class TestCardinalityEstimation(unittest.TestCase):
def setUp(self):
self.hll = HyperLogLog(5)
def test_small_range_correction_all_registers_set_to_zero(self):
self.assertEqual(self.hll.cardinality(), 0.0)
def test_small_range_correction_not_all_registers_set_to_zero(self):
self.hll.set_register(0, 1)
c = self.hll.cardinality()
correction= 1.46571806761 <= c and c <= 1.46571806762
self.assertTrue(correction)
def test_medium_range_no_correction(self):
for i in range(32):
self.hll.set_register(i, 2)
c = self.hll.cardinality()
no_correction = 89.216 <= c and c <= 89.217
self.assertTrue(no_correction)
@unittest.skip("correction value needs to be re-computed")
def test_large_range_correction(self):
hll = HyperLogLog(16)
for i in range(hll.size() - 1):
hll.set_register(i, 16)
c = hll.cardinality()
correction = 7916284520 <= c and c <= 7916284521
self.assertTrue(correction)
class TestCardinalityEstimation(unittest.TestCase):
def setUp(self):
self.hll = HyperLogLog(5)
def test_small_range_correction_all_registers_set_to_zero(self):
self.assertEqual(self.hll.cardinality(), 0.0)
def test_small_range_correction_not_all_registers_set_to_zero(self):
self.hll.set_register(0, 1)
c = self.hll.cardinality()
correction = 1.46571806761 <= c and c <= 1.46571806762
self.assertTrue(correction)
def test_medium_range_no_correction(self):
for i in range(32):
self.hll.set_register(i, 2)
c = self.hll.cardinality()
no_correction = 89.216 <= c and c <= 89.217
self.assertTrue(no_correction)
@unittest.skip("correction value needs to be re-computed")
def test_large_range_correction(self):
hll = HyperLogLog(16)
for i in range(hll.size() - 1):
hll.set_register(i, 16)
c = hll.cardinality()
correction = 7916284520 <= c and c <= 7916284521
self.assertTrue(correction)
def test_large_range_correction(self):
hll = HyperLogLog(16)
for i in range(hll.size() - 1):
hll.set_register(i, 16)
c = hll.cardinality()
correction = 7916284520 <= c and c <= 7916284521
self.assertTrue(correction)
def test_large_range_correction(self):
hll = HyperLogLog(16)
for i in range(hll.size() - 1):
hll.set_register(i, 16)
c = hll.cardinality()
correction = 7916284520 <= c and c <= 7916284521
self.assertTrue(correction)
def test_merge(self):
expected = bytearray(4)
expected[0] = 1
expected[3] = 1
hll = HyperLogLog(2)
hll2 = HyperLogLog(2)
hll.set_register(0, 1)
hll2.set_register(3, 1)
hll.merge(hll2)
self.assertEqual(hll.registers(), expected)
def test_merge(self):
expected = bytearray(4)
expected[0] = 1
expected[3] = 1
hll = HyperLogLog(2)
hll2 = HyperLogLog(2)
hll.set_register(0, 1)
hll2.set_register(3, 1)
hll.merge(hll2)
self.assertEqual(hll.registers(), expected)
class TestRegisterFunctions(unittest.TestCase):
def setUp(self):
self.k = 5
self.hll = HyperLogLog(5)
def test_set_last_register(self):
self.hll.set_register(self.k - 1, 1)
self.assertTrue(self.hll.registers()[self.k - 1] == 1)
def test_set_first_register(self):
self.hll.set_register(0, 1)
self.assertTrue(self.hll.registers()[0] == 1)
def test_set_register_with_negative_value_fails(self):
with self.assertRaises(ValueError):
self.hll.set_register(0, -1)
def test_set_register_with_greater_than_max_rank_fails(self):
with self.assertRaises(ValueError):
self.hll.set_register(0, 33)
def test_set_register_with_index_out_of_bounds(self):
with self.assertRaises(IndexError):
self.hll.set_register(32, 1)
def test_set_register_with_negative_index_fails(self):
with self.assertRaises(ValueError):
self.hll.set_register(0, -1)
def test_bytesarray_has_correct_values(self):
expected = bytearray(randint(0, 16) for x in range(32))
for i in range(32):
self.hll.set_register(i, expected[i])
registers = self.hll.registers()
self.assertEqual(expected, registers)
def test_registers_returns_bytesarray(self):
self.assertTrue(type(self.hll.registers()) is bytearray)
def test_bytesarray_has_correct_length(self):
self.assertTrue(len(self.hll.registers()) == pow(2, self.k))
def test_set_registers(self):
expected = bytearray(randint(0, 16) for x in range(32))
self.hll.set_registers(expected)
registers = self.hll.registers()
self.assertEqual(expected, registers)
class TestRegisterFunctions(unittest.TestCase):
def setUp(self):
self.k = 5
self.hll = HyperLogLog(5)
def test_set_last_register(self):
self.hll.set_register(self.k - 1, 1)
self.assertTrue(self.hll.registers()[self.k - 1] == 1)
def test_set_first_register(self):
self.hll.set_register(0, 1)
self.assertTrue(self.hll.registers()[0] == 1)
def test_set_register_with_negative_value_fails(self):
with self.assertRaises(ValueError):
self.hll.set_register(0, -1)
def test_set_register_with_greater_than_max_rank_fails(self):
with self.assertRaises(ValueError):
self.hll.set_register(0, 33)
def test_set_register_with_index_out_of_bounds(self):
with self.assertRaises(IndexError):
self.hll.set_register(32, 1)
def test_set_register_with_negative_index_fails(self):
with self.assertRaises(ValueError):
self.hll.set_register(0, -1)
def test_bytesarray_has_correct_values(self):
expected = bytearray(32)
for i in range(31):
expected[i] = randint(0, 16)
for i in range(31):
self.hll.set_register(i, expected[i])
registers = self.hll.registers()
for i in range(31):
self.assertEqual(expected[i], registers[i])
def test_registers_returns_bytesarray(self):
self.assertTrue(type(self.hll.registers()) is bytearray)
def test_bytesarray_has_correct_length(self):
self.assertTrue(len(self.hll.registers()) == pow(2, self.k))
