UserSnack

A minimalistic pizza ordering API service built with FastAPI and TortoiseORM.

✍🏾 Project Philosophy

• The project is based on the service-repository pattern.
• The idea behind using a repository despite using a library like Tortoise-ORM is to be able to abstract complex database queries and do easy mocking as scale during unit tests.

💻 Development Setup

• Copy .env.sample to .env and set the values as necessary.

1️⃣ (Option 1) All in Docker Compose

Run:

docker compose up

2️⃣ (Option 2) App outside of Docker

This could be useful if you need to attach a debugger to your application. As a requirement, you'd need to have Poetry installed. Then you may follow the steps:

• Start only the Database in docker if you don't have any existing server on your machine.

docker compose up db -d

• Override the default database config by adding the following to your gitignored .env:

POSTGRES_HOST=localhost
POSTGRES_PORT=5438

The host port map is changed in this project to avoid conflict with existing Postgres on host machine.

export PYTHONPATH=src # you only need to set this once in your terminal session
python -m uvicorn main:app --reload

♼ Migrations

Migrations are managed with Aerich (a migration tool by Tortoise-ORM). To run migrations:

# Outside docker setup
aerich upgrade

# Full docker compose setup
docker compose exec app aerich upgrade

🌱 Seed Data

Some default pizzas and pizza extras can be available by default over the API. To seed your new database with them:

# Outside docker setup (assumes the terminal session still has the PYTHONPATH=src)
python scripts/load_fixtures.py

# Full docker compose setup
docker compose exec app python scripts/load_fixtures.py

Once started up, ideally, you should be able to access the project documentation over http://localhost:8000/docs if no custom port settings were done.

🥷🏼 Linting & Formatting

Flake8 is used for linting while Black is used for code formatting. To run both respectively:

# Format code
./scripts/format.sh

# Lint code
./scripts/lint.sh

🧪 Testing

Pytest is used for testing in the project, and currently both unit tests and integration tests are ran together (can be improved to split into individuals). To run tests:

pytest -s

You would need your database to be running in order to execute the integration tests successfully.

🌌 Media Files

In development/test environment, FastAPI serves the media files, but these requests are served by the webserver in Production.

🧢 Project Assumptions

• No authentication required.
• Scope does not include the logistics flow after ordering action.

🪂 Deployment

You can set up the project on a single node machine with docker.

• Clone the project
• Copy .env.sample to .env and set the values as necessary.
• Set ENV to prod, so the application loads the correct configurations.
• Modify HOST_DOMAIN environment in the caddy service of docker-compose.yml; set to your domain, and ensure that the domain is equally pointing to your remote machine IP address.
• Run migrations and seed data as specified above.
• Your API documentation should be accessible over https://<your-domain>/docs, and media over https://<your-domain>/media/<path-to-media>.

🪡 Improvements / Scale Production

• Phone number validation
• Add CI
• Use an container orchestration tool like k8s at scale
• Manage postgres connections with tool like PgBouncer to reduce the time spent opening and closing connections (reuse connections).
• Generate typescript client that the react FE can use.
• Caching using Redis/memcache or similar tool.
• JSON logging
• More unit tests for the services / repositories
• Use factories to generate test data
• Split unit and integration tests