def test_update_item(self):
table = Table("test", ['a'])
table['a'] = dict(b='B')
table['a'] = dict(b='C')
self.assertTrue('a' in table)
self.assertEqual(table['a'], {'a': 'a', 'b': 'C'})
self.assertEqual(1, len(table))
self.assertEqual([('a', 'C')], list(table.to_tuples()))
def test_insert_item(self):
table = Table("test", ['a'])
table['A'] = dict(b='B')
table['a'] = dict(b='B')
self.assertTrue('A' in table)
self.assertTrue('a' in table)
self.assertEqual(table['a'], {'a': 'a', 'b': 'B'})
self.assertEqual(table['A'], {'a': 'A', 'b': 'B'})
self.assertEqual(2, len(table))
self.assertEqual([('A', 'B'), ('a', 'B')], list(table.to_tuples()))
def test_pandas_huge_pack_item(self):
table = Table("test", ['simulation_id', 'learner_id'])
for i in range(2):
table[(i, 2)] = dict(C=5,
A=5,
N=1,
_packed=dict(reward=[2] * 9000))
time = min(timeit.repeat(lambda: table.to_pandas(), repeat=6,
number=1))
#best time on my laptop was 0.15
self.assertLess(time, 1)
def test_pandas_with_dict_column(self):
import pandas as pd #type: ignore
import pandas.testing #type: ignore
table = Table("test", ['a'])
table['A'] = dict(b='B', c={'z': 10}, d='d')
table['B'] = dict(e='E')
expected_df = pd.DataFrame(
[dict(a='A', b='B', c={'z': 10}, d='d'),
dict(a='B', e='E')])
actual_df = table.to_pandas()
pandas.testing.assert_frame_equal(expected_df, actual_df)
def test_tuples_with_array_column(self):
table = Table("test", ['a'])
table['A'] = dict(b='B', c=[1, 2], d='d')
table['B'] = dict(e='E')
expected_tuples = [('A', 'B', [1, 2], 'd', float('nan')),
('B', float('nan'), float('nan'), float('nan'), 'E')
]
actual_tuples = table.to_tuples()
for expected_tuple, actual_tuple in zip(expected_tuples,
actual_tuples):
self.assertTrue(
all([
v1 == v2 or (math.isnan(v1) and math.isnan(v2))
for v1, v2 in zip(expected_tuple, actual_tuple)
]))
def test_insert_two_pack_item(self):
table = Table("test", ['a'])
table['A'] = dict(c=1, _packed=dict(b=['B', 'b'], d=['D', 'd']))
self.assertTrue('A' in table)
self.assertEqual(table['A'], {
'a': 'A',
'index': [1, 2],
'b': ['B', 'b'],
'c': 1,
'd': ['D', 'd']
})
self.assertEqual(2, len(table))
self.assertEqual([('A', 1, 1, 'B', 'D'), ('A', 2, 1, 'b', 'd')],
list(table.to_tuples()))
def test_pandas_two_pack_item(self):
import pandas as pd
import pandas.testing
table = Table("test", ['a'])
table['A'] = dict(c=1, _packed=dict(b=['B', 'b'], d=['D', 'd']))
table['B'] = dict(e='E')
expected_df = pd.DataFrame([
dict(a='A', index=1, c=1, b='B', d='D'),
dict(a='A', index=2, c=1, b='b', d='d'),
dict(a='B', index=1, e='E')
])
actual_df = table.to_pandas()
pandas.testing.assert_frame_equal(expected_df,
actual_df,
check_dtype=False)
def test_missing_columns(self):
table = Table("test", ['a'])
table['A'] = dict(b='B')
table['a'] = dict(c='C')
self.assertTrue('A' in table)
self.assertTrue('a' in table)
self.assertEqual(table['A'], {'a': 'A', 'b': 'B'})
self.assertEqual(table['a'], {'a': 'a', 'c': 'C'})
self.assertEqual(2, len(table))
expected = [('A', 'B', float('nan')), ('a', float('nan'), 'C')]
actual = table.to_tuples()
for tuple1, tuple2 in zip(expected, actual):
for val1, val2 in zip(tuple1, tuple2):
if isinstance(val1, float) and math.isnan(val1):
self.assertTrue(math.isnan(val2))
else:
self.assertEqual(val1, val2)
def test_pandas_with_packed_array_column(self):
import pandas as pd #type: ignore
import pandas.testing #type: ignore
table = Table("test", ['a'])
table['A'] = dict(b=1.,
c=[1, 2],
d='d',
_packed={'z': [[1, 2], [3, 4]]})
table['B'] = dict(b=2., e='E')
expected_df = pd.DataFrame([
dict(a='A', index=1, b=1., c=[1, 2], d='d', z=[1, 2]),
dict(a='A', index=2, b=1., c=[1, 2], d='d', z=[3, 4]),
dict(a='B', index=1, b=2., e='E')
])
actual_df = table.to_pandas()
pandas.testing.assert_frame_equal(expected_df,
actual_df,
check_dtype=False)
def test_unequal_pack_exception(self):
with self.assertRaises(Exception):
table = Table("test", ['a'])
table['A'] = dict(c=1,
_packed=dict(b=['B', 'b'], d=['D', 'd', 'e']))