut W) -> core::fmt::Result { |
| 543 | self.path().write_to(sink) |
| 544 | } |
| 545 | |
| 546 | fn writeable_length_hint(&self) -> LengthHint { |
| 547 | self.path().writeable_length_hint() |
| 548 | } |
| 549 | |
| 550 | fn write_to_string(&self) -> Cow<str> { |
| 551 | Cow::Borrowed(self.path().get()) |
| 552 | } |
| 553 | } |
| 554 | |
| 555 | writeable::impl_display_with_writeable!(DataKey); |
| 556 | |
| 557 | #[test ] |
| 558 | fn test_path_syntax() { |
| 559 | // Valid keys: |
| 560 | DataKey::construct_internal(tagged!("hello/world@1" ), Default::default()).unwrap(); |
| 561 | DataKey::construct_internal(tagged!("hello/world/foo@1" ), Default::default()).unwrap(); |
| 562 | DataKey::construct_internal(tagged!("hello/world@999" ), Default::default()).unwrap(); |
| 563 | DataKey::construct_internal(tagged!("hello_world/foo@1" ), Default::default()).unwrap(); |
| 564 | DataKey::construct_internal(tagged!("hello_458/world@1" ), Default::default()).unwrap(); |
| 565 | DataKey::construct_internal(tagged!("hello_world@1" ), Default::default()).unwrap(); |
| 566 | |
| 567 | // No version: |
| 568 | assert_eq!( |
| 569 | DataKey::construct_internal(tagged!("hello/world" ), Default::default()), |
| 570 | Err(( |
| 571 | "[a-zA-z0-9_/@]" , |
| 572 | concat!(leading_tag!(), "hello/world" ).len() |
| 573 | )) |
| 574 | ); |
| 575 | |
| 576 | assert_eq!( |
| 577 | DataKey::construct_internal(tagged!("hello/world@" ), Default::default()), |
| 578 | Err(("[0-9]" , concat!(leading_tag!(), "hello/world@" ).len())) |
| 579 | ); |
| 580 | assert_eq!( |
| 581 | DataKey::construct_internal(tagged!("hello/world@foo" ), Default::default()), |
| 582 | Err(("[0-9]" , concat!(leading_tag!(), "hello/world@" ).len())) |
| 583 | ); |
| 584 | assert_eq!( |
| 585 | DataKey::construct_internal(tagged!("hello/world@1foo" ), Default::default()), |
| 586 | Err(("[0-9]" , concat!(leading_tag!(), "hello/world@1" ).len())) |
| 587 | ); |
| 588 | |
| 589 | // Meta no longer accepted: |
| 590 | assert_eq!( |
| 591 | DataKey::construct_internal(tagged!("foo@1[R]" ), Default::default()), |
| 592 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 593 | ); |
| 594 | assert_eq!( |
| 595 | DataKey::construct_internal(tagged!("foo@1[u-ca]" ), Default::default()), |
| 596 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 597 | ); |
| 598 | assert_eq!( |
| 599 | DataKey::construct_internal(tagged!("foo@1[R][u-ca]" ), Default::default()), |
| 600 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 601 | ); |
| 602 | |
| 603 | // Invalid meta: |
| 604 | assert_eq!( |
| 605 | DataKey::construct_internal(tagged!("foo@1[U]" ), Default::default()), |
| 606 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 607 | ); |
| 608 | assert_eq!( |
| 609 | DataKey::construct_internal(tagged!("foo@1[uca]" ), Default::default()), |
| 610 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 611 | ); |
| 612 | assert_eq!( |
| 613 | DataKey::construct_internal(tagged!("foo@1[u-" ), Default::default()), |
| 614 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 615 | ); |
| 616 | assert_eq!( |
| 617 | DataKey::construct_internal(tagged!("foo@1[u-caa]" ), Default::default()), |
| 618 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 619 | ); |
| 620 | assert_eq!( |
| 621 | DataKey::construct_internal(tagged!("foo@1[R" ), Default::default()), |
| 622 | Err(("[0-9]" , concat!(leading_tag!(), "foo@1" ).len())) |
| 623 | ); |
| 624 | |
| 625 | // Invalid characters: |
| 626 | assert_eq!( |
| 627 | DataKey::construct_internal(tagged!("你好/世界@1" ), Default::default()), |
| 628 | Err(("[a-zA-Z0-9_]" , leading_tag!().len())) |
| 629 | ); |
| 630 | |
| 631 | // Invalid tag: |
| 632 | assert_eq!( |
| 633 | DataKey::construct_internal( |
| 634 | concat!("hello/world@1" , trailing_tag!()), |
| 635 | Default::default() |
| 636 | ), |
| 637 | Err(("tag" , 0)) |
| 638 | ); |
| 639 | assert_eq!( |
| 640 | DataKey::construct_internal(concat!(leading_tag!(), "hello/world@1" ), Default::default()), |
| 641 | Err(("tag" , concat!(leading_tag!(), "hello/world@1" ).len())) |
| 642 | ); |
| 643 | assert_eq!( |
| 644 | DataKey::construct_internal("hello/world@1" , Default::default()), |
| 645 | Err(("tag" , 0)) |
| 646 | ); |
| 647 | } |
| 648 | |
| 649 | #[test ] |
| 650 | fn test_key_to_string() { |
| 651 | struct KeyTestCase { |
| 652 | pub key: DataKey, |
| 653 | pub expected: &'static str, |
| 654 | } |
| 655 | |
| 656 | for cas in [ |
| 657 | KeyTestCase { |
| 658 | key: data_key!("core/cardinal@1" ), |
| 659 | expected: "core/cardinal@1" , |
| 660 | }, |
| 661 | KeyTestCase { |
| 662 | key: data_key!("core/maxlengthsubcatg@1" ), |
| 663 | expected: "core/maxlengthsubcatg@1" , |
| 664 | }, |
| 665 | KeyTestCase { |
| 666 | key: data_key!("core/cardinal@65535" ), |
| 667 | expected: "core/cardinal@65535" , |
| 668 | }, |
| 669 | ] { |
| 670 | writeable::assert_writeable_eq!(&cas.key, cas.expected); |
| 671 | assert_eq!(cas.expected, &*cas.key.path()); |
| 672 | } |
| 673 | } |
| 674 | |
| 675 | #[test ] |
| 676 | fn test_hash_word_32() { |
| 677 | assert_eq!(0, fxhash_32(b"" , 0, 0)); |
| 678 | assert_eq!(0, fxhash_32(b"a" , 1, 0)); |
| 679 | assert_eq!(0, fxhash_32(b"a" , 0, 1)); |
| 680 | assert_eq!(0, fxhash_32(b"a" , 0, 10)); |
| 681 | assert_eq!(0, fxhash_32(b"a" , 10, 0)); |
| 682 | assert_eq!(0, fxhash_32(b"a" , 1, 1)); |
| 683 | assert_eq!(0xF3051F19, fxhash_32(b"a" , 0, 0)); |
| 684 | assert_eq!(0x2F9DF119, fxhash_32(b"ab" , 0, 0)); |
| 685 | assert_eq!(0xCB1D9396, fxhash_32(b"abc" , 0, 0)); |
| 686 | assert_eq!(0x8628F119, fxhash_32(b"abcd" , 0, 0)); |
| 687 | assert_eq!(0xBEBDB56D, fxhash_32(b"abcde" , 0, 0)); |
| 688 | assert_eq!(0x1CE8476D, fxhash_32(b"abcdef" , 0, 0)); |
| 689 | assert_eq!(0xC0F176A4, fxhash_32(b"abcdefg" , 0, 0)); |
| 690 | assert_eq!(0x09AB476D, fxhash_32(b"abcdefgh" , 0, 0)); |
| 691 | assert_eq!(0xB72F5D88, fxhash_32(b"abcdefghi" , 0, 0)); |
| 692 | } |
| 693 | |
| 694 | #[test ] |
| 695 | fn test_key_hash() { |
| 696 | struct KeyTestCase { |
| 697 | pub key: DataKey, |
| 698 | pub hash: DataKeyHash, |
| 699 | } |
| 700 | |
| 701 | for cas in [ |
| 702 | KeyTestCase { |
| 703 | key: data_key!("core/cardinal@1" ), |
| 704 | hash: DataKeyHash([172, 207, 42, 236]), |
| 705 | }, |
| 706 | KeyTestCase { |
| 707 | key: data_key!("core/maxlengthsubcatg@1" ), |
| 708 | hash: DataKeyHash([193, 6, 79, 61]), |
| 709 | }, |
| 710 | KeyTestCase { |
| 711 | key: data_key!("core/cardinal@65535" ), |
| 712 | hash: DataKeyHash([176, 131, 182, 223]), |
| 713 | }, |
| 714 | ] { |
| 715 | assert_eq!(cas.hash, cas.key.hashed(), "{}" , cas.key); |
| 716 | } |
| 717 | } |
| 718 | |