Great-expectations

DataContexts are an opinionated framework for deploying Great Expectations within real data projects.

* Namespaced DataSources
* Batches as first-class citizens
* Tooling for validation
* Helper notebooks
* Compile to Docs

This branch is definitely the future of Great Expectations. Also, these features are in beta, so expect small-to-medium-sized changes in API, behavior, and design.

If you'd like help getting started, please reach out on Slack: https://tinyurl.com/great-expectations-slack

The core team will be more than happy to sit with you (probably via video call) and work with you to install, get started, and answer questions.

* Re-add testing (and support) for py2
* NOTE: Support for SqlAlchemyDataset and SparkDFDataset is enabled via optional install
(e.g. `pip install great_expectations[sqlalchemy]` or `pip install great_expectations[spark]`)

* Add support for SparkDFDataset and caching (HUGE work from cselig)
* Migrate distributional expectations to new testing framework
* Add support for two new expectations: expect_column_distinct_values_to_contain_set
and expect_column_distinct_values_to_equal_set (thanks RoyalTS)
* FUTURE BREAKING CHANGE: The new cache mechanism for Datasets, \
when enabled, causes GE to assume that dataset does not change between evaluation of individual expectations. \
We anticipate this will become the future default behavior.
* BREAKING CHANGE: Drop official support for python 2 and pandas < 0.22

* Fix issue where no result_format available for expect_column_values_to_be_null caused error
* Use vectorized computation in pandas (443, 445; thanks RoyalTS)

* Restructured class hierarchy to have a more generic DataAsset parent that maintains expectation logic separate from the tabular organization of Dataset expectations
* Added new FileDataAsset and associated expectations (416 thanks anhollis)
* Added support for date/datetime type columns in some SQLAlchemy expectations (413)
* Added support for a multicolumn expectation, expect multicolumn values to be unique (408)
* Optimization: You can now disable `partial_unexpected_counts` by setting the `partial_unexpected_count` value to 0 in the result_format argument, and we do not compute it when it would not be returned. (431, thanks eugmandel)
* Fix: Correct error in unexpected_percent computations for sqlalchemy when unexpected values exceed limit (424)
* Fix: Pass meta object to expectation result (415, thanks jseeman)
* Add support for multicolumn expectations, with `expect_multicolumn_values_to_be_unique` as an example (406)
* Add dataset class to from_pandas to simplify using custom datasets (404, thanks jtilly)
* Add schema support for sqlalchemy data context (410, thanks rahulj51)
* Minor documentation, warning, and testing improvements (thanks zdog).

* Add a new autoinspect API and remove default expectations.
* Improve details for expect_table_columns_to_match_ordered_list (379, thanks rlshuhart)
* Linting fixes (thanks elsander)
* Add support for dataset_class in from_pandas (thanks jtilly)
* Improve redshift compatibility by correcting faulty isnull operator (thanks avanderm)
* Adjust partitions to use tail_weight to improve JSON compatibility and
support special cases of KL Divergence (thanks anhollis)
* Enable custom_sql datasets for databases with multiple schemas, by
adding a fallback for column reflection (387, thanks elsander)
* Remove `IF NOT EXISTS` check for custom sql temporary tables, for
Redshift compatibility (372, thanks elsander)
* Allow users to pass args/kwargs for engine creation in
SqlAlchemyDataContext (369, thanks elsander)
* Add support for custom schema in SqlAlchemyDataset (370, thanks elsander)
* Use getfullargspec to avoid deprecation warnings.
* Add expect_column_values_to_be_unique to SqlAlchemyDataset
* Fix map expectations for categorical columns (thanks eugmandel)
* Improve internal testing suite (thanks anhollis and ccnobbli)
* Consistently use value_set instead of mixing value_set and values_set (thanks njsmith8)