Cacheable URL retriever for Django

cachereq.py

"""
(c) 2015 github.com/miraculixx
MIT License

Cacheable URL getter for  Django

Uses Django's cache framework to cache responses with the same URL.

Usage:
   url = 'http://www.domain.com/static/content'
   req = CachedRequest()
   resp = req.get(url)
   # 1st time resp.was_cached == False
   resp = req.get(url)
   # 2nd time resp.was_cached == True
   
   req = CachedRequest(cache_timeout=15) # in seconds
   resp = req.get(url)
   # 1st time resp.was_cached == False
   sleep(20)
   resp = req.get(url)
   # 2nd time resp.was_cached == False
"""
import logging
import urllib
import urllib2
import urlparse

from django.core.cache import get_cache

logger = logging.getLogger(__name__)

class CachedRequest(object):  
    def __init__(self, url, cache_timeout=None):
      self.url = url
      self.was_cached = False
      self.cache_timeout = cache_timeout or 24 * 3600
      try:
          self.cache = get_cache(self.config['cache'])
          logger.debug('GeoAddress using cache %s' % self.cache)
      except KeyError:
          self.cache = None
          logger.debug('No cache enabled')
      
    def get(self, url=None):
        """
        will GET call the url. if a cache was set in GeoAddress.cache
        (usually by 'cache' in GEOADDRESS_CONFIG), will get the response
        that was previously returned for the same query
        """
        url = url or self.url
        if self.cache:
            key = hashlib.md5(url).hexdigest()
            response =self.cache.get(key)
            if response:
                self.was_cached = True
                logger.debug('served url=%s from cache' % url)
                return response
        #opener = urllib2.build_opener(urllib2.HTTPHandler(debuglevel=1))
        opener = urllib2.build_opener(urllib2.HTTPHandler())
        response = opener.open(url).read()
        logger.debug('retrieved url=%s' % url)
        self.was_cached = False
        if self.cache:
            key = hashlib.md5(url).hexdigest()
            # cache response for 24 hours
            self.cache.set(key, response, self.cache_timeout)
            logger.debug('cached url=%s' % url)
        return response