def test_cuckoo_filter_check(self):
"""test checking if element in cuckoo filter"""
cko = CuckooFilter()
cko.add("this is a test")
cko.add("this is another test")
cko.add("this is yet another test")
self.assertEqual(cko.check("this is a test"), True)
self.assertEqual(cko.check("this is another test"), True)
self.assertEqual(cko.check("this is yet another test"), True)
self.assertEqual(cko.check("this is not another test"), False)
self.assertEqual(cko.check("this is not a test"), False)
def test_cuckoo_filter_check(self):
''' test checking if element in cuckoo filter '''
cko = CuckooFilter()
cko.add('this is a test')
cko.add('this is another test')
cko.add('this is yet another test')
self.assertEqual(cko.check('this is a test'), True)
self.assertEqual(cko.check('this is another test'), True)
self.assertEqual(cko.check('this is yet another test'), True)
self.assertEqual(cko.check('this is not another test'), False)
self.assertEqual(cko.check('this is not a test'), False)
def test_cuckoo_filter_remove(self):
"""test removing from the cuckoo filter"""
cko = CuckooFilter()
cko.add("this is a test")
self.assertEqual(cko.elements_added, 1)
cko.add("this is another test")
self.assertEqual(cko.elements_added, 2)
cko.add("this is yet another test")
self.assertEqual(cko.elements_added, 3)
res = cko.remove("this is a test")
self.assertTrue(res)
self.assertEqual(cko.elements_added, 2)
self.assertFalse(cko.check("this is a test"))
self.assertTrue(cko.check("this is another test"))
self.assertTrue(cko.check("this is yet another test"))
def test_cuckoo_filter_remove_miss(self):
''' test removing from the cuckoo filter when not present '''
cko = CuckooFilter()
cko.add('this is a test')
self.assertEqual(cko.elements_added, 1)
cko.add('this is another test')
self.assertEqual(cko.elements_added, 2)
cko.add('this is yet another test')
self.assertEqual(cko.elements_added, 3)
res = cko.remove('this is still a test')
self.assertFalse(res)
self.assertEqual(cko.elements_added, 3)
self.assertTrue(cko.check('this is a test'))
self.assertTrue(cko.check('this is another test'))
self.assertTrue(cko.check('this is yet another test'))
def test_cuckoo_filter_auto_expand(self):
"""test inserting until cuckoo filter is full"""
cko = CuckooFilter(capacity=100, bucket_size=2, max_swaps=100)
for i in range(375): # this would fail if it doesn't expand
cko.add(str(i))
self.assertEqual(400, cko.capacity)
self.assertEqual(375, cko.elements_added)
for i in range(375):
self.assertTrue(cko.check(str(i)))
def test_cuckoo_filter_expand_els(self):
"""test out the expansion of the cuckoo filter"""
cko = CuckooFilter()
for i in range(200):
cko.add(str(i))
cko.expand()
for i in range(200):
self.assertTrue(cko.check(str(i)))
self.assertEqual(20000, cko.capacity)
def test_cuckoo_filter_load(self):
''' test loading a saved cuckoo filter '''
filename = './test.cko'
md5sum = '49b947ddf364d27934570a6b33076b93'
cko = CuckooFilter()
for i in range(1000):
cko.add(str(i))
cko.export(filename)
md5_out = calc_file_md5(filename)
self.assertEqual(md5sum, md5_out)
ckf = CuckooFilter(filepath=filename)
for i in range(1000):
self.assertTrue(ckf.check(str(i)))
self.assertEqual(10000, ckf.capacity)
self.assertEqual(4, ckf.bucket_size)
self.assertEqual(500, ckf.max_swaps)
self.assertEqual(0.025, ckf.load_factor())
os.remove(filename)
def test_cuckoo_filter_load(self):
"""test loading a saved cuckoo filter"""
md5sum = "1371760d4ee9ccbe83e0144919750140"
with NamedTemporaryFile(dir=os.getcwd(),
suffix=".cko",
delete=DELETE_TEMP_FILES) as fobj:
cko = CuckooFilter()
for i in range(1000):
cko.add(str(i))
cko.export(fobj.name)
md5_out = calc_file_md5(fobj.name)
self.assertEqual(md5sum, md5_out)
ckf = CuckooFilter(filepath=fobj.name)
for i in range(1000):
self.assertTrue(ckf.check(str(i)))
self.assertEqual(10000, ckf.capacity)
self.assertEqual(4, ckf.bucket_size)
self.assertEqual(500, ckf.max_swaps)
self.assertEqual(0.025, ckf.load_factor())
from probables import (CuckooFilter)
from probables import (CountingCuckooFilter)
cko = CuckooFilter(capacity=100, max_swaps=10)
cko.add('google.com')
cko.check('facebook.com') # should return False
cko.check('google.com') # should return True
cck = CountingCuckooFilter(capacity=100, max_swaps=10)
cck.add("google")
cck.add("google")
cck.add("google")
print(cck.check("google"))
cck.remove("google")
print(cck.check("google"))
