Examples of Python Testing

README.md

These can be used as pre-commit tests or as part of a broader testing strategy.

While I consider these examples to be sane, there are a plethora of additional options available. I'm still using the pytest library. But this should still work when using the unittest module now part of the standard library.

I've also kept these reasonably quiet and only checking for the most egregious problems.

The assumption is that we're using Poetry as a builder.

Additionally, the output from these tests will be saved as a JUnit XML file for future parsing.

We'll be testing for:

• Code coverage - Coverage
• Type checks - MyPy
• PEP-8 style - Style Guide
• Code complexity - xenon and radon
• Docstring coverage - interrogate

This is a pretty bare bones set of examples. The actual implementation will vary. Please don't simply copy and paste.

pyproject.toml

[tool.pytest.ini_options]
testpaths = [
    "tests"
]

[tool.pylint.format]
max-line-length = 120

[tool.pylint.MASTER]
fail-under = 9.0
ignore-patterns = ''
jobs = 0
ignore = '.git'

[tool.pylint."MESSAGES CONTROL"]
disable = 'E,W,C,R'   # Clean slate. No implicit checks. Error, Warning, Convention, Refactor
enable = 'W0612,W0611,W0120,C'  # We care about a few Warnings and all Convention messages

[tool.pylint.REPORTS]
output-format = 'json' # Plain, simple, JSON. https://pylint.pycqa.org/en/latest/user_guide/usage/output.html

# Examples of overriding the global settings on a per message type level.
[tool.pylint.REFACTORING]
max-nested-blocks = 3
never-returning-functions = 'sys.exit'

[tool.pylint.LOGGING]
logging-format-style = 'new'
logging-modules = 'logging'

[tool.pylint.MISCELLANEOUS]
notes = '''FIXME,
           TODO,
        '''
notes-rgx = '[A-Z]+:.*'  # Capture any all caps string, terminated by a ':'.

[tool.mypy]
disallow_untyped_defs = true
disallow_incomplete_defs = true
disallow_untyped_calls = true
#exclude = []
#ignore_missing_imports = true

runner.sh

#!/bin/bash

poetry run pylint path/to/src --exit-zero
poetry run pytest --junitxml=test-results/pytest.xml

poetry run coverage run -m pytest
"""
We can leverage the pytest (or unittest) checks to also
check for code coverage as, in theory, the unit tests 
should be executing as much of the codebase as cna be
executed
"""
poetry run coverage report