Testing WordPress sites can be a challenge. We should strive to make use of different testing techniques to ensure we're getting good test coverage across our sites. Here are the different types of tests we recommend and how to achieve each.
Unit tests should test your code decoupled and independent the core logic from WordPress. For this to be true we should avoid any use of global functions or hidden dependencies in any classes/functions we use for our core logic. We should then be able to unit test these easily.
We use WP_Mock to help mock some WP core functions, with PHPUnit as the test runner.
Integration tests should test where your code depends on intergration with other codebases, including WordPress or other plugins.
We also use PHPUnit for integration testing along with WP PHPUnit which will allow you to interact with the WordPress install easily. It also provides a number of helpers to allow for easy generation of synthetic data for testing.
Unit tests are should live within each mu-plugin in a tests directory.
Integration tests are to be stored in /tests/integration though it would be sensible to structure this folder with names consistent with the class under test.
bin/docker/composer run phpunit
bin/docker/composer run phpintegration
(or composer run phpunit if not using docker)
Sometimes hard to find bugs can be present even though all our automated tests are passing, this could be due to something that's fallen through the cracks, a JS/CSS issue, an update to a plugin or theme, an update to a browser or something else entirely. It's therefore useful to run Visual Regression (VR) tests to catch anything that may have changed how the site appears to the end user.
VR tests work by taking screenshots of given scenarios (typically defined pages/posts), approving the screenshots to a known state and then comparing future runs against this approved state to see if anything has changed. Tests will then fail if the difference between the screenshots is above a set tolerance.
We use Playwright for VR testing.