BUG: instantiation using a dict with a period scalar

[justinessert]

• closes BUG: DataFrame creates an object-dtype array for extension type scalars in dictionaries #35965
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• whatsnew entry

Fixing bug discussed in issue 35965 where pd.DataFrame({'a': pd.Period('2020-01')}) created a as an object column instead of a period[m] column.

Changing the functionality of infer_dtype_from_scalar isn't necessarily required here, but the fact that infer_dtype_from_scalar would return the period.ordinal value seems inconsistent with the behavior for other dtypes in this function. Additionally, that functionality was only used in a single place within the code (interval.py), which I fixed accordingly.

[pep8speaks]

Hello @justinessert! 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 2020-09-06 22:33:19 UTC

[jreback]

can u update tests i think in pandas/tests/test/test_dtypes.py (thathit infer_scalar_from_dtype for the new behavior) it's possible we were not testing a period before

[justinessert]

@jreback there's one that I updated in test_infer_dtype_from_period, is that what you mean or are you after something else?

[justinessert]

@jreback What are you looking to be tested outside of test_infer_dtype_from_period?

[jreback]

@jreback What are you looking to be tested outside of test_infer_dtype_from_period?

yep saw that, your change is good

[jreback]

thanks @justinessert very nice

[justinessert]

@jreback Thanks for all your helpful and timely replies, it's much appreciated!