⇽ back to luminar.dev


I hope that after reading this you will be excited about writing tests. Writing tests should not be this chore that you do after you finished your project. It should a be a tool that helps you while you are programming. It should automate repetitive tasks and raise confidence in your code.

You are testing software even if you are not writing integration, unit and other smart-word tests. Everyone does. How else would we know that our code works? The most basic type of test is the manual test and everyone is doing those — we just run the code and see what happens.

I would like to show you, that pytest is a great tool that helps you to automate your manual tests. And at the same time pytest will not put obstacles in your way — you don’t have to write any boilerplate code to start testing. In fact, if you don’t need to, you won’t be writing any boilerplate code. Ever.

But automating manual testing is just the first step. pytest can do so much more for you — it can help you improve your code while you are writing it. How? It forces you to use your code. Doesn’t matter if you are testing one function or the whole application. Writing the test can show you how your design can be improved. A simple rule of thumb:

It it’s hard to test, it’s hard to use. If it’s hard to use, it can be designed better.

No jargon

There are many articles and videos about writing tests with pytest. I would like this article to be a bit different — I won’t try to describe all features of pytest, the documentation is there for that. I also would like to refrain from using the testing jargon like "unit under test", "setup/teardown" and "unit test". I especially won’t be talking about "test-driven development" and "test coverage". I feel like these phrases are not helpful.

To people who are not writing tests, I would like to show how to start writing some tests and how tests can help them.

To people who are writing tests as an afterthought or are used to writing tests in other languages, I would like to show (maybe) a different way of thinking about tests.

To people who are already proficient in using pytest, well, maybe they can teach me something.


I would like to show how to start writing pytest from the first line of code — not before, not after, but while you are writing code.

Then I would like to show a couple of tricks that pytest has to help you write readable, e.g. "fixtures" and "parametrization".

Finaly, I would like to show, how to better structure your tests and integrate them into your application and workflow.


Readability counts in test too — tests should be as easy to review as code. Hard to understand tests are hard to fix. Badly structured and unreadable tests should not pass codereview (how often we skip tests during code review?).

How fixtures help: Test should not contain any logic that is required to set up the test. A test should contain only asserts — clearly described conditions when the test passes and when in fails. Everything should go into fixtures.

Fixtures are help with "code reuse", you can combine fixtures.

Structuring tests: a very good article.