django-wurfl use the python-Levenshtein package. However, the hosting site is defunct, so

john@doe:~# sudo easy_install python-Levenshtein

probably won't work. You can however still get the source code from Michael Noll's website and install it manually.

Put the wurfl folder somewhere on your python path. Then add in your settings file :

# in settings.py
INSTALLED_APPS = (
    ...
    'wurfl',
    ...
)

Create the database tables with syncdb command. Then import WURFL data in the database using the custom command:

john@doe:~/mydjangoproject/# ./manage.py wurfl-load <path/to/wurfl.xml>

django-wurfl provides a middleware, a decorator, and a context processor for you to use in your django powered application.

How to use the middleware :

# Add the middleware in your settings.py file
MIDDLEWARE_CLASSES = (
    ...
    'wurfl.middleware.DeviceMiddleware',
    ...
)

# You can now access the detected device in your views
def myview(request):
    markup_dict = request.device['markup'] # this is a group of
            # capabilities stored represented as a dictionnary

    if request.device['html_wi_w3_xhtmlbasic']:
        # this is a single capability
        # in this case, equivalent to markup_dict['html_wi_w3_xhtmlbasic']
        pass

The decorators allows you to activate the device detection only in specific views, whereas the middleware is application wide.

# In your views
from wurfl.decorators import device

@device
def myview(request):
    if request.device['html_wi_w3_xhtmlbasic']:
        # bla bla bla

And the context processor is helpful to have the device variable directly accessible in your template (if the device has been detected before hand using the middleware or a decorator).

# Add the context processor in your settings.py file
TEMPLATE_CONTEXT_PROCESSORS = (
    ...
    "wurfl.context_processors.device",
    ...
)

# Use the device in your templates
<p>Markup xhtmlbasic : {{ device.html_wi_w3_xhtmlbasic }}</p>

• WURFL_USE_CACHE : if set to True, django-wurfl will use Django cache backend to cache user agents and associated capabilities. That will give a significant performance boost to your mobile application, and is highly recommended. See also WURFL_CACHE_PREFIX and WURFL_CACHE_TIMEOUT. Default to False.
• WURFL_CACHE_PREFIX : Cache key in django-wurfl are user agent's MD5 hexadecimal digest. In order to prevent cache key collision, all keys are prefixed with the string in this setting (default to 'wurfl_').
• WURFL_CACHE_TIMEOUT : timeout (in seconds) for cached user agent capabilities. If set to None, it will use the default timeout of your application. If set to 0, the cached values will never expire. Default to None.

Note:

The cache is not automatically invalidated when you update your WURFL data (patches, or main files), it is your responsibility to do so, or to set a relevant timeout value.

• WURFL_USE_PATCH : if set to True, django-wurfl will take in account the patches to calculate the capabilities of the detected device. The patched capabilities are pre-processed using the wurfl-hybrid command (see below) Default to False.
• WURFL_URL : the url to the XML wurfl file, not used in this version.
• WURFL_UA_PREFIX_MATCHING : if set to True, django-wurfl will search for the device using only a prefix of the user agent in case it can't find the exact user agent in the database. Default to False.
• WURFL_UA_PREFIX_MATCHING_MAX_DISTANCE : when searching with only a prefix of the user_agent, this setting is used to limit the sensitivity of the search. It indicates the maximum Levenshtein Distance that will be tolerated between the original user agent and the search results. Default to 20.

• wurfl-load : load the data from a wurfl file and store it in the database. Takes 1 argument, the path to the WURLF XML file.
• wurfl-hybrid : apply the patch to the current wurfl data and store the result in the hybrid table.
• wurfl-add-patch : create a new patch from an XML file. You can also do that using the administration back office. See the documentation of this command for more information.