def test_join_with_index(self):
"""
Test Join operation with index
"""
R = Table.inputfromfile('test_csv/real_sales1.csv')
S = Table.inputfromfile('test_csv/real_sales1.csv')
start = time.time()
R1 = Table.join(R, S, 'R', 'S', 'R.pricerange = S.pricerange')
end = time.time()
self.assertIsNotNone(R1.table)
for row in R1.table:
self.assertEqual(row['R_pricerange'], row['S_pricerange'])
without_index = end - start
Table.Hash(R, 'pricerange')
Table.Hash(S, 'pricerange')
start = time.time()
R1 = Table.join(R, S, 'R', 'S', 'R.pricerange = S.pricerange')
end = time.time()
with_index = end - start
self.assertIsNotNone(R1.table)
for row in R1.table:
self.assertEqual(row['R_pricerange'], row['S_pricerange'])
self.assertLess(with_index, without_index*0.9)
start = time.time()
R1 = Table.join(R, S, 'R', 'S', 'S.pricerange = R.pricerange')
end = time.time()
with_index = end - start
self.assertIsNotNone(R1.table)
for row in R1.table:
self.assertEqual(row['R_pricerange'], row['S_pricerange'])
self.assertLess(with_index, without_index*0.9)
def test_hash(self):
"""Test building hash index on column: saleid"""
self.R = Table.inputfromfile('test_csv/sales1.csv')
Table.Hash(self.R, 'saleid')
self.assertIsNotNone(self.R.index)
self.assertIsNotNone(self.R.index['saleid'])
test_index = self.R.index['saleid']
self.assertEqual([0], test_index[36])
self.assertEqual([1], test_index[784])
self.assertEqual([2], test_index[801])
self.assertEqual([3], test_index[905])
def test_hash_with_pricerange(self):
"""Test building hash index on column: pricerange"""
R = Table.inputfromfile('test_csv/sales1.csv')
Table.Hash(R, 'pricerange')
self.assertIsNotNone(R.index)
self.assertIsNotNone(R.index['pricerange'])
test_index = R.index['pricerange']
self.assertEqual([0, 1], test_index.get('moderate'))
self.assertEqual([2, 3, 5, 6, 7, 10, 11], test_index.get('outrageous'))
self.assertEqual([4], test_index.get('supercheap'))
self.assertEqual([9], test_index.get('cheap'))
self.assertEqual([8], test_index.get('expensive'))
def test_select_with_index_flip(self):
"""Test select with hash index on column pricerage: 'cheap' = pricerange """
R = Table.inputfromfile('test_csv/real_sales1.csv')
condition = "'cheap' = pricerange"
start = time.time()
for i in range(10):
R1 = Table.select(R, condition)
end = time.time()
without_index = end - start
Table.Hash(R, 'pricerange')
start = time.time()
for i in range(10):
R1 = Table.select(R, condition)
end = time.time()
with_index = end - start
for row in R1.table:
self.assertEqual(row['pricerange'], 'cheap')
self.assertLess(with_index, without_index/5)