Flask JsonTools

JSON API tools for Flask

Table of Contents

• View Utilities
  • @jsonapi
    • JsonResponse
    • make_json_response()
• FlaskJsonClient
• Class-Based Views
  • MethodView
  • RestfulView

View Utilities

@jsonapi

Decorate a view function that talks JSON.

Such function can return:

• tuples of (response, status[, headers]): to set custom status code and optionally - headers
• Instances of JsonResponse
• The result of helper function make_json_response

Example:

from flask.ext.jsontools import jsonapi

@app.route('/users')
@jsonapi
def list_users():
    return [
        {'id': 1, 'login': 'kolypto'},
        #...
    ]
   
@app.route('/user/<int:id>', methods=['DELETE'])
def delete_user(id):
    return {'error': 'Access denied'}, 403

JsonResponse

Extends flask.Request and encodes the response with JSON. Views decorated with @jsonapi return these objects.

Arguments:

• response: response data
• status: status code. Optional, defaults to 200
• headers: additional headers dict. Optional.
• **kwargs: additional argumets for Response

Methods:

• preprocess_response_data(response): Override to get custom response behavior.
• get_json(): Get the original response data.
• __getitem__(key): Get an item from the response data

The extra methods allows to reuse views:

from flask.ext.jsontools import jsonapi

@app.route('/user', methods=['GET'])
@jsonapi
def list_users():
    return [ { 1: 'first', 2: 'second' } ]
    
@app.route('/user/<int:id>', methods=['GET'])
@jsonapi
def get_user(id):
    return list_users().get_json()[id]  # Long form
    return list_users()[id]  # Shortcut

make_json_response()

Helper function that actually preprocesses view return value into JsonResponse.

Accepts rv as any of:

• tuple of (response, status[, headers])
• Object to encode as JSON

FlaskJsonClient

FlaskJsonClient is a JSON-aware test client: it can post JSON and parse JSON responses into JsonResponse.

from myapplication import Application
from flask.ext.jsontools import FlaskJsonClient

def JsonTest(unittest.TestCase):
    def setUp(self):
        self.app = Application(__name__)
        self.app.test_client_class = FlaskJsonClient
        
    def testCreateUser(self):
        with self.app.test_client() as c:
            rv = c.post('/user/', json={'name': 'kolypto'})
            # rv is JsonResponse
            rv.status_code
            rv.get_json()['user']  # Long form for the previous
            rv['user']  # Shortcut for the previous

Formatting Utils

DynamicJSONEncoder

In python, de-facto standard for encoding objects of custom classes is the __json__() method which returns the representation of the object.

DynamicJSONEncoder is the implementation of this protocol: if an object has the __json__() method, its result if used for the representation.

You'll definitely want to subclass it to support other types, e.g. dates and times:

from flask.ext.jsontools import DynamicJSONEncoder

class ApiJSONEncoder(DynamicJSONEncoder):
    def default(self, o):
        # Custom formats
        if isinstance(o, datetime.datetime):
            return o.isoformat(' ')
        if isinstance(o, datetime.date):
            return o.isoformat()
        if isinstance(o, set):
            return list(o)
        
        # Fallback
        return super(ApiJSONEncoder, self).default(o)

Now, just install the encoder to your Flask:

from flask import Flask

app = Flask(__name__)
app.json_encoder = DynamicJSONEncoder

JsonSerializableBase

Serializing SqlAlchemy models to JSON is a headache: if an attribute is present on an instance, this does not mean it's loaded from the database.

JsonSerializableBase is a mixin for SqlAlchemy Declarative Base that adds a magic __json__() method, compatible with DynamicJSONEncoder. When serializing, it makes sure that entity serialization will never issue additional requests.

Example:

from sqlalchemy.ext.declarative import declarative_base
from flask.ext.jsontools import JsonSerializableBase

Base = declarative_base(cls=(JsonSerializableBase,))

class User(Base):
    #...

Now, you can safely respond with SqlAlchemy models in your JSON views, and jsontools will handle the rest :)

Class-Based Views

Module flask.ext.jsontools.views contains a couple of classes that allow to build class-based views which dispatch to different methods.

In contrast to MethodView, this gives much higher flexibility.

MethodView

Using MethodView class for methods, decorate them with @methodview(), which takes the following arguments:

• methods=(): Iterable of HTTP methods to use with this method.
• ifnset=None: Conditional matching. List of route parameter names that should not be set for this method to match.
• ifset=None: Conditional matching. List of route parameter names that should be set for this method to match.

This allows to map HTTP methods to class methods, and in addition define when individual methods should match.

Quick example:

from flask.ext.jsontools import jsonapi, MethodView, methodview

class UserView(MethodView):
    # Canonical way to specify decorators for class-based views
    decorators = (jsonapi, )

    @methodview
    def list(self):
        """ List users """
        return db.query(User).all()
       
    @methodview
    def get(self, user_id):
        """ Load a user by id """
        return db.query(User).get(user_id)

userview = CrudView.as_view('user')
app.add_url_rule('/user/', view_func=userview)
app.add_url_rule('/user/<int:user_id>', view_func=userview)

Now, GET HTTP method is routed to two different methods depending on conditions. Keep defining more methods to get good routing :)

To simplify the last step of creating the view, there's a helper:

UserView.route_as_view(app, 'user', ('/user/', '/user/<int:user_id>'))

RestfulView

Since MethodView is mostly useful to expose APIs over collections of entities, there is a RESTful helper which automatically decorates some special methods with @methodview.

View method	HTTP method	URL
list()	GET	/
create()	POST	/
get()	GET	/<pk>
replace()	PUT	/<pk>
update()	POST	/<pk>
delete()	DELETE	/<pk>

By subclassing RestfulView and implementing some of these methods, you'll get a complete API endpoint with a single class.

It's also required to define the list of primary key fields by defining the primary_key property:

from flask.ext.jsontools import jsonapi, RestfulView

class User(RestfulView):
    decorators = (jsonapi, )
    primary_key = ('id',)
    
    #region Operation on the collection
    
    def list():
        return db.query(User).all()
    
    def create():
        db.save(user)
        return user
        
    #endregion
    
    #region Operation on entities
    
    def get(id):
        return db.query(User).get(id)
    
    def replace(id):
        db.save(user, id)
    
    def update(id):
        db.save(user)
       
    def delete(id):
        db.delete(user)
    
    #endregion

When a class like this is defined, its metaclass goes through the methods and decorates them with @methodview. This way, list() gets @methodview('GET', ifnset=('id',)), and get() gets @methodview('GET', ifset=('id',)).