pytest-regtest 2.3.0

pytest plugin for snapshot regression testing

The full documentatin for this package are available at https://pytest-regtest.readthedocs.org

About

Introduction

pytest-regtest is a plugin for pytest to implement regression testing.

Unlike functional testing, regression testing does not test whether the software produces the correct results, but whether it behaves as it did before changes were introduced.

More specifically, pytest-regtest provides snapshot testing, which implements regression testing by recording data within a test function and comparing this recorded output to a previously recorded reference output.

Installation

To install and activate this plugin execute:

$ pip install pytest-regtest

!!! note

`pyteset-regtest` offers some functionalities which are tailored
 for `NumPy`, `pandas` and `polars`. These depencies are not
 installed if you install `pytest-regtest`. In case you are using
 e.g. `NumPy` snapshots, we assume that your productive code (the
 code under test) uses `NumPy` and thus should be part of the setup
 of your project.

Use case 1: Changing code with no or little testing setup yet

If you're working with code that has little or no unit testing, you can use regression testing to ensure that your changes don't break or alter previous results.

Example: This can be useful when working with data analysis scripts, which often start as one long script and then are restructured into different functions as they evolve.

Use case 2: Testing complex data

If a unit tests contains many assert statements to check a complex data structure you can use regression tests instead.

Example: To test code which ingests data into a database one can use regression tests on textual database dumps.

Use case 3: Testing NumPy arrays or pandas data frames

If your code generates numerical results, such as NumPy arrays, pandas or polars data frames, you can use pytest-regtest to simply record such results and test them later, taking into account relative and absolute tolerances.

Example: A function creates a 10 x 10 matrix. Either you have to write 100 assert statements or you use summary statistics to test your result. In both cases, you may get little debugging information if a test fails.