Django Generic JSON Views

Class based generic views that render JSON data.

The django.views.generic package has several generic class based views that render HTML for a model or a set of applications and provide an API to use those views and modify querysets and context data. However, Django recommends that to specify the output of JSON data, you use a mixin or another approach to render_to_response the JSON.

Many AJAX applications, however, have the need for quick views like the ones specified the in the generic package to either serialize a model or to provide other context data in the View logic. These generic views allow that, and correctly render JSON data with the correct HTTP Headers.

More documentation can be found at Read the Docs: Django Generic JSON Views Documentation

Installation

The easiest way to install this library is to use pip as follows:

$ pip install django-generic-json-views

Alternatively you can download the source code and run the following:

$ python setup.py install

You can import from the json_views.views module.

Otherwise copy the views.py file to a utils directory in your Django Project and import it correctly from there.

Basic Usage

There are several provided classes that work right off the bat:

• JSONDataView
• JSONDetailView
• JSONListView
• PaginatedJSONListView
• JSONFormView

These classes all provide the following functionality:

1. Render the output of get_context_data as a JSON object
2. Serialize Django Models and Querysets with a Lazy Encoder
3. Correctly add the application/json HTTP header
4. Remove duplicate objects and middleware context built into Django

The most simple example is the JSONDataView which will allow you to output generic JSON data as needed:

from datetime import datetime
from json_views.views import JSONDataView

class JSONTime(JSONDataView):

    def get_context_data(self, **kwargs):
        context = super(JSONTime, self).get_context_data(**kwargs)
        context['current_time'] = datetime.now().strftime("%c")
        return context

In order to render the detail view of a model as a JSON serialization, use the JSONDetailView -- note that this view can be used for both serialization and deserialization as noted in the section below.

from json_views.views import JSONDetailView

class CustomerJSON(JSONDetailView):

    model = Customer

Many JSON views are centered around lists, hence the JSONListView and the PaginatedJSONListView.

from json_views.views import JSONListView

class ProductsJSON(JSONListView):

    model = Product

The PaginatedJSONListView looks up the special parameter page=n to decide which part of the list to return. It also removes the Django paginator object and returns the following data:

{
    'pages': Number,
    'count': Number,
    'is_paginated': Bool,
    'per_page': Number,
    'current': Number,
    'object_list': List
}

An example of using this view is:

from json_views.views import PaginatedJSONListView

class PaginatedProducts(PaginatedJSONListView):

    paginate_by = 10
    count_query = 'count'
    count_only  = False

    def get_querset(self):
        return Product.objects.filter(instores=True)

count_query specifies the name of the property to pass in as kwargs in a GET request that will force the view to only return the total number of expected items, and not fetch them from the db. E.g. if you use /?count=true in your url, the view will respond only with the total number.

Finally the JSONFormView accepts incoming JSON data to populate a form rather than waiting for HTTP POST data, this is how you would do AJAX login forms. Note that GET requests are not allowed.

from json_views.views import JSONFormView

class LoginForm(JSONFormView):

    form = LoginForm

In the form view, a parameter success is passed back, as either True or False. This can be customized by overriding get_context_data.

Serialization & Deserialization

The lazy encoder attempts to serialze the model into a dictionary or other compatible JSON-Python type, and does so in a recursive fashion, continuing to serialize objects, the default serialization is this:

1. If the object is an iterable- construct a list
2. If the object extends ModelBase attempt to call a serialize method
3. Utilize force_unicode to force the object to unicode data.

Therefore if your Models provide a serialize method that returns a dictionary, you can very easily control what fields and what format the model is serialized to.

Additionally you can specify a @classmethod deserialize that accepts as input a dictionary and returns an instance of the Model.

Here is an example of the serialization of a UserProfile, as an extension of the django.contrib.auth package.

class UserProfile(models.Model):

    user  = models.OneToOneField( 'django.contrib.auth.User', editable=False, related_name='profile' )
    bio   = models.CharField( max_length=255, null=True, blank=True, default=None )
    birth = models.DateField( )

    def serialize(self):
        return {
            'username': self.user.username,
            'email': self.user.email,
            'name': " ".join((self.user.first_name, self.user.last_name)),
            'birthday': self.birth.strftime("%d %M, %Y"),
        }

    @classmethod
    def deserialize(klass, data):
        username = data.get('username', None)
        if username:
            try:
                return klass.objects.get(username=username)
            except klass.DoesNotExist:
                pass
        return klass(**data)

Pushing to PyPi

A couple of notes on pushing this software to PyPi:

1. Version Bump
   1. Change the version in __init__.py
   2. Change the version in tests
   3. Change the download URL in setup.py
2. Tag the release git tag -a v0.x.x
3. Push to Github
4. Register python setup.py register -r pypi
5. Upload python setup.py sdist upload -r pypi

Contributors

Thanks to @philippbosch for helping provide Python 3 support with later versions of Django.

Thanks to @freeworlder who provided Django 1.10 compatibility.