100W个随机数，取值范围[0,10W)，要求统计出重复数目前1000的数值。内存占用不得超过60M。

100W个随机数，取值范围[0,10W)，要求统计出重复数目前1000的数值。内存占用不得超过60M。

原始数据保存在文本文档，输出数据同样要求保存到文本。

首先是读取原始数据。我使用内存映射将整个文件载入到内存

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

HANDLE hFile = CreateFile(buf, GENERIC_READ|GENERIC_WRITE, 0, NULL, OPEN_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL); if (hFile == INVALID_HANDLE_VALUE){     return 1; }   HANDLE hFileMapping = CreateFileMapping(hFile, NULL, PAGE_READWRITE, 0, 0, NULL); if (hFileMapping == NULL){     int err = GetLastError();     printf("err code %d", err);     return 1; }   LPTSTR lpMapAddr = (LPTSTR)MapViewOfFile(hFileMapping, PAGE_READWRITE, 0, 0, 0); if (lpMapAddr == NULL){     int err = GetLastError();     printf("err code %d", err);     return 1; }

调用atoi将字符串转换到数值。

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

void statistics(void * data){     char * buff = (char*)data;     char tmp[32];       memset(nstatistics, 0, 100000);     int i = 0;     while (buff[i] != '\0'){         int j = 0;         while (buff[i] != '\r'){             if (buff[i] == '\0'){                 return;             }                           i++;             j++;         }           memset(tmp, 0, 32);         memcpy(tmp, &buff[i-j], j);           if (buff[++i] != '\n'){             assert(0);         }           unsigned int n = atoi(tmp);         nstatistics[n]++;     } }

填充前1000个统计数据到数组并排序，然后遍历剩余数值，将重复次数更多的数值插入到排序好的数组void count(void * data){ memset(ncountdata, 0, sizeof(countdata)* 1000);

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

for (int i = 0; i < 1000; i++){     ncountdata[i].count = i;     ncountdata[i].num = nstatistics[i]; } qsort(ncountdata, 1000, sizeof(countdata), cmp);   for (int i = 1000; i < 100000; i++){     for (int j = 0; j < 1000; j++){         //if (nstatistics[i] > ncountdata[j].num && nstatistics[i] <= ncountdata[j+1].num){         //  ncountdata[j].count = i;         //  ncountdata[j].num = nstatistics[i];         //  break;         //}           if (nstatistics[i] > ncountdata[j].num){             if (nstatistics[i] <= ncountdata[j+1].num){                 ncountdata[j].count = i;                 ncountdata[j].num = nstatistics[i];                 break;             }         }else{             break;         }     } }

　　

修改一个显而易见的可优化之处，总算比直接qsort快了。