DGIM算法

/*****************************************************
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>


typedef struct
{
    unsigned int count;
    unsigned int time_stamp;
}BARR;//桶的属性
/*函数声明*/
int accurate(int *wins);
void merge(BARR*, int);
int estimate(BARR*);
void judge_file_pointer_null(FILE *fp);

int window;//窗口大小
int size = 0;//桶的个数

int main()
{
    //BARR barrel[276];
    BARR *barrel = (BARR *)malloc(150000 * sizeof(BARR));
    int i = 0;
    int data;//接收0 1
    int *wins;//分配窗口大小的数组

    printf("please input window size:\n");
    scanf("%d", &window);
    wins = (int *)malloc(window * sizeof(int));
    
    //FILE *fp = fopen("./01stream.txt", "r");//打开文件
    FILE *fp = fopen("./01stream_sample.txt", "r");//打开文件

    judge_file_pointer_null(fp);//文件打开是否失败

    int win_count = 0;//窗口计数

    while (!feof(fp))//扫描流中的数据
    {
        fscanf(fp, "%d", &data);
        win_count++;

        if (win_count < window)
        {
            wins[win_count] = data;
        }
        else//大于窗口时
        {
            wins[win_count % window] = data;//数组下标求模
        }
#if 1
        if (data == 1)//放入桶中
        {
            barrel[size].time_stamp = win_count;
            barrel[size].count = 1;
            size++;
            merge(barrel, size);
        }
#endif
    }
    printf("1 estimate count :%d\n", estimate(barrel));
    printf("1 accurate count :%d\n", accurate(wins));

    fclose(fp);
    free(wins);
    free(barrel);
    system("pause");
    return 0;
}

/*合并桶*/
void merge(BARR barrel[], int length)
{
    int i, j;
    int count = 0;//记录最多可以存在几个相同
    int tmp;
#if 1
    //判断有没有桶过期
    if ((barrel[length - 1].time_stamp > window))
    {
        for (i = 0; i < length - 1; i++)
        {
            if (barrel[i].time_stamp < (barrel[length - 1].time_stamp - window))//
            {
                barrel[i].count = 0;
            }
        }
    }
#endif

    //合并
    for (i = 0; i < length ; i++)
    {    
        count = 0;//记录窗口中的相同大小的桶的个数
        for (j = i + 1; j < length ; j++)
        {
            if (barrel[i].count == barrel[j].count && (barrel[i].count != 0))
            {
                count++;
                if (count == 1)
                {
                    tmp = j;//保存时间较久的桶的下标
                }
            }
        }//end of inner for
        if (count == 2)//如果窗口中有三个相同大小的桶则合并时间最久的两个桶
        {
                barrel[i].count = 0;//时间戳小的舍弃，置零
                barrel[tmp].count *= 2;
        }
    }//end of external for
}

/*DGIM估计窗口中1的数目*/
int estimate(BARR *barrel)
{
    int i;
    int estimate_count = 0;

    //打印窗口中的桶
    for (i = 0; i < size; i++)
    {
        if (barrel[i].count != 0)
        {
            printf("%d, %d\n",barrel[i].time_stamp, barrel[i].count);
        }
    }

    int count = 0;//计数第一个窗口内的值
    for (i = 0; i < size; i++)
    {
        if (barrel[i].time_stamp >= 1000 - window && barrel[i].count != 0)
        {
            count++;
            if (count == 1)
            {
                estimate_count += 0.5 * barrel[i].count;//最小时间戳的桶值*0.5
            }
            else
            {
                estimate_count +=  barrel[i].count;
            }
        }
    }

    return estimate_count;
}

/*精确计数窗口中1的数目*/
int accurate(int *wins)
{
    int accurate_count = 0;
    for (int i = 0; i < window; i++)
    {
        if (*(wins + i) == 1)
        {
            accurate_count++;
        }
    }
    return accurate_count;
}

/*判断文件指针是否为空*/
void judge_file_pointer_null(FILE *fp)
{
    if (fp == NULL)
    {
        printf("file open failed!\n");
        exit(0);
    }
}