布隆过滤器

/*****************************************************
copyright (C), 2014-2015, Lighting Studio. Co.,     Ltd. 
File name：
Author：Jerey_Jobs    Version:0.1    Date: 
Description：
Funcion List: 
*****************************************************/

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
#include <time.h>

#define S_size 10000
#define barrel_size 5000
#define Hash_count 4

int hashvalue(int data,double rand_num)
{
//    double rand_num;
    int hash;
//    rand_num=(rand()%barrel_size)/(double)barrel_size;
    hash=(int)(barrel_size*(fmod(rand_num*data,1.0)));

    return hash;
}

void creat_randnum(double *hash)
{
    int i;
    //srand(time(0));
    for(i=0;i<Hash_count;i++)
    {
        *(hash+i)=(rand()%(barrel_size/10))/(double)barrel_size;        
    }
}

int main()
{
    int S[S_size];//S集合
    int barrel[barrel_size]={0};//桶的大小
    int data;
    int i=0;
    int count_estimate=0;//估计值
    int count_only=0;//query中在S集合中的个数
    double hash[Hash_count];//hash的值
    srand((int)time(NULL));
    FILE *fp=fopen("./stream_for_bm.txt","r");//
    creat_randnum(hash);
    if(fp == NULL)//stream_for_bm.txt打开失败
    {
        printf("stream_for_bm can't open
'");
        return 0;
    }

    while(!feof(fp))
    {
        fscanf(fp,"%d",&data);//存入S集合
        S[i++]=data;
        
        for(int j=0;j<Hash_count;j++)//Hash_count数目个hash函数
        {
            barrel[hashvalue(data,hash[j])]=1;//存入桶中
        }
    }

    FILE *fp_query=fopen("./stream_for_query.txt","r");
    if(fp_query == NULL)
    {
        printf("stream_for_query.txt can't open
");
        return 0;
    }
    while(!feof(fp_query))
    {
        fscanf(fp_query,"%d",&data);
        for(int j=0;j<S_size;j++)
        {
            if(data == S[j])
                count_only++;//精确值
        }
        int tmp_count=0;
        for(int j=0;j<Hash_count;j++)//Hash_count数目个hash函数
        {
            if(barrel[hashvalue(data,hash[j])] == 0)
                break;
            else
                tmp_count++;
        }
        if(tmp_count == Hash_count)
            count_estimate++;
    }
    printf("count_only=%d
count_estimate=%d
",count_only,count_estimate);
    printf("error rate:%.2lf
",(count_estimate-count_only)/(double)20000);
    
    fclose(fp);
    fclose(fp_query);
    system("pause");
    return 0;
}