Write unit test from scratch
0 likes751 views
Unit tests validate that individual units of code are working as expected. Pytest is a testing framework for Python that makes writing unit tests simple. Some key features of Pytest include writing tests as normal functions, using fixtures to set up and tear down resources for tests, and plugins that add functionality like test coverage reporting. Writing unit tests for existing code can start with easy to test functions or interfaces and help catch bugs, though unit testing requires extra development time.
Write unit test from scratch
- 1. Writing Unit Test from Scratch Using Pytest Wen-Shih Chao a.k.a Yen3
- 2. Overview ● What is unit test ● PyUnit (unittest) and pytest ● Pytest basic usage ● Some useful pytest usage and plugin ● How to start writing your own unit test ?
- 3. The talk would uncover ● Mock (because I still learn how to use it XD) ● Test-Driven Development ● Integration Test ● How to become a test master XD
- 4. What it unit test ? ● A unit test is a program to test your target program ● A program is possible to have wrong behaviour and so does unit test. DEFINITION: A unit test is a piece of a code (usually a method) that invokes another piece of code and checks the correctness of some assumptions after- ward. If the assumptions turn out to be wrong, the unit test has failed. A unit is a method or function.
- 5. Example
- 9. Unittest library ● Unittest (PyUnit) - python standard library ○ xUnit style ○ No need to install ○ Mock support (python version >= 3.3) ● Pytest ○ Need to install ○ Pythonic style ○ Mock support (through plugin)
- 10. Why pytest ? ● You can write `assert XXX == YYY` rather than `self.assertZZZ(XXX, YYY)` ○ For example, if you want to compare two dictionaries x and y are equal. ■ PyUnit: `self.assertDictEqual(x, y)` ■ Pytest: `assert x == y` ● You can write fixture (pytest term) to replace `setUp()` and `tearDown()` function. A test case could use multiple fixtures. ● Use function notation `@` to mark attribute of each test case. ● You can define manual command line option to control the test flow. ● Some useful plugins can reduce the pain to write unit tests. You can also write/maintain plugins for your requirement.
- 13. Pytest basic 1. Write the test function 2. Run `pytest` 3. Check the result
- 14. Pytest folder structure ● A test folder contains ○ `__init__.py`: Please write pytest as a python module ○ `conftest.py` (optional): config and settings for the pytest module (Only valid in pytest) ○ `test_xxx.py`: Test cases ● You can make submodule for tests. ○ The submodule can use upper modules config
- 16. Pytest - Fixture ● The fixture provides resources which test cases need. ● 2 types fixture ○ A normal fixture - Just return ○ A yield fixture - Declare a resource and free it after finished ● The fixture concept is like context manager ● A test case can use multiple fixtures. ● A fixture can rely on other fixture
- 17. Pytest fixture ● Normal fixture Other fixture
- 18. Pytest fixture ● Yield fixture
- 19. Pytest - monkeypatch ● A simple approach for mock ● You can replace any method to your method
- 20. Example ● `check_worker_alive` would call `ping_worker` to get the ping result. For avoiding to access the real internet, the test fakes a function that always return true value to test the main function part of `check_worker_alive`
- 21. Pytest - mark ● Mark is like a tag. Before starting the test, you can check the mark to do what you want to do.
- 22. Mark example ● In the example, when you mark some test as slow `@pytest.makr.slow` and you run pytest without `--slow` option (e.g. `pytest`). The pytest would ignore the test
- 23. Pytest - parametrize test ● Parameterize the argument
- 24. Pytest plugin ● pytest-tornado/pytest-tornado-yen3: The plugin lets you to test coroutine function as native function and provides some useful fixture. ● pytest-mock: wrapper for unittest.mock ● pytest-cov: Generate test coverage report ● pytest-pep8: Generate pep8 style checking report ● pytest-selenium: I have no experience about the plugin, maybe sw2 needs it
- 25. Pytest-tornado/Pytest-tornado-yen3 ● Pytest-tornado provides some marks to run your test as a coroutine ○ It would start a new ioloop and use `IOLoop.run_sync` to run each test case ● The last update of pytest-tornado is 2 years ago. I have some needs and change for the plugin so I maintain a patched version for my project. You can get more details from https://github.com/yen3/pytest-tornado
- 26. Pytest-cov ● Generate coverage ● Usage is simple `pytest --cov=<module_name>`. You can use `--cov` multiple time
- 27. Coverage Example docker-compose -f docker-compose-test.yml exec test-server sh -c "cd /app && pytest --cov=common --cov=api_server --cov=worker --slow"
- 28. Coverage example docker-compose -f docker-compose-test.yml exec test-server sh -c "cp -r /app/* /app_cov && cd /app_cov && pytest -sv --cov-report=html --cov=common --cov=api_server --cov=worker --slow --pep8 | tee htmlcov/test.log"
- 29. Should I write unit test for my program ? ● The answer is probably. In actually, some program is hard to test. ● If your program is easy to test, you should try to write some tests for it. ● Writing unit test would cost your extra time. If you have plan to write unit test, remember to request more develop time to your manager. ● Unit test is also a program. It’s possible to go wrong. ● Unit test can not find all bugs. But it can find many bugs if you write tests well.
- 30. Should I write unit test for my program ? ● How to write unit tests for existing program ? ○ It’s up-on your situation. If the time schedule of the project/program has no enough time, maybe you have to quit the idea. ○ If you have enough time, there are several possible ways. ■ Start from easy function (e.g. utility functions). You can feel how to write unit tests. ■ Start from function that are easy to be buggy. It’s usually hard to test. If you can finish, maybe you can get rid of the buggy function (for a while) ■ Start from the interface. ● I take the policy for middleware-network. The policy is like to write integration tests. It can make sure your program works after each modification.
- 31. My experience about unit test ● If you are a front-end developer, I am sorry that I have no experience about it. ● If you are a back-end developer and use python, you could write test with pytest. ● Consider writing unit test to replace testing function/module manually ● Run test in docker container is very important. It can avoid reset the testing environment every time. ● Take the advantage of fixture to initial/release resource for test cases
- 32. My experience about unit test ● The textbook says writing unit test is easy. In fact, a lot of problems need to be resolve. ○ It costs lots of effort to maintain a good unit test. ○ Resource problems ■ Interactive/Isolate with real world environment like database, file system … etc ■ Resource initialize and restore ● I also write tests for third-party packages (e.g. tornadis, sqlalchemy … etc) ○ learn how to use these packages and avoid the unexpected behaviour when it updates. ○ The type of test is also called learning test.
- 33. My experience about unit test ● Prepare test environment. The test environment must isolate the production environment. The straightforward way is to use docker to achieve the effect. ● If you have multiple service need to start, you can ○ Use docker-compose to start several containers ○ User docker with supervisor to start several services in a container ○ Mock all services ○ … etc. It’s upon your requirement and situation
- 34. Questions ?
- 35. Reference ● Software test concept a. Software Testing - https://en.wikipedia.org/wiki/Software_testing#The_box_approach b. The Art of unit testing with C# 2nd edition, ISBN: 9781617290893 (中譯: 單元測試的藝術 2/e, ISBN: 9789864342471) ● Pytest a. Pytest - https://docs.pytest.org/en/latest/ b. Pytest 還有他的快樂夥伴 - https://www.slideshare.net/excusemejoe/pytest-and-friends c. Python Testing with pytest, ISBN: 9781680502404 ● Practical test experience sharing a. Clean code, ISBN: 9780132350884 (中譯: 無瑕的程式碼, ISBN: 9789862017050) b. Code Complete 2/e, ISBN: 9780735619678