Including external data with your unit tests

We generally preach that you should not use real data in your unit tests. You should instead mock said data out in an effort to lesson potential side-effects, including degrading test performance and adversely affecting live data.

Sometimes, you’d like to test actual file handling in a unit test. If you do, you should provide a sample data set packaged with your code in a data directory within your test directory.

Your package would then resemble the following structure:

/package
    /tests
        /data
            data.txt
        __init__.py
        test_module.py
    __init__.py
    module.py

In your test, you’ll need to provide a relative path to the data file. To do this, use the special __file__ attribute of the package to provide a starting point. This will let you know where the module lives on disk regardless of where the package is installed.

import os
from package.tests import __file__ as test_directory
 
def _data_dir():
    return os.path.join(os.path.dirname(test_directory), 'data')
 
def test_sample_data():
    file_path = os.path.join(_data_dir(), 'data.txt')
    with open(file_path) as f:
        package.module.do_something(f)

There you have it!