def test_bloomfilter(self):
m, k = Bloomfilter.calculate_parameters(100000, 0.01)
b = Bloomfilter(m, k)
keys = [random.randint(0, 80000) for i in xrange(40000)]
b.add(keys)
self.assertTrue(keys[0] in b)
self.assertTrue(all(b.match(keys)))
self.assertTrue(
len(filter(None, b.match(xrange(80000, 100000)))) < 100000 * 0.01)
def add(self, value, position):
if self.index_type == BITMAP_INDEX and len(self.index) > MAX_BIT_INDEX_SIZE:
index = self.index
self.index_type = BLOOMFILTER_INDEX
self.index = Bloomfilter(BF_SIZE, BF_HASHES)
for v in index:
for pos in index[v].positions():
self.index.add([(v, pos)])
if self.index_type == BITMAP_INDEX:
if value not in self.index:
self.index[value] = BitIndex()
index = self.index[value]
index.set(position)
if self.index_type == BLOOMFILTER_INDEX:
self.index.add([(value, position)])
assert (value, position) in self.index
