位图法

用位图法实现海量数据排序和重复问题

C

#include "stdafx.h"
#include <time.h>
#define BITWORD 32
#define ARRNUM 10000

int mmin = 10000000;
int mmax = 99999999;
int N = (mmax-mmin+1);   //存储的最大数字
#define  BITS_PER_WORD 32
#define WORD_OFFSET(b) ((b) / BITS_PER_WORD)
#define BIT_OFFSET(b)  ((b) % BITS_PER_WORD)
void SetBit(int *words, int n)
{
    n -= mmin;
    words[WORD_OFFSET(n)] |= (1 << BIT_OFFSET(n));
}

 void ClearBit(int *words, int n)
{
    words[WORD_OFFSET(n)] &= ~(1 << BIT_OFFSET(n));//将逻辑位置为n的二进制位置为0
}

 int GetBit(int *words, int n)
{
     int bit = words[WORD_OFFSET(n)] & (1 << BIT_OFFSET(n));//测试逻辑位置为n的二进制位是否为1 
     return bit != 0;
}

 int main( )
{
    int i;
    int j;
    int arr[ARRNUM];
    int* words = new int[1 + N/BITS_PER_WORD]; //建立的位图的大小=存储的最大数字/每个字节的位数+1
    if(words == NULL) {
        cout << "new error\n" << endl;
        exit(0);
    }
    int count = 0;
    /*
    for (i = 0; i < N; i++)
   {
       ClearBit(words, i); //将逻辑位置为n的二进制位置为0
   }
   */
    for (j = 0; j < 1 + N/BITS_PER_WORD; j++)
   {
        words[j]=0;  //直接将该int置0，应该也是可以的
   }
    srand( (unsigned)time( NULL ) );
    printf("数组大小:%d\n", ARRNUM);
    for (j = 0; j < ARRNUM; j++)
   {
        arr[j]= rand( )%N;
        arr[j] += mmin;
        printf("%d\t", arr[j]);
    }
    for (j = 0; j < ARRNUM; j++)
   {
        SetBit(words, arr[j]);
   }

    printf("排序后a为:\n");

    for (i = 0; i < N; i++) 
    {
        if (GetBit(words, i)) {
            printf("%d\t", i+mmin);
            count++;
        }
     }

    printf("总个数为:%d\n",count);
    delete[] words;
    words = NULL;
	system("pause");
    return 0;
 }

C++

#include "stdafx.h"
#include <bitset>
#include <iostream>
#include <time.h>
#include <iomanip>
#define ARRNUM 2500
#define N ARRNUM/(sizeof(int)*8)+1
using namespace std;
 int main( )
{
    int i;
    int cnt=0;
    bitset<ARRNUM> bit;
    
    cout<<"数组大小: "<<ARRNUM<<endl;    
 
    srand((unsigned)time(NULL));
   for (i = 0; i < ARRNUM; i++)
    {
        int tmp= rand( )%ARRNUM;
		bit.set(tmp,1);
        cout<<setw(5)<<tmp<<" ";
    }
	cout<<endl<<"排序后: "<<endl;

	for(i=0;i<ARRNUM;i++){		
	    if(bit[i]==1){
	       cout<<setw(5)<<i<<" ";
	       cnt++;
	    }
	}
    cout<<endl<<"不重复的数目为: "<<cnt<<endl;
    system("pause");
    return 0;
 }