2 from __future__ import unicode_literals
6 from .common import InfoExtractor
9 class TeleTaskIE(InfoExtractor):
10 _VALID_URL = r'http?://(?:www\.)?tele-task\.de/archive/video/html5/(?P<id>[0-9]+)/'
12 'url': 'http://www.tele-task.de/archive/video/html5/26168/',
14 'title': 'Duplicate Detection',
17 'md5': '290ef69fb2792e481169c3958dbfbd57',
19 'title': 'Duplicate Detection',
20 'upload_date': '20141218',
21 'id': 'speaker_26168',
26 'md5': 'e1e7218c5f0e4790015a437fcf6c71b4',
28 'title': 'Duplicate Detection',
29 'upload_date': '20141218',
36 def _real_extract(self, url):
37 lecture_id = self._match_id(url)
38 webpage = self._download_webpage(url, lecture_id)
40 title = self._html_search_regex(
41 r'itemprop="name">([^"]+)</a>', webpage, 'title')
42 url_speaker = self._html_search_regex(
43 r'class="speaker".*?src="([^"]+)"', webpage, 'video_url_speaker', flags=re.DOTALL)
44 url_slides = self._html_search_regex(
45 r'class="slides".*?src="([^"]+)"', webpage, 'video_url_slides', flags=re.DOTALL)
46 date = self._html_search_regex(
47 r'<td class="label">Date:</td><td>([^"]+)</td>', webpage, 'date')
48 date = datetime.datetime.strptime(date, '%d.%m.%Y').strftime('%Y%m%d')
53 'id': "speaker_"+lecture_id,
59 'id': "slides_"+lecture_id,