Simple Bloom Filter Implementation

bloom.py

# Implementation of a simple Bloom Filter
# @author Matt McCollow <[email protected]>

import hashlib

class Filter():
	
	def __init__(self, debug=False):
		self.bits = 0x0
		self.mask = 0xFFFF
		self.elements = []
		self.debug = debug

	def hash(self, string):
		md5 = hashlib.md5(string).hexdigest()
		sha1 = hashlib.sha1(string).hexdigest()
		return (md5[:1], sha1[-1:])

	def add(self, string):
		hashes = self.hash(string)
		self.bits = self.bits | (1 << int(hashes[0], 16))
		self.bits = self.bits | (1 << int(hashes[1], 16))
		self.elements.append(string)
		if self.debug:
			print bin(self.bits)

	def contains(self, string):
		hashes = self.hash(string)
		mask1 = 1 << int(hashes[0], 16)
		mask2 = 1 << int(hashes[1], 16)
		if (((mask1 | mask2) & self.bits) == (mask1 | mask2)):
			return True
		return False

	def items(self):
		print self.elements