mygpo/data/mimetype.py

   1 from mygpo.api.models import Episode
   2 from collections import defaultdict
   3 import mimetypes
   4
   5 # If 20% of the episodes of a podcast are of a given type,
   6 # then the podcast is considered to be of that type, too
   7 TYPE_THRESHOLD=.2
   8
   9
  10 _ = lambda s: s
  11
  12 CONTENT_TYPES = (_('image'), _('audio'), _('video'))
  13
  14 def get_podcast_types(podcast):
  15     """Returns the types of a podcast
  16
  17     A podcast is considered to be of a given types if the ratio of episodes that are of that type equals TYPE_THRESHOLD
  18     """
  19     episodes = Episode.objects.filter(podcast=podcast, mimetype__isnull=False)
  20     types = defaultdict()
  21     for e in episodes:
  22         t = get_type(e.mimetype)
  23         types[t] = types.get(t, 0) + 1
  24
  25     max_episodes = sum(types.itervalues())
  26     l = list(types.iteritems())
  27     l.sort(key=lambda x: x[1], reverse=True)
  28
  29     return [x[0] for x in filter(lambda x: max_episodes / float(x[1]) >= TYPE_THRESHOLD, l)]
  30
  31
  32 def get_type(mimetype):
  33     """Returns the simplified type for the given mimetype
  34
  35     All "wanted" mimetypes are mapped to one of audio/video/image
  36     Everything else returns None
  37     """
  38     if not mimetype:
  39         return None
  40
  41     if '/' in mimetype:
  42         category, type = mimetype.split('/', 1)
  43         if category in ('audio', 'video', 'image'):
  44             return category
  45         elif type == 'ogg':
  46             return 'audio'
  47         elif type == 'x-youtube':
  48             return 'video'
  49     return None
  50
  51 def check_mimetype(mimetype):
  52     """Checks if the given mimetype can be processed by mygpo
  53     """
  54     if '/' in mimetype:
  55         category, type = mimetype.split('/', 1)
  56         if category in ('audio', 'video', 'image'):
  57             return True
  58
  59         # application/ogg is a valid mime type for Ogg files
  60         # but we do not want to accept all files with application category
  61         if type in ('ogg', ):
  62             return True
  63
  64         return False
  65     else:
  66         return False
  67
  68
  69 def get_mimetype(mimetype, url):
  70     """Returns the mimetype; if None is given it tries to guess it"""
  71
  72     if not mimetype:
  73         mimetype, _encoding = mimetypes.guess_type(url)
  74
  75     return mimetype
  76