1 from .common
import InfoExtractor
13 class URPlayIE(InfoExtractor
):
14 _VALID_URL
= r
'https?://(?:www\.)?ur(?:play|skola)\.se/(?:program|Produkter)/(?P<id>[0-9]+)'
16 'url': 'https://urplay.se/program/203704-ur-samtiden-livet-universum-och-rymdens-markliga-musik-om-vetenskap-kritiskt-tankande-och-motstand',
17 'md5': 'ff5b0c89928f8083c74bbd5099c9292d',
21 'title': 'UR Samtiden - Livet, universum och rymdens märkliga musik : Om vetenskap, kritiskt tänkande och motstånd',
22 'description': 'md5:5344508a52aa78c1ced6c1b8b9e44e9a',
23 'timestamp': 1513292400,
24 'upload_date': '20171214',
25 'series': 'UR Samtiden - Livet, universum och rymdens märkliga musik',
27 'categories': ['Vetenskap & teknik'],
28 'tags': ['Kritiskt tänkande', 'Vetenskap', 'Vetenskaplig verksamhet'],
29 'episode': 'Om vetenskap, kritiskt tänkande och motstånd',
33 'url': 'https://urskola.se/Produkter/190031-Tripp-Trapp-Trad-Sovkudde',
37 'title': 'Tripp, Trapp, Träd : Sovkudde',
38 'description': 'md5:b86bffdae04a7e9379d1d7e5947df1d1',
39 'timestamp': 1440086400,
40 'upload_date': '20150820',
41 'series': 'Tripp, Trapp, Träd',
44 'episode': 'Sovkudde',
47 'url': 'http://urskola.se/Produkter/155794-Smasagor-meankieli-Grodan-i-vida-varlden',
48 'only_matching': True,
51 def _real_extract(self
, url
):
52 video_id
= self
._match
_id
(url
)
53 url
= url
.replace('skola.se/Produkter', 'play.se/program')
54 webpage
= self
._download
_webpage
(url
, video_id
)
55 urplayer_data
= self
._search
_nextjs
_data
(webpage
, video_id
, fatal
=False) or {}
57 urplayer_data
= try_get(urplayer_data
, lambda x
: x
['props']['pageProps']['program'], dict)
59 raise ExtractorError('Unable to parse __NEXT_DATA__')
61 accessible_episodes
= self
._parse
_json
(self
._html
_search
_regex
(
62 r
'data-react-class="routes/Product/components/ProgramContainer/ProgramContainer"[^>]+data-react-props="({.+?})"',
63 webpage
, 'urplayer data'), video_id
)['accessibleEpisodes']
64 urplayer_data
= next(e
for e
in accessible_episodes
if e
.get('id') == int_or_none(video_id
))
65 episode
= urplayer_data
['title']
67 host
= self
._download
_json
('http://streaming-loadbalancer.ur.se/loadbalancer.json', video_id
)['redirect']
69 urplayer_streams
= urplayer_data
.get('streamingInfo', {})
71 for k
, v
in urplayer_streams
.get('raw', {}).items():
72 if not (k
in ('sd', 'hd') and isinstance(v
, dict)):
74 file_http
= v
.get('location')
76 formats
.extend(self
._extract
_wowza
_formats
(
77 'http://%s/%splaylist.m3u8' % (host
, file_http
),
78 video_id
, skip_protocols
=['f4m', 'rtmp', 'rtsp']))
79 self
._sort
_formats
(formats
)
83 def parse_lang_code(code
):
84 "3-character language code or None (utils candidate)"
88 if not ISO639Utils
.long2short(lang
):
89 lang
= ISO639Utils
.short2long(lang
)
92 for k
, v
in (urplayer_data
['streamingInfo'].get('sweComplete') or {}).items():
93 if (k
in ('sd', 'hd') or not isinstance(v
, dict)):
95 lang
, sttl_url
= (v
.get(kk
) for kk
in ('language', 'location', ))
98 lang
= parse_lang_code(lang
)
101 sttl
= subtitles
.get(lang
) or []
102 sttl
.append({'ext': k, 'url': sttl_url, }
)
103 subtitles
[lang
] = sttl
105 image
= urplayer_data
.get('image') or {}
107 for k
, v
in image
.items():
115 'width': int_or_none(wh
[0]),
116 'height': int_or_none(wh
[1]),
120 series
= urplayer_data
.get('series') or {}
121 series_title
= dict_get(series
, ('seriesTitle', 'title')) or dict_get(urplayer_data
, ('seriesTitle', 'mainTitle'))
125 'title': '%s : %s' % (series_title
, episode
) if series_title
else episode
,
126 'description': urplayer_data
.get('description'),
127 'thumbnails': thumbnails
,
128 'timestamp': unified_timestamp(urplayer_data
.get('publishedAt')),
129 'series': series_title
,
131 'duration': int_or_none(urplayer_data
.get('duration')),
132 'categories': urplayer_data
.get('categories'),
133 'tags': urplayer_data
.get('keywords'),
134 'season': series
.get('label'),
136 'episode_number': int_or_none(urplayer_data
.get('episodeNumber')),
137 'age_limit': parse_age_limit(min(try_get(a
, lambda x
: x
['from'], int) or 0
138 for a
in urplayer_data
.get('ageRanges', []))),
139 'subtitles': subtitles
,