Effortless REST w/ FLask

There are 5 chapter folders. CD into each one and follow the below Chapter outline.

Pre-requisites

1. pip install requirements

pip install -r requirements.txt

2. (Optional) to work with a Heroku DB, create an application and attach PostgreSQL to the application. See chapters 2-5 in task.py and set the application name

   • You may use a local sqlLite DB instead of a Heroku deployed DB.

Getting Started

• cd into each chapter and run the invoke commands to start such as:

invoke start

invoke init-db

invoke seed-db

---

Chapter 1: "Say Hello to the REAL world"

A word about dependencies

• Python uses pip as a package manager
• Many data scientist use conda to install packages.

example dependency export

pip freeze > requirements.txt

example dependency import/load

pip install -r requirements.txt

app/__init__.py

• The main Flask application
• Application factory
  • Allows for easier testing

Q: How do you configure a Flask application?

app/config.py

• Place configs for Flask and its plugins here

Q: How do you start the server?

export FLASK_APP="app:create_app('dev')"
export FLASK_ENV="development" &&
flask run

⚠️ DO NOT USE 'flask' cli for produciton

Q: How do you start the server for PRODUCTION?

gunicorn --bind 0.0.0.0 --workers 2 "app:create_app('prod')"

• 0.0.0.0: An alias IP to say expose on all network interfaces.
• workers: number of unique web application processes to spin up.

Q: Do I have to type that EVERY TIME?

• Check out Invoke
• Build pre-defined tasks with easy configuration for options
• see ./tasks.py

Example

invoke start

---

Chapter 2: Database: SQLAlchemy

app/__init__.py

• Add SQLAlchemy INIT
• Add route to get users

app/models/user.py

• Create a model for the user table

app/config.py

• Add the SQLALCHEMY_DATABASE_URI config

app/tasks.py

• Add task to init db tables
• Add task to seed users
• run the seed on both local SQLite and PSQL

Q: Why does the /uhoh route not return?

jsonify() doesn’t know how to convert a SQLAlchemy class to JSON.

If only there were a way to convert a serialized Python class into a deserialized dict. Then we could convert the dict with jsonify()…

In comes...

app/__init__.py

• The /marsh route uses a marshmallow Schema to deserialize a Python class.
• We can control what’s being deserialized with the different Schemas. (UserSchema vs UserSchemaWithPassword)
• Marshmallow can also serialize and validate a dict with .load()
  • e.g. an ISO date time stamp can be converted to a datetime object easily

---

Chapter 3: Security: FlaskPraetorian

app/config.py

• Add the proper configs: JWT_ACCESS_LIFESPAN JWT_REFRESH_LIFESPAN

app/__init__.py

• INIT Praetorian()
• Add login() route
• Protect routes with @auth_required decorator

tasks.py

• Use utils for password hashing

---

Chapter 4: Documentation & API Scale

app/__init__.py

• INIT Flask Rest Plus
• Define prefix for all routes
• Convert routes to Flask Rest Plus
• Add Namespaces for segmenting API routes within swagger
• Define authorizations section to easily add a JWT header to requests

Open up Swagger docs!

---

Chapter 5: Flask-Accepts

app/schemas/user.py

• Add UserLoginSchema

app/__init__.py

• Flask_accepts allows for easy Marshmallow schema conversion to JSON as well as adding Schema to Swagger docs
• Decorate endpoints with @accepts & @responds