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

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3 import re
   4
   5 from .common import InfoExtractor
   6 from .brightcove import BrightcoveIE
   7 from .discovery import DiscoveryIE
   8
   9
  10 class TlcIE(DiscoveryIE):
  11     IE_NAME = 'tlc.com'
  12     _VALID_URL = r'http://www\.tlc\.com\/[a-zA-Z0-9\-]*/[a-zA-Z0-9\-]*/videos/(?P<id>[a-zA-Z0-9\-]*)(.htm)?'
  13
  14     _TEST = {
  15         'url': 'http://www.tlc.com/tv-shows/cake-boss/videos/too-big-to-fly.htm',
  16         'md5': 'c4038f4a9b44d0b5d74caaa64ed2a01a',
  17         'info_dict': {
  18             'id': '853232',
  19             'ext': 'mp4',
  20             'title': 'Cake Boss: Too Big to Fly',
  21             'description': 'Buddy has taken on a high flying task.',
  22             'duration': 119,
  23         },
  24     }
  25
  26
  27 class TlcDeIE(InfoExtractor):
  28     IE_NAME = 'tlc.de'
  29     _VALID_URL = r'http://www\.tlc\.de/sendungen/[^/]+/videos/(?P<title>[^/?]+)'
  30
  31     _TEST = {
  32         'url': 'http://www.tlc.de/sendungen/breaking-amish/videos/#3235167922001',
  33         'info_dict': {
  34             'id': '3235167922001',
  35             'ext': 'mp4',
  36             'title': 'Breaking Amish: Die Welt da draußen',
  37             'uploader': 'Discovery Networks - Germany',
  38             'description': 'Vier Amische und eine Mennonitin wagen in New York'
  39                 '  den Sprung in ein komplett anderes Leben. Begleitet sie auf'
  40                 ' ihrem spannenden Weg.',
  41         },
  42     }
  43
  44     def _real_extract(self, url):
  45         mobj = re.match(self._VALID_URL, url)
  46         title = mobj.group('title')
  47         webpage = self._download_webpage(url, title)
  48         iframe_url = self._search_regex(
  49             '<iframe src="(http://www\.tlc\.de/wp-content/.+?)"', webpage,
  50             'iframe url')
  51         # Otherwise we don't get the correct 'BrightcoveExperience' element,
  52         # example: http://www.tlc.de/sendungen/cake-boss/videos/cake-boss-cannoli-drama/
  53         iframe_url = iframe_url.replace('.htm?', '.php?')
  54         iframe = self._download_webpage(iframe_url, title)
  55
  56         return {
  57             '_type': 'url',
  58             'url': BrightcoveIE._extract_brightcove_url(iframe),
  59             'ie': BrightcoveIE.ie_key(),
  60         }