27 |
andreas |
1 |
/*
|
|
|
2 |
* -------------------------------------------------------------------------------
|
|
|
3 |
* lookup3.c, by Bob Jenkins, May 2006, Public Domain.
|
|
|
4 |
*
|
|
|
5 |
* These are functions for producing 32-bit hashes for hash table lookup.
|
|
|
6 |
* hashword(), hashlittle(), hashlittle2(), hashbig(), mix(), and final()
|
|
|
7 |
* are externally useful functions. Routines to test the hash are included
|
|
|
8 |
* if SELF_TEST is defined. You can use this free for any purpose. It's in
|
|
|
9 |
* the public domain. It has no warranty.
|
|
|
10 |
*
|
|
|
11 |
* You probably want to use hashlittle(). hashlittle() and hashbig()
|
|
|
12 |
* hash byte arrays. hashlittle() is is faster than hashbig() on
|
|
|
13 |
* little-endian machines. Intel and AMD are little-endian machines.
|
|
|
14 |
* On second thought, you probably want hashlittle2(), which is identical to
|
|
|
15 |
* hashlittle() except it returns two 32-bit hashes for the price of one.
|
|
|
16 |
* You could implement hashbig2() if you wanted but I haven't bothered here.
|
|
|
17 |
*
|
|
|
18 |
* If you want to find a hash of, say, exactly 7 integers, do
|
|
|
19 |
* a = i1; b = i2; c = i3;
|
|
|
20 |
* mix(a,b,c);
|
|
|
21 |
* a += i4; b += i5; c += i6;
|
|
|
22 |
* mix(a,b,c);
|
|
|
23 |
* a += i7;
|
|
|
24 |
* final(a,b,c);
|
|
|
25 |
* then use c as the hash value. If you have a variable length array of
|
|
|
26 |
* 4-byte integers to hash, use hashword(). If you have a byte array (like
|
|
|
27 |
* a character string), use hashlittle(). If you have several byte arrays, or
|
|
|
28 |
* a mix of things, see the comments above hashlittle().
|
|
|
29 |
*
|
|
|
30 |
* Why is this so big? I read 12 bytes at a time into 3 4-byte integers,
|
|
|
31 |
* then mix those integers. This is fast (you can do a lot more thorough
|
|
|
32 |
* mixing with 12*3 instructions on 3 integers than you can with 3 instructions
|
|
|
33 |
* on 1 byte), but shoehorning those bytes into integers efficiently is messy.
|
|
|
34 |
* -------------------------------------------------------------------------------
|
|
|
35 |
*/
|
|
|
36 |
|
|
|
37 |
#ifndef __LOOKUP3_H__
|
|
|
38 |
#define __LOOKUP3_H__
|
|
|
39 |
|
55 |
andreas |
40 |
#ifndef __APPLE__
|
27 |
andreas |
41 |
#ifndef uint32_t
|
|
|
42 |
typedef unsigned long uint32_t;
|
|
|
43 |
#endif
|
55 |
andreas |
44 |
#endif
|
27 |
andreas |
45 |
#ifndef uint8_t
|
|
|
46 |
typedef unsigned char uint8_t;
|
|
|
47 |
#endif
|
|
|
48 |
|
|
|
49 |
/*
|
|
|
50 |
* --------------------------------------------------------------------
|
|
|
51 |
* This works on all machines. To be useful, it requires
|
|
|
52 |
* -- that the key be an array of uint32_t's, and
|
|
|
53 |
* -- that the length be the number of uint32_t's in the key
|
|
|
54 |
*
|
|
|
55 |
* The function hashword() is identical to hashlittle() on little-endian
|
|
|
56 |
* machines, and identical to hashbig() on big-endian machines,
|
|
|
57 |
* except that the length has to be measured in uint32_ts rather than in
|
|
|
58 |
* bytes. hashlittle() is more complicated than hashword() only because
|
|
|
59 |
* hashlittle() has to dance around fitting the key bytes into registers.
|
|
|
60 |
* --------------------------------------------------------------------
|
|
|
61 |
*/
|
|
|
62 |
uint32_t hashword(
|
|
|
63 |
const uint32_t *k, /* the key, an array of uint32_t values */
|
|
|
64 |
size_t length, /* the length of the key, in uint32_ts */
|
|
|
65 |
uint32_t initval); /* the previous hash, or an arbitrary value */
|
|
|
66 |
|
|
|
67 |
/*
|
|
|
68 |
* --------------------------------------------------------------------
|
|
|
69 |
* hashword2() -- same as hashword(), but take two seeds and return two
|
|
|
70 |
* 32-bit values. pc and pb must both be nonnull, and *pc and *pb must
|
|
|
71 |
* both be initialized with seeds. If you pass in (*pb)==0, the output
|
|
|
72 |
* (*pc) will be the same as the return value from hashword().
|
|
|
73 |
* --------------------------------------------------------------------
|
|
|
74 |
*/
|
|
|
75 |
void hashword2 (
|
|
|
76 |
const uint32_t *k, /* the key, an array of uint32_t values */
|
|
|
77 |
size_t length, /* the length of the key, in uint32_ts */
|
|
|
78 |
uint32_t *pc, /* IN: seed OUT: primary hash value */
|
|
|
79 |
uint32_t *pb); /* IN: more seed OUT: secondary hash value */
|
|
|
80 |
|
|
|
81 |
/*
|
|
|
82 |
* -------------------------------------------------------------------------------
|
|
|
83 |
* hashlittle() -- hash a variable-length key into a 32-bit value
|
|
|
84 |
* k : the key (the unaligned variable-length array of bytes)
|
|
|
85 |
* length : the length of the key, counting by bytes
|
|
|
86 |
* initval : can be any 4-byte value
|
|
|
87 |
* Returns a 32-bit value. Every bit of the key affects every bit of
|
|
|
88 |
* the return value. Two keys differing by one or two bits will have
|
|
|
89 |
* totally different hash values.
|
|
|
90 |
*
|
|
|
91 |
* The best hash table sizes are powers of 2. There is no need to do
|
|
|
92 |
* mod a prime (mod is sooo slow!). If you need less than 32 bits,
|
|
|
93 |
* use a bitmask. For example, if you need only 10 bits, do
|
|
|
94 |
* h = (h & hashmask(10));
|
|
|
95 |
* In which case, the hash table should have hashsize(10) elements.
|
|
|
96 |
*
|
|
|
97 |
* If you are hashing n strings (uint8_t **)k, do it like this:
|
|
|
98 |
* for (i=0, h=0; i<n; ++i) h = hashlittle( k[i], len[i], h);
|
|
|
99 |
*
|
|
|
100 |
* By Bob Jenkins, 2006. bob_jenkins@burtleburtle.net. You may use this
|
|
|
101 |
* code any way you wish, private, educational, or commercial. It's free.
|
|
|
102 |
*
|
|
|
103 |
* Use for hash table lookup, or anything where one collision in 2^^32 is
|
|
|
104 |
* acceptable. Do NOT use for cryptographic purposes.
|
|
|
105 |
* -------------------------------------------------------------------------------
|
|
|
106 |
*/
|
|
|
107 |
|
|
|
108 |
uint32_t hashlittle( const void *key, size_t length, uint32_t initval);
|
|
|
109 |
|
|
|
110 |
/*
|
|
|
111 |
* hashlittle2: return 2 32-bit hash values
|
|
|
112 |
*
|
|
|
113 |
* This is identical to hashlittle(), except it returns two 32-bit hash
|
|
|
114 |
* values instead of just one. This is good enough for hash table
|
|
|
115 |
* lookup with 2^^64 buckets, or if you want a second hash if you're not
|
|
|
116 |
* happy with the first, or if you want a probably-unique 64-bit ID for
|
|
|
117 |
* the key. *pc is better mixed than *pb, so use *pc first. If you want
|
|
|
118 |
* a 64-bit value do something like "*pc + (((uint64_t)*pb)<<32)".
|
|
|
119 |
*/
|
|
|
120 |
void hashlittle2(
|
|
|
121 |
const void *key, /* the key to hash */
|
|
|
122 |
size_t length, /* length of the key */
|
|
|
123 |
uint32_t *pc, /* IN: primary initval, OUT: primary hash */
|
|
|
124 |
uint32_t *pb); /* IN: secondary initval, OUT: secondary hash */
|
|
|
125 |
|
|
|
126 |
/*
|
|
|
127 |
* hashbig():
|
|
|
128 |
* This is the same as hashword() on big-endian machines. It is different
|
|
|
129 |
* from hashlittle() on all machines. hashbig() takes advantage of
|
|
|
130 |
* big-endian byte ordering.
|
|
|
131 |
*/
|
|
|
132 |
uint32_t hashbig( const void *key, size_t length, uint32_t initval);
|
|
|
133 |
|
|
|
134 |
|
|
|
135 |
#endif
|