seiflotfy / dlcbf

d-left Counting Bloom Filter

dlCBF

A d-left Counting Bloom Filter (dlCBF) is simple hashing-based Counting Bloom Filter (CBF) alternative based on d-left hashing that offers the same functionality as a CBF, but uses less space, generally saving a factor of two or more.

For details about the algorithm and citations please use this article for now

"Bloom Filters via d-Left Hashing and Dynamic Bit Reassignment Extended Abstract" by Flavio Bonomi, Michael Mitzenmacher, Rina Panigrahy, Sushil Singh and George Varghese

Note:

• This implmentation currently does not implement the counter feature. (coming soon)
• Insertions of items that already exists in the filter will fail (item can only exist once).
• fingerprint size set to a static 16 Bit (also to be changed soon)

Usage

dlcbf, err := NewDlcbfForCapacity(1000000)
	
// Add Item
dlcbf.Add("jon snow is alive")

count := dlcbf.GetCount() // count >> 1

member := dlcbf.IsMember("jon snow is alive") // member = true
	
// Remove Item
dlcbf.Delete("jon snow is alive") // returns true

// Remove Item again
dlcbf.Delete("jon snow is alive") // returns false
	
count := dlcbf.GetCount() // count >> 0

dlcbf's Issues

Count test is wrong

count := dlcbf.GetCount()
	if float64(count)*100/235886 < 1 {
		t.Error("Expected error < 1 percent, got", float64(count)*100/235886)
	}

In this logic,
float64(count)*100/235886 < 1 is count*100/235886 < 1,
and this is equal to count/235886 < 1/100, count < 235886/100.
This mean, the count is bigger than 2359, is all OK! This is strange.

Maybe '235886' is web2's size, I will call this Web2Size.
The condition of the if statement should be like this.

if float64(Web2Size - count) * 100 / Web2Size > 1 {
	t.Error(...)
}

or

if ( float64(count) * 100 / Web2Size ) < ( 100 - 1 ) {
	t.Error(...)
}