Integration Testing Guide with Pytest

[laurent-laporte-pro]

Integration Testing Guide with Pytest

This guide explains how to organize and execute integration tests for your project using Pytest. The primary goal is to enhance test maintainability by following a clear structure and adhering to naming conventions.

Organization of Test Modules

To ensure a close correspondence between the implementation modules' hierarchy and the test modules, follow these guidelines:

• Integration tests for the AntaREST application reside in the tests/integration directory.

• This directory contains assets used for tests (resources such as studies, application configuration for testing, etc.) and fixtures¹ used for dependency injection in your tests.

• Within this package, create a sub-module named after the tested component. For example, if the component is "watcher," name this sub-module test_watcher.

Naming Test Functions

For each component you test that involves multiple endpoints², name test functions as follows:

• For the "watcher" component, suppose there is an endpoint /watcher/_scan implemented by a function named scan_dir in the antarest.study.web.watcher_blueprint module. In this case, name your test function test_scan_dir.

• If you have multiple test scenarios for the same endpoint, add a descriptive suffix to the test function, for example, test_scan_dir__scenario1, test_scan_dir__scenario2, etc.

Using Fixtures

In your test module, use fixtures defined in the conftest.py file in the tests/integration package. For instance, you can use the user_access_token fixture (access token for the "admin" user's API) and the study_id fixture (ID of the "STA-mini" study used as a working base for tests).

Limiting the Number of Test Functions

Since integration tests can be time-consuming due to initial setup (app fixture), it is recommended to limit the number of test functions per endpoint. For some components, you may consider grouping multiple aspects into a single test, such as test_lifecycle, which covers the entire component lifecycle (creation, update, deletion, error handling).

Footnotes

1. A fixture in Pytest is a Python function used to set up a specific test environment and provide data or objects necessary for running tests. ↩

2. An endpoint is a URL associated with an HTTP method, used to access specific functionalities of a web application. For instance, the POST /watcher/_scan endpoint allows creating a workspace analysis task in Antares Web. ↩