]>
Commit | Line | Data |
---|---|---|
ee456252 | 1 | from __future__ import unicode_literals |
885e4384 | 2 | |
33b53b60 | 3 | import re |
ee456252 M |
4 | |
5 | from .common import InfoExtractor | |
885e4384 | 6 | from ..utils import unified_strdate |
ee456252 M |
7 | |
8 | ||
9 | class TeleTaskIE(InfoExtractor): | |
885e4384 | 10 | _VALID_URL = r'https?://(?:www\.)?tele-task\.de/archive/video/html5/(?P<id>[0-9]+)' |
ee456252 | 11 | _TEST = { |
33b53b60 | 12 | 'url': 'http://www.tele-task.de/archive/video/html5/26168/', |
ee456252 | 13 | 'info_dict': { |
ee456252 | 14 | 'title': 'Duplicate Detection', |
33b53b60 M |
15 | }, |
16 | 'playlist': [{ | |
17 | 'md5': '290ef69fb2792e481169c3958dbfbd57', | |
18 | 'info_dict': { | |
885e4384 S |
19 | 'id': '26168-speaker', |
20 | 'ext': 'mp4', | |
33b53b60 M |
21 | 'title': 'Duplicate Detection', |
22 | 'upload_date': '20141218', | |
33b53b60 | 23 | } |
885e4384 | 24 | }, { |
33b53b60 M |
25 | 'md5': 'e1e7218c5f0e4790015a437fcf6c71b4', |
26 | 'info_dict': { | |
885e4384 S |
27 | 'id': '26168-slides', |
28 | 'ext': 'mp4', | |
33b53b60 M |
29 | 'title': 'Duplicate Detection', |
30 | 'upload_date': '20141218', | |
33b53b60 M |
31 | } |
32 | }] | |
ee456252 M |
33 | } |
34 | ||
35 | def _real_extract(self, url): | |
33b53b60 | 36 | lecture_id = self._match_id(url) |
885e4384 | 37 | |
33b53b60 M |
38 | webpage = self._download_webpage(url, lecture_id) |
39 | ||
ee456252 | 40 | title = self._html_search_regex( |
885e4384 S |
41 | r'itemprop="name">([^<]+)</a>', webpage, 'title') |
42 | upload_date = unified_strdate(self._html_search_regex( | |
43 | r'Date:</td><td>([^<]+)</td>', webpage, 'date', fatal=False)) | |
ee456252 | 44 | |
33b53b60 | 45 | entries = [{ |
885e4384 S |
46 | 'id': '%s-%s' % (lecture_id, format_id), |
47 | 'url': video_url, | |
33b53b60 | 48 | 'title': title, |
885e4384 S |
49 | 'upload_date': upload_date, |
50 | } for format_id, video_url in re.findall( | |
51 | r'<video class="([^"]+)"[^>]*>\s*<source src="([^"]+)"', webpage)] | |
ee456252 | 52 | |
885e4384 | 53 | return self.playlist_result(entries, lecture_id, title) |