The bloomfilter's discuss from hugh2632

bloomfilter's Issues

结果不准确

基于内存的模式添加500万数据，拿任何数据，碰撞得出的结果都是存在。

func TestMemoryFilter(t *testing.T) {
	bf := bloomfilter.NewMemoryFilter(make([]byte, 10240), bloomfilter.DefaultHash[:5]...)
	bf.Clear()

	//file, err := os.Open("./1.txt")
	file, err := os.Open("/Users/pandy/go/src/go-sdk/data/domain.txt")
	defer file.Close()
	if err != nil {
		fmt.Printf("Error: %s\n", err)
		return
	}

	reader := bufio.NewReader(file)
	countWrite := 0
	for {
		a, _, c := reader.ReadLine()
		if c == io.EOF {
			break
		}

		bf.Push(a)
		countWrite++
	}

	bf.Write()
	t.Log("write num:", countWrite)

	// 入数据之后 进行碰撞检测
	countRead := 0
	//file2, err := os.Open("/Users/pandy/go/src/go-sdk/data/domain_read.txt")
	file2, err := os.Open("./2.txt")
	defer file.Close()
	if err != nil {
		fmt.Printf("Error: %s\n", err)
		return
	}

	reader2 := bufio.NewReader(file2)
	for {
		a, _, c := reader2.ReadLine()
		if c == io.EOF {
			break
		}

		if bf.Exists(a) {
			countRead++
		} else {
			t.Log("not hash", string(a))
		}
	}

	t.Log("hashed num:", countRead)

	bf.Clear()
	bf.Close()
}

Recommend Projects