]> jfr.im git - yt-dlp.git/blame - yt_dlp/extractor/puls4.py
[ie/orf:on] Improve extraction (#9677)
[yt-dlp.git] / yt_dlp / extractor / puls4.py
CommitLineData
9d54b02b 1from .prosiebensat1 import ProSiebenSat1BaseIE
14f25df2 2from ..compat import compat_str
3from ..utils import parse_duration, unified_strdate
7862ad88
H
4
5
9d54b02b 6class Puls4IE(ProSiebenSat1BaseIE):
74394b5e 7 _VALID_URL = r'https?://(?:www\.)?puls4\.com/(?P<id>[^?#&]+)'
7862ad88 8 _TESTS = [{
9d54b02b
RA
9 'url': 'http://www.puls4.com/2-minuten-2-millionen/staffel-3/videos/2min2miotalk/Tobias-Homberger-von-myclubs-im-2min2miotalk-118118',
10 'md5': 'fd3c6b0903ac72c9d004f04bc6bb3e03',
7862ad88 11 'info_dict': {
9d54b02b
RA
12 'id': '118118',
13 'ext': 'flv',
14 'title': 'Tobias Homberger von myclubs im #2min2miotalk',
15 'description': 'md5:f9def7c5e8745d6026d8885487d91955',
16 'upload_date': '20160830',
6c87c2ee
S
17 'uploader': 'PULS_4',
18 },
74394b5e
S
19 }, {
20 'url': 'http://www.puls4.com/pro-und-contra/wer-wird-prasident/Ganze-Folgen/Wer-wird-Praesident.-Norbert-Hofer',
21 'only_matching': True,
22 }, {
23 'url': 'http://www.puls4.com/pro-und-contra/wer-wird-prasident/Ganze-Folgen/Wer-wird-Praesident-Analyse-des-Interviews-mit-Norbert-Hofer-416598',
24 'only_matching': True,
6c87c2ee 25 }]
9d54b02b
RA
26 _TOKEN = 'puls4'
27 _SALT = '01!kaNgaiNgah1Ie4AeSha'
28 _CLIENT_NAME = ''
7862ad88
H
29
30 def _real_extract(self, url):
9d54b02b
RA
31 path = self._match_id(url)
32 content_path = self._download_json(
33 'http://www.puls4.com/api/json-fe/page/' + path, path)['content'][0]['url']
34 media = self._download_json(
35 'http://www.puls4.com' + content_path,
36 content_path)['mediaCurrent']
37 player_content = media['playerContent']
38 info = self._extract_video_info(url, player_content['id'])
39 info.update({
40 'id': compat_str(media['objectId']),
41 'title': player_content['title'],
42 'description': media.get('description'),
43 'thumbnail': media.get('previewLink'),
44 'upload_date': unified_strdate(media.get('date')),
45 'duration': parse_duration(player_content.get('duration')),
46 'episode': player_content.get('episodePartName'),
47 'show': media.get('channel'),
48 'season_id': player_content.get('seasonId'),
49 'uploader': player_content.get('sourceCompany'),
50 })
51 return info