flask-app-template

Simple Flask application template for HTTP API backend with decoupled business logic, Alembic, pytest, uWSGI, Sentry, statsd server and docker.

Features

• Clean architecture. Flask request/app objects are limited to web package only. It narrows possibilities for deep penetration of framework and HTTP layer into an aplication core and leads to better design.

• Ready to use API routes with errors and fast json serialization.

• Fast start, you only need to replace my_project occurrences and ready to code models and endpoints.

• Alembic migrations.

• Configured pytest with dbsession fixture to prepare tables for your unit tests.

• Sentry integration.

• Graphite/Carbon/Grafana integration.

• Flexible configuration.

• Multistage dockerfiles (gcc, dev libs and other tooling are stripped). Alpine by default with Ubuntu as alternative.

• Ready to deploy to production. Without serious scalability in mind, it's a DevOps work anyway.

Local run in virtualenv

1. Your favourite way to make virtualenv. (I like virtualenvwrapper).

2. Install python requirements:

   pip install -r requirements.txt -r requirements-dev.txt
   

3. Run migrations:

   alembic upgrade head
   

4. Run server:

   uwsgi --ini etc/uwsgi.ini --py-autoreload=2
   

5. Call some endpoints:

   curl -X POST 127.0.0.1:5000/add?value=10
   curl -X POST 127.0.0.1:5000/add?value=10
   curl -X GET 127.0.0.1:5000/sum
   

6. Run tests:

   py.test
   

Local run with docker-compose

1. Build image:

   ./docker/build.sh
   

   You can use multi-stage dockerfiles for newer (>= 17.03) docker runtimes. See docker directory. By default template assumes "legacy" versions and emulates multi-stage build via special script.

2. Run migrations:

   USER_ID=$UID:$GROUPS docker-compose run http alembic upgrade head
   

3. Run server:

   USER_ID=$UID:$GROUPS docker-compose up
   

4. Call some endpoints:

   curl -X POST 127.0.0.1:5000/add?value=10
   curl -X POST 127.0.0.1:5000/add?value=10
   curl -X GET 127.0.0.1:5000/sum
   

Deploy to some docker-aware server

You need ssh login to host and working docker environment there also you need locally installed Fabric3. For first deploy run:

fab -H some.host init push_image upload migrate restart

Following deploys need:

fab -H some.host upload restart

If you changed docker/Dockerfile.* or requirements.txt you should update image on a server:

fab -H some.host push_image

You can tune default host, http port and config file location in fabfile.py. For full list of commands you can run fab --list.