| 1 | //! # simd-adler32 |
| 2 | //! |
| 3 | //! A SIMD-accelerated Adler-32 hash algorithm implementation. |
| 4 | //! |
| 5 | //! ## Features |
| 6 | //! |
| 7 | //! - No dependencies |
| 8 | //! - Support `no_std` (with `default-features = false`) |
| 9 | //! - Runtime CPU feature detection (when `std` enabled) |
| 10 | //! - Blazing fast performance on as many targets as possible (currently only x86 and x86_64) |
| 11 | //! - Default to scalar implementation when simd not available |
| 12 | //! |
| 13 | //! ## Quick start |
| 14 | //! |
| 15 | //! > Cargo.toml |
| 16 | //! |
| 17 | //! ```toml |
| 18 | //! [dependencies] |
| 19 | //! simd-adler32 = "*" |
| 20 | //! ``` |
| 21 | //! |
| 22 | //! > example.rs |
| 23 | //! |
| 24 | //! ```rust |
| 25 | //! use simd_adler32::Adler32; |
| 26 | //! |
| 27 | //! let mut adler = Adler32::new(); |
| 28 | //! adler.write(b"rust is pretty cool, man" ); |
| 29 | //! let hash = adler.finish(); |
| 30 | //! |
| 31 | //! println!("{}" , hash); |
| 32 | //! // 1921255656 |
| 33 | //! ``` |
| 34 | //! |
| 35 | //! ## Feature flags |
| 36 | //! |
| 37 | //! * `std` - Enabled by default |
| 38 | //! |
| 39 | //! Enables std support, see [CPU Feature Detection](#cpu-feature-detection) for runtime |
| 40 | //! detection support. |
| 41 | //! * `nightly` |
| 42 | //! |
| 43 | //! Enables nightly features required for avx512 support. |
| 44 | //! |
| 45 | //! * `const-generics` - Enabled by default |
| 46 | //! |
| 47 | //! Enables const-generics support allowing for user-defined array hashing by value. See |
| 48 | //! [`Adler32Hash`] for details. |
| 49 | //! |
| 50 | //! ## Support |
| 51 | //! |
| 52 | //! **CPU Features** |
| 53 | //! |
| 54 | //! | impl | arch | feature | |
| 55 | //! | ---- | ---------------- | ------- | |
| 56 | //! | ✅ | `x86`, `x86_64` | avx512 | |
| 57 | //! | ✅ | `x86`, `x86_64` | avx2 | |
| 58 | //! | ✅ | `x86`, `x86_64` | ssse3 | |
| 59 | //! | ✅ | `x86`, `x86_64` | sse2 | |
| 60 | //! | 🚧 | `arm`, `aarch64` | neon | |
| 61 | //! | | `wasm32` | simd128 | |
| 62 | //! |
| 63 | //! **MSRV** `1.36.0`\*\* |
| 64 | //! |
| 65 | //! Minimum supported rust version is tested before a new version is published. [**] Feature |
| 66 | //! `const-generics` needs to disabled to build on rustc versions `<1.51` which can be done |
| 67 | //! by updating your dependency definition to the following. |
| 68 | //! |
| 69 | //! ## CPU Feature Detection |
| 70 | //! simd-adler32 supports both runtime and compile time CPU feature detection using the |
| 71 | //! `std::is_x86_feature_detected` macro when the `Adler32` struct is instantiated with |
| 72 | //! the `new` fn. |
| 73 | //! |
| 74 | //! Without `std` feature enabled simd-adler32 falls back to compile time feature detection |
| 75 | //! using `target-feature` or `target-cpu` flags supplied to rustc. See [https://rust-lang.github.io/packed_simd/perf-guide/target-feature/rustflags.html](https://rust-lang.github.io/packed_simd/perf-guide/target-feature/rustflags.html) |
| 76 | //! for more information. |
| 77 | //! |
| 78 | //! Feature detection tries to use the fastest supported feature first. |
| 79 | #![cfg_attr (not(feature = "std" ), no_std)] |
| 80 | #![cfg_attr (feature = "nightly" , feature(stdsimd, avx512_target_feature))] |
| 81 | |
| 82 | #[doc (hidden)] |
| 83 | pub mod hash; |
| 84 | #[doc (hidden)] |
| 85 | pub mod imp; |
| 86 | |
| 87 | pub use hash::*; |
| 88 | use imp::{get_imp, Adler32Imp}; |
| 89 | |
| 90 | /// An adler32 hash generator type. |
| 91 | #[derive (Clone)] |
| 92 | pub struct Adler32 { |
| 93 | a: u16, |
| 94 | b: u16, |
| 95 | update: Adler32Imp, |
| 96 | } |
| 97 | |
| 98 | impl Adler32 { |
| 99 | /// Constructs a new `Adler32`. |
| 100 | /// |
| 101 | /// Potential overhead here due to runtime feature detection although in testing on 100k |
| 102 | /// and 10k random byte arrays it was not really noticeable. |
| 103 | /// |
| 104 | /// # Examples |
| 105 | /// ```rust |
| 106 | /// use simd_adler32::Adler32; |
| 107 | /// |
| 108 | /// let mut adler = Adler32::new(); |
| 109 | /// ``` |
| 110 | pub fn new() -> Self { |
| 111 | Default::default() |
| 112 | } |
| 113 | |
| 114 | /// Constructs a new `Adler32` using existing checksum. |
| 115 | /// |
| 116 | /// Potential overhead here due to runtime feature detection although in testing on 100k |
| 117 | /// and 10k random byte arrays it was not really noticeable. |
| 118 | /// |
| 119 | /// # Examples |
| 120 | /// ```rust |
| 121 | /// use simd_adler32::Adler32; |
| 122 | /// |
| 123 | /// let mut adler = Adler32::from_checksum(0xdeadbeaf); |
| 124 | /// ``` |
| 125 | pub fn from_checksum(checksum: u32) -> Self { |
| 126 | Self { |
| 127 | a: checksum as u16, |
| 128 | b: (checksum >> 16) as u16, |
| 129 | update: get_imp(), |
| 130 | } |
| 131 | } |
| 132 | |
| 133 | /// Computes hash for supplied data and stores results in internal state. |
| 134 | pub fn write(&mut self, data: &[u8]) { |
| 135 | let (a, b) = (self.update)(self.a, self.b, data); |
| 136 | |
| 137 | self.a = a; |
| 138 | self.b = b; |
| 139 | } |
| 140 | |
| 141 | /// Returns the hash value for the values written so far. |
| 142 | /// |
| 143 | /// Despite its name, the method does not reset the hasher’s internal state. Additional |
| 144 | /// writes will continue from the current value. If you need to start a fresh hash |
| 145 | /// value, you will have to use `reset`. |
| 146 | pub fn finish(&self) -> u32 { |
| 147 | (u32::from(self.b) << 16) | u32::from(self.a) |
| 148 | } |
| 149 | |
| 150 | /// Resets the internal state. |
| 151 | pub fn reset(&mut self) { |
| 152 | self.a = 1; |
| 153 | self.b = 0; |
| 154 | } |
| 155 | } |
| 156 | |
| 157 | /// Compute Adler-32 hash on `Adler32Hash` type. |
| 158 | /// |
| 159 | /// # Arguments |
| 160 | /// * `hash` - A Adler-32 hash-able type. |
| 161 | /// |
| 162 | /// # Examples |
| 163 | /// ```rust |
| 164 | /// use simd_adler32::adler32; |
| 165 | /// |
| 166 | /// let hash = adler32(b"Adler-32" ); |
| 167 | /// println!("{}" , hash); // 800813569 |
| 168 | /// ``` |
| 169 | pub fn adler32<H: Adler32Hash>(hash: &H) -> u32 { |
| 170 | hash.hash() |
| 171 | } |
| 172 | |
| 173 | /// A Adler-32 hash-able type. |
| 174 | pub trait Adler32Hash { |
| 175 | /// Feeds this value into `Adler32`. |
| 176 | fn hash(&self) -> u32; |
| 177 | } |
| 178 | |
| 179 | impl Default for Adler32 { |
| 180 | fn default() -> Self { |
| 181 | Self { |
| 182 | a: 1, |
| 183 | b: 0, |
| 184 | update: get_imp(), |
| 185 | } |
| 186 | } |
| 187 | } |
| 188 | |
| 189 | #[cfg (feature = "std" )] |
| 190 | pub mod read { |
| 191 | //! Reader-based hashing. |
| 192 | //! |
| 193 | //! # Example |
| 194 | //! ```rust |
| 195 | //! use std::io::Cursor; |
| 196 | //! use simd_adler32::read::adler32; |
| 197 | //! |
| 198 | //! let mut reader = Cursor::new(b"Hello there" ); |
| 199 | //! let hash = adler32(&mut reader).unwrap(); |
| 200 | //! |
| 201 | //! println!("{}" , hash) // 800813569 |
| 202 | //! ``` |
| 203 | use crate::Adler32; |
| 204 | use std::io::{Read, Result}; |
| 205 | |
| 206 | /// Compute Adler-32 hash on reader until EOF. |
| 207 | /// |
| 208 | /// # Example |
| 209 | /// ```rust |
| 210 | /// use std::io::Cursor; |
| 211 | /// use simd_adler32::read::adler32; |
| 212 | /// |
| 213 | /// let mut reader = Cursor::new(b"Hello there" ); |
| 214 | /// let hash = adler32(&mut reader).unwrap(); |
| 215 | /// |
| 216 | /// println!("{}" , hash) // 800813569 |
| 217 | /// ``` |
| 218 | pub fn adler32<R: Read>(reader: &mut R) -> Result<u32> { |
| 219 | let mut hash = Adler32::new(); |
| 220 | let mut buf = [0; 4096]; |
| 221 | |
| 222 | loop { |
| 223 | match reader.read(&mut buf) { |
| 224 | Ok(0) => return Ok(hash.finish()), |
| 225 | Ok(n) => { |
| 226 | hash.write(&buf[..n]); |
| 227 | } |
| 228 | Err(err) => return Err(err), |
| 229 | } |
| 230 | } |
| 231 | } |
| 232 | } |
| 233 | |
| 234 | #[cfg (feature = "std" )] |
| 235 | pub mod bufread { |
| 236 | //! BufRead-based hashing. |
| 237 | //! |
| 238 | //! Separate `BufRead` trait implemented to allow for custom buffer size optimization. |
| 239 | //! |
| 240 | //! # Example |
| 241 | //! ```rust |
| 242 | //! use std::io::{Cursor, BufReader}; |
| 243 | //! use simd_adler32::bufread::adler32; |
| 244 | //! |
| 245 | //! let mut reader = Cursor::new(b"Hello there" ); |
| 246 | //! let mut reader = BufReader::new(reader); |
| 247 | //! let hash = adler32(&mut reader).unwrap(); |
| 248 | //! |
| 249 | //! println!("{}" , hash) // 800813569 |
| 250 | //! ``` |
| 251 | use crate::Adler32; |
| 252 | use std::io::{BufRead, ErrorKind, Result}; |
| 253 | |
| 254 | /// Compute Adler-32 hash on buf reader until EOF. |
| 255 | /// |
| 256 | /// # Example |
| 257 | /// ```rust |
| 258 | /// use std::io::{Cursor, BufReader}; |
| 259 | /// use simd_adler32::bufread::adler32; |
| 260 | /// |
| 261 | /// let mut reader = Cursor::new(b"Hello there" ); |
| 262 | /// let mut reader = BufReader::new(reader); |
| 263 | /// let hash = adler32(&mut reader).unwrap(); |
| 264 | /// |
| 265 | /// println!("{}" , hash) // 800813569 |
| 266 | /// ``` |
| 267 | pub fn adler32<R: BufRead>(reader: &mut R) -> Result<u32> { |
| 268 | let mut hash = Adler32::new(); |
| 269 | |
| 270 | loop { |
| 271 | let consumed = match reader.fill_buf() { |
| 272 | Ok(buf) => { |
| 273 | if buf.is_empty() { |
| 274 | return Ok(hash.finish()); |
| 275 | } |
| 276 | |
| 277 | hash.write(buf); |
| 278 | buf.len() |
| 279 | } |
| 280 | Err(err) => match err.kind() { |
| 281 | ErrorKind::Interrupted => continue, |
| 282 | ErrorKind::UnexpectedEof => return Ok(hash.finish()), |
| 283 | _ => return Err(err), |
| 284 | }, |
| 285 | }; |
| 286 | |
| 287 | reader.consume(consumed); |
| 288 | } |
| 289 | } |
| 290 | } |
| 291 | |
| 292 | #[cfg (test)] |
| 293 | mod tests { |
| 294 | #[test ] |
| 295 | fn test_from_checksum() { |
| 296 | let buf = b"rust is pretty cool man" ; |
| 297 | let sum = 0xdeadbeaf; |
| 298 | |
| 299 | let mut simd = super::Adler32::from_checksum(sum); |
| 300 | let mut adler = adler::Adler32::from_checksum(sum); |
| 301 | |
| 302 | simd.write(buf); |
| 303 | adler.write_slice(buf); |
| 304 | |
| 305 | let simd = simd.finish(); |
| 306 | let scalar = adler.checksum(); |
| 307 | |
| 308 | assert_eq!(simd, scalar); |
| 309 | } |
| 310 | } |
| 311 | |