gistfile1.py

class SearchEngineReferrerMiddleware(object):
    """
    Usage example:
    ==============
    
    {% if request.session.search_engine %}
        You searched for {{ request.session.search_term }} using {{ request.session.search_engine }}.
    {% endif %}
    """
    SEARCH_PARAMS = {
        'AltaVista': 'q',
        'Ask': 'q',
        'Google': 'q',
        'Live': 'q',
        'Lycos': 'query',
        'MSN': 'q',
        'Yahoo': 'p',
        'Cuil': 'q',
    }

    NETWORK_RE = r"""^
        (?P<subdomain>[-.a-z\d]+\.)?
        (?P<engine>%s)
        (?P<top_level>(?:\.[a-z]{2,3}){1,2})
        (?P<port>:\d+)?
        $(?ix)"""
    
    @classmethod
    def parse_search(cls, url):
        
        """
        Extract the search engine, domain, and search term from `url`
        and return them as (engine, domain, term). For example,
        ('Google', 'www.google.co.uk', 'django framework'). Note that
        the search term will be converted to lowercase and have normalized
        spaces.

        The first tuple item will be None if the referrer is not a
        search engine.
        """
        try:
            parsed = urlparse.urlsplit(url)
            network = parsed[1]
            query = parsed[3]
        except (AttributeError, IndexError):
            return (None, None, None)
        for engine, param in cls.SEARCH_PARAMS.iteritems():
            match = re.match(cls.NETWORK_RE % engine, network)
            if match and match.group(2):
                term = cgi.parse_qs(query).get(param)
                if term and term[0]:
                    term = ' '.join(term[0].split()).lower()
                    return (engine, network, term)
        return (None, network, None)
    
    
    def process_request(self, request):
        referrer = request.META.get('HTTP_REFERER')
        engine, domain, term = self.parse_search(referrer)
        request.session['search_engine'] = engine
        request.session['search_domain'] = domain
        request.session['search_term'] = term