59.排序好的大数据创建索引文件,并实现大文件的二分查找,根据索引百万数据秒读数据...

• 创建索引

  //创建索引
  struct index
  {
      //保存每行偏移的位置
      int *pindex;
      //文件的总长度
      int length;
  }allindex;//索引

   

• 初始化索引数组并把索引写入到文件

  //初始化索引数组,并把索引写入到文件
  void init(char *path)
  {
      printf("\n索引数组开始分配");
      allindex.length = N;
      //分配内存
      allindex.pindex = calloc(N, sizeof(int));
      printf("\n索引数组完成分配");
  
      printf("\n开始读取");
      //二进制读取文件 避免/r/n读取成/n
      FILE *pf = fopen("filesort.txt", "rb");
      if (pf == NULL)
      {
          return -1;
      }
      else
      {
          int alllength = 0;
          for (int i = 0; i < N; i++)
          {
              char str[50] = { 0 };
              fgets(str, 50, pf);
              //每一个首地址的偏移
              allindex.pindex[i] = alllength;
  
              int length = strlen(str);
              alllength += length;
          }
          fclose(pf);
      }
      printf("\n结束读取");
  
      printf("\n开始写入");
      //二进制方式打开文件,并写入索引
      FILE *pfw = fopen("index.txt", "wb");
      //写入
      fwrite(allindex.pindex, sizeof(int), allindex.length, pfw);
      //关闭文件
      fclose(pfw);
      printf("\n结束写入");
  
      //释放内存
      free(allindex.pindex);
  }

   

• 从文件中读取索引到索引数组中

  //从文件中读取索引
  void qucik()
  {
      printf("\n索引数组开始分配");
      allindex.length = N;
      allindex.pindex = calloc(N, sizeof(int));//分配内存
      printf("\n索引数组完成分配");
  
  
      printf("\n开始读取");
      //以二进制读的方式读取索引
      FILE *pfw = fopen("index.txt", "rb");
      //读取
      fread(allindex.pindex, sizeof(int), allindex.length, pfw);
      //关闭文件
      fclose(pfw);
      printf("\n结束读取");
  }

   

• 测试函数

   1　　 FILE *pf1 = fopen("index.txt", "rb");
      FILE *pf2 = fopen("filesort.txt", "rb");
      while (1)
      {
          printf("\n请输入要读取的行数");
          int num = 0;
          scanf("%d", &num);
  
          int indexnum = 0;
          fseek(pf1, num*sizeof(int), SEEK_SET);
          fread(&indexnum, sizeof(int), 1, pf1);//读索引到indexnum
  
          fseek(pf2, indexnum, SEEK_SET);
          char str[128] = { 0 };
          fgets(str, 128, pf2);//读取
          printf("\n%s", str);
  
      }
      fclose(pf1);
      fclose(pf2);

   

   

• 根据索引文件对已经排序好的文件进行二分查找

  void binsearch(char *searchstr)
  {
      //头部
      int tou = 0;
      //尾部
      int wei = N - 1;
      //是否找到的标识
      int flag = 0;
      //如果头小于尾
      while (tou <= wei)
      {
          //获取中部
          int zhong = (tou + wei) / 2;
          //读取中部索引的内容
          char zhongstr[256] = { 0 };
          {
              //打开索引文件
              FILE *pf1 = fopen("index.txt", "rb");
              //打开排序好的文件
              FILE *pf2 = fopen("filesort.txt", "rb");
  
              //读zhong对应的地址存到indexnum中
              int indexnum = 0;
              fseek(pf1, zhong * sizeof(int), SEEK_SET);
              fread(&indexnum, sizeof(int), 1, pf1);
  
              //根据读取的位置读取文件到zhongstr中
              fseek(pf2, indexnum, SEEK_SET);
              fgets(zhongstr, 128, pf2);
  
              fclose(pf1);
              fclose(pf2);
          }
          //消除'\r或者\n'
          eatN(zhongstr);
          char pnewzhongstr[256] = { 0 };
          sprintf(pnewzhongstr, zhongstr);
          //进行处理,遇到-终止
          eatg(pnewzhongstr);
          //比较是否找到
          int res = strcmp(pnewzhongstr, searchstr);//1 0  -1
          if (res == 0)
          {
              flag = 1;
              printf("%s", zhongstr);
              break;
          }
          //如果中比searchstr要大
          else if (res == 1)
          {
              wei = zhong - 1;
          }
          //如果中比searchstr小
          else
          {
              tou = zhong + 1;
          }
  
  
      }
      //判断是否找到
      if (flag)
      {
          printf("\nfind");
      } 
      else
      {
          printf("\n not find");
      }
  }

   

• 遇到'-'结束

  //遇到'-'结束
  void eatg(char *str)
  {
      while (*str!='\0')
      {
  
          if (*str=='-')
          {
              *str = '\0';
          }
          str++;
      }
  
  }

   

• 测试函数

  void main()
  {
  
  
      char str[256] = { 0 };
      scanf("%s", str);
      binsearch(str);
  
  
  
      system("pause");
  }