Fix regression for loc and __setitem__ when one-dimensional tuple was given for MultiIndex #37787
Conversation
… given for MultiIndex
phofl
added
Indexing
| @@ -288,6 +288,20 @@ def convert_nested_indexer(indexer_type, keys): | |||
|
|
|||
| tm.assert_series_equal(result, expected) | |||
|
|
|||
| def test_multiindex_loc_one_dimensional_tuple(self): | |||
|
@phofl you're opening PRs faster than they can be reviewed (which is awesome, keep up the good work!). Maybe you can tackle one of the more challenging and immediate issues: the darns windows builds |
|
Thank you very much @jbrockmendel Have a bit of free time this week. Are you referring to the pipeline windows py38_no18? Unfortunately I am not that familiar with CI and even less with CI on windows. |
� Conflicts: � doc/source/whatsnew/v1.1.5.rst
| @@ -288,6 +288,14 @@ def convert_nested_indexer(indexer_type, keys): | |||
|
|
|||
| tm.assert_series_equal(result, expected) | |||
|
|
|||
| def test_multiindex_loc_one_dimensional_tuple(self, frame_or_series): | |||
There was a problem hiding this comment.
can you parameterize over all of the selectors in the OP e.g. obj.loc[('a',), :] if its a DataFrame
There was a problem hiding this comment.
Added tests for df, I think one of the cases mentioned there should raise and not work
| tm.assert_equal(obj, expected) | ||
|
|
||
| @pytest.mark.parametrize("indexer", [("a",), ("a")]) | ||
| def test_multiindex_one_dimensiona_tuple_columns(self, indexer): |
|
lgtm ex @jbrockmendel comment, ping on green. |
� Conflicts: � doc/source/whatsnew/v1.1.5.rst
|
thanks @phofl keep em coming! |
|
@meeseeksdev backport 1.1.x |
…when one-dimensional tuple was given for MultiIndex
CI is just another learning curve we all have to get up at some point. i still regularly ask for help with it. As for windows, im not actually convinced anyone uses it ever. It might be some kind of cruel prank played by a trickster god. see #37746 where i only-half-kidding suggest we drop windows altogether. More generally, I'm suggesting you look for areas that will have a higher ratio of (time to create PR) / (time to review PR).* * Though im really excited about the indexing fixes you've been making. |
… given for MultiIndex (pandas-dev#37787) (cherry picked from commit b80691c)
…tem__ when one-dimensional tuple was given for MultiIndex) (#37849)
.loc[(slice(...), ), ]when modifying a subset of rows in a pandas dataframe/series in 1.1.4 #37711black pandasgit diff upstream/master -u -- "*.py" | flake8 --diff