BUG: mask in test_mask_stringdtype would always return the same result regardless of cond #15
Description
Reproducible Example
import pandas as pd
test_mask_stringdtype
obj = pd.DataFrame(
{"A": ["foo", "bar", "baz", pd.NA]},
index=["id1", "id2", "id3", "id4"],
dtype=pd.StringDtype(),
)
filtered_obj = pd.DataFrame(
{"A": ["this", "that"]}, index=["id2", "id3"], dtype=pd.StringDtype()
)
expected = pd.DataFrame(
{"A": [pd.NA, "this", "that", pd.NA]},
index=["id1", "id2", "id3", "id4"],
dtype=pd.StringDtype(),
)
filter_ser = pd.Series([False, True, True, False])
obj.mask(filter_ser, filtered_obj)
A
id1
id2 this
id3 that
id4
filter_ser = pd.Series([True, False, False, True])
obj.mask(filter_ser, filtered_obj)
A
id1
id2 this
id3 that
id4
filter_ser = pd.Series([False, False, False, False])
obj.mask(filter_ser, filtered_obj)
A
id1
id2 this
id3 that
id4
filter_ser = pd.Series([True, True, True, True])
obj.mask(filter_ser, filtered_obj)
A
id1
id2 this
id3 that
id4
Issue Description
Found during pandas-dev#60772 .
I suppose the purpose of this test is to check if mask works as expected with pd.StringDtype() (See pandas-dev#40824 ), but the test seems to return the same result regardless of cond since it fails to align in _where.
If we want to check if mask replaces with other only where cond is True and let cond propagate where cond is False, I think filter_ser should have index so that mask can recognize the corresponding other value.
Expected Behavior
filter_ser = pd.Series([False, True, True, False], index=["id1", "id2", "id3", "id4"])
obj.mask(filter_ser, filtered_obj)