60.大数据创建索引,并实现大文件的二分查找,迁移实现分层

• index.h

  #define  _CRT_SECURE_NO_WARNINGS
  #include<stdio.h>
  #include<stdlib.h>
  #define N 10000000
  
  struct index
  {
      int *pindex;
      int length;
  };
  
  char **g_pp;//保存指针数组
  char filepath[256];
  char sortpath[256];
  char indexpath[256];
  struct index allindex;//索引
  
  int getN();//函数声明
  void eatg(char *str);
  void eatN(char *str);

   

• index.c

  #include"index.h"
  
  char **g_pp = NULL;//保存指针数组
  char filepath[256] = { 0 };
  char sortpath[256] = { 0 };
  char indexpath[256] = { 0 };
  struct index allindex = { 0 };//索引
  
  int getN()
  {
      FILE *pf = fopen("file.txt", "r");
      if (pf == NULL)
      {
          return -1;
      }
      else
      {
          int i = 0;
          while (!feof(pf))
          {
              char str[50] = { 0 };
              fgets(str, 50, pf);//读取
              i++;
          }
          fclose(pf);
          return i;
  
      }
  }
  void eatg(char *str)
  {
      while (*str != '\0')
      {
  
          if (*str == '-')
          {
              *str = '\0';
          }
          str++;
      }
  
  }
  void eatN(char *str)
  {
      while (*str != '\0')
      {
          if (*str == '\r' || *str == '\n')
          {
              *str = '\0';
          }
  
          str++;
      }
  
  }

   

• createsort.h

  #include "index.h"
  
  void initmem();
  int com(void *p1, void*p2);
  void sort();
  void show();
  void writetofile();

   

• createsort.cp

  #include "createsort.h"
  void initmem()
  {
      g_pp = calloc(N, sizeof(char*));//分配指针数组
      FILE *pf = fopen(filepath, "r");
      if (pf == NULL)
      {
          return -1;
      }
      else
      {
          for (int i = 0; i < N; i++)
          {
              char str[50] = { 0 };
              fgets(str, 50, pf);//读取
              g_pp[i] = calloc(strlen(str) + 1, sizeof(char));//分配
              if (g_pp[i]!=NULL)
              {
                  //sprintf(g_pp[i], str);//打印进去
                  strcpy(g_pp[i], str);
                  eatN(g_pp[i]);
              }
              
              //printf("%s", g_pp[i]);//显示测试
  
  
          }
  
  
          fclose(pf);
  
  
      }
  
  
  
  
  
  
  }
  
  int com(void *p1, void*p2)
  {
      char **pp1 = p1;
      char **pp2 = p2;
  
      return strcmp(*pp1, *pp2);
  
  }
  
  void sort()
  {
      qsort(g_pp, N, sizeof(char*), com);
  
  
  }
  void show()
  {
      printf("\n此时状态\n");
      for (int i = 0; i < N; i++)
      {
          printf("\n%s", g_pp[i]);
      }
  }
  void writetofile()
  {
      FILE *pf = fopen(sortpath, "w");
      for (int i = 0; i < N; i++)
      {
          char temp[100] = { 0 };
      //    printf("\n%s", g_pp[i]);
          sprintf(temp, "%s\n", g_pp[i]);
      //    printf("\n%s", temp);
          fputs(temp, pf);
      }
  
      fclose(pf);
  }

   

• createindex.h

  #include "index.h"
  void init();
  void qucik();

   

• createindex.c

  #include "createindex.h"
  
  
  void init()
  {
      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");
      FILE *pf = fopen(sortpath, "rb");//\r\n->\n
      if (pf == NULL)
      {
          return -1;
      }
      else
      {
          int alllength = 0;
          for (int i = 0; i < N; i++)
          {
              char str[50] = { 0 };
              fgets(str, 50, pf);
              allindex.pindex[i] = alllength;//错位从0开始
  
              int length = strlen(str);
              alllength += length;
  
          }
  
          fclose(pf);
      }
      printf("\n结束读取");
  
      printf("\n开始写入");
      FILE *pfw = fopen(indexpath, "wb");//写入索引
      fwrite(allindex.pindex, sizeof(int), allindex.length, pfw);
      fclose(pfw);//关闭
      printf("\n结束写入");
  
  
      free(allindex.pindex);
  
  }
  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");
      FILE *pfw = fopen("index.txt", "rb");//写入索引
      fread(allindex.pindex, sizeof(int), allindex.length, pfw);
      fclose(pfw);//关闭
      printf("\n结束读取");
  }

   

• binsearch.h

  #include "index.h"
  void binsearch(char *searchstr);

   

• binsearch.c

  #include "binsearch.h"
  
  void binsearch(char *searchstr)
  {
      int tou = 0;
      int wei = N - 1;
      int flag = 0;
      while (tou <= wei)
      {
          int zhong = (tou + wei) / 2;
          char zhongstr[256] = { 0 };
          {
              FILE *pf1 = fopen(indexpath, "rb");
              FILE *pf2 = fopen(sortpath, "rb");
  
  
              int indexnum = 0;
              fseek(pf1, zhong*sizeof(int), SEEK_SET);
              fread(&indexnum, sizeof(int), 1, pf1);//读索引zhong到indexnum
  
              fseek(pf2, indexnum, SEEK_SET);
              fgets(zhongstr, 128, pf2);//读取
  
              fclose(pf1);
              fclose(pf2);
          }
          eatN(zhongstr);
          char pnewzhongstr[256] = { 0 };
          sprintf(pnewzhongstr, zhongstr);
          eatg(pnewzhongstr);//遇到-终止
          int res = strcmp(pnewzhongstr, searchstr);//1 0  -1
  
  
          if (res == 0)
          {
              flag = 1;
              printf("%s", zhongstr);
              break;
          }
          else if (res == 1)
          {
              wei = zhong - 1;
          }
          else
          {
              tou = zhong + 1;
          }
  
  
      }
  
  
      if (flag)
      {
          printf("\nfind");
      }
      else
      {
          printf("\n not find");
      }
  
  
  }

   

• main.c

  #include "binsearch.h"
  void initall()
  {
      strcpy(filepath, "1E~001OK.txt");
      strcpy(sortpath, "1E~001sort.txt");
      strcpy(indexpath, "1E~001index.txt");
  
  }
  
  void main()
  {
      initall();
      //初始化内存
      initmem();
      //排序
      sort();
      //写入文件
      writetofile();
  
      //初始化索引
      init();
      
      //二分查找
      while (1)
      {
          char str[256] = { 0 };
          scanf("%s", str);
          binsearch(str);
      }
      system("pause");
  
  }