import sys
from math import log, ceil
import mmh3

class BloomFilter:

	def __init__(self, n, p=0.01):
		''' n is size of unique lines.
		    p is false positive rate
		'''
		self.size = int(ceil((-n*log(p))/ (log(2)**2)))
		self.hash_count = int(ceil((self.size/n) * log(2)))
		self.filter = [False for _ in range(self.size)]

	def add(self, item):
		# Adds data to bloom filter
		hash_keys = []
		for seed in range(self.hash_count):
			key = mmh3.hash(item, seed) % self.size
			hash_keys.append(key)

		if all([self.filter[i] for i in hash_keys]):
			return False
		# If above doesn't return it means it can be added
		for key in hash_keys:
			self.filter[key] = True
		return True