Notes on testing

The PyOWM has a strong focus on testing: we want it to be an open source project with industrial quality!

Unit tests

Unit tests aim to test the SW API of each class/module.

We make use of monkey patching (which is more convenient than mocking because Python is a dynamically typed) whenever there is the need to test the behavior of a SW component which is bound to an external SW entity/system (ie: the parts of the PyOWM library that call the real PyOWM web API via HTTP).

The default unit testing environment is Python's unittest: simple and clean enough, so no additional dependencies are needed.

Unit tests can be easily run using Tox along with py.test.

Running:

$ tox

triggers unit tests execution on all Python platfoms that are supported by PyOWM.

Also, unit tests can be run with:

$ pythonX.Y setup.py test -s tests.unit

Integration tests

Export your API Key as an environment variable named OWM_API_KEY:

$ export OWM_API_KEY='<your-api-key>'

and then you can run the integration tests with:

$ cd tests/integration
$ tox

or - on Linux - use the test runner:

$ cd scripts
$ bash run_integration_tests.sh

Please note that depending on your subscription type some of the tests may be neglected, eg: if you have a free subscription tier, the test cases that invoke the OWM API to get historical weather data will fail as these data can only be retrieved using a paid account.

Testing using Docker

A Docker image is available for developing/testing PyOWM.

Continuous Integration

PyOWM is continuously built with Travis-CI CI builds every time a Git push and a Github pull request are issued.

Code coverage

Code coverage is checked on each CI build with Coveralls.io