Creating API endpointsΒΆ

To use this extension, you must have defined your database models using either SQLAlchemy or Flask-SQLALchemy.

The basic setup for Flask-SQLAlchemy is the same. First, create your flask.Flask object, flask.ext.sqlalchemy.SQLAlchemy object, and model classes as usual but with the following two (reasonable) restrictions on models:

  1. They must have a primary key column of type sqlalchemy.Integer or type sqlalchemy.Unicode.
  2. They must have an __init__ method which accepts keyword arguments for all columns (the constructor in flask.ext.sqlalchemy.SQLAlchemy.Model supplies such a method, so you don’t need to declare a new one).
import flask
import flask.ext.sqlalchemy

app = flask.Flask(__name__)
app.config['SQLALCHEMY_DATABASE_URI'] = 'sqlite:////tmp/test.db'
db = flask.ext.sqlalchemy.SQLAlchemy(app)

class Person(db.Model):
    id = db.Column(db.Integer, primary_key=True)
    name = db.Column(db.Unicode, unique=True)
    birth_date = db.Column(db.Date)
    computers = db.relationship('Computer',

class Computer(db.Model):
    id = db.Column(db.Integer, primary_key=True)
    name = db.Column(db.Unicode, unique=True)
    vendor = db.Column(db.Unicode)
    owner_id = db.Column(db.Integer, db.ForeignKey(''))
    purchase_time = db.Column(db.DateTime)


If you are using pure SQLAlchemy:

from flask import Flask
from sqlalchemy import Column, Date, DateTime, Float, Integer, Unicode
from sqlalchemy import ForeignKey
from sqlalchemy import create_engine
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy.orm import backref, relationship
from sqlalchemy.orm import scoped_session, sessionmaker

app = Flask(__name__)
engine = create_engine('sqlite:////tmp/testdb.sqlite', convert_unicode=True)
Session = sessionmaker(autocommit=False, autoflush=False, bind=engine)
mysession = scoped_session(Session)

Base = declarative_base()
Base.metadata.bind = engine

class Computer(Base):
    __tablename__ = 'computer'
    id = Column(Integer, primary_key=True)
    name = Column(Unicode, unique=True)
    vendor = Column(Unicode)
    buy_date = Column(DateTime)
    owner_id = Column(Integer, ForeignKey(''))

class Person(Base):
    __tablename__ = 'person'
    id = Column(Integer, primary_key=True)
    name = Column(Unicode, unique=True)
    age = Column(Float)
    other = Column(Float)
    birth_date = Column(Date)
    computers = relationship('Computer',
                             backref=backref('owner', lazy='dynamic'))



Attributes of these entities must not have a name containing two underscores. For example, this class definition is no good:

class Person(db.Model):
    __mysecretfield = db.Column(db.Unicode)

This restriction is necessary because the search feature (see Making search queries) uses double underscores as a separator. This may change in future versions.

Second, instantiate a flask.ext.restless.APIManager object with the Flask and SQLAlchemy objects:

import flask.ext.restless

manager = flask.ext.restless.APIManager(app, flask_sqlalchemy_db=db)

Or if you are using pure SQLAlchemy, specify the session you created above instead:

manager = flask.ext.restless.APIManager(app, session=mysession)

Third, create the API endpoints which will be accessible to web clients:

person_blueprint = manager.create_api(Person,
                                      methods=['GET', 'POST', 'DELETE'])
computer_blueprint = manager.create_api(Computer)

Note that you can specify which HTTP methods are available for each API endpoint. There are several more customization options; for more information, see Customizing the ReSTful interface.

Due to the design of Flask, these APIs must be created before your application handles any requests. The return value of APIManager.create_api() is the blueprint in which the endpoints for the specified database model live. The blueprint has already been registered on the Flask application, so you do not need to register it yourself. It is provided so that you can examine its attributes, but if you don’t need it then just ignore it:

manager.create_api(Person, methods=['GET', 'POST', 'DELETE'])

If you wish to create the blueprint for the API without registering it (for example, if you wish to register it later in your code), use the APIManager.create_api_blueprint() method instead:

blueprint = manager.create_api_blueprint(Person, methods=['GET', 'POST'])
# later...

By default, the API for Person, in the above code samples, will be accessible at http://<host>:<port>/api/person, where the person part of the URL is the value of Person.__tablename__:

>>> import json  # import simplejson as json, if on Python 2.5
>>> import requests  # python-requests is installable from PyPI...
>>> newperson = {'name': u'Lincoln', 'age': 23}
>>> r ='/api/person', data=json.dumps(newperson),
...                   headers={'content-type': 'application/json'})
>>> r.status_code, r.headers['content-type'],
(201, 'application/json', '{"id": 1}')
>>> newid = json.loads(['id']
>>> r = requests.get('/api/person/%s' % newid,
...                  headers={'content-type': 'application/json'})
>>> r.status_code, r.headers['content-type']
(200, 'application/json')
  "other": null,
  "name": "Lincoln",
  "birth_date": null,
  "age": 23.0,
  "computers": [],
  "id": 1

If the primary key is a Unicode instead of an Integer, the instances will be accessible at URL endpoints like http://<host>:<port>/api/person/foo instead of http://<host>:<port>/api/person/1.