def testexists(self):
bf = bloomfilter()
bf.setData(100000000)
bf.setData(100005000)
bf.setData(100004000)
bf.setData(100002000)
self.assertEqual(True,bf.exists(100000000))
self.assertEqual(True,bf.exists(100002000))
self.assertEqual(True,bf.exists(100004000))
self.assertEqual(True,bf.exists(100005000))
self.assertEqual(False,bf.exists(100005001))
def add(self, element):
"""
Add a new element to the set. Return False if element is already in the set.
Otherwise return True and the element is added in the filter.
"""
element = str(element)
if len(self.filter_list) == 0:
# initialize the filter list
one_filter = bloomfilter(capacity = self.total_capacity, error = self.error*(1. - self.r))
one_filter.add(element)
self.filter_list.append(one_filter)
self.n_filters = self.n_filters + 1
self.cnt = self.cnt + 1
self.total_size = self.total_size + one_filter.size()
return True
elif self.check(element) == True:
print("The element to be added:" + str(element) + " already exists.")
return False
else:
num_filters = len(self.filter_list)
last_filter = self.filter_list[num_filters - 1]
# create a new filter
if last_filter.count() == last_filter.capacity():
one_filter = bloomfilter(capacity = last_filter.capacity()*self.s, error = last_filter.error*self.r)
one_filter.add(element)
self.filter_list.append(one_filter)
self.n_filters = self.n_filters + 1
self.cnt = self.cnt + 1
self.total_capacity = self.total_capacity + one_filter.capacity()
self.total_size = self.total_size + one_filter.size()
return True
else:
last_filter.add(element)
self.cnt = self.cnt + 1
return True
def __init__(self, config):
super(dupcheck, self).__init__()
self.config = config
self.bf = bloomfilter(self.config.bf_config())
def testgetHashList(self):
bf = bloomfilter()
l = bf.getHashList(200000000)
self.assertEqual( 10,len(l))
self.assertEqual([98, 128, 162, 200, 242, 288, 338, 392, 450, 512], l)
