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 = ;//桶的个数

 int main()
 {
     //BARR barrel[276];
     BARR *barrel = (BARR *)malloc( * sizeof(BARR));
     int i = ;
     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 = ;//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 == )//放入桶中
         {
             barrel[size].time_stamp = win_count;
             barrel[size].count = ;
             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 ;
 }

 /*合并桶*/
 void merge(BARR barrel[], int length)
 {
     int i, j;
     int count = ;//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 - ].time_stamp > window))
     {
         for (i = ; i < length - ; i++)
         {
             if (barrel[i].time_stamp < (barrel[length - ].time_stamp - window))//
             {
                 barrel[i].count = ;
             }
         }
     }
 #endif

     //合并
     for (i = ; i < length ; i++)
     {
         count = ;//记录窗口中的相同大小的桶的个数
         for (j = i + ; j < length ; j++)
         {
             if (barrel[i].count == barrel[j].count && (barrel[i].count != ))
             {
                 count++;
                 if (count == )
                 {
                     tmp = j;//保存时间较久的桶的下标
                 }
             }
         }//end of inner for
         if (count == )//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 = ;//时间戳小的舍弃，置零
                 barrel[tmp].count *= ;
         }
     }//end of external for
 }

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

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

     int count = ;//计数第一个窗口内的值
     for (i = ; i < size; i++)
     {
         if (barrel[i].time_stamp >=  - window && barrel[i].count != )
         {
             count++;
             if (count == )
             {
                 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 = ;
     for (int i = ; i < window; i++)
     {
         if (*(wins + i) == )
         {
             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();
     }
 }