Bloom_filter算法(zz)

（转自: http://www.felix021.com/blog/read.php?1882）

布隆过滤器的详细介绍和典型用途，可参见
Wikipedia：http://en.wikipedia.org/wiki/Bloom_filter
谷歌黑板报（数学之美）：http://www.google.cn/ggblog/googlechinablog/2007/07/bloom-filter_7469.html

下面是一个简单的布隆过滤器的C/C++实现，以及使用例程。使用sdbmhash字符串hash方法来进行hash。

  1 #include <stdio.h>
  2 #include <stdlib.h>
  3 #include <string.h>
  4 
  5 unsigned int jshash(const char *s, unsigned size);
  6 unsigned int sdbmhash(const char *s, unsigned size);
  7 
  8 /* ------------- bloom types and funcs --------------- */
  9 const unsigned char masks[8] = {0x01,0x02,0x04,0x08,0x10,0x20,0x40,0x80};
 10 
 11 typedef unsigned (*hash_func_ptr)(const char *buffer, unsigned size);
 12 struct __bloom_filter
 13 {
 14     unsigned n;
 15     unsigned size;
 16     unsigned char *bits;
 17     hash_func_ptr hash;
 18 };
 19 typedef struct __bloom_filter* bloom_filter;
 20 
 21 bloom_filter bloom_init (unsigned n, hash_func_ptr hash);
 22 int bloom_insert(bloom_filter b, void *data, unsigned size);
 23 int bloom_check(bloom_filter b, void *data, unsigned size);
 24 void bloom_destroy(bloom_filter b);
 25 /* ------------- end of bloom types and funcs --------------- */
 26 
 27 int main()
 28 {
 29     const int size = 655371;
 30     bloom_filter b1 = bloom_init(size, sdbmhash);
 31     for (int i = 0; i < size / 2; i += 2)
 32     {
 33         if (!bloom_insert(b1, &i, sizeof(i)))
 34         {
 35             fprintf(stderr, "err insert %d\n", i);
 36             exit(1);
 37         }
 38     }
 39     printf("insert ok\n");
 40 
 41     int cnt = 0;
 42     for (int i = 0; i < size / 2; i++)
 43     {
 44         if (bloom_check(b1, &i, sizeof(i)))
 45         {
 46             if (i & 1)
 47             {
 48                 //printf("i = %d should not be checked, tolerable.\n", i);
 49                 cnt++;
 50             }
 51         }
 52         else
 53         {
 54             if (!(i & 1))
 55             {
 56                 printf("i = %d should be checked! BUG!\n", i);
 57             }
 58         }
 59     }
 60     printf("cnt = %d\n", cnt);
 61     return 0;
 62 }
 63 
 64 bloom_filter bloom_init (unsigned n, hash_func_ptr hash)
 65 {
 66     bloom_filter b = (bloom_filter)malloc(sizeof(__bloom_filter));
 67     if (b == NULL)
 68     {
 69         fprintf(stderr, "bloom_init: err malloc bloom_filter\n");
 70         return NULL;
 71     }
 72 
 73     b->n    = n;
 74     b->size = (n + 7) / 8;
 75     b->hash = hash;
 76 
 77     b->bits = (unsigned char *)malloc(b->size);
 78     memset(b->bits, 0, b->size);
 79     if (b->bits == NULL)
 80     {
 81         fprintf(stderr, "bloom_init: err malloc bits\n");
 82         return NULL;
 83     }
 84     return b;
 85 }
 86 
 87 int bloom_insert(bloom_filter b, void *data, unsigned size)
 88 {
 89     unsigned h = b->hash((const char *)data, size) % (b->n);
 90     unsigned idx = h / 8;
 91     if (idx >= b->size)
 92     {
 93         fprintf(stderr, "bloom_insert: hash value overflow\n");
 94         return 0;
 95     }
 96     b->bits[idx] |= masks[h % 8];
 97     //printf("h = %2d, idx = %2d, bit = %2d\n", h, idx, h % 8);
 98     return 1;
 99 }
100 
101 int bloom_check(bloom_filter b, void *data, unsigned size)
102 {
103     unsigned h = b->hash((const char *)data, size) % (b->n);
104     unsigned idx = h / 8;
105     if (idx >= b->size)
106     {
107         fprintf(stderr, "bloom_insert: hash value overflow\n");
108         exit(1);
109     }
110     return !!(b->bits[idx] & masks[h % 8]);
111 }
112 
113 void bloom_destroy(bloom_filter b)
114 {
115     if (b != NULL)
116     {
117         if (b->bits != NULL)
118             free(b->bits);
119         free(b);
120     }
121 }
122 
123 //-----------------------------------------------
124 
125 unsigned int jshash(const char *s, unsigned size)
126 {
127     int hash = 1315423911;
128     unsigned len = 0;
129     while (len < size)
130     {
131         hash ^= (hash << 5) + s[len] + (hash >> 2);
132         len++;
133     }
134     return (hash & 0x7fffffffl);
135 }
136 
137 unsigned int sdbmhash(const char *s, unsigned size)
138 {
139     int hash = 0;
140     unsigned len = 0;
141     while (len < size)
142     {
143         hash = (hash << 6) + (hash << 16) - hash + s[len];
144         len++;
145     }
146     return (hash & 0x7fffffffl);
147 }
148