]> jfr.im git - yt-dlp.git/blame - youtube_dl/extractor/puls4.py
[thisoldhouse] Add new extractor(closes #10837)
[yt-dlp.git] / youtube_dl / extractor / puls4.py
CommitLineData
7862ad88 1# -*- coding: utf-8 -*-
7862ad88
H
2from __future__ import unicode_literals
3
9d54b02b 4from .prosiebensat1 import ProSiebenSat1BaseIE
6c87c2ee 5from ..utils import (
6c87c2ee 6 unified_strdate,
9d54b02b
RA
7 parse_duration,
8 compat_str,
6c87c2ee 9)
7862ad88
H
10
11
9d54b02b
RA
12class Puls4IE(ProSiebenSat1BaseIE):
13 _VALID_URL = r'https?://(?:www\.)?puls4\.com/(?P<id>(?:[^/]+/)*?videos/[^?#]+)'
7862ad88 14 _TESTS = [{
9d54b02b
RA
15 'url': 'http://www.puls4.com/2-minuten-2-millionen/staffel-3/videos/2min2miotalk/Tobias-Homberger-von-myclubs-im-2min2miotalk-118118',
16 'md5': 'fd3c6b0903ac72c9d004f04bc6bb3e03',
7862ad88 17 'info_dict': {
9d54b02b
RA
18 'id': '118118',
19 'ext': 'flv',
20 'title': 'Tobias Homberger von myclubs im #2min2miotalk',
21 'description': 'md5:f9def7c5e8745d6026d8885487d91955',
22 'upload_date': '20160830',
6c87c2ee
S
23 'uploader': 'PULS_4',
24 },
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