README.md

api-testing

This repo outlines a few techniques for unit testing external APIs with pytest.

Prerequisites

• poetry
• an API key for OpenWeather

Installation

git clone https://github.com/jonwhittlestone/today-i-learned.git

cd pytest/api-testing

poetry install

Usage

• In the virtualenv, add API key as env var and run the Flask-based implementation

  poetry shell;
  
  export API_KEY=changeme;
  
  python3 api_testing/__init__.py

  visit http://127.0.0.0.1:8000

• Run tests individually

  1. Using mocks, specifically pytest-mock

     export API_KEY=changeme;
     
     alias t1="pytest tests -k 'test_retrieve_weather_using_mocks';"
     
     t1

  2. Using responses which intercepts requests calls and patches them avoiding the length set up.

     export API_KEY=changeme;
     
     alias t2="pytest tests -k 'test_retrieve_weather_using_responses';"
     
     t2

  3. Using dependency injection to inject a fake adapter during test time.

     This way the behaviour doesn't change if we decide to change the implementation because all we've done is change the adapters (from one that uses responses to one that uses urllib for example)

     export API_KEY=changeme;
     
     alias t3="pytest tests -k 'test_retrieve_weather_using_adapter'";
     
     t3

  4. Using VCR.py

     change the casette stored API Key

     # cassettes/test_retrieve_weather_using_vcr.yaml 
     
     -     uri: https://api.openweathermap.org/data/2.5/weather?q=London&appid={changeme}&units=metric
     
     +     uri: https://api.openweathermap.org/data/2.5/weather?q=London&appid={actual_api_key}&units=metric

     export API_KEY=changeme;
     
     alias t4="pytest tests -k 'test_retrieve_weather_using_vcr'"
     
     t4

Resources

• https://opensource.com/article/21/9/unit-test-python