willsather / python-tdd-with-django

seven different stages of testing a python and django rest framework app

Python TDD with Django REST Framework

Python REST DRF Learning Path

The goal of this repository is to create a series of content about building Python application using Django REST Framework.

Also, the intent was to use this DRF application to service a data science / machine learning model, and create the whole thing by TDDing in Python. The hope was to not only build out various stages of an application, but also demonstrate one strategy for writing tests for machine learning. In doing so, I goaled for creating as close to a _ real-world_ example to a data science project as possible, while documenting all the work I did building it.

/docs contains the 7 different Markdown documents that respectively align with the 7 different directories in the repository. Each directory contains an individual virtual environment and should be able to be opened in any editor to easily run the tests and server.

Section 0: Introduction

• Why Python is important
• Primary Goals / Objectives
  • Maybe list out
• Overview on what is being built
• What is the endpoint about (Penguins)
• List of each section / respective topics

Section 1: Setup Environment

• Create directories / git repo
• Install Python using Homebrew
• Create virtual environment with venv
• Install dependencies REST dependencies Django and rest_framework
• Install pytest-django

Section 2: Setup Django

• Differences in Python web frameworks
• Database Discussion
• Structure of a Django application
• Creating project / app
• What is manage.py
• What is urls.py and views.py (Class Based Views)

Section 3: TDD for Django Rest Framework GET Endpoint

• TDD
  • Write GET test with the GET endpoint (hardcoded value)
  • Create model and serializer
  • Database migration
    • python manage.py makemigrations
    • python manage.py migrate
  • Update GET test with the GET endpoint to get all penguins

Section 4: TDD for Django Rest Framework CRUD Application

• Write POST test with the POST endpoint
• Add PenguinDetailsController using Generic
• Differences with GenericAPIView and ListCreateAPIView
  • Tests pass with either implementation (useful for refactoring)
• Ends with basic CRUD application that can GET,POST,PUT,DELETE

Section 5: Machine Learning / Decision Trees

• Introduction to machine learning
  • Important terms and definitions
  • Common steps to create a model
• Create model using SciKit learn

Section 6: TDD DRF Predict (using Decision Tree)

• Skips forward and adds 5 more properties to a penguin *

Note: Previously the Penguin object had only two properties, now there exists 5 more properties. To add properties, you must do a db migration.

• TDD
  • Write POST /predict test with the POST /predict endpoint
    • Basic POST and returns a static value
  • Plop in exported model into code and use sci-kit learn sk.predict method
  • Write unit test for different cases of our custom business logic
    • Write test in test_service.py using @pytest.mark.unit
  • Write business logic in service.py that takes in input, and returns value using sk.predict
  • Update test_views.py to have test for /predict for a given penguin, returns type of penguin

Section 7: Deploy Python Application using Docker and Kubernetes

• Python Application Containerization using Docker
  • Maybe using Cloud Native build packs?
• K8S configuration
• Simplify YAML using Tanzu Applications