def test_category(self):
result = render(
pd.DataFrame({"A": [np.nan, "", "c"]}, dtype="category"),
P(["A"]), # drops both '' and np.nan
)
expected = pd.DataFrame({"A": ["c"]}, dtype="category")
assert_frame_equal(result, expected)
def test_category(self):
result = render(
pd.DataFrame({'A': [np.nan, '', 'c']}, dtype='category'),
P(['A']) # drops both '' and np.nan
)
expected = pd.DataFrame({'A': ['c']}, dtype='category')
assert_frame_equal(result, expected)
def test_multiple_columns_drops_union(self):
result = render(
pd.DataFrame({'A': [1, np.nan, 3.3], 'B': [np.nan, 'b', 'c']}),
P(['A', 'B'])
)
expected = pd.DataFrame({'A': [3.3], 'B': ['c']})
assert_frame_equal(result, expected)
def test_string(self):
result = render(
pd.DataFrame({'A': [1, 2, 3, 4], 'B': [np.nan, None, '', 'c']}),
P(['B']) # drops both '' and np.nan
)
expected = pd.DataFrame({'A': [4], 'B': ['c']})
assert_frame_equal(result, expected)
def test_numeric(self):
result = render(
pd.DataFrame({'A': [1, np.nan, 3.3], 'B': [np.nan, 'b', 'c']}),
P(['A'])
)
expected = pd.DataFrame({'A': [1, 3.3], 'B': [np.nan, 'c']})
assert_frame_equal(result, expected)
def test_datetime(self):
result = render(
pd.DataFrame({"A": [pd.NaT, "2019-05-01"]},
dtype="datetime64[ns]"),
P(["A"]),
)
expected = pd.DataFrame({"A": ["2019-05-01"]}, dtype="datetime64[ns]")
assert_frame_equal(result, expected)
def test_numeric(self):
result = render(
pd.DataFrame({
"A": [1, np.nan, 3.3],
"B": [np.nan, "b", "c"]
}), P(["A"]))
expected = pd.DataFrame({"A": [1, 3.3], "B": [np.nan, "c"]})
assert_frame_equal(result, expected)
def test_no_colnames(self):
# No colnames -> do nothing
result = render(
pd.DataFrame({'A': ['', np.nan, 'x']}),
P([])
)
expected = pd.DataFrame({'A': ['', np.nan, 'x']})
assert_frame_equal(result, expected)
def test_datetime(self):
result = render(
pd.DataFrame({'A': [pd.NaT, '2019-05-01']},
dtype='datetime64[ns]'),
P(['A'])
)
expected = pd.DataFrame({'A': ['2019-05-01']}, dtype='datetime64[ns]')
assert_frame_equal(result, expected)
def test_colnames_comma_separated(self):
result = pd.DataFrame({
'A': ['', 'b', 'c'],
'B': [1.1, np.nan, 3.3],
'C': ['x', 'y', 'z'],
})
result = render(result, {'colnames': 'A,B'})
expected = pd.DataFrame({'A': ['c'], 'B': [3.3], 'C': ['z']})
assert_frame_equal(result, expected)
def test_multiple_columns_drops_union(self):
result = render(
pd.DataFrame({
"A": [1, np.nan, 3.3],
"B": [np.nan, "b", "c"]
}),
P(["A", "B"]),
)
expected = pd.DataFrame({"A": [3.3], "B": ["c"]})
assert_frame_equal(result, expected)
def test_string(self):
result = render(
pd.DataFrame({
"A": [1, 2, 3, 4],
"B": [np.nan, None, "", "c"]
}),
P(["B"]), # drops both '' and np.nan
)
expected = pd.DataFrame({"A": [4], "B": ["c"]})
assert_frame_equal(result, expected)
def test_missing_colname(self):
result = pd.DataFrame({'A': [1]})
result = render(result, {'colnames': 'B'})
self.assertEqual(result, 'You chose a missing column')