Dead Parrot is a RESTful framework with the declarative approach of Django Models in mind. Describing a class and its fields, you can tailor the serialization/deserialization of your model in several formats, including xml and json.
The project has been conceived to allow for custom-tailored format-styles and/or serializers, model-aware data validation and other extensibility points. One key goal for the project is removing the dependency that current tools have on the large Django stack.
Dead Parrot uses a much thinner layer to serve and consume its resources. A standalone web server, based on the CherryPy project is included to serve resources, as well as a client layer based in the Django queryset manager.
It's a tribute to one of my favorite Monty Python's sketch. In the sketch a customer tries to exchange his parrot, that have been bought dead, and then he argues with the seller.
- Provide a plugin-based pool of serializers, so you write yout model once, but deals with {de}serialization in many formats.
- Dead Parrot is a library that reaches both server and client tiers, so that it can be the "seller" as the server providing a RESTful API, and also the "customer" consuming the API.
- Dead Parrot also means to support database access through SQLAlchemy, but for the programmer, it will be transparent, and will feel like coding in Django.
As long as Dead Parrot is also a Monty Python's sketch, its release names are sentences spoken by the actors, the odd numbered releases are "the customer" phrases, and the even ones are from "the seller".
- Release HelloPolly (0.1) - Support all django-like model fields, serialization/deserialization on xml/json
- Release JustResting (0.2) - Embed a CherryPy serving RESTful resources, with a model-like declarative approach to resources, also support a django-like self-consuming its resources.
- Release PassedOn (0.3) - Support SQLAlchemy querying.
Dead Parrot builds on some very well-known python libraries. The consequence is that in order to use it, you need those installed. They can be obtained from the following sites:
- lxml for XML serialization: http://codespeak.net/lxml/
- simplejson for JSON serialization: http://pypi.python.org/pypi/simplejson/
Or, to install those in Debian/Ubuntu:
aptitude install python-lxml
aptitude install python-simplejsonWhat about doing some serialization ? :
>>> from deadparrot import models
>>>
>>> class Car(models.Model):
... brand = models.CharField(max_length=20)
... color = models.CharField(max_length=15)
... website = models.CharField(max_length=0, validate=False)
... def __unicode__(self):
... return '<Car of brand "%s">' % self.brand
...
>>> car1 = Car(brand="Chevy", color="blue")
>>> car2 = Car(brand="OSCar", color="red", website="http://www.theoscarproject.org")
>>> print car1.brand
Chevy
>>> print car1.color
blue
>>> car2.website
u'http://www.theoscarproject.org'
>>> json_cars = '{"Cars": [{"Car": {"website": "http://www.theoscarproject.org", "color": "red", "brand": "OSCar", "id": 2}}]}'
>>> cars = Car.Set().deserialize(json_cars, format="json")
>>> cars
Car.Set([<Car of brand "OSCar">])
>>> car2.serialize(to="xml")
'<Car><color>red</color><website>http://www.theoscarproject.org</website><brand>OSCar</brand></Car>'- In GNU/Linux: make build
- In other systems: python setup.py test && python setup.py build