| author | |
| committer | |
| log | 7d9ee5d6d53011e9edc7ca10b76d64af807a833f |
| tree | 36ad24d05bbd2384748a4028a92a59922e64e74c |
| parent | 1547692d188ddc0c6f2dfbe4a1f943b380717306 |
| parent | 69129c2e932bd51dff02578e67ea79799380a3cc |
| signature |
Implementation of MurmurHash and CityHash4 files changed, 745 insertions(+), 0 deletions(-)
CMakeLists.txt+2| ... | @@ -523,6 +523,8 @@ set(ZIG_STD_FILES | ... | @@ -523,6 +523,8 @@ set(ZIG_STD_FILES |
| 523 | "hash/crc.zig" | 523 | "hash/crc.zig" |
| 524 | "hash/fnv.zig" | 524 | "hash/fnv.zig" |
| 525 | "hash/siphash.zig" | 525 | "hash/siphash.zig" |
| 526 | "hash/murmur.zig" | ||
| 527 | "hash/cityhash.zig" | ||
| 526 | "hash_map.zig" | 528 | "hash_map.zig" |
| 527 | "heap.zig" | 529 | "heap.zig" |
| 528 | "heap/logging_allocator.zig" | 530 | "heap/logging_allocator.zig" |
std/hash.zig+11| ... | @@ -14,9 +14,20 @@ const siphash = @import("hash/siphash.zig"); | ... | @@ -14,9 +14,20 @@ const siphash = @import("hash/siphash.zig"); |
| 14 | pub const SipHash64 = siphash.SipHash64; | 14 | pub const SipHash64 = siphash.SipHash64; |
| 15 | pub const SipHash128 = siphash.SipHash128; | 15 | pub const SipHash128 = siphash.SipHash128; |
| 16 | 16 | ||
| 17 | pub const murmur = @import("hash/murmur.zig"); | ||
| 18 | pub const Murmur2_32 = murmur.Murmur2_32; | ||
| 19 | pub const Murmur2_64 = murmur.Murmur2_64; | ||
| 20 | pub const Murmur3_32 = murmur.Murmur3_32; | ||
| 21 | |||
| 22 | pub const cityhash = @import("hash/cityhash.zig"); | ||
| 23 | pub const CityHash32 = cityhash.CityHash32; | ||
| 24 | pub const CityHash64 = cityhash.CityHash64; | ||
| 25 | |||
| 17 | test "hash" { | 26 | test "hash" { |
| 18 | _ = @import("hash/adler.zig"); | 27 | _ = @import("hash/adler.zig"); |
| 19 | _ = @import("hash/crc.zig"); | 28 | _ = @import("hash/crc.zig"); |
| 20 | _ = @import("hash/fnv.zig"); | 29 | _ = @import("hash/fnv.zig"); |
| 21 | _ = @import("hash/siphash.zig"); | 30 | _ = @import("hash/siphash.zig"); |
| 31 | _ = @import("hash/murmur.zig"); | ||
| 32 | _ = @import("hash/cityhash.zig"); | ||
| 22 | } | 33 | } |
std/hash/cityhash.zig created+387| ... | @@ -0,0 +1,387 @@ | ||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | |||
| 4 | pub const CityHash32 = struct { | ||
| 5 | const Self = @This(); | ||
| 6 | |||
| 7 | // Magic numbers for 32-bit hashing. Copied from Murmur3. | ||
| 8 | const c1: u32 = 0xcc9e2d51; | ||
| 9 | const c2: u32 = 0x1b873593; | ||
| 10 | |||
| 11 | fn fetch32(ptr: [*]const u8) u32 { | ||
| 12 | var v: u32 = undefined; | ||
| 13 | @memcpy(@ptrCast([*]u8, &v), ptr, 4); | ||
| 14 | if (builtin.endian == builtin.Endian.Big) | ||
| 15 | return @byteSwap(u32, v); | ||
| 16 | return v; | ||
| 17 | } | ||
| 18 | |||
| 19 | // A 32-bit to 32-bit integer hash copied from Murmur3. | ||
| 20 | fn fmix(h: u32) u32 { | ||
| 21 | var h1: u32 = h; | ||
| 22 | h1 ^= h1 >> 16; | ||
| 23 | h1 *%= 0x85ebca6b; | ||
| 24 | h1 ^= h1 >> 13; | ||
| 25 | h1 *%= 0xc2b2ae35; | ||
| 26 | h1 ^= h1 >> 16; | ||
| 27 | return h1; | ||
| 28 | } | ||
| 29 | |||
| 30 | // Rotate right helper | ||
| 31 | fn rotr32(x: u32, comptime r: u32) u32 { | ||
| 32 | return (x >> r) | (x << (32 - r)); | ||
| 33 | } | ||
| 34 | |||
| 35 | // Helper from Murmur3 for combining two 32-bit values. | ||
| 36 | fn mur(a: u32, h: u32) u32 { | ||
| 37 | var a1: u32 = a; | ||
| 38 | var h1: u32 = h; | ||
| 39 | a1 *%= c1; | ||
| 40 | a1 = rotr32(a1, 17); | ||
| 41 | a1 *%= c2; | ||
| 42 | h1 ^= a1; | ||
| 43 | h1 = rotr32(h1, 19); | ||
| 44 | return h1 *% 5 +% 0xe6546b64; | ||
| 45 | } | ||
| 46 | |||
| 47 | fn hash32Len0To4(str: []const u8) u32 { | ||
| 48 | const len: u32 = @truncate(u32, str.len); | ||
| 49 | var b: u32 = 0; | ||
| 50 | var c: u32 = 9; | ||
| 51 | for (str) |v| { | ||
| 52 | b = b *% c1 +% @bitCast(u32, @intCast(i32, @bitCast(i8, v))); | ||
| 53 | c ^= b; | ||
| 54 | } | ||
| 55 | return fmix(mur(b, mur(len, c))); | ||
| 56 | } | ||
| 57 | |||
| 58 | fn hash32Len5To12(str: []const u8) u32 { | ||
| 59 | var a: u32 = @truncate(u32, str.len); | ||
| 60 | var b: u32 = a *% 5; | ||
| 61 | var c: u32 = 9; | ||
| 62 | const d: u32 = b; | ||
| 63 | |||
| 64 | a +%= fetch32(str.ptr); | ||
| 65 | b +%= fetch32(str.ptr + str.len - 4); | ||
| 66 | c +%= fetch32(str.ptr + ((str.len >> 1) & 4)); | ||
| 67 | |||
| 68 | return fmix(mur(c, mur(b, mur(a, d)))); | ||
| 69 | } | ||
| 70 | |||
| 71 | fn hash32Len13To24(str: []const u8) u32 { | ||
| 72 | const len: u32 = @truncate(u32, str.len); | ||
| 73 | const a: u32 = fetch32(str.ptr + (str.len >> 1) - 4); | ||
| 74 | const b: u32 = fetch32(str.ptr + 4); | ||
| 75 | const c: u32 = fetch32(str.ptr + str.len - 8); | ||
| 76 | const d: u32 = fetch32(str.ptr + (str.len >> 1)); | ||
| 77 | const e: u32 = fetch32(str.ptr); | ||
| 78 | const f: u32 = fetch32(str.ptr + str.len - 4); | ||
| 79 | |||
| 80 | return fmix(mur(f, mur(e, mur(d, mur(c, mur(b, mur(a, len))))))); | ||
| 81 | } | ||
| 82 | |||
| 83 | pub fn hash(str: []const u8) u32 { | ||
| 84 | if (str.len <= 24) { | ||
| 85 | if (str.len <= 4) { | ||
| 86 | return hash32Len0To4(str); | ||
| 87 | } else { | ||
| 88 | if (str.len <= 12) | ||
| 89 | return hash32Len5To12(str); | ||
| 90 | return hash32Len13To24(str); | ||
| 91 | } | ||
| 92 | } | ||
| 93 | |||
| 94 | const len: u32 = @truncate(u32, str.len); | ||
| 95 | var h: u32 = len; | ||
| 96 | var g: u32 = c1 *% len; | ||
| 97 | var f: u32 = g; | ||
| 98 | |||
| 99 | const a0: u32 = rotr32(fetch32(str.ptr + str.len - 4) *% c1, 17) *% c2; | ||
| 100 | const a1: u32 = rotr32(fetch32(str.ptr + str.len - 8) *% c1, 17) *% c2; | ||
| 101 | const a2: u32 = rotr32(fetch32(str.ptr + str.len - 16) *% c1, 17) *% c2; | ||
| 102 | const a3: u32 = rotr32(fetch32(str.ptr + str.len - 12) *% c1, 17) *% c2; | ||
| 103 | const a4: u32 = rotr32(fetch32(str.ptr + str.len - 20) *% c1, 17) *% c2; | ||
| 104 | |||
| 105 | h ^= a0; | ||
| 106 | h = rotr32(h, 19); | ||
| 107 | h = h *% 5 +% 0xe6546b64; | ||
| 108 | h ^= a2; | ||
| 109 | h = rotr32(h, 19); | ||
| 110 | h = h *% 5 +% 0xe6546b64; | ||
| 111 | g ^= a1; | ||
| 112 | g = rotr32(g, 19); | ||
| 113 | g = g *% 5 +% 0xe6546b64; | ||
| 114 | g ^= a3; | ||
| 115 | g = rotr32(g, 19); | ||
| 116 | g = g *% 5 +% 0xe6546b64; | ||
| 117 | f +%= a4; | ||
| 118 | f = rotr32(f, 19); | ||
| 119 | f = f *% 5 +% 0xe6546b64; | ||
| 120 | var iters = (str.len - 1) / 20; | ||
| 121 | var ptr = str.ptr; | ||
| 122 | while (iters != 0) : (iters -= 1) { | ||
| 123 | const b0: u32 = rotr32(fetch32(ptr) *% c1, 17) *% c2; | ||
| 124 | const b1: u32 = fetch32(ptr + 4); | ||
| 125 | const b2: u32 = rotr32(fetch32(ptr + 8) *% c1, 17) *% c2; | ||
| 126 | const b3: u32 = rotr32(fetch32(ptr + 12) *% c1, 17) *% c2; | ||
| 127 | const b4: u32 = fetch32(ptr + 16); | ||
| 128 | |||
| 129 | h ^= b0; | ||
| 130 | h = rotr32(h, 18); | ||
| 131 | h = h *% 5 +% 0xe6546b64; | ||
| 132 | f +%= b1; | ||
| 133 | f = rotr32(f, 19); | ||
| 134 | f = f *% c1; | ||
| 135 | g +%= b2; | ||
| 136 | g = rotr32(g, 18); | ||
| 137 | g = g *% 5 +% 0xe6546b64; | ||
| 138 | h ^= b3 +% b1; | ||
| 139 | h = rotr32(h, 19); | ||
| 140 | h = h *% 5 +% 0xe6546b64; | ||
| 141 | g ^= b4; | ||
| 142 | g = @byteSwap(u32, g) *% 5; | ||
| 143 | h +%= b4 *% 5; | ||
| 144 | h = @byteSwap(u32, h); | ||
| 145 | f +%= b0; | ||
| 146 | const t: u32 = h; | ||
| 147 | h = f; | ||
| 148 | f = g; | ||
| 149 | g = t; | ||
| 150 | ptr += 20; | ||
| 151 | } | ||
| 152 | g = rotr32(g, 11) *% c1; | ||
| 153 | g = rotr32(g, 17) *% c1; | ||
| 154 | f = rotr32(f, 11) *% c1; | ||
| 155 | f = rotr32(f, 17) *% c1; | ||
| 156 | h = rotr32(h +% g, 19); | ||
| 157 | h = h *% 5 +% 0xe6546b64; | ||
| 158 | h = rotr32(h, 17) *% c1; | ||
| 159 | h = rotr32(h +% f, 19); | ||
| 160 | h = h *% 5 +% 0xe6546b64; | ||
| 161 | h = rotr32(h, 17) *% c1; | ||
| 162 | return h; | ||
| 163 | } | ||
| 164 | }; | ||
| 165 | |||
| 166 | pub const CityHash64 = struct { | ||
| 167 | const Self = @This(); | ||
| 168 | |||
| 169 | // Some primes between 2^63 and 2^64 for various uses. | ||
| 170 | const k0: u64 = 0xc3a5c85c97cb3127; | ||
| 171 | const k1: u64 = 0xb492b66fbe98f273; | ||
| 172 | const k2: u64 = 0x9ae16a3b2f90404f; | ||
| 173 | |||
| 174 | fn fetch32(ptr: [*]const u8) u32 { | ||
| 175 | var v: u32 = undefined; | ||
| 176 | @memcpy(@ptrCast([*]u8, &v), ptr, 4); | ||
| 177 | if (builtin.endian == builtin.Endian.Big) | ||
| 178 | return @byteSwap(u32, v); | ||
| 179 | return v; | ||
| 180 | } | ||
| 181 | |||
| 182 | fn fetch64(ptr: [*]const u8) u64 { | ||
| 183 | var v: u64 = undefined; | ||
| 184 | @memcpy(@ptrCast([*]u8, &v), ptr, 8); | ||
| 185 | if (builtin.endian == builtin.Endian.Big) | ||
| 186 | return @byteSwap(u64, v); | ||
| 187 | return v; | ||
| 188 | } | ||
| 189 | |||
| 190 | // Rotate right helper | ||
| 191 | fn rotr64(x: u64, comptime r: u64) u64 { | ||
| 192 | return (x >> r) | (x << (64 - r)); | ||
| 193 | } | ||
| 194 | |||
| 195 | fn shiftmix(v: u64) u64 { | ||
| 196 | return v ^ (v >> 47); | ||
| 197 | } | ||
| 198 | |||
| 199 | fn hashLen16(u: u64, v: u64) u64 { | ||
| 200 | return @inlineCall(hash128To64, u, v); | ||
| 201 | } | ||
| 202 | |||
| 203 | fn hashLen16Mul(low: u64, high: u64, mul: u64) u64 { | ||
| 204 | var a: u64 = (low ^ high) *% mul; | ||
| 205 | a ^= (a >> 47); | ||
| 206 | var b: u64 = (high ^ a) *% mul; | ||
| 207 | b ^= (b >> 47); | ||
| 208 | b *%= mul; | ||
| 209 | return b; | ||
| 210 | } | ||
| 211 | |||
| 212 | fn hash128To64(low: u64, high: u64) u64 { | ||
| 213 | return @inlineCall(hashLen16Mul, low, high, 0x9ddfea08eb382d69); | ||
| 214 | } | ||
| 215 | |||
| 216 | fn hashLen0To16(str: []const u8) u64 { | ||
| 217 | const len: u64 = @truncate(u64, str.len); | ||
| 218 | if (len >= 8) { | ||
| 219 | const mul: u64 = k2 +% len *% 2; | ||
| 220 | const a: u64 = fetch64(str.ptr) +% k2; | ||
| 221 | const b: u64 = fetch64(str.ptr + str.len - 8); | ||
| 222 | const c: u64 = rotr64(b, 37) *% mul +% a; | ||
| 223 | const d: u64 = (rotr64(a, 25) +% b) *% mul; | ||
| 224 | return hashLen16Mul(c, d, mul); | ||
| 225 | } | ||
| 226 | if (len >= 4) { | ||
| 227 | const mul: u64 = k2 +% len *% 2; | ||
| 228 | const a: u64 = fetch32(str.ptr); | ||
| 229 | return hashLen16Mul(len +% (a << 3), fetch32(str.ptr + str.len - 4), mul); | ||
| 230 | } | ||
| 231 | if (len > 0) { | ||
| 232 | const a: u8 = str[0]; | ||
| 233 | const b: u8 = str[str.len >> 1]; | ||
| 234 | const c: u8 = str[str.len - 1]; | ||
| 235 | const y: u32 = @intCast(u32, a) +% (@intCast(u32, b) << 8); | ||
| 236 | const z: u32 = @truncate(u32, str.len) +% (@intCast(u32, c) << 2); | ||
| 237 | return shiftmix(@intCast(u64, y) *% k2 ^ @intCast(u64, z) *% k0) *% k2; | ||
| 238 | } | ||
| 239 | return k2; | ||
| 240 | } | ||
| 241 | |||
| 242 | fn hashLen17To32(str: []const u8) u64 { | ||
| 243 | const len: u64 = @truncate(u64, str.len); | ||
| 244 | const mul: u64 = k2 +% len *% 2; | ||
| 245 | const a: u64 = fetch64(str.ptr) *% k1; | ||
| 246 | const b: u64 = fetch64(str.ptr + 8); | ||
| 247 | const c: u64 = fetch64(str.ptr + str.len - 8) *% mul; | ||
| 248 | const d: u64 = fetch64(str.ptr + str.len - 16) *% k2; | ||
| 249 | |||
| 250 | return hashLen16Mul(rotr64(a +% b, 43) +% rotr64(c, 30) +% d, a +% rotr64(b +% k2, 18) +% c, mul); | ||
| 251 | } | ||
| 252 | |||
| 253 | fn hashLen33To64(str: []const u8) u64 { | ||
| 254 | const len: u64 = @truncate(u64, str.len); | ||
| 255 | const mul: u64 = k2 +% len *% 2; | ||
| 256 | const a: u64 = fetch64(str.ptr) *% k2; | ||
| 257 | const b: u64 = fetch64(str.ptr + 8); | ||
| 258 | const c: u64 = fetch64(str.ptr + str.len - 24); | ||
| 259 | const d: u64 = fetch64(str.ptr + str.len - 32); | ||
| 260 | const e: u64 = fetch64(str.ptr + 16) *% k2; | ||
| 261 | const f: u64 = fetch64(str.ptr + 24) *% 9; | ||
| 262 | const g: u64 = fetch64(str.ptr + str.len - 8); | ||
| 263 | const h: u64 = fetch64(str.ptr + str.len - 16) *% mul; | ||
| 264 | |||
| 265 | const u: u64 = rotr64(a +% g, 43) +% (rotr64(b, 30) +% c) *% 9; | ||
| 266 | const v: u64 = ((a +% g) ^ d) +% f +% 1; | ||
| 267 | const w: u64 = @byteSwap(u64, (u +% v) *% mul) +% h; | ||
| 268 | const x: u64 = rotr64(e +% f, 42) +% c; | ||
| 269 | const y: u64 = (@byteSwap(u64, (v +% w) *% mul) +% g) *% mul; | ||
| 270 | const z: u64 = e +% f +% c; | ||
| 271 | const a1: u64 = @byteSwap(u64, (x +% z) *% mul +% y) +% b; | ||
| 272 | const b1: u64 = shiftmix((z +% a1) *% mul +% d +% h) *% mul; | ||
| 273 | return b1 +% x; | ||
| 274 | } | ||
| 275 | |||
| 276 | const WeakPair = struct { | ||
| 277 | first: u64, | ||
| 278 | second: u64, | ||
| 279 | }; | ||
| 280 | |||
| 281 | fn weakHashLen32WithSeedsHelper(w: u64, x: u64, y: u64, z: u64, a: u64, b: u64) WeakPair { | ||
| 282 | var a1: u64 = a; | ||
| 283 | var b1: u64 = b; | ||
| 284 | a1 +%= w; | ||
| 285 | b1 = rotr64(b1 +% a1 +% z, 21); | ||
| 286 | var c: u64 = a1; | ||
| 287 | a1 +%= x; | ||
| 288 | a1 +%= y; | ||
| 289 | b1 +%= rotr64(a1, 44); | ||
| 290 | return WeakPair{ .first = a1 +% z, .second = b1 +% c }; | ||
| 291 | } | ||
| 292 | |||
| 293 | fn weakHashLen32WithSeeds(ptr: [*]const u8, a: u64, b: u64) WeakPair { | ||
| 294 | return @inlineCall(weakHashLen32WithSeedsHelper, fetch64(ptr), fetch64(ptr + 8), fetch64(ptr + 16), fetch64(ptr + 24), a, b); | ||
| 295 | } | ||
| 296 | |||
| 297 | pub fn hash(str: []const u8) u64 { | ||
| 298 | if (str.len <= 32) { | ||
| 299 | if (str.len <= 16) { | ||
| 300 | return hashLen0To16(str); | ||
| 301 | } else { | ||
| 302 | return hashLen17To32(str); | ||
| 303 | } | ||
| 304 | } else if (str.len <= 64) { | ||
| 305 | return hashLen33To64(str); | ||
| 306 | } | ||
| 307 | |||
| 308 | var len: u64 = @truncate(u64, str.len); | ||
| 309 | |||
| 310 | var x: u64 = fetch64(str.ptr + str.len - 40); | ||
| 311 | var y: u64 = fetch64(str.ptr + str.len - 16) +% fetch64(str.ptr + str.len - 56); | ||
| 312 | var z: u64 = hashLen16(fetch64(str.ptr + str.len - 48) +% len, fetch64(str.ptr + str.len - 24)); | ||
| 313 | var v: WeakPair = weakHashLen32WithSeeds(str.ptr + str.len - 64, len, z); | ||
| 314 | var w: WeakPair = weakHashLen32WithSeeds(str.ptr + str.len - 32, y +% k1, x); | ||
| 315 | |||
| 316 | x = x *% k1 +% fetch64(str.ptr); | ||
| 317 | len = (len - 1) & ~@intCast(u64, 63); | ||
| 318 | |||
| 319 | var ptr: [*]const u8 = str.ptr; | ||
| 320 | while (true) { | ||
| 321 | x = rotr64(x +% y +% v.first +% fetch64(ptr + 8), 37) *% k1; | ||
| 322 | y = rotr64(y +% v.second +% fetch64(ptr + 48), 42) *% k1; | ||
| 323 | x ^= w.second; | ||
| 324 | y +%= v.first +% fetch64(ptr + 40); | ||
| 325 | z = rotr64(z +% w.first, 33) *% k1; | ||
| 326 | v = weakHashLen32WithSeeds(ptr, v.second *% k1, x +% w.first); | ||
| 327 | w = weakHashLen32WithSeeds(ptr + 32, z +% w.second, y +% fetch64(ptr + 16)); | ||
| 328 | const t: u64 = z; | ||
| 329 | z = x; | ||
| 330 | x = t; | ||
| 331 | |||
| 332 | ptr += 64; | ||
| 333 | len -= 64; | ||
| 334 | if (len == 0) | ||
| 335 | break; | ||
| 336 | } | ||
| 337 | |||
| 338 | return hashLen16(hashLen16(v.first, w.first) +% shiftmix(y) *% k1 +% z, hashLen16(v.second, w.second) +% x); | ||
| 339 | } | ||
| 340 | |||
| 341 | pub fn hashWithSeed(str: []const u8, seed: u64) u64 { | ||
| 342 | return @inlineCall(Self.hashWithSeeds, str, k2, seed); | ||
| 343 | } | ||
| 344 | |||
| 345 | pub fn hashWithSeeds(str: []const u8, seed0: u64, seed1: u64) u64 { | ||
| 346 | return hashLen16(hash(str) -% seed0, seed1); | ||
| 347 | } | ||
| 348 | }; | ||
| 349 | |||
| 350 | fn SMHasherTest(comptime hash_fn: var, comptime hashbits: u32) u32 { | ||
| 351 | const hashbytes = hashbits / 8; | ||
| 352 | var key: [256]u8 = undefined; | ||
| 353 | var hashes: [hashbytes * 256]u8 = undefined; | ||
| 354 | var final: [hashbytes]u8 = undefined; | ||
| 355 | |||
| 356 | @memset(@ptrCast([*]u8, &key[0]), 0, @sizeOf(@typeOf(key))); | ||
| 357 | @memset(@ptrCast([*]u8, &hashes[0]), 0, @sizeOf(@typeOf(hashes))); | ||
| 358 | @memset(@ptrCast([*]u8, &final[0]), 0, @sizeOf(@typeOf(final))); | ||
| 359 | |||
| 360 | var i: u32 = 0; | ||
| 361 | while (i < 256) : (i += 1) { | ||
| 362 | key[i] = @intCast(u8, i); | ||
| 363 | |||
| 364 | var h = hash_fn(key[0..i], 256 - i); | ||
| 365 | if (builtin.endian == builtin.Endian.Big) | ||
| 366 | h = @byteSwap(@typeOf(h), h); | ||
| 367 | @memcpy(@ptrCast([*]u8, &hashes[i * hashbytes]), @ptrCast([*]u8, &h), hashbytes); | ||
| 368 | } | ||
| 369 | |||
| 370 | return @truncate(u32, hash_fn(hashes, 0)); | ||
| 371 | } | ||
| 372 | |||
| 373 | fn CityHash32hashIgnoreSeed(str: []const u8, seed: u32) u32 { | ||
| 374 | return CityHash32.hash(str); | ||
| 375 | } | ||
| 376 | |||
| 377 | test "cityhash32" { | ||
| 378 | // Note: SMHasher doesn't provide a 32bit version of the algorithm. | ||
| 379 | // Note: The implementation was verified against the Google Abseil version. | ||
| 380 | std.testing.expectEqual(SMHasherTest(CityHash32hashIgnoreSeed, 32), 0x68254F81); | ||
| 381 | } | ||
| 382 | |||
| 383 | test "cityhash64" { | ||
| 384 | // Note: This is not compliant with the SMHasher implementation of CityHash64! | ||
| 385 | // Note: The implementation was verified against the Google Abseil version. | ||
| 386 | std.testing.expectEqual(SMHasherTest(CityHash64.hashWithSeed, 64), 0x5FABC5C5); | ||
| 387 | } | ||
std/hash/murmur.zig created+345| ... | @@ -0,0 +1,345 @@ | ||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | const testing = std.testing; | ||
| 4 | |||
| 5 | const default_seed: u32 = 0xc70f6907; | ||
| 6 | |||
| 7 | pub const Murmur2_32 = struct { | ||
| 8 | const Self = @This(); | ||
| 9 | |||
| 10 | pub fn hash(str: []const u8) u32 { | ||
| 11 | return @inlineCall(Self.hashWithSeed, str, default_seed); | ||
| 12 | } | ||
| 13 | |||
| 14 | pub fn hashWithSeed(str: []const u8, seed: u32) u32 { | ||
| 15 | const m: u32 = 0x5bd1e995; | ||
| 16 | const len = @truncate(u32, str.len); | ||
| 17 | var h1: u32 = seed ^ len; | ||
| 18 | for (@ptrCast([*]allowzero align(1) const u32, str.ptr)[0..(len >> 2)]) |v| { | ||
| 19 | var k1: u32 = v; | ||
| 20 | if (builtin.endian == builtin.Endian.Big) | ||
| 21 | k1 = @byteSwap(u32, k1); | ||
| 22 | k1 *%= m; | ||
| 23 | k1 ^= k1 >> 24; | ||
| 24 | k1 *%= m; | ||
| 25 | h1 *%= m; | ||
| 26 | h1 ^= k1; | ||
| 27 | } | ||
| 28 | const offset = len & 0xfffffffc; | ||
| 29 | const rest = len & 3; | ||
| 30 | if (rest >= 3) { | ||
| 31 | h1 ^= @intCast(u32, str[offset + 2]) << 16; | ||
| 32 | } | ||
| 33 | if (rest >= 2) { | ||
| 34 | h1 ^= @intCast(u32, str[offset + 1]) << 8; | ||
| 35 | } | ||
| 36 | if (rest >= 1) { | ||
| 37 | h1 ^= @intCast(u32, str[offset + 0]); | ||
| 38 | h1 *%= m; | ||
| 39 | } | ||
| 40 | h1 ^= h1 >> 13; | ||
| 41 | h1 *%= m; | ||
| 42 | h1 ^= h1 >> 15; | ||
| 43 | return h1; | ||
| 44 | } | ||
| 45 | |||
| 46 | pub fn hashUint32(v: u32) u32 { | ||
| 47 | return @inlineCall(Self.hashUint32WithSeed, v, default_seed); | ||
| 48 | } | ||
| 49 | |||
| 50 | pub fn hashUint32WithSeed(v: u32, seed: u32) u32 { | ||
| 51 | const m: u32 = 0x5bd1e995; | ||
| 52 | const len: u32 = 4; | ||
| 53 | var h1: u32 = seed ^ len; | ||
| 54 | var k1: u32 = undefined; | ||
| 55 | k1 = v *% m; | ||
| 56 | k1 ^= k1 >> 24; | ||
| 57 | k1 *%= m; | ||
| 58 | h1 *%= m; | ||
| 59 | h1 ^= k1; | ||
| 60 | h1 ^= h1 >> 13; | ||
| 61 | h1 *%= m; | ||
| 62 | h1 ^= h1 >> 15; | ||
| 63 | return h1; | ||
| 64 | } | ||
| 65 | |||
| 66 | pub fn hashUint64(v: u64) u32 { | ||
| 67 | return @inlineCall(Self.hashUint64WithSeed, v, default_seed); | ||
| 68 | } | ||
| 69 | |||
| 70 | pub fn hashUint64WithSeed(v: u64, seed: u32) u32 { | ||
| 71 | const m: u32 = 0x5bd1e995; | ||
| 72 | const len: u32 = 8; | ||
| 73 | var h1: u32 = seed ^ len; | ||
| 74 | var k1: u32 = undefined; | ||
| 75 | k1 = @truncate(u32, v) *% m; | ||
| 76 | k1 ^= k1 >> 24; | ||
| 77 | k1 *%= m; | ||
| 78 | h1 *%= m; | ||
| 79 | h1 ^= k1; | ||
| 80 | k1 = @truncate(u32, v >> 32) *% m; | ||
| 81 | k1 ^= k1 >> 24; | ||
| 82 | k1 *%= m; | ||
| 83 | h1 *%= m; | ||
| 84 | h1 ^= k1; | ||
| 85 | h1 ^= h1 >> 13; | ||
| 86 | h1 *%= m; | ||
| 87 | h1 ^= h1 >> 15; | ||
| 88 | return h1; | ||
| 89 | } | ||
| 90 | }; | ||
| 91 | |||
| 92 | pub const Murmur2_64 = struct { | ||
| 93 | const Self = @This(); | ||
| 94 | |||
| 95 | pub fn hash(str: []const u8) u64 { | ||
| 96 | return @inlineCall(Self.hashWithSeed, str, default_seed); | ||
| 97 | } | ||
| 98 | |||
| 99 | pub fn hashWithSeed(str: []const u8, seed: u64) u64 { | ||
| 100 | const m: u64 = 0xc6a4a7935bd1e995; | ||
| 101 | const len = @truncate(u64, str.len); | ||
| 102 | var h1: u64 = seed ^ (len *% m); | ||
| 103 | for (@ptrCast([*]allowzero align(1) const u64, str.ptr)[0..(len >> 3)]) |v| { | ||
| 104 | var k1: u64 = v; | ||
| 105 | if (builtin.endian == builtin.Endian.Big) | ||
| 106 | k1 = @byteSwap(u64, k1); | ||
| 107 | k1 *%= m; | ||
| 108 | k1 ^= k1 >> 47; | ||
| 109 | k1 *%= m; | ||
| 110 | h1 ^= k1; | ||
| 111 | h1 *%= m; | ||
| 112 | } | ||
| 113 | const rest = len & 7; | ||
| 114 | const offset = len - rest; | ||
| 115 | if (rest > 0) { | ||
| 116 | var k1: u64 = 0; | ||
| 117 | @memcpy(@ptrCast([*]u8, &k1), @ptrCast([*]const u8, &str[offset]), rest); | ||
| 118 | if (builtin.endian == builtin.Endian.Big) | ||
| 119 | k1 = @byteSwap(u64, k1); | ||
| 120 | h1 ^= k1; | ||
| 121 | h1 *%= m; | ||
| 122 | } | ||
| 123 | h1 ^= h1 >> 47; | ||
| 124 | h1 *%= m; | ||
| 125 | h1 ^= h1 >> 47; | ||
| 126 | return h1; | ||
| 127 | } | ||
| 128 | |||
| 129 | pub fn hashUint32(v: u32) u64 { | ||
| 130 | return @inlineCall(Self.hashUint32WithSeed, v, default_seed); | ||
| 131 | } | ||
| 132 | |||
| 133 | pub fn hashUint32WithSeed(v: u32, seed: u32) u64 { | ||
| 134 | const m: u64 = 0xc6a4a7935bd1e995; | ||
| 135 | const len: u64 = 4; | ||
| 136 | var h1: u64 = seed ^ (len *% m); | ||
| 137 | var k1: u64 = v; | ||
| 138 | h1 ^= k1; | ||
| 139 | h1 *%= m; | ||
| 140 | h1 ^= h1 >> 47; | ||
| 141 | h1 *%= m; | ||
| 142 | h1 ^= h1 >> 47; | ||
| 143 | return h1; | ||
| 144 | } | ||
| 145 | |||
| 146 | pub fn hashUint64(v: u64) u64 { | ||
| 147 | return @inlineCall(Self.hashUint64WithSeed, v, default_seed); | ||
| 148 | } | ||
| 149 | |||
| 150 | pub fn hashUint64WithSeed(v: u64, seed: u32) u64 { | ||
| 151 | const m: u64 = 0xc6a4a7935bd1e995; | ||
| 152 | const len: u64 = 8; | ||
| 153 | var h1: u64 = seed ^ (len *% m); | ||
| 154 | var k1: u64 = undefined; | ||
| 155 | k1 = v *% m; | ||
| 156 | k1 ^= k1 >> 47; | ||
| 157 | k1 *%= m; | ||
| 158 | h1 ^= k1; | ||
| 159 | h1 *%= m; | ||
| 160 | h1 ^= h1 >> 47; | ||
| 161 | h1 *%= m; | ||
| 162 | h1 ^= h1 >> 47; | ||
| 163 | return h1; | ||
| 164 | } | ||
| 165 | }; | ||
| 166 | |||
| 167 | pub const Murmur3_32 = struct { | ||
| 168 | const Self = @This(); | ||
| 169 | |||
| 170 | fn rotl32(x: u32, comptime r: u32) u32 { | ||
| 171 | return (x << r) | (x >> (32 - r)); | ||
| 172 | } | ||
| 173 | |||
| 174 | pub fn hash(str: []const u8) u32 { | ||
| 175 | return @inlineCall(Self.hashWithSeed, str, default_seed); | ||
| 176 | } | ||
| 177 | |||
| 178 | pub fn hashWithSeed(str: []const u8, seed: u32) u32 { | ||
| 179 | const c1: u32 = 0xcc9e2d51; | ||
| 180 | const c2: u32 = 0x1b873593; | ||
| 181 | const len = @truncate(u32, str.len); | ||
| 182 | var h1: u32 = seed; | ||
| 183 | for (@ptrCast([*]allowzero align(1) const u32, str.ptr)[0..(len >> 2)]) |v| { | ||
| 184 | var k1: u32 = v; | ||
| 185 | if (builtin.endian == builtin.Endian.Big) | ||
| 186 | k1 = @byteSwap(u32, k1); | ||
| 187 | k1 *%= c1; | ||
| 188 | k1 = rotl32(k1, 15); | ||
| 189 | k1 *%= c2; | ||
| 190 | h1 ^= k1; | ||
| 191 | h1 = rotl32(h1, 13); | ||
| 192 | h1 *%= 5; | ||
| 193 | h1 +%= 0xe6546b64; | ||
| 194 | } | ||
| 195 | { | ||
| 196 | var k1: u32 = 0; | ||
| 197 | const offset = len & 0xfffffffc; | ||
| 198 | const rest = len & 3; | ||
| 199 | if (rest == 3) { | ||
| 200 | k1 ^= @intCast(u32, str[offset + 2]) << 16; | ||
| 201 | } | ||
| 202 | if (rest >= 2) { | ||
| 203 | k1 ^= @intCast(u32, str[offset + 1]) << 8; | ||
| 204 | } | ||
| 205 | if (rest >= 1) { | ||
| 206 | k1 ^= @intCast(u32, str[offset + 0]); | ||
| 207 | k1 *%= c1; | ||
| 208 | k1 = rotl32(k1, 15); | ||
| 209 | k1 *%= c2; | ||
| 210 | h1 ^= k1; | ||
| 211 | } | ||
| 212 | } | ||
| 213 | h1 ^= len; | ||
| 214 | h1 ^= h1 >> 16; | ||
| 215 | h1 *%= 0x85ebca6b; | ||
| 216 | h1 ^= h1 >> 13; | ||
| 217 | h1 *%= 0xc2b2ae35; | ||
| 218 | h1 ^= h1 >> 16; | ||
| 219 | return h1; | ||
| 220 | } | ||
| 221 | |||
| 222 | pub fn hashUint32(v: u32) u32 { | ||
| 223 | return @inlineCall(Self.hashUint32WithSeed, v, default_seed); | ||
| 224 | } | ||
| 225 | |||
| 226 | pub fn hashUint32WithSeed(v: u32, seed: u32) u32 { | ||
| 227 | const c1: u32 = 0xcc9e2d51; | ||
| 228 | const c2: u32 = 0x1b873593; | ||
| 229 | const len: u32 = 4; | ||
| 230 | var h1: u32 = seed; | ||
| 231 | var k1: u32 = undefined; | ||
| 232 | k1 = v *% c1; | ||
| 233 | k1 = rotl32(k1, 15); | ||
| 234 | k1 *%= c2; | ||
| 235 | h1 ^= k1; | ||
| 236 | h1 = rotl32(h1, 13); | ||
| 237 | h1 *%= 5; | ||
| 238 | h1 +%= 0xe6546b64; | ||
| 239 | h1 ^= len; | ||
| 240 | h1 ^= h1 >> 16; | ||
| 241 | h1 *%= 0x85ebca6b; | ||
| 242 | h1 ^= h1 >> 13; | ||
| 243 | h1 *%= 0xc2b2ae35; | ||
| 244 | h1 ^= h1 >> 16; | ||
| 245 | return h1; | ||
| 246 | } | ||
| 247 | |||
| 248 | pub fn hashUint64(v: u64) u32 { | ||
| 249 | return @inlineCall(Self.hashUint64WithSeed, v, default_seed); | ||
| 250 | } | ||
| 251 | |||
| 252 | pub fn hashUint64WithSeed(v: u64, seed: u32) u32 { | ||
| 253 | const c1: u32 = 0xcc9e2d51; | ||
| 254 | const c2: u32 = 0x1b873593; | ||
| 255 | const len: u32 = 8; | ||
| 256 | var h1: u32 = seed; | ||
| 257 | var k1: u32 = undefined; | ||
| 258 | k1 = @truncate(u32, v) *% c1; | ||
| 259 | k1 = rotl32(k1, 15); | ||
| 260 | k1 *%= c2; | ||
| 261 | h1 ^= k1; | ||
| 262 | h1 = rotl32(h1, 13); | ||
| 263 | h1 *%= 5; | ||
| 264 | h1 +%= 0xe6546b64; | ||
| 265 | k1 = @truncate(u32, v >> 32) *% c1; | ||
| 266 | k1 = rotl32(k1, 15); | ||
| 267 | k1 *%= c2; | ||
| 268 | h1 ^= k1; | ||
| 269 | h1 = rotl32(h1, 13); | ||
| 270 | h1 *%= 5; | ||
| 271 | h1 +%= 0xe6546b64; | ||
| 272 | h1 ^= len; | ||
| 273 | h1 ^= h1 >> 16; | ||
| 274 | h1 *%= 0x85ebca6b; | ||
| 275 | h1 ^= h1 >> 13; | ||
| 276 | h1 *%= 0xc2b2ae35; | ||
| 277 | h1 ^= h1 >> 16; | ||
| 278 | return h1; | ||
| 279 | } | ||
| 280 | }; | ||
| 281 | |||
| 282 | fn SMHasherTest(comptime hash_fn: var, comptime hashbits: u32) u32 { | ||
| 283 | const hashbytes = hashbits / 8; | ||
| 284 | var key: [256]u8 = undefined; | ||
| 285 | var hashes: [hashbytes * 256]u8 = undefined; | ||
| 286 | var final: [hashbytes]u8 = undefined; | ||
| 287 | |||
| 288 | @memset(@ptrCast([*]u8, &key[0]), 0, @sizeOf(@typeOf(key))); | ||
| 289 | @memset(@ptrCast([*]u8, &hashes[0]), 0, @sizeOf(@typeOf(hashes))); | ||
| 290 | @memset(@ptrCast([*]u8, &final[0]), 0, @sizeOf(@typeOf(final))); | ||
| 291 | |||
| 292 | var i: u32 = 0; | ||
| 293 | while (i < 256) : (i += 1) { | ||
| 294 | key[i] = @truncate(u8, i); | ||
| 295 | |||
| 296 | var h = hash_fn(key[0..i], 256 - i); | ||
| 297 | if (builtin.endian == builtin.Endian.Big) | ||
| 298 | h = @byteSwap(@typeOf(h), h); | ||
| 299 | @memcpy(@ptrCast([*]u8, &hashes[i * hashbytes]), @ptrCast([*]u8, &h), hashbytes); | ||
| 300 | } | ||
| 301 | |||
| 302 | return @truncate(u32, hash_fn(hashes, 0)); | ||
| 303 | } | ||
| 304 | |||
| 305 | test "murmur2_32" { | ||
| 306 | testing.expectEqual(SMHasherTest(Murmur2_32.hashWithSeed, 32), 0x27864C1E); | ||
| 307 | var v0: u32 = 0x12345678; | ||
| 308 | var v1: u64 = 0x1234567812345678; | ||
| 309 | var v0le: u32 = v0; | ||
| 310 | var v1le: u64 = v1; | ||
| 311 | if (builtin.endian == builtin.Endian.Big) { | ||
| 312 | v0le = @byteSwap(u32, v0le); | ||
| 313 | v1le = @byteSwap(u64, v1le); | ||
| 314 | } | ||
| 315 | testing.expectEqual(Murmur2_32.hash(@ptrCast([*]u8, &v0le)[0..4]), Murmur2_32.hashUint32(v0)); | ||
| 316 | testing.expectEqual(Murmur2_32.hash(@ptrCast([*]u8, &v1le)[0..8]), Murmur2_32.hashUint64(v1)); | ||
| 317 | } | ||
| 318 | |||
| 319 | test "murmur2_64" { | ||
| 320 | std.testing.expectEqual(SMHasherTest(Murmur2_64.hashWithSeed, 64), 0x1F0D3804); | ||
| 321 | var v0: u32 = 0x12345678; | ||
| 322 | var v1: u64 = 0x1234567812345678; | ||
| 323 | var v0le: u32 = v0; | ||
| 324 | var v1le: u64 = v1; | ||
| 325 | if (builtin.endian == builtin.Endian.Big) { | ||
| 326 | v0le = @byteSwap(u32, v0le); | ||
| 327 | v1le = @byteSwap(u64, v1le); | ||
| 328 | } | ||
| 329 | testing.expectEqual(Murmur2_64.hash(@ptrCast([*]u8, &v0le)[0..4]), Murmur2_64.hashUint32(v0)); | ||
| 330 | testing.expectEqual(Murmur2_64.hash(@ptrCast([*]u8, &v1le)[0..8]), Murmur2_64.hashUint64(v1)); | ||
| 331 | } | ||
| 332 | |||
| 333 | test "murmur3_32" { | ||
| 334 | std.testing.expectEqual(SMHasherTest(Murmur3_32.hashWithSeed, 32), 0xB0F57EE3); | ||
| 335 | var v0: u32 = 0x12345678; | ||
| 336 | var v1: u64 = 0x1234567812345678; | ||
| 337 | var v0le: u32 = v0; | ||
| 338 | var v1le: u64 = v1; | ||
| 339 | if (builtin.endian == builtin.Endian.Big) { | ||
| 340 | v0le = @byteSwap(u32, v0le); | ||
| 341 | v1le = @byteSwap(u64, v1le); | ||
| 342 | } | ||
| 343 | testing.expectEqual(Murmur3_32.hash(@ptrCast([*]u8, &v0le)[0..4]), Murmur3_32.hashUint32(v0)); | ||
| 344 | testing.expectEqual(Murmur3_32.hash(@ptrCast([*]u8, &v1le)[0..8]), Murmur3_32.hashUint64(v1)); | ||
| 345 | } | ||