]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/urplay.py
2 from __future__
import unicode_literals
4 from .common
import InfoExtractor
12 class URPlayIE(InfoExtractor
):
13 _VALID_URL
= r
'https?://(?:www\.)?ur(?:play|skola)\.se/(?:program|Produkter)/(?P<id>[0-9]+)'
15 'url': 'https://urplay.se/program/203704-ur-samtiden-livet-universum-och-rymdens-markliga-musik-om-vetenskap-kritiskt-tankande-och-motstand',
16 'md5': 'ff5b0c89928f8083c74bbd5099c9292d',
20 'title': 'UR Samtiden - Livet, universum och rymdens märkliga musik : Om vetenskap, kritiskt tänkande och motstånd',
21 'description': 'md5:5344508a52aa78c1ced6c1b8b9e44e9a',
22 'timestamp': 1513292400,
23 'upload_date': '20171214',
26 'url': 'https://urskola.se/Produkter/190031-Tripp-Trapp-Trad-Sovkudde',
30 'title': 'Tripp, Trapp, Träd : Sovkudde',
31 'description': 'md5:b86bffdae04a7e9379d1d7e5947df1d1',
32 'timestamp': 1440086400,
33 'upload_date': '20150820',
36 'url': 'http://urskola.se/Produkter/155794-Smasagor-meankieli-Grodan-i-vida-varlden',
37 'only_matching': True,
40 def _real_extract(self
, url
):
41 video_id
= self
._match
_id
(url
)
42 url
= url
.replace('skola.se/Produkter', 'play.se/program')
43 webpage
= self
._download
_webpage
(url
, video_id
)
44 urplayer_data
= self
._parse
_json
(self
._html
_search
_regex
(
45 r
'data-react-class="routes/Product/components/ProgramContainer/ProgramContainer"[^>]+data-react-props="({.+?})"',
46 webpage
, 'urplayer data'), video_id
)['accessibleEpisodes'][0]
47 episode
= urplayer_data
['title']
49 host
= self
._download
_json
('http://streaming-loadbalancer.ur.se/loadbalancer.json', video_id
)['redirect']
51 urplayer_streams
= urplayer_data
.get('streamingInfo', {})
53 for k
, v
in urplayer_streams
.get('raw', {}).items():
54 if not (k
in ('sd', 'hd') and isinstance(v
, dict)):
56 file_http
= v
.get('location')
58 formats
.extend(self
._extract
_wowza
_formats
(
59 'http://%s/%splaylist.m3u8' % (host
, file_http
),
60 video_id
, skip_protocols
=['f4m', 'rtmp', 'rtsp']))
61 self
._sort
_formats
(formats
)
64 subs
= urplayer_streams
.get("sweComplete", {}).get("tt", {}
).get("location")
66 subtitles
.setdefault('Svenska', []).append({
70 image
= urplayer_data
.get('image') or {}
72 for k
, v
in image
.items():
80 'width': int_or_none(wh
[0]),
81 'height': int_or_none(wh
[1]),
85 series
= urplayer_data
.get('series') or {}
86 series_title
= dict_get(series
, ('seriesTitle', 'title')) or dict_get(urplayer_data
, ('seriesTitle', 'mainTitle'))
90 'subtitles': subtitles
,
91 'title': '%s : %s' % (series_title
, episode
) if series_title
else episode
,
92 'description': urplayer_data
.get('description'),
93 'thumbnails': thumbnails
,
94 'timestamp': unified_timestamp(urplayer_data
.get('publishedAt')),
95 'series': series_title
,
97 'duration': int_or_none(urplayer_data
.get('duration')),
98 'categories': urplayer_data
.get('categories'),
99 'tags': urplayer_data
.get('keywords'),
100 'season': series
.get('label'),
102 'episode_number': int_or_none(urplayer_data
.get('episodeNumber')),