REpresentational State Transfer (REST) is a web development architecture design style which refers to logically separating your API resources so as to enable easy access, manipulation and scaling. Reusable components are written in a way so that they can be easily managed via simple and intuitive HTTP requests which can be GET, POST, PUT, PATCH, and DELETE (there can be more, but above are the most commonly used ones).
Despite what it looks like, REST does not command a protocol or a standard. It just sets a software architectural style for writing web applications and APIs, and results in simplification of the interfaces within and outside the application. Web service APIs which are written so as to follow the REST principles, they are called RESTful APIs.
In this three-part tutorial series, I will cover different ways in which RESTful APIs can be created using Flask as a web framework. In the first part, I will cover how to create class-based REST APIs which are more like DIY (Do it yourself), i.e. implementing them all by yourself without using any third-party extensions. In the latter parts of this series, I will cover how to leverage various Flask extensions to build more effective REST APIs in an easier way.
I assume that you have a basic understanding of Flask and environment setup best practices using virtualenv to be followed while developing a Python application.
Installing Dependencies
The following packages need to installed for the application that we'll be developing.
$ pip install flask $ pip install flask-sqlalchemy
The above commands should install all the required packages that are needed for this application to work.
The Flask Application
For this tutorial, I will create a small application in which I will create a trivial model for Product. Then I will demonstrate how we can write a RESTful API for the same. Below is the structure of the application.
flask_app/
my_app/
- __init__.py
product/
- __init__.py // Empty file
- models.py
- views.py
- run.py
I won't be creating a front-end for this application as RESTful APIs endpoints can be tested directly by making HTTP calls using various other methods.
flask_app/my_app/__init__.py
from flask import Flask from flask.ext.sqlalchemy import SQLAlchemy app = Flask(__name__) app.config['SQLALCHEMY_DATABASE_URI'] = 'sqlite:////tmp/test.db' db = SQLAlchemy(app) from my_app.catalog.views import catalog app.register_blueprint(catalog) db.create_all()
In the file above, the application has been configured with the initialisation of the extensions and finally creation of the database. The last statement creates a new database at the location provided against SQLALCHEMY_DATABASE_URI if a database does not already exist at that location, otherwise it loads the application with the same database.
flask_app/my_app/catalog/models.py
from my_app import db
class Product(db.Model):
id = db.Column(db.Integer, primary_key=True)
name = db.Column(db.String(255))
price = db.Column(db.Float(asdecimal=True))
def __init__(self, name, price):
self.name = name
self.price = price
def __repr__(self):
return '<Product %d>' % self.id
In the file above, I have created a very trivial model for storing the name and price of a Product. This will create a table in SQLite corresponding to the details provided in the model.
flask_app/my_app/catalog/views.py
import json
from flask import request, jsonify, Blueprint, abort
from flask.views import MethodView
from my_app import db, app
from my_app.catalog.models import Product
catalog = Blueprint('catalog', __name__)
@catalog.route('/')
@catalog.route('/home')
def home():
return "Welcome to the Catalog Home."
class ProductView(MethodView):
def get(self, id=None, page=1):
if not id:
products = Product.query.paginate(page, 10).items
res = {}
for product in products:
res[product.id] = {
'name': product.name,
'price': str(product.price),
}
else:
product = Product.query.filter_by(id=id).first()
if not product:
abort(404)
res = {
'name': product.name,
'price': str(product.price),
}
return jsonify(res)
def post(self):
name = request.form.get('name')
price = request.form.get('price')
product = Product(name, price)
db.session.add(product)
db.session.commit()
return jsonify({product.id: {
'name': product.name,
'price': str(product.price),
}})
def put(self, id):
# Update the record for the provided id
# with the details provided.
return
def delete(self, id):
# Delete the record for the provided id.
return
product_view = ProductView.as_view('product_view')
app.add_url_rule(
'/product/', view_func=product_view, methods=['GET', 'POST']
)
app.add_url_rule(
'/product/<int:id>', view_func=product_view, methods=['GET']
)
The major crux of this tutorial is dealt with in the file above. Flask provides a utility called pluggable views, which allows you to create views in the form of classes instead of normally as functions. Method-based dispatching (MethodView) is an implementation of pluggable views which allows you to write methods corresponding to the HTTP methods in lower case. In the example above, I have written methods get() and post() corresponding to HTTP's GET and POST respectively.
Routing is also implemented in a different manner, in the last few lines of the above file. We can specify the methods that will be supported by any particular rule. Any other HTTP call would be met by Error 405 Method not allowed.
Running the Application
To run the application, execute the script run.py. The contents of this script are:
from my_app import app app.run(debug=True)
Now just execute from the command line:
$ python run.py
To check if the application works, fire up http://127.0.0.1:5000/ in your browser, and a simple screen with a welcome message should greet you.
Testing the RESTful API
To test this API, we can simply make HTTP calls using any of the many available methods. GET calls can be made directly via the browser. POST calls can be made using a Chrome extension like Postman or from the command line using curl, or we can use Python's requests library to do the job for us. I'll use the requests library here for demonstration purposes.
Let's make a GET call first to assure that we don't have any products created yet. As per RESTful API's design, a get call which looks something like /product/ should list all products. Then I will create a couple of products by making POST calls to /product/ with some data. Then a GET call to /product/ should list all the products created. To fetch a specific product, a GET call to /product/<product id> should do the job. Below is a sample of all the calls that can be made using this example.
$ pip install requests
$ python
>>> import requests
>>> r = requests.get('http://localhost:5000/product/')
>>> r.json()
{}
>>> r = requests.post('http://localhost:5000/product/', data={'name': 'iPhone 6s', 'price': 699})
>>> r.json()
{u'1': {u'price': u'699.0000000000', u'name': u'iPhone 6s'}}
>>> r = requests.post('http://localhost:5000/product/', data={'name': 'iPad Pro', 'price': 999})
>>> r.json()
{u'2': {u'price': u'999.0000000000', u'name': u'iPad Pro'}}
>>> r = requests.get('http://localhost:5000/product/')
>>> r.json()
{u'1': {u'price': u'699.0000000000', u'name': u'iPhone 6s'}, u'2': {u'price': u'999.0000000000', u'name': u'iPad Pro'}}
>>> r = requests.get('http://localhost:5000/product/1')
>>> r.json()
{u'price': u'699.0000000000', u'name': u'iPhone 6s'}
Conclusion
In this tutorial, you saw how to create RESTful interfaces all by yourself using Flask's pluggable views utility. This is the most flexible approach while writing REST APIs but involves much more code to be written.
There are extensions which make life a bit easier and automate the implementation of RESTful APIs to a huge extent. I will be covering these in the next couple of parts of this tutorial series.
Comments