ENH: snapshot testing of dataframes

[fkromer]

Is your feature request related to a problem?

When testing dataframe manipulating functions defining explicit expected dataframes takes a lot of time.

Describe the solution you'd like

I've used snapshottest to test GraphQl APIs. The same approach could be implemented in a pytest plugin for pandas as well.

API breaking implications

[this should provide a description of how this feature will affect the API]

Describe alternatives you've considered

[this should provide a description of any alternative solutions or features you've considered]

Additional context

[add any other context, code examples, or references to existing implementations about the feature request here]

# Your code here, if applicable

[MarcoGorelli]

Thanks @fkromer for the report - could you please provide the complete code showing what you're doing? https://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports

[fkromer]

@MarcoGorelli That does not really make sense in this case. This issue does not apply to testing the DataFrame functionality of pandas itself but applications making use of pandas DataFrame functionality. For example I've functions which manipulate DataFrames in a data manipulation pipeline and have a signature like this: def df_manipulation_func_a(df: pd.DataFrame, another_df: pdDataFrame) -> pd.DataFrame. Right now when unit testing these functions I've to define the expected returned dataframe myself. This is a lot of boilerplate. Instead in analogy to snapshottesting a tool/pytest plugin could run the tests in snapshot mode first to run the tests, capture the functions output and write it into files. In usual execution mode thetoll/plugin compares the actual output of the tests with the output previously captured.

[MarcoGorelli]

OK, thanks for the details @fkromer .

So, from what I've understood, this doesn't need to be implemented in pandas itself, it's a request for a pytest plugin

[fkromer]

Exactly. I know this issue is misplaced. But there was no better place to propose this. The issue is related to #16567 (comment) where other functionality is discussed which would fit into pandas.testing or some pandas-dev/pytest-pandas. Support for snapshot testing however would have to be put into pandas-dev/pytest-pandasfor sure cause it requires interaction with a test runner.

[MarcoGorelli]

Closing as this wouldn't go in pandas and there doesn't seem to have been any uptake from the folks here, but please do let us know if you implement this

[fkromer]

@MarcoGorelli Sure. This won't be soon but in case I can prioritize and implement this I'll let you know.

[clintmod]

It's worth noting that this works pretty well with pytest and snapshottest for a dataframe:

def test_something_works(snapshot): # snapshot is a pytest fixture from snapshottest
    data_frame = calc_something_and_return_pandas_dataframe()
    snapshot.assert_match(data_frame.to_csv(index=False), 'some_module_level_unique_name_for_the_snapshot')