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 BrightcoveLegacyIE
   7 from ..compat import compat_urlparse
   8
   9
  10 class TlcDeIE(InfoExtractor):
  11     IE_NAME = 'tlc.de'
  12     _VALID_URL = r'http://www\.tlc\.de/sendungen/[^/]+/videos/(?P<title>[^/?]+)'
  13
  14     _TEST = {
  15         'url': 'http://www.tlc.de/sendungen/breaking-amish/videos/#3235167922001',
  16         'info_dict': {
  17             'id': '3235167922001',
  18             'ext': 'mp4',
  19             'title': 'Breaking Amish: Die Welt da draußen',
  20             'uploader': 'Discovery Networks - Germany',
  21             'description': (
  22                 'Vier Amische und eine Mennonitin wagen in New York'
  23                 '  den Sprung in ein komplett anderes Leben. Begleitet sie auf'
  24                 ' ihrem spannenden Weg.'),
  25         },
  26     }
  27
  28     def _real_extract(self, url):
  29         mobj = re.match(self._VALID_URL, url)
  30         title = mobj.group('title')
  31         webpage = self._download_webpage(url, title)
  32         iframe_url = self._search_regex(
  33             '<iframe src="(http://www\.tlc\.de/wp-content/.+?)"', webpage,
  34             'iframe url')
  35         # Otherwise we don't get the correct 'BrightcoveExperience' element,
  36         # example: http://www.tlc.de/sendungen/cake-boss/videos/cake-boss-cannoli-drama/
  37         iframe_url = iframe_url.replace('.htm?', '.php?')
  38         url_fragment = compat_urlparse.urlparse(url).fragment
  39         if url_fragment:
  40             # Since the fragment is not send to the server, we always get the same iframe
  41             iframe_url = re.sub(r'playlist=(\d+)', 'playlist=%s' % url_fragment, iframe_url)
  42         iframe = self._download_webpage(iframe_url, title)
  43
  44         return {
  45             '_type': 'url',
  46             'url': BrightcoveLegacyIE._extract_brightcove_url(iframe),
  47             'ie': BrightcoveLegacyIE.ie_key(),
  48         }