Fix mypy error in pandas/tests.indexes.test_base.py

[rushabh-v]

• relates to Fix mypy errors for pandas\tests\* #28926
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

[WillAyd]

What error does this give?

[rushabh-v]

Hello @WillAyd
Thanks for responding.
It was giving error: Need type annotation for 'x'.

[WillAyd]

OK lgtm. Can you merge master one more time and push? Fixed some things with CI recently

@simonjayhawkins

[rushabh-v]

I have merged it again. Please review the changes.
@WillAyd @simonjayhawkins

[simonjayhawkins]

mypy checks are failing see https://dev.azure.com/pandas-dev/pandas/_build/results?buildId=19636

[rushabh-v]

I have started working on them.

[rushabh-v]

Hello @simonjayhawkins @WillAyd
It is giving the error: Slice index must be an integer or None.
This error is coming from mypy packages. And I think there is a mypy version conflict. Because this bug is solved with (python/mypy#2410)
The lines giving this error are (https://github.com/RushabhVasani/pandas/blob/fcddcfcda2de3306191067afe19bc86859665091/pandas/tests/indexes/test_base.py#L1741-L1756)
So I tried to run the same code on my local machine. And it all works fine. The code I run was,

@pytest.mark.parametrize( 
       "in_slice,expected",
        [
            (pd.IndexSlice[::-1], "yxdcb"),
            (pd.IndexSlice["b":"y":-1], ""),
            (pd.IndexSlice["b"::-1], "b"),
            (pd.IndexSlice[:"b":-1], "yxdcb"),
            (pd.IndexSlice[:"y":-1], "y"),
            (pd.IndexSlice["y"::-1], "yxdcb"),
            (pd.IndexSlice["y"::-4], "yb"),
            # absent labels
            (pd.IndexSlice[:"a":-1], "yxdcb"),
            (pd.IndexSlice[:"a":-2], "ydb"),
            (pd.IndexSlice["z"::-1], "yxdcb"),
            (pd.IndexSlice["z"::-3], "yc"),
            (pd.IndexSlice["m"::-1], "dcb"),
            (pd.IndexSlice[:"m":-1], "yx"),
            (pd.IndexSlice["a":"a":-1], ""),
            (pd.IndexSlice["z":"z":-1], ""),
            (pd.IndexSlice["m":"m":-1], ""),
        ],
    )

And the output it gave was,

MarkDecorator(mark=Mark(name='parametrize', args=('in_slice,expected', [(slice(None, None, -1), 'yxdcb'), (slice('b', 'y', -1), ''), (slice('b', None, -1), 'b'), (slice(None, 'b', -1), 'yxdcb'), (slice(None, 'y', -1), 'y'), (slice('y', None, -1), 'yxdcb'), (slice('y', None, -4), 'yb'), (slice(None, 'a', -1), 'yxdcb'), (slice(None, 'a', -2), 'ydb'), (slice('z', None, -1), 'yxdcb'), (slice('z', None, -3), 'yc'), (slice('m', None, -1), 'dcb'), (slice(None, 'm', -1), 'yx'), (slice('a', 'a', -1), ''), (slice('z', 'z', -1), ''), (slice('m', 'm', -1), '')]), kwargs={}))

So I think there is a mypy version conflict.

[simonjayhawkins]

So I think there is a mypy version conflict.

yes these errors are not reported in 0.730. It was a typeshed fix though, see python/typeshed#3024 which was pulled into mypy for version 0.730

we are currently pinned to 0.720 on ci.

[rushabh-v]

we are currently pinned to 0.720 on ci.

Okay, then what should be done now.

[simonjayhawkins]

we are currently pinned to 0.720 on ci.

Okay, then what should be done now.

I guess two options.

1. seperate PR to bump mypy to 0.730 as a precursor to this
2. add # type: ignore with comments for error message and link to typeshed issue.

eg.

pandas/pandas/io/formats/printing.py

Lines 490 to 494 in 2ca2161

	# https://github.com/python/mypy/issues/4975
	# error: Incompatible return value type (got "Tuple[List[Sequence[str]],
	# List[Sequence[str]]]", expected "Tuple[List[Tuple[str, ...]],
	# List[Tuple[str, ...]]]")
	return head, tail # type: ignore

@WillAyd any preference?

[simonjayhawkins]

@rushabh-v can you merge master. mypy has been bumped to 0.730 on ci.

[rushabh-v]

It is giving the same error(Slice index must be an integer or None) again.
@simonjayhawkins

[WillAyd]

Yea I see the same locally - can you investigate further @rushabh-v ?

[rushabh-v]

Yes, I am looking into it.

[rushabh-v]

I am sorry, I mistakenly committed directly to master.

[rushabh-v]

It is giving this error for every version of mypy on my local machine. Even in 0.740.

I guess two options.

1. seperate PR to bump mypy to 0.730 as a precursor to this

2. add # type: ignore with comments for error message and link to typeshed issue.

Should we go with the 2nd option? @simonjayhawkins @WillAyd
I have checked locally adding, #type: ignore works fine on mypy 0.730.

[simonjayhawkins]

replacing (pd.IndexSlice["b":"y":-1], ""), with (pd.IndexSlice[slice("b", "y", -1)], ""), silences mypy. so it appears that the current mypy errors are not directly related to/ been solved by python/typeshed#3024

the mypy issue, python/mypy#2410 is still open, so a # type: ignore with a comment referencing the mypy issue is probably the most practical solution here.

[pep8speaks]

Hello @rushabh-v! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2019-11-27 12:05:49 UTC

[simonjayhawkins]

@rushabh-v Thanks for sticking with this. @WillAyd

[WillAyd]

Thanks @rushabh-v