forked from reusee/mmh3
/
mmh3.go
412 lines (362 loc) · 8.38 KB
/
mmh3.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
package mmh3
import (
"encoding/binary"
"math"
"math/bits"
"reflect"
"unsafe"
)
const (
h32c1 uint32 = 0xcc9e2d51
h32c2 uint32 = 0x1b873593
h64c1 uint64 = 0x87c37b91114253d5
h64c2 uint64 = 0x4cf5ad432745937f
h128BlockSize = 16
)
type Hash128Value struct {
high, low uint64
}
// Values returns the two 8 byte values that compose the hash
func (h Hash128Value) Values() (uint64, uint64) {
return h.high, h.low
}
// Write writes the hash to given buffer which should be 16 bytes long
func (h Hash128Value) Write(out []byte) {
if len(out) != 16 {
panic("provided return buffer should be 16 bytes")
}
rh := *(*reflect.SliceHeader)(unsafe.Pointer(&out))
rh.Len = 2
b := *(*[]uint64)(unsafe.Pointer(&rh))
b[0] = h.high
b[1] = h.low
rh.Len = 16
}
// Bytes returns the 16 bytes that compose the hash
func (h Hash128Value) Bytes() []byte {
b := make([]byte, 16)
h.Write(b)
return b
}
func Hash32(key []byte) uint32 {
length := len(key)
if length == 0 {
return 0
}
nblocks := length / 4
var h, k uint32
for i := 0; i < nblocks; i++ {
k = binary.LittleEndian.Uint32(key[i*4:])
k *= h32c1
k = (k << 15) | (k >> (32 - 15))
k *= h32c2
h ^= k
h = (h << 13) | (h >> (32 - 13))
h = (h * 5) + 0xe6546b64
}
k = 0
tailIndex := nblocks * 4
switch length & 3 {
case 3:
k ^= uint32(key[tailIndex+2]) << 16
fallthrough
case 2:
k ^= uint32(key[tailIndex+1]) << 8
fallthrough
case 1:
k ^= uint32(key[tailIndex])
k *= h32c1
k = (k << 15) | (k >> (32 - 15))
k *= h32c2
h ^= k
}
h ^= uint32(length)
h ^= h >> 16
h *= 0x85ebca6b
h ^= h >> 13
h *= 0xc2b2ae35
h ^= h >> 16
return h
}
// Hash128 s a version of MurmurHash which is designed to run only on little-endian processors
func Hash128(key []byte) Hash128Value {
length := len(key)
if length == 0 {
return Hash128Value{}
}
nblocks := length / 16
var h1, h2 uint64
values := (*(*[]uint64)(unsafe.Pointer(&key)))[: nblocks*2 : nblocks*2]
for i := 0; i < len(values); i += 2 {
h1, h2 = merge128(h1, h2, values[i], values[i+1])
}
var k1, k2 uint64
tailIndex := nblocks * 16
switch length & 15 {
case 15:
k2 ^= uint64(key[tailIndex+14]) << 48
fallthrough
case 14:
k2 ^= uint64(key[tailIndex+13]) << 40
fallthrough
case 13:
k2 ^= uint64(key[tailIndex+12]) << 32
fallthrough
case 12:
k2 ^= uint64(key[tailIndex+11]) << 24
fallthrough
case 11:
k2 ^= uint64(key[tailIndex+10]) << 16
fallthrough
case 10:
k2 ^= uint64(key[tailIndex+9]) << 8
fallthrough
case 9:
k2 ^= uint64(key[tailIndex+8])
k2 *= h64c2
k2 = bits.RotateLeft64(k2, 33)
k2 *= h64c1
h2 ^= k2
fallthrough
case 8:
k1 ^= uint64(key[tailIndex+7]) << 56
fallthrough
case 7:
k1 ^= uint64(key[tailIndex+6]) << 48
fallthrough
case 6:
k1 ^= uint64(key[tailIndex+5]) << 40
fallthrough
case 5:
k1 ^= uint64(key[tailIndex+4]) << 32
fallthrough
case 4:
k1 ^= uint64(key[tailIndex+3]) << 24
fallthrough
case 3:
k1 ^= uint64(key[tailIndex+2]) << 16
fallthrough
case 2:
k1 ^= uint64(key[tailIndex+1]) << 8
fallthrough
case 1:
k1 ^= uint64(key[tailIndex])
k1 *= h64c1
k1 = bits.RotateLeft64(k1, 31)
k1 *= h64c2
h1 ^= k1
}
h1 ^= uint64(length)
h2 ^= uint64(length)
h1 += h2
h2 += h1
h1 ^= h1 >> 33
h1 *= 0xff51afd7ed558ccd
h1 ^= h1 >> 33
h1 *= 0xc4ceb9fe1a85ec53
h1 ^= h1 >> 33
h2 ^= h2 >> 33
h2 *= 0xff51afd7ed558ccd
h2 ^= h2 >> 33
h2 *= 0xc4ceb9fe1a85ec53
h2 ^= h2 >> 33
h1 += h2
h2 += h1
return Hash128Value{high: h1, low: h2}
}
// Hash128x64 calls WriteHash128x64 with an allocated output buffer
func Hash128x64(key []byte) []byte {
return Hash128(key).Bytes()
}
// WriteHash128x64 creates a hash of key and writes it to ret
func WriteHash128x64(key, ret []byte) {
Hash128(key).Write(ret)
}
type HashWriter128 struct {
h1 uint64
h2 uint64
tail [h128BlockSize]byte
tailLength int
length int
}
func (hw *HashWriter128) Reset() {
hw.h1 = 0
hw.h2 = 0
hw.length = 0
hw.tailLength = 0
}
func (hw *HashWriter128) Size() int {
return h128BlockSize
}
func (hw *HashWriter128) BlockSize() int {
return h128BlockSize
}
// Write writes the given bytes to the hash
func (hw *HashWriter128) Write(b []byte) (int, error) {
hw.AddBytes(b)
return len(b), nil
}
// WriteString writes the given string to the hash
func (hw *HashWriter128) WriteString(s string) (int, error) {
hw.AddString(s)
return len(s), nil
}
// Sum completes the hash and _appends_ the hash to the given input buffer
// This is kept for backwards compatibility
func (hw *HashWriter128) Sum(b []byte) []byte {
h1, h2 := hw.Sum128().Values()
var retbuf [8]byte
binary.LittleEndian.PutUint64(retbuf[:], h1)
b = append(b, retbuf[:]...)
binary.LittleEndian.PutUint64(retbuf[:], h2)
b = append(b, retbuf[:]...)
return b
}
// AddBytes adds the given bytes to the hash
func (hw *HashWriter128) AddBytes(input []byte) {
chunkStart := 0 // The index in to the input where we will begin reading full chunks of data
inputLength := len(input)
hw.length += inputLength
// If there are bytes remaining in the tail, attempt to form
// a full block and process that first
if hw.tailLength > 0 {
bytesNeeded := h128BlockSize - hw.tailLength
// If the input isn't large enough to form a full block with the tail
// copy it in to the tail and return
if bytesNeeded > inputLength {
copy(hw.tail[hw.tailLength:], input)
hw.tailLength += inputLength
return
}
copy(hw.tail[hw.tailLength:], input[0:bytesNeeded])
chunkStart += bytesNeeded
inputLength -= bytesNeeded
k1Buf := hw.tail[0:8]
k2Buf := hw.tail[8:]
k1 := *(*uint64)(unsafe.Pointer(&k1Buf[0]))
k2 := *(*uint64)(unsafe.Pointer(&k2Buf[0]))
hw.h1, hw.h2 = merge128(hw.h1, hw.h2, k1, k2)
}
chunks := inputLength / h128BlockSize
if chunks > 0 {
chunkInput := input
if chunkStart > 0 {
chunkInput = input[chunkStart:]
}
values := (*(*[]uint64)(unsafe.Pointer(&chunkInput)))[: chunks*2 : chunks*2]
for i := 0; i < len(values); i += 2 {
hw.h1, hw.h2 = merge128(hw.h1, hw.h2, values[i], values[i+1])
}
}
tailLength := inputLength % h128BlockSize
hw.tailLength = tailLength
if tailLength == 0 {
return
}
// Copy remaining bytes in to tail
tailStart := inputLength - tailLength + chunkStart
copy(hw.tail[:], input[tailStart:])
}
// AddString adds the given string to the hash
func (hw *HashWriter128) AddString(key string) {
if len(key) == 0 {
return
}
sh := (*reflect.StringHeader)(unsafe.Pointer(&key))
byteSlice := (*[math.MaxInt32 - 1]byte)(unsafe.Pointer(sh.Data))[:sh.Len:sh.Len]
hw.AddBytes(byteSlice)
}
// Sum completes the hash value and returns it. Reset should be called after calling Sum
func (hw *HashWriter128) Sum128() Hash128Value {
h1, h2 := hw.h1, hw.h2
k1, k2 := uint64(0), uint64(0)
switch hw.tailLength {
case 15:
k2 ^= uint64(hw.tail[14]) << 48
fallthrough
case 14:
k2 ^= uint64(hw.tail[13]) << 40
fallthrough
case 13:
k2 ^= uint64(hw.tail[12]) << 32
fallthrough
case 12:
k2 ^= uint64(hw.tail[11]) << 24
fallthrough
case 11:
k2 ^= uint64(hw.tail[10]) << 16
fallthrough
case 10:
k2 ^= uint64(hw.tail[9]) << 8
fallthrough
case 9:
k2 ^= uint64(hw.tail[8])
k2 *= h64c2
k2 = bits.RotateLeft64(k2, 33)
k2 *= h64c1
h2 ^= k2
fallthrough
case 8:
k1 ^= uint64(hw.tail[7]) << 56
fallthrough
case 7:
k1 ^= uint64(hw.tail[6]) << 48
fallthrough
case 6:
k1 ^= uint64(hw.tail[5]) << 40
fallthrough
case 5:
k1 ^= uint64(hw.tail[4]) << 32
fallthrough
case 4:
k1 ^= uint64(hw.tail[3]) << 24
fallthrough
case 3:
k1 ^= uint64(hw.tail[2]) << 16
fallthrough
case 2:
k1 ^= uint64(hw.tail[1]) << 8
fallthrough
case 1:
k1 ^= uint64(hw.tail[0])
k1 *= h64c1
k1 = bits.RotateLeft64(k1, 31)
k1 *= h64c2
h1 ^= k1
}
h1 ^= uint64(hw.length)
h2 ^= uint64(hw.length)
h1 += h2
h2 += h1
h1 ^= h1 >> 33
h1 *= 0xff51afd7ed558ccd
h1 ^= h1 >> 33
h1 *= 0xc4ceb9fe1a85ec53
h1 ^= h1 >> 33
h2 ^= h2 >> 33
h2 *= 0xff51afd7ed558ccd
h2 ^= h2 >> 33
h2 *= 0xc4ceb9fe1a85ec53
h2 ^= h2 >> 33
h1 += h2
h2 += h1
return Hash128Value{high: h1, low: h2}
}
// merge128 merges k1 and k2 with h1 and h2 and returns the updated (h1, h2) values
func merge128(h1, h2, k1, k2 uint64) (uint64, uint64) {
k1 *= h64c1
k1 = bits.RotateLeft64(k1, 31)
k1 *= h64c2
h1 ^= k1
h1 = bits.RotateLeft64(h1, 27)
h1 += h2
h1 = h1*5 + 0x52dce729
k2 *= h64c2
k2 = bits.RotateLeft64(k2, 33)
k2 *= h64c1
h2 ^= k2
h2 = bits.RotateLeft64(h2, 31)
h2 += h1
h2 = h2*5 + 0x38495ab5
return h1, h2
}