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
|
// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
package com.yahoo.collections;
import com.yahoo.text.Utf8;
/**
* A Java port of Michael Susag's BobHash in FastLib. This version is
* specifically done to be bit compatible with the one in FastLib, as it
* is used in decoding packets from FastServer.
*
* Hash function based on
* <a href="http://burtleburtle.net/bob/hash/index.html">http://burtleburtle.net/bob/hash/index.html</a>
* by Bob Jenkins, 1996. bob_jenkins@burtleburtle.net. You may use this
* code any way you wish, private, educational, or commercial. It's free.
*
* @author Michael Susag
* @author Steinar Knutsen
*/
public class BobHash {
/**
* mix -- mix 3 32-bit values reversibly.
* For every delta with one or two bits set, and the deltas of all three
* high bits or all three low bits, whether the original value of a,b,c
* is almost all zero or is uniformly distributed,
* If mix() is run forward or backward, at least 32 bits in a,b,c
* have at least 1/4 probability of changing.
* If mix() is run forward, every bit of c will change between 1/3 and
* 2/3 of the time. (Well, 22/100 and 78/100 for some 2-bit deltas.)
* mix() was built out of 36 single-cycle latency instructions in a
* structure that could supported 2x parallelism, like so:
*
* <pre>
* a -= b;
* a -= c; x = (c>>13);
* b -= c; a ^= x;
* b -= a; x = (a<<8);
* c -= a; b ^= x;
* c -= b; x = (b>>13);
* ...
* </pre>
*
* <p>
* Unfortunately, superscalar Pentiums and Sparcs can't take advantage
* of that parallelism. They've also turned some of those single-cycle
* latency instructions into multi-cycle latency instructions. Still,
* this is the fastest good hash I could find. There were about 2^^68
* to choose from. I only looked at a billion or so.
*/
private static int[] mix(int a, int b, int c) {
a -= b; a -= c; a ^= (c >>> 13);
b -= c; b -= a; b ^= (a << 8);
c -= a; c -= b; c ^= (b >>> 13);
a -= b; a -= c; a ^= (c >>> 12);
b -= c; b -= a; b ^= (a << 16);
c -= a; c -= b; c ^= (b >>> 5);
a -= b; a -= c; a ^= (c >>> 3);
b -= c; b -= a; b ^= (a << 10);
c -= a; c -= b; c ^= (b >>> 15);
return new int[]{ a, b, c };
}
/**
* Transform a byte to an int viewed as an unsigned byte.
*/
private static int unsign(byte x) {
int y;
y = 0xFF & x;
return y;
}
/**
* Hashes a string, by calling hash(byte[] key,int initval) with
* the utf-8 bytes of the string as key and 0 as initval.
* Note: This is copying the string content, change implementation to
* use efficiently on large strings.
*
* bratseth
*/
public static int hash(String key) {
return hash(Utf8.toBytes(key), 0);
}
/**
* The hash function
*
* <p>
* hash() -- hash a variable-length key into a 32-bit value<br>
* k : the key (the unaligned variable-length array of bytes)<br>
* len : the length of the key, counting by bytes<br>
* initval : can be any 4-byte value
*
* <p>
* Returns a 32-bit value. Every bit of the key affects every bit of
* the return value. Every 1-bit and 2-bit delta achieves avalanche.
* About 6*len+35 instructions.
*
* <p>
* The best hash table sizes are powers of 2. There is no need to do
* mod a prime (mod is sooo slow!). If you need less than 32 bits,
* use a bitmask. For example, if you need only 10 bits, do
* h = (h & hashmask(10));
* In which case, the hash table should have hashsize(10) elements.
*
* If you are hashing n strings (ub1 **)k, do it like this:
* for (i=0, h=0; i<n; ++i) h = hash( k[i], len[i], h);
*
* <p>
* By Bob Jenkins, 1996. bob_jenkins@burtleburtle.net. You may use this
* code any way you wish, private, educational, or commercial. It's free.
*
* <p>
* See http://burtleburtle.net/bob/hash/evahash.html
* Use for hash table lookup, or anything where one collision in 2^^32 is
* acceptable. Do NOT use for cryptographic purposes.
*
* @param k the key
* @param initval the previous hash, or an arbitrary value
* @return A 32 bit hash value
*/
@SuppressWarnings("fallthrough")
public static int hash(byte[] k, int initval) {
int a, b, c, len;
int offset = 0;
int[] abcBuffer;
/* Set up the internal state */
len = k.length;
a = b = 0x9e3779b9; /* the golden ratio; an arbitrary value */
c = initval; /* the previous hash value */
// handle most of the key
while (len >= 12) {
a += (unsign(k[offset + 0]) + (unsign(k[offset + 1]) << 8)
+ (unsign(k[offset + 2]) << 16)
+ (unsign(k[offset + 3]) << 24));
b += (unsign(k[offset + 4]) + (unsign(k[offset + 5]) << 8)
+ (unsign(k[offset + 6]) << 16)
+ (unsign(k[offset + 7]) << 24));
c += (unsign(k[offset + 8]) + (unsign(k[offset + 9]) << 8)
+ (unsign(k[offset + 10]) << 16)
+ (unsign(k[offset + 11]) << 24));
abcBuffer = mix(a, b, c);
a = abcBuffer[0];
b = abcBuffer[1];
c = abcBuffer[2];
offset += 12;
len -= 12;
}
// handle the last 11 bytes
c += k.length;
switch (len) {
// all the case statements fall through
case 11:
c += (unsign(k[offset + 10]) << 24);
case 10:
c += (unsign(k[offset + 9]) << 16);
case 9:
c += (unsign(k[offset + 8]) << 8);
/* the first byte of c is reserved for the length */
case 8:
b += (unsign(k[offset + 7]) << 24);
case 7:
b += (unsign(k[offset + 6]) << 16);
case 6:
b += (unsign(k[offset + 5]) << 8);
case 5:
b += unsign(k[offset + 4]);
case 4:
a += (unsign(k[offset + 3]) << 24);
case 3:
a += (unsign(k[offset + 2]) << 16);
case 2:
a += (unsign(k[offset + 1]) << 8);
case 1:
a += unsign(k[offset + 0]);
/* case 0: nothing left to add */
}
abcBuffer = mix(a, b, c);
return abcBuffer[2];
}
}
|
