2-bitmap实现海量数据查找不重复数据

#include<iostream>
#include<cstring>
using namespace std;

#define MAX 100000
unsigned array[MAX];

void init()
{
	memset(array, 0, MAX * sizeof(unsigned));
}

void set(unsigned i)
{
	array[i >> 5] |= 0x01 << (i & 31);
}

bool test(unsigned i)
{
	return array[i >> 5] & (0x01 << (i & 31));	
}

void clear(unsigned i)
{
	array[i >> 5] &= ~(0x01 << (i & 31));
}

int test_bit(unsigned i)
{
	unsigned e = i << 1;
	if(!test(e) && !test(e + 1))
	  return 0;
	else if(test(e) && !test(e + 1))
	  return 1;
	else if(!test(e) && test(e + 1))
	  return 2;
	else
	  return 3;
}

void set_bit(unsigned i)
{
	unsigned e = i << 1;
	if(test_bit(i) == 0)
	  set(e);
	else if(test_bit(i) == 1)
	{
		set(e + 1);
		clear(e);
	}
}

void clear_bit(unsigned i)
{
	unsigned e = i << 1;
	clear(e);
	clear(e + 1);
}

int main()
{
	init();
	int a[] = {4, 1, 1, 38, 999, 43, 999, 3, 48, 48};
	int i;
	for(i = 0; i < 10; i++)
	  set_bit(a[i]);
	for(i = 0; i < 1000; i++)
	{
		if(test_bit(i) == 1)
		  cout << i << endl;
	}

	return 0;
}