def test_set_index_twice(sample_df):
dt = DataTable(sample_df, index='id', time_index='signup_date')
original_df = dt.df.copy()
dt_index_twice = dt.set_index('id')
assert 'index' in dt_index_twice['id'].semantic_tags
assert dt_index_twice.index == 'id'
assert dt_index_twice == dt
pd.testing.assert_frame_equal(to_pandas(original_df),
to_pandas(dt_index_twice.df))
dt_time_index_twice = dt.set_time_index('signup_date')
assert 'time_index' in dt_time_index_twice['signup_date'].semantic_tags
assert dt_time_index_twice.time_index == 'signup_date'
assert dt_time_index_twice == dt
pd.testing.assert_frame_equal(to_pandas(original_df),
to_pandas(dt_time_index_twice.df))
dt.index = 'id'
assert 'index' in dt['id'].semantic_tags
assert dt.index == 'id'
pd.testing.assert_frame_equal(to_pandas(original_df), to_pandas(dt.df))
dt.time_index = 'signup_date'
assert 'time_index' in dt['signup_date'].semantic_tags
assert dt.time_index == 'signup_date'
pd.testing.assert_frame_equal(to_pandas(original_df), to_pandas(dt.df))
def test_datatable_init_with_numpy(sample_df_pandas):
numpy_df = sample_df_pandas.to_numpy()
dt = DataTable(numpy_df, index=0)
assert set(dt.columns.keys()) == {i for i in range(len(numpy_df[0]))}
assert dt.index == 0
assert dt[0].logical_type == Categorical
assert dt[1].logical_type == NaturalLanguage
assert dt[5].logical_type == Datetime
np_ints = np.array([[1, 0], [2, 4], [3, 6], [4, 1]])
dt = DataTable(np_ints)
assert dt[0].logical_type == Integer
assert dt[1].logical_type == Integer
dt = dt.set_index(0)
assert dt.index == 0
dt = DataTable(np_ints,
time_index=0,
logical_types={
0: 'Double',
1: Datetime
},
semantic_tags={1: 'numeric_datetime'})
assert dt.time_index == 0
assert dt[0].logical_type == Double
assert dt[0].semantic_tags == {'numeric', 'time_index'}
assert dt[1].logical_type == Datetime
assert dt[1].semantic_tags == {'numeric_datetime'}
def test_set_index(sample_df):
# Test setting index with set_index()
dt = DataTable(sample_df)
new_dt = dt.set_index('id')
assert new_dt is not dt
assert new_dt.index == 'id'
assert dt.index is None
assert new_dt.columns['id'].semantic_tags == {'index'}
non_index_cols = [
col for col in new_dt.columns.values() if col.name != 'id'
]
assert all(['index' not in col.semantic_tags for col in non_index_cols])
# Test changing index with set_index()
new_dt2 = new_dt.set_index('full_name')
assert new_dt.index == 'id'
assert new_dt2.columns['full_name'].semantic_tags == {'index'}
non_index_cols = [
col for col in new_dt2.columns.values() if col.name != 'full_name'
]
assert all(['index' not in col.semantic_tags for col in non_index_cols])
# Test setting index using setter
dt = DataTable(sample_df)
dt.index = 'id'
assert dt.index == 'id'
assert 'index' in dt.columns['id'].semantic_tags
non_index_cols = [col for col in dt.columns.values() if col.name != 'id']
assert all(['index' not in col.semantic_tags for col in non_index_cols])
# Test changing index with setter
dt.index = 'full_name'
assert 'index' in dt.columns['full_name'].semantic_tags
non_index_cols = [
col for col in dt.columns.values() if col.name != 'full_name'
]
assert all(['index' not in col.semantic_tags for col in non_index_cols])
# Test changing index also changes underlying DataFrame - pandas only
if isinstance(sample_df, pd.DataFrame):
dt = DataTable(sample_df)
dt.index = 'id'
assert (dt.to_dataframe().index == [0, 1, 2, 3]).all()
assert (dt._dataframe.index == [0, 1, 2, 3]).all()
dt.index = 'full_name'
assert (
dt.to_dataframe().index == dt.to_dataframe()['full_name']).all()
assert (dt._dataframe.index == dt.to_dataframe()['full_name']).all()
def test_underlying_index(sample_df):
if dd and isinstance(sample_df, dd.DataFrame):
pytest.xfail(
'Setting underlying index is not supported with Dask input')
if ks and isinstance(sample_df, ks.DataFrame):
pytest.xfail(
'Setting underlying index is not supported with Koalas input')
unspecified_index = pd.RangeIndex
specified_index = pd.Index
dt = DataTable(sample_df.copy(), index='id')
assert dt._dataframe.index.name is None
assert (dt._dataframe.index == [0, 1, 2, 3]).all()
assert type(dt._dataframe.index) == specified_index
assert type(dt.to_dataframe().index) == specified_index
dt = DataTable(sample_df.copy())
dt = dt.set_index('full_name')
assert (dt._dataframe.index == dt.to_dataframe()['full_name']).all()
assert dt._dataframe.index.name is None
assert type(dt._dataframe.index) == specified_index
assert type(dt.to_dataframe().index) == specified_index
dt.index = 'id'
assert (dt._dataframe.index == [0, 1, 2, 3]).all()
assert dt._dataframe.index.name is None
assert type(dt._dataframe.index) == specified_index
assert type(dt.to_dataframe().index) == specified_index
# test removing index removes the dataframe's index
dt.index = None
assert type(dt._dataframe.index) == unspecified_index
assert type(dt.to_dataframe().index) == unspecified_index
dt = DataTable(sample_df.copy(), index='made_index', make_index=True)
assert (dt._dataframe.index == [0, 1, 2, 3]).all()
assert dt._dataframe.index.name is None
assert type(dt._dataframe.index) == specified_index
assert type(dt.to_dataframe().index) == specified_index
dt_dropped = dt.drop('made_index')
assert 'made_index' not in dt_dropped.columns
assert 'made_index' not in dt_dropped._dataframe.columns
assert type(dt_dropped._dataframe.index) == unspecified_index
assert type(dt_dropped.to_dataframe().index) == unspecified_index
