Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/funnyordie.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6
   7
   8 class FunnyOrDieIE(InfoExtractor):
   9     _VALID_URL = r'^(?:https?://)?(?:www\.)?funnyordie\.com/videos/(?P<id>[0-9a-f]+)/.*$'
  10     _TEST = {
  11         'url': 'http://www.funnyordie.com/videos/0732f586d7/heart-shaped-box-literal-video-version',
  12         'file': '0732f586d7.mp4',
  13         'md5': 'f647e9e90064b53b6e046e75d0241fbd',
  14         'info_dict': {
  15             'description': ('Lyrics changed to match the video. Spoken cameo '
  16                 'by Obscurus Lupa (from ThatGuyWithTheGlasses.com). Based on a '
  17                 'concept by Dustin McLean (DustFilms.com). Performed, edited, '
  18                 'and written by David A. Scott.'),
  19             'title': 'Heart-Shaped Box: Literal Video Version',
  20         },
  21     }
  22
  23     def _real_extract(self, url):
  24         mobj = re.match(self._VALID_URL, url)
  25
  26         video_id = mobj.group('id')
  27         webpage = self._download_webpage(url, video_id)
  28
  29         video_url = self._search_regex(
  30             [r'type="video/mp4" src="(.*?)"', r'src="([^>]*?)" type=\'video/mp4\''],
  31             webpage, 'video URL', flags=re.DOTALL)
  32
  33         return {
  34             'id': video_id,
  35             'url': video_url,
  36             'ext': 'mp4',
  37             'title': self._og_search_title(webpage),
  38             'description': self._og_search_description(webpage),
  39         }