BUG: setitem with boolean mask and series as value is broken for Series with EA type #37676
Conversation
phofl
added
Indexing
| {"a": [0, 0, np.nan, np.nan], "b": array(range(4), dtype="Int64")} | ||
| ) | ||
| s = Series(array([1] * 4, dtype="Int64")) | ||
| s[df["a"].isna()] = df.loc[df["a"].isna(), "b"] |
There was a problem hiding this comment.
can you do this without constructing a DataFrame
There was a problem hiding this comment.
Avoided the DataFrame construction. Checked, that this raised the same error as reported in the issue.
jreback
changed the title
| def test_setitem_boolean_ea_type(self): | ||
| # GH: 26468 | ||
| s = Series(array([5, 6, 7, 8], dtype="Int64")) | ||
| s[s > 6] = Series(range(4), dtype="Int64") |
There was a problem hiding this comment.
generally looks good, couple of nitpicks: can you import array as pd_array, name this ser instead of s (doesnt have to be ser, the important thing is to avoid 1-letter variable names), instead of "ea_type" use "Int64_values" (or better yet, parametrize over nullable int dtypes)
| ser[ser > 6] = Series(range(4), dtype="Int64") | ||
| expected = Series([5, 6, 2, 3], dtype="Int64") | ||
| tm.assert_series_equal(ser, expected) | ||
| ser = Series(pd_array([5, 6, 7, 8], dtype="Int64")) |
There was a problem hiding this comment.
can you add a blank line between cases
| @@ -126,6 +127,16 @@ def test_setitem_boolean_different_order(self, string_series): | |||
|
|
|||
| tm.assert_series_equal(copy, expected) | |||
|
|
|||
| def test_setitem_boolean_nullable_int_types(self, any_nullable_int_dtype): | |||
| # GH: 26468 | |||
| ser = Series(pd_array([5, 6, 7, 8], dtype="Int64")) | |||
There was a problem hiding this comment.
shouldn't you be using the any_nullable_int_dtype fixture here?
There was a problem hiding this comment.
Sorry, somehow missed that. Thanks for pointing it out. Changed it
� Conflicts: � pandas/tests/series/indexing/test_setitem.py
| @@ -127,6 +127,15 @@ def test_setitem_boolean_different_order(self, string_series): | |||
|
|
|||
| tm.assert_series_equal(copy, expected) | |||
|
|
|||
| @pytest.mark.parametrize("value", [None, NaT, np.nan]) | |||
There was a problem hiding this comment.
can you also check that np.datetime64("NaT") doesn't get cast to td64
There was a problem hiding this comment.
Thats not from my change, that got introduced by 5c4f737
| @@ -135,6 +136,17 @@ def test_setitem_boolean_td64_values_cast_na(self, value): | |||
| expected = Series([NaT, 1, 2], dtype="timedelta64[ns]") | |||
| tm.assert_series_equal(series, expected) | |||
|
|
|||
| def test_setitem_boolean_nullable_int_types(self, any_nullable_int_dtype): | |||
| # GH: 26468 | |||
| ser = Series(pd_array([5, 6, 7, 8], dtype=any_nullable_int_dtype)) | |||
There was a problem hiding this comment.
No need for the pd.array here, since the Series constructor will handle the nullable dtype as well.
| tm.assert_series_equal(ser, expected) | ||
|
|
||
| ser = Series([5, 6, 7, 8], dtype=any_numeric_dtype) | ||
| ser.loc[ser > 6] = Series(range(4), dtype=any_numeric_dtype) |
There was a problem hiding this comment.
is the failure on 3.9 something else? https://dev.azure.com/pandas-dev/pandas/_build/results?buildId=47857&view=logs&j=f016abb9-7827-5fa2-935a-22bd9b1477b6&t=c86edbe1-8c1d-5e5c-5b6f-d970fa4acf6d
did you merge master recently?
There was a problem hiding this comment.
AFAICT thats from the reduction bugfix yesterday. im troubleshooting it now
There was a problem hiding this comment.
Tuesday, so not really recently, merged now
|
thanks @phofl |
black pandasgit diff upstream/master -u -- "*.py" | flake8 --diffWas fixed already, added a test