def test_simple_bloomfilter():
bf = BloomFilter(100)
bf.add('test')
assert 'test' in bf
bf.add('anothertest')
assert 'anothertest' in bf
assert 'notthere' not in bf
def test_bloomfilter_urls(n, r):
bf = BloomFilter(n, error_rate=r)
misses = []
for key in xrange(n):
url = generate_url(key)
if url in bf:
# false positive!
misses.append(key)
else:
bf.add(url)
observed_count = len(misses)
observed_rate = observed_count / float(n)
assert observed_rate <= r
def test_bloom_filter(capacity, max_keys):
replays = randint(0, max_keys)
bf = BloomFilter(capacity)
keys = []
errors = []
for i in range(replays):
key = 'bloom-filter-key-%s' % i
keys.append(key)
bf.add(key)
for key in keys:
assert key in bf
if 'this-should-not-%s' % key in bf:
errors.append(1)
actual = len(errors) / float(capacity)
assert actual <= 0.01
def __init__(self, capacity, error_rate=0.001, redis_host='localhost',
redis_port=6379, redis_prefix='bloomfilter'):
RedisHashing.__init__(self, redis_host, redis_port, redis_prefix)
BloomFilter.__init__(self, capacity, error_rate)
