mygpo/data/mimetype.py

   1 from collections import defaultdict
   2
   3 import mimetypes
   4
   5 from django.utils.translation import ugettext_lazy as _
   6
   7
   8 # If 20% of the episodes of a podcast are of a given type,
   9 # then the podcast is considered to be of that type, too
  10 TYPE_THRESHOLD=.2
  11
  12
  13 CONTENT_TYPES = (_('image'), _('audio'), _('video'))
  14
  15 def get_podcast_types(episodes):
  16     """Returns the types of a podcast
  17
  18     A podcast is considered to be of a given types if the ratio of episodes that are of that type equals TYPE_THRESHOLD
  19     """
  20     has_mimetype = lambda e: e.mimetypes
  21     episodes = filter(has_mimetype, episodes)
  22     types = defaultdict()
  23     for e in episodes:
  24         for mimetype in e.mimetypes:
  25             t = get_type(mimetype)
  26             if not t:
  27                 continue
  28             types[t] = types.get(t, 0) + 1
  29
  30     max_episodes = sum(types.values())
  31     l = list(types.items())
  32     l.sort(key=lambda x: x[1], reverse=True)
  33
  34     return [x[0] for x in
  35         filter(lambda x: max_episodes / float(x[1]) >= TYPE_THRESHOLD, l)]
  36
  37
  38 def get_type(mimetype):
  39     """Returns the simplified type for the given mimetype
  40
  41     All "wanted" mimetypes are mapped to one of audio/video/image
  42     Everything else returns None
  43
  44     >>> get_type('audio/mpeg3')
  45     'audio'
  46
  47     >>> get_type('video/mpeg')
  48     'video'
  49
  50     >>> get_type('image/jpeg')
  51     'image'
  52
  53     >>> get_type('application/ogg')
  54     'audio'
  55
  56     >>> get_type('application/x-youtube')
  57     'video'
  58
  59     >>> get_type('application/x-vimeo')
  60     'video'
  61
  62     >>> get_type('application/octet-stream') == None
  63     True
  64
  65     >>> get_type('') == None
  66     True
  67
  68     >>> get_type('music') == None
  69     True
  70     """
  71     if not mimetype:
  72         return None
  73
  74     if '/' not in mimetype:
  75         return None
  76
  77     category, type = mimetype.split('/', 1)
  78     if category in ('audio', 'video', 'image'):
  79         return category
  80     elif type == 'ogg':
  81         return 'audio'
  82     elif type == 'x-youtube':
  83         return 'video'
  84     elif type == 'x-vimeo':
  85         return 'video'
  86
  87
  88 def get_mimetype(mimetype, url):
  89     """Returns the mimetype; if None is given it tries to guess it"""
  90
  91     if not mimetype:
  92         mimetype, _encoding = mimetypes.guess_type(url)
  93
  94     return mimetype