/
xxhash.go
185 lines (154 loc) · 3.44 KB
/
xxhash.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
// package xxhash implements a fast non-cryptographic hashing algorithm ported
// from https://code.google.com/p/xxhash/.
package xxhash
import (
"encoding/binary"
"hash"
"unsafe"
)
const (
prime1 = 2654435761
prime2 = 2246822519
prime3 = 3266489917
prime4 = 668265263
prime5 = 374761393
)
type digest struct {
size int
seed uint32
v1 uint32
v2 uint32
v3 uint32
v4 uint32
over int
tail []byte
}
// New returns a new 32-bit xxHash hash.Hash initialized with seed.
func New(seed uint32) hash.Hash32 {
state := new(digest)
state.seed = seed
state.Reset()
return state
}
func (d *digest) BlockSize() int {
return 4
}
func (d *digest) Reset() {
d.size = 0
d.v1, d.v2, d.v3, d.v4 = initialize(d.seed)
d.over = 0
d.tail = make([]byte, 16)
}
func (d *digest) Size() int {
return 4
}
func (d *digest) Sum(b []byte) []byte {
h := d.Sum32()
r := [4]byte{}
binary.BigEndian.PutUint32(r[:], h)
return append(b, r[:]...)
}
func (d *digest) Sum32() uint32 {
var h uint32
if d.size < 16 {
h = d.seed + prime5
} else {
h = (d.v1 << 1) | (d.v1 >> 31)
h += (d.v2 << 7) | (d.v2 >> 25)
h += (d.v3 << 12) | (d.v3 >> 20)
h += (d.v4 << 18) | (d.v4 >> 14)
}
h += uint32(d.size)
h = tail(h, d.tail, d.over)
return fmix(h)
}
func (d *digest) Write(data []byte) (int, error) {
i := 0
n := len(data)
d.size += n
// A blocksize of 16 is required in order to Write.
if n+d.over < 16 {
copy(d.tail[d.over:], data)
d.over += n
return n, nil
}
// Write full blocksize of previous and current data.
copy(d.tail[d.over:], data)
i = 16 - d.over
j := uintptr(unsafe.Pointer(&d.tail[0]))
k := j + 16
d.v1, d.v2, d.v3, d.v4 = body(d.v1, d.v2, d.v3, d.v4, j, k)
d.over = 0
// Write all 16 byte blocks of data.
j = uintptr(unsafe.Pointer(&data[i]))
k = j + uintptr(n-i)
d.v1, d.v2, d.v3, d.v4 = body(d.v1, d.v2, d.v3, d.v4, j, k)
i = n - (n & 15)
// Place any remaining data in memory.
copy(d.tail, data[i:])
d.over = n - i
return n, nil
}
// Checksum returns the xxHash checksum of data.
func Checksum(data []byte, seed uint32) uint32 {
n := len(data)
v1, v2, v3, v4 := initialize(seed)
if n > 0 {
i := uintptr(unsafe.Pointer(&data[0]))
j := i + uintptr(n)
v1, v2, v3, v4 = body(v1, v2, v3, v4, i, j)
}
var h uint32
if n < 16 {
h = seed + prime5
} else {
h = (v1 << 1) | (v1 >> 31)
h += (v2 << 7) | (v2 >> 25)
h += (v3 << 12) | (v3 >> 20)
h += (v4 << 18) | (v4 >> 14)
}
h += uint32(n)
q := data[n-(n&15):]
h = tail(h, q, len(q))
return fmix(h)
}
func initialize(seed uint32) (v1, v2, v3, v4 uint32) {
v1 = seed + prime1 + prime2
v2 = seed + prime2
v3 = seed
v4 = seed - prime1
return
}
func update(v, x uint32) uint32 {
v += x * prime2
return ((v << 13) | (v >> 19)) * prime1
}
func body(v1, v2, v3, v4 uint32, i, n uintptr) (uint32, uint32, uint32, uint32) {
for i <= n-16 {
v1, i = update(v1, *(*uint32)(unsafe.Pointer(i))), i+4
v2, i = update(v2, *(*uint32)(unsafe.Pointer(i))), i+4
v3, i = update(v3, *(*uint32)(unsafe.Pointer(i))), i+4
v4, i = update(v4, *(*uint32)(unsafe.Pointer(i))), i+4
}
return v1, v2, v3, v4
}
func tail(h uint32, p []byte, n int) uint32 {
i := 0
for ; i <= n-4; i += 4 {
h += *(*uint32)(unsafe.Pointer(&p[i])) * prime3
h = ((h << 17) | (h >> 15)) * prime4
}
for ; i < n; i++ {
h += uint32(p[i]) * prime5
h = ((h << 11) | (h >> 21)) * prime1
}
return h
}
func fmix(h uint32) uint32 {
h ^= h >> 15
h *= prime2
h ^= h >> 13
h *= prime3
h ^= h >> 16
return h
}