Pandera

New Features

- Allow attaching registered dataframe checks by using Config field names (478)

Bugfixes

- alias propagation works correctly on empty subclass (446)
- Add missing inplace arg to SchemaModel's validate (450)
- fix check_types decorator should return results from validate (458)
- Dataframe schemas in yaml do not require any field (479)
- coerce=True and pandas_dtype=None should be a noop (476)

Doc Improvement

- update documentation css to fit mobile (447)
- add copy button to docs (448)
- link documentation to github (449)

Infrastructure Changes

- add bugfixes and release branches to github actions eb38173290b5a6bd40d847bc94bcb8a767a04294
- fix github action triggers 3191be9c1237eb07780c5c4b50c245d3a767afb3
- update bug report template b0db5b05f8a8d114e18f2c622a33b6546bebdbd9
- bump cache 0ee703fe11971f2b094ffaf6adec76cae6bb74e3
- noxfile fixes 489695df6efaedfbf08e8e758a03f52c9ce4e04c
- update pylint (477)

New Features

- add new method `SchemaModel.to_yaml` to serialize `SchemaModel`s to yaml 428

Bugfixes

- preserve pandas extension types during validation (443)
- Fix to_yaml serialization dropping global checks (428) 🎉 first contribution antonl 🎉
- fix empty data type not supported for serialization (435)
- fix empty SchemaModel (434)
- add doc about attributes excluded by SchemaModel (436) jeffzi
- fix DataFrameSchema comparison with non-DataFrameSchema (431) jeffzi
- schema serialization handles non-PandasDtype (424)
- pa.Object coerce should preserve object type (423)

Documentation

- Update documentation theme to use [furo](https://github.com/pradyunsg/furo) #444
- Add favicon e3540f1

New Feature

- Add SchemaModel column name access through class attributes (388) jespercodes jeffzi 🎉
- Parametrized PandasExtensionType types (389) jeffzi 🎉
- adding filter argument to strict parameter (401) ktroutman
- feature/341: improve __str__ and __repr__ methods for schemas (413)

Bugfixes

- fix py3.6 optional + literal dtypes in SchemaModel (379) jeffzi 🎉
- Fix minimally required packaging version (380) **contribution 1️⃣ probberechts 🎉**
- prevent mypy Check getattr error for registered checks 920a98cb6113bf002fb1df11951196b0130f524a
- Compatibility with numpy 1.20 (395) jeffzi
- dataframe strategies can generate regex columns (402)
- bugfix: df data synthesis with size=None, fix CI (410)
- bugfix: SeriesSchema raises SchemaErrors on lazy validation (412)

Repo Improvements

- improvements to local CI (409) jeffzi
- feature/414: improve contributing docs and add to sphinx docs (416)

Bugfix Release

This release contains two bugfixes:

- coerce nullable str column handles all na (366)
- non-required columns that are not in dataframe are not coerced (368)

🎉🎉🎉 Thanks to jeffzi, ktroutman, m1so for your contributions! 🎉🎉🎉

Enhancements

- Improve memory efficiency of validation process (360)
- Add column order validation (352)
- Implement data synthesis strategies using [hypothesis](https://hypothesis.readthedocs.io/en/latest/) (#344)
- Add support for aliases in `SchemaModel` (329)
- Add support for optional name validation of single-index (326)
- Move columns to multiindex: add `reset_index`, `set_index` method to `DataFrameSchema` (319)
- Add support for Python 3.9 (307)

Bugfixes

- typing.DataFrame should expect annotation input (318)

Deprecations

- `SchemaErrors.schema_errors` has been changed to `failure_cases`, and the `schema_errors` attribute now contains a list of dicts containing schema errors and reason codes. This is a breaking change, but is a minor part of the API and is fairly straightforward to fix (360).

Documentation Improvements

- Add required columns documentation for schema models (362)
- Fix docs: schema examples (347)
- Add documentation for dataframeschema transformations (333)
- Fix deprecated SchemaErrorReport references in docs (310)
- Fix SchemaModel dtype example (309)

Repo Improvements

- Update logo 69c6e56a0961bb0a9f37947f040209fd71dd3826
- Add `flynt` to pre-commit hooks (325)
- Use generic zenodo link for citation information c4f4fe780729ee82444709910b5f2a7f206d3c81

pandera relied on the packaging package to get version information to determine pandas legacy status. This was an implicit sub-dependency of one of pandera's dependencies, which was apparently dropped and led to a bug: 335. This bugfix version explicitly adds packaging.