mygpo/data/podcast.py

   1 from collections import Counter
   2 import random
   3 import logging
   4
   5 from django.conf import settings
   6
   7 from mygpo.podcasts.models import Podcast
   8 from mygpo.subscriptions.models import Subscription
   9 from mygpo.pubsub import utils
  10
  11 logger = logging.getLogger(__name__)
  12
  13
  14 def calc_similar_podcasts(podcast, num=20, user_sample=100):
  15     """Get a list of podcasts that seem to be similar to the given one.
  16
  17     The similarity is based on similar subscriptions; for performance
  18     reasons, only a sample of subscribers is considered"""
  19
  20     logger.info("Calculating podcasts similar to {podcast}".format(podcast=podcast))
  21
  22     # get all users that subscribe to this podcast
  23     user_ids = (
  24         Subscription.objects.filter(podcast=podcast)
  25         .order_by("user")
  26         .distinct("user")
  27         .values_list("user", flat=True)
  28     )
  29     logger.info(
  30         "Found {num_subscribers} subscribers, taking a sample "
  31         "of {sample_size}".format(
  32             num_subscribers=len(user_ids), sample_size=user_sample
  33         )
  34     )
  35
  36     # take a random sample of ``user_sample`` subscribers
  37     user_ids = list(user_ids)  # evaluate ValuesQuerySet
  38     random.shuffle(user_ids)
  39     user_ids = user_ids[:user_sample]
  40
  41     # get other podcasts that the user sample subscribes to
  42     podcasts = Counter()
  43     for user_id in user_ids:
  44         subscriptions = (
  45             Podcast.objects.filter(subscription__user__id__in=user_ids)
  46             .distinct("pk")
  47             .exclude(pk=podcast.pk)
  48         )
  49         podcasts.update(Counter(subscriptions))
  50     logger.info(
  51         "Found {num_podcasts}, returning top {num_results}".format(
  52             num_podcasts=len(podcasts), num_results=num
  53         )
  54     )
  55
  56     return podcasts.most_common(num)
  57
  58
  59 def subscribe_at_hub(podcast):
  60     """Tries to subscribe to the given podcast at its hub"""
  61
  62     if not podcast.hub:
  63         return
  64
  65     base_url = settings.DEFAULT_BASE_URL
  66
  67     if not base_url:
  68         logger.warning(
  69             "Could not subscribe to podcast {podcast} "
  70             "at hub {hub} because DEFAULT_BASE_URL is not "
  71             "set.".format(podcast=podcast, hub=podcast.hub)
  72         )
  73         return
  74
  75     logger.info(
  76         "subscribing to {podcast} at {hub}.".format(podcast=podcast, hub=podcast.hub)
  77     )
  78     utils.subscribe(podcast, podcast.url, podcast.hub, base_url)