_ Git - youtube-dl/blob - youtube_dl/extractor/funnyordie.py

   1 from __future__ import unicode_literals
   2
   3 import json
   4 import re
   5
   6 from .common import InfoExtractor
   7
   8
   9 class FunnyOrDieIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?funnyordie\.com/(?P<type>embed|videos)/(?P<id>[0-9a-f]+)(?:$|[?#/])'
  11     _TEST = {
  12         'url': 'http://www.funnyordie.com/videos/0732f586d7/heart-shaped-box-literal-video-version',
  13         'file': '0732f586d7.mp4',
  14         'md5': 'f647e9e90064b53b6e046e75d0241fbd',
  15         'info_dict': {
  16             'description': ('Lyrics changed to match the video. Spoken cameo '
  17                 'by Obscurus Lupa (from ThatGuyWithTheGlasses.com). Based on a '
  18                 'concept by Dustin McLean (DustFilms.com). Performed, edited, '
  19                 'and written by David A. Scott.'),
  20             'title': 'Heart-Shaped Box: Literal Video Version',
  21         },
  22     }
  23
  24     def _real_extract(self, url):
  25         mobj = re.match(self._VALID_URL, url)
  26
  27         video_id = mobj.group('id')
  28         webpage = self._download_webpage(url, video_id)
  29
  30         video_url = self._search_regex(
  31             [r'type="video/mp4" src="(.*?)"', r'src="([^>]*?)" type=\'video/mp4\''],
  32             webpage, 'video URL', flags=re.DOTALL)
  33
  34         if mobj.group('type') == 'embed':
  35             post_json = self._search_regex(
  36                 r'fb_post\s*=\s*(\{.*?\});', webpage, 'post details')
  37             post = json.loads(post_json)
  38             title = post['name']
  39             description = post.get('description')
  40             thumbnail = post.get('picture')
  41         else:
  42             title = self._og_search_title(webpage)
  43             description = self._og_search_description(webpage)
  44             thumbnail = None
  45
  46         return {
  47             'id': video_id,
  48             'url': video_url,
  49             'ext': 'mp4',
  50             'title': title,
  51             'description': description,
  52             'thumbnail': thumbnail,
  53         }