Skip to content

REGR: Avoid regression warning with ea dtype and assert_index_equal order False #47325

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Jun 15, 2022
Merged

REGR: Avoid regression warning with ea dtype and assert_index_equal order False #47325

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4e98bad
REGR: Avoid regression warning with ea dtype and assert_index_equal o…
phofl Jun 13, 2022
aa6e165
Add test
phofl Jun 14, 2022
5879fb4
Remove fixture
phofl Jun 15, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions doc/source/whatsnew/v1.4.3.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@ Fixed regressions
- Fixed regression in :func:`read_csv` with ``index_col=False`` identifying first row as index names when ``header=None`` (:issue:`46955`)
- Fixed regression in :meth:`.DataFrameGroupBy.agg` when used with list-likes or dict-likes and ``axis=1`` that would give incorrect results; now raises ``NotImplementedError`` (:issue:`46995`)
- Fixed regression in :meth:`DataFrame.resample` and :meth:`DataFrame.rolling` when used with list-likes or dict-likes and ``axis=1`` that would raise an unintuitive error message; now raises ``NotImplementedError`` (:issue:`46904`)
- Fixed regression in :func:`assert_index_equal` when ``check_order=False`` and :class:`Index` has extension dtype (:issue:`47207`)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think I may have been wrong to add the regression label to #47207

Not strictly a regression since 1.3.5 did not have EA backed Indexes and the code sample created an object dtype Index. maybe this should be in the bugfix section. wdyt?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You are correct wrt ea back indexes, but you can also trigger this with

idx1 = pd.Index([1, 3], dtype="object")
idx2 = pd.Index([3, 1], dtype="object")
tm.assert_index_equal(idx1, idx2, check_order=False)

I'll add a test and will adjust the release note

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Added both

- Fixed regression in :func:`read_excel` returning ints as floats on certain input sheets (:issue:`46988`)
- Fixed regression in :meth:`DataFrame.shift` when ``axis`` is ``columns`` and ``fill_value`` is absent, ``freq`` is ignored (:issue:`47039`)

Expand Down
4 changes: 2 additions & 2 deletions pandas/_testing/asserters.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -367,8 +367,8 @@ def _get_ilevel_values(index, level):

# If order doesn't matter then sort the index entries
if not check_order:
left = Index(safe_sort(left))
right = Index(safe_sort(right))
left = Index(safe_sort(left), dtype=left.dtype)
right = Index(safe_sort(right), dtype=right.dtype)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe another issue, but the check_names check is later. It maybe that would need to resuse the code pattern you added in #47206. Not here, as this PR is targeted 1.4.3, just an observation.

Copy link
Member Author

@phofl phofl Jun 13, 2022 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good point, I am not using that myself, so I am a bit fuzzy on all the keywords :)

It is actually a bit more comlex, because These calls return regular indexes, even if a MultiIndex was given. I'll open an issue about it

Edit: Sorry misunderstood you. You are correct, we have to use exactly the same pattern

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

my thinking was that your fix in #47206 looked quite useful to robustly sort indexes. and maybe a safe_sort_index would be handy. (IIRC we cannot support pandas objects in safe_sort but would need to check this). I wonder can the fix in #47206 suffer from the rogue warning too without the dtype specified there?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yep you are completely right. Opened #47330 for 1.5


# MultiIndex special comparison for little-friendly error messages
if left.nlevels > 1:
Expand Down
7 changes: 7 additions & 0 deletions pandas/tests/util/test_assert_index_equal.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -242,3 +242,10 @@ def test_assert_index_equal_mixed_dtype():
# GH#39168
idx = Index(["foo", "bar", 42])
tm.assert_index_equal(idx, idx, check_order=False)


def test_assert_index_equal_ea_dtype_order_false(any_numeric_ea_dtype):
# GH#47207
idx1 = Index([1, 3], dtype=any_numeric_ea_dtype)
idx2 = Index([3, 1], dtype=any_numeric_ea_dtype)
tm.assert_index_equal(idx1, idx2, check_order=False)