| 1 | // Copyright 2015 The Rust Project Developers. See the COPYRIGHT |
| 2 | // file at the top-level directory of this distribution and at |
| 3 | // http://rust-lang.org/COPYRIGHT. |
| 4 | // |
| 5 | // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
| 6 | // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
| 7 | // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your |
| 8 | // option. This file may not be copied, modified, or distributed |
| 9 | // except according to those terms. |
| 10 | |
| 11 | //! # Fx Hash |
| 12 | //! |
| 13 | //! This hashing algorithm was extracted from the Rustc compiler. This is the same hashing |
| 14 | //! algorithm used for some internal operations in Firefox. The strength of this algorithm |
| 15 | //! is in hashing 8 bytes at a time on 64-bit platforms, where the FNV algorithm works on one |
| 16 | //! byte at a time. |
| 17 | //! |
| 18 | //! ## Disclaimer |
| 19 | //! |
| 20 | //! It is **not a cryptographically secure** hash, so it is strongly recommended that you do |
| 21 | //! not use this hash for cryptographic purproses. Furthermore, this hashing algorithm was |
| 22 | //! not designed to prevent any attacks for determining collisions which could be used to |
| 23 | //! potentially cause quadratic behavior in `HashMap`s. So it is not recommended to expose |
| 24 | //! this hash in places where collissions or DDOS attacks may be a concern. |
| 25 | |
| 26 | use core::convert::TryInto; |
| 27 | use core::ops::BitXor; |
| 28 | |
| 29 | const ROTATE: u32 = 5; |
| 30 | const SEED64: u64 = 0x517cc1b727220a95; |
| 31 | const SEED32: u32 = (SEED64 & 0xFFFF_FFFF) as u32; |
| 32 | |
| 33 | #[cfg (target_pointer_width = "32" )] |
| 34 | const SEED: usize = SEED32 as usize; |
| 35 | #[cfg (target_pointer_width = "64" )] |
| 36 | const SEED: usize = SEED64 as usize; |
| 37 | |
| 38 | trait HashWord { |
| 39 | fn hash_word(&mut self, word: Self); |
| 40 | } |
| 41 | |
| 42 | impl HashWord for usize { |
| 43 | #[inline ] |
| 44 | fn hash_word(&mut self, word: Self) { |
| 45 | *self = self.rotate_left(ROTATE).bitxor(word).wrapping_mul(SEED); |
| 46 | } |
| 47 | } |
| 48 | |
| 49 | impl HashWord for u32 { |
| 50 | #[inline ] |
| 51 | fn hash_word(&mut self, word: Self) { |
| 52 | *self = self.rotate_left(ROTATE).bitxor(word).wrapping_mul(SEED32); |
| 53 | } |
| 54 | } |
| 55 | |
| 56 | impl HashWord for u64 { |
| 57 | #[inline ] |
| 58 | fn hash_word(&mut self, word: Self) { |
| 59 | *self = self.rotate_left(ROTATE).bitxor(word).wrapping_mul(SEED64); |
| 60 | } |
| 61 | } |
| 62 | |
| 63 | #[cfg (target_endian = "little" )] |
| 64 | fn read_u32(buf: &[u8]) -> u32 { |
| 65 | u32::from_le_bytes(buf[..4].try_into().unwrap()) |
| 66 | } |
| 67 | |
| 68 | #[cfg (target_endian = "little" )] |
| 69 | fn read_u64(buf: &[u8]) -> u64 { |
| 70 | u64::from_le_bytes(buf[..8].try_into().unwrap()) |
| 71 | } |
| 72 | |
| 73 | #[cfg (target_endian = "big" )] |
| 74 | fn read_u32(buf: &[u8]) -> u32 { |
| 75 | u32::from_be_bytes(buf[..4].try_into().unwrap()) |
| 76 | } |
| 77 | |
| 78 | #[cfg (target_endian = "big" )] |
| 79 | fn read_u64(buf: &[u8]) -> u64 { |
| 80 | u64::from_be_bytes(buf[..8].try_into().unwrap()) |
| 81 | } |
| 82 | |
| 83 | #[inline ] |
| 84 | #[cfg (target_pointer_width = "32" )] |
| 85 | fn write(initial_state: usize, mut bytes: &[u8]) -> usize { |
| 86 | let mut hash = initial_state as u32; |
| 87 | while bytes.len() >= 4 { |
| 88 | let n = read_u32(bytes); |
| 89 | hash.hash_word(n); |
| 90 | bytes = bytes.split_at(4).1; |
| 91 | } |
| 92 | |
| 93 | for byte in bytes { |
| 94 | hash.hash_word(*byte as u32); |
| 95 | } |
| 96 | hash as usize |
| 97 | } |
| 98 | |
| 99 | #[inline ] |
| 100 | #[cfg (target_pointer_width = "64" )] |
| 101 | fn write(initial_state: usize, mut bytes: &[u8]) -> usize { |
| 102 | let mut hash: u64 = initial_state as u64; |
| 103 | while bytes.len() >= 8 { |
| 104 | let n: u64 = read_u64(buf:bytes); |
| 105 | hash.hash_word(n); |
| 106 | bytes = bytes.split_at(mid:8).1; |
| 107 | } |
| 108 | |
| 109 | if bytes.len() >= 4 { |
| 110 | let n: u32 = read_u32(buf:bytes); |
| 111 | hash.hash_word(n as u64); |
| 112 | bytes = bytes.split_at(mid:4).1; |
| 113 | } |
| 114 | |
| 115 | for byte: &u8 in bytes { |
| 116 | hash.hash_word(*byte as u64); |
| 117 | } |
| 118 | hash as usize |
| 119 | } |
| 120 | |
| 121 | pub fn hash(bytes: &[u8]) -> usize { |
| 122 | write(initial_state:0usize, bytes) |
| 123 | } |
| 124 | |