NYU DevOps lab on Test Driven Development

One of my favorite quotes is:

“If it's worth building, it's worth testing. If it's not worth testing, why are you wasting your time working on it?”

As Software Engineers we need to have the discipline to ensure that our code works as expected and continues to do so regardless of any changes, refactoring, or the introduction of new functionality.

You can read more about my thoughts on TDD in the article: A Case for Test Driven Development

This lab introduces Test Driven Development using PyUnit and PyTest. It also demonstrates how to create a simple RESTful service using Python Flask and PostgreSQL. The resource model is persistence using SQLAlchemy to keep the application simple. It's purpose is to show the correct API calls and return codes that should be used for a REST API.

Note: The base service code is contained in routes.py while the business logic for manipulating Pets is in the models.py file. This follows the popular Model View Controller (MVC) separation of duties by keeping the model separate from the controller. As such, we have two test suites: one for the model (test_models.py) and one for the service itself (test_routes.py)

This lab uses Docker and Visual Studio Code with the Remote Containers extension to provide a consistent repeatable disposable development environment for all of the labs in this course.

You will need the following software installed:

• Docker Desktop
• Visual Studio Code
• Remote Containers extension from the Visual Studio Marketplace

All of these can be installed manually by clicking on the links above or you can use a package manager like Homebrew on Mac of Chocolatey on Windows.

Alternately, you can use Vagrant and VirtualBox to create a consistent development environment in a virtual machine (VM).

You can read more about creating these environments in my article: Creating Reproducible Development Environments

To bring up the development environment you should clone this repo, change into the repo directory:

git clone https://github.com/nyu-devops/lab-flask-tdd.git
cd lab-flask-tdd

Depending on which development environment you created, pick from the following:

Open Visual Studio Code using the code . command. VS Code will prompt you to reopen in a container and you should say yes. This will take a while as it builds the Docker image and creates a container from it to develop in.

code .

Note that there is a period . after the code command. This tells Visual Studio Code to open the editor and load the current folder of files.

Once the environment is loaded you should be placed at a bash prompt in the /app folder inside of the development container. This folder is mounted to the current working directory of your repository on your computer. This means that any file you edit while inside of the /app folder in the container is actually being edited on your computer. You can then commit your changes to git from either inside or outside of the container.

Bring up the virtual machine using Vagrant.

vagrant up
vagrant ssh
cd /vagrant

This will place you in the virtual machine in the /vagrant folder which has been shared with your computer so that your source files can be edited outside of the VM and run inside of the VM.

As developers we always want to run the tests before we change any code. That way we know if we broke the code or if someone before us did. Always run the test cases first!

Run the unit tests using pytest

make test

PyTest is configured via the included setup.cfg file to automatically include the --pspec flag so that red-green-refactor is meaningful. If you are in a command shell that supports colors, passing tests will be green while failing tests will be red.

PyTest is also configured to automatically run the coverage tool and you should see a percentage-of-coverage report at the end of your tests. If you want to see what lines of code were not tested use:

coverage report -m

This is particularly useful because it reports the line numbers for the code that have not been covered so you know which lines you want to target with new test cases to get higher code coverage.

You can also manually run pytest with coverage (but settings in pyporojrct.toml do this already)

$ pytest --pspec --cov=service --cov-fail-under=95

Try and get as close to 100% coverage as you can.

It's also a good idea to make sure that your Python code follows the PEP8 standard. Both flake8 and pylint have been included in the pyproject.toml file so that you can check if your code is compliant like this:

make lint

Which does the equivalent of these commands:

flake8 service tests --count --select=E9,F63,F7,F82 --show-source --statistics
flake8 service tests --count --max-complexity=10 --max-line-length=127 --statistics
pylint service tests --max-line-length=127

Visual Studio Code is configured to use pylint while you are editing. This catches a lot of errors while you code that would normally be caught at runtime. It's a good idea to always code with pylint active.

The project uses honcho which gets it's commands from the Procfile. To start the service simply use:

honcho start

As a convenience you can aso use:

make run

You should be able to reach the service at: http://localhost:8000. The port that is used is controlled by an environment variable defined in the .flaskenv file which Flask uses to load it's configuration from the environment by default.

If you are using Visual Studio Code with Docker, simply existing Visual Studio Code will stop the docker containers. They will start up again the next time you need to develop as long as you don't manually delete them.

If you are using Vagrant and VirtualBox, when you are done, you can exit and shut down the vm with:

exit
vagrant halt

If the VM is no longer needed you can remove it with:

vagrant destroy

• service/__init__.py -- establishes the Flask app factory
• service/routes.py -- the main Service routes using Python Flask
• service/models.py -- the data model using SQLAlchemy
• tests/test_routes.py -- test cases against the Pet service
• tests/test_models.py -- test cases against the Pet model

Copyright (c) 2016, 2024 John Rofrano. All rights reserved.

Licensed under the Apache License. See LICENSE

This repository is part of the New York University (NYU) masters class: CSCI-GA.2820-001 DevOps and Agile Methodologies created and taught by John Rofrano, Adjunct Instructor, NYU Courant Institute, Graduate Division, Computer Science, and NYU Stern School of Business.