BUG: loc.setitem raising ValueError when df has duplicate columns #39278
Conversation
phofl
commented
Jan 19, 2021
phofl
commented
- closes BUG: Setting values to slice fails with duplicated column name #38521
- tests added / passed
- Ensure all linting tests pass, see here for how to run them
- whatsnew entry
phofl
added
the
Indexing
|
Is there a more elegant way to check if an element occurs only once in an index? |
| @@ -373,6 +373,15 @@ def test_setitem_string_column_numpy_dtype_raising(self): | |||
| expected = DataFrame([[1, 2, 5], [3, 4, 6]], columns=[0, 1, "0 - Name"]) | |||
| tm.assert_frame_equal(df, expected) | |||
|
|
|||
| def test_setitem_empty_df_duplicate_columns(self): | |||
There was a problem hiding this comment.
do we need tests for Series?
There was a problem hiding this comment.
Don't think so, indexer has to be a tuple to land there, which is only valid for dataframes I think?
There was a problem hiding this comment.
its weird, but you could do series.loc[("foo",)]
There was a problem hiding this comment.
Yep, definitely weird :)
But this is already converted before reaching that point.
df = Series(index=["a", "b", "b"], dtype="float64")
df.loc[("a", )] = 1
This results in indexer=0. Tested the same with a MultiIndex Series, also 0 there
pandas/core/indexing.py
Outdated
| @@ -1850,7 +1850,8 @@ def _setitem_single_block(self, indexer, value, name: str): | |||
| for i, idx in enumerate(indexer) | |||
| if i != info_axis | |||
| ) | |||
| and item_labels.is_unique | |||
| and len(item_labels.get_indexer_for([item_labels[indexer[info_axis]]])) | |||
| == 1 | |||
There was a problem hiding this comment.
lets separate this out into a nested condition and define item_labels[indexer[info_axis]] once, then reuse it on L1856
|
Failure unrelated |
|
thanks @phofl |
