February 25, 2010

How to Store Arbitrary Data in a Django Model

I have a number of different places where I have wanted to store arbitrary data along with structured data in a Django model. This is data that I wouldn't necessarily care to use a value to retrieve data but when displaying or working with a record or records of this data, being able to have this data available in a manner that I didn't have to parse was nice.

My solution was to serialize/deserialize in and out of JSON using simplejson and a Field class that derives from a TextField. i think it is easier to just read the code and the example of how to use it in the gist below than for me to continue with my rambles.

The Code

from django.db import models
from django.utils import simplejson as json
from django.conf import settings
from datetime import datetime


class JSONEncoder(json.JSONEncoder):
    def default(self, obj):
        if isinstance(obj, datetime):
            return obj.strftime('%Y-%m-%d %H:%M:%S')
        elif isinstance(obj, datetime.date):
            return obj.strftime('%Y-%m-%d')
        elif isinstance(obj, datetime.time):
            return obj.strftime('%H:%M:%S')
        return json.JSONEncoder.default(self, obj)


class JSONField(models.TextField):
    def _dumps(self, data):
        return JSONEncoder().encode(data)

    def _loads(self, str):
        return json.loads(str, encoding=settings.DEFAULT_CHARSET)

    def db_type(self):
        return 'text'

    def pre_save(self, model_instance, add):
        value = getattr(model_instance, self.attname, None)
        return self._dumps(value)

    def contribute_to_class(self, cls, name):
        self.class_name = cls
        super(JSONField, self).contribute_to_class(cls, name)
        models.signals.post_init.connect(self.post_init)

        def get_json(model_instance):
            return self._dumps(getattr(model_instance, self.attname, None))
        setattr(cls, 'get_%s_json' % self.name, get_json)

        def set_json(model_instance, json):
            return setattr(model_instance, self.attname, self._loads(json))
        setattr(cls, 'set_%s_json' % self.name, set_json)

    def post_init(self, **kwargs):
        if 'sender' in kwargs and 'instance' in kwargs:
            if kwargs['sender'] == self.class_name and hasattr(kwargs['instance'], self.attname):
                value = self.value_from_object(kwargs['instance'])
                if (value):
                    setattr(kwargs['instance'], self.attname, self._loads(value))
                else:
                    setattr(kwargs['instance'], self.attname, None)


class SampleModel(models.Model):
    first_name = models.CharField(max_length=50)
    last_name = models.CharField(max_length=50)
    data = JSONField(null=True, blank=True)


sample = SampleModel(first_name='Patrick', last_name='Altman')
sample.data = {'pets': None, 'children': ['Benjamin', 'Clare', 'Joshua'], 'some_date': datetime(2010, 02, 01)}
sample.save()

UPDATE: It wasn't clear in this post. This wasn't my code, as I had previously pointed out. Just wanted to be clear that I have found this solution useful and have used the snippet so much that I feel like it's mine -- but it's not. :)