2 from __future__ import unicode_literals
7 from .common import InfoExtractor
10 compat_urllib_parse_unquote,
26 class NRKBaseIE(InfoExtractor):
27 _GEO_COUNTRIES = ['NO']
29 def _extract_nrk_formats(self, asset_url, video_id):
30 return self._extract_m3u8_formats(
31 re.sub(r'(?:bw_(?:low|high)=\d+|no_audio_only)&?', '', asset_url),
32 video_id, 'mp4', 'm3u8_native', fatal=False)
34 def _raise_error(self, data):
36 'ProgramRightsAreNotReady': 'Du kan dessverre ikke se eller høre programmet',
37 'ProgramRightsHasExpired': 'Programmet har gått ut',
38 'NoProgramRights': 'Ikke tilgjengelig',
39 'ProgramIsGeoBlocked': 'NRK har ikke rettigheter til å vise dette programmet utenfor Norge',
41 message_type = data.get('messageType', '')
42 # Can be ProgramIsGeoBlocked or ChannelIsGeoBlocked*
43 if 'IsGeoBlocked' in message_type or try_get(data, lambda x: x['usageRights']['isGeoBlocked']) is True:
44 self.raise_geo_restricted(
45 msg=MESSAGES.get('ProgramIsGeoBlocked'),
46 countries=self._GEO_COUNTRIES)
47 message = data.get('endUserMessage') or MESSAGES.get(message_type, message_type)
48 raise ExtractorError('%s said: %s' % (self.IE_NAME, message), expected=True)
51 class NRKIE(NRKBaseIE):
57 (?:www\.)?nrk\.no/video/(?:PS\*|[^_]+_)|
58 v8[-.]psapi\.nrk\.no/mediaelement/
66 'url': 'http://www.nrk.no/video/PS*150533',
67 'md5': '706f34cdf1322577589e369e522b50ef',
71 'title': 'Dompap og andre fugler i Piip-Show',
72 'description': 'md5:d9261ba34c43b61c812cb6b0269a5c8f',
77 'url': 'http://www.nrk.no/video/PS*154915',
82 'title': 'Slik høres internett ut når du er blind',
83 'description': 'md5:a621f5cc1bd75c8d5104cb048c6b8568',
87 'url': 'nrk:ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
88 'only_matching': True,
90 'url': 'nrk:clip/7707d5a3-ebe7-434a-87d5-a3ebe7a34a70',
91 'only_matching': True,
93 'url': 'https://v8-psapi.nrk.no/mediaelement/ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
94 'only_matching': True,
96 'url': 'https://www.nrk.no/video/dompap-og-andre-fugler-i-piip-show_150533',
97 'only_matching': True,
99 'url': 'https://www.nrk.no/video/humor/kommentatorboksen-reiser-til-sjos_d1fda11f-a4ad-437a-a374-0398bc84e999',
100 'only_matching': True,
103 def _extract_from_playback(self, video_id):
104 manifest = self._download_json(
105 'http://psapi.nrk.no/playback/manifest/%s' % video_id,
106 video_id, 'Downloading manifest JSON')
108 if manifest.get('playability') == 'nonPlayable':
109 self._raise_error(manifest['nonPlayable'])
111 playable = manifest['playable']
114 for asset in playable['assets']:
115 if not isinstance(asset, dict):
117 if asset.get('encrypted'):
119 format_url = url_or_none(asset.get('url'))
122 if asset.get('format') == 'HLS' or determine_ext(format_url) == 'm3u8':
123 formats.extend(self._extract_nrk_formats(format_url, video_id))
124 self._sort_formats(formats)
126 data = self._download_json(
127 'http://psapi.nrk.no/playback/metadata/%s' % video_id,
128 video_id, 'Downloading metadata JSON')
130 preplay = data['preplay']
131 titles = preplay['titles']
132 title = titles['title']
133 alt_title = titles.get('subtitle')
135 description = preplay.get('description')
136 duration = parse_duration(playable.get('duration')) or parse_duration(data.get('duration'))
139 for image in try_get(
140 preplay, lambda x: x['poster']['images'], list) or []:
141 if not isinstance(image, dict):
143 image_url = url_or_none(image.get('url'))
148 'width': int_or_none(image.get('pixelWidth')),
149 'height': int_or_none(image.get('pixelHeight')),
155 'alt_title': alt_title,
156 'description': description,
157 'duration': duration,
158 'thumbnails': thumbnails,
162 def _real_extract(self, url):
163 video_id = self._match_id(url)
164 return self._extract_from_playback(video_id)
167 class NRKTVIE(NRKBaseIE):
168 IE_DESC = 'NRK TV and NRK Radio'
169 _EPISODE_RE = r'(?P<id>[a-zA-Z]{4}\d{8})'
170 _VALID_URL = r'https?://(?:tv|radio)\.nrk(?:super)?\.no/(?:[^/]+/)*%s' % _EPISODE_RE
171 _API_HOSTS = ('psapi-ne.nrk.no', 'psapi-we.nrk.no')
173 'url': 'https://tv.nrk.no/program/MDDP12000117',
174 'md5': '8270824df46ec629b66aeaa5796b36fb',
176 'id': 'MDDP12000117AA',
178 'title': 'Alarm Trolltunga',
179 'description': 'md5:46923a6e6510eefcce23d5ef2a58f2ce',
184 'url': 'https://tv.nrk.no/serie/20-spoersmaal-tv/MUHH48000314/23-05-2014',
185 'md5': '9a167e54d04671eb6317a37b7bc8a280',
187 'id': 'MUHH48000314AA',
189 'title': '20 spørsmål 23.05.2014',
190 'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
192 'series': '20 spørsmål',
193 'episode': '23.05.2014',
195 'skip': 'NoProgramRights',
197 'url': 'https://tv.nrk.no/program/mdfp15000514',
199 'id': 'MDFP15000514CA',
201 'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014',
202 'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db',
204 'series': 'Kunnskapskanalen',
205 'episode': '24.05.2014',
208 'skip_download': True,
211 # single playlist video
212 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2',
214 'id': 'MSPO40010515-part2',
216 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
217 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
220 'skip_download': True,
222 'expected_warnings': ['Video is geo restricted'],
223 'skip': 'particular part is not supported currently',
225 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015',
228 'id': 'MSPO40010515AH',
230 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 1)',
231 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
233 'series': 'Tour de Ski',
234 'episode': '06.01.2015',
237 'skip_download': True,
241 'id': 'MSPO40010515BH',
243 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 2)',
244 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
246 'series': 'Tour de Ski',
247 'episode': '06.01.2015',
250 'skip_download': True,
254 'id': 'MSPO40010515',
255 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015',
256 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
258 'expected_warnings': ['Video is geo restricted'],
260 'url': 'https://tv.nrk.no/serie/anno/KMTE50001317/sesong-3/episode-13',
262 'id': 'KMTE50001317AA',
264 'title': 'Anno 13:30',
265 'description': 'md5:11d9613661a8dbe6f9bef54e3a4cbbfa',
270 'episode_number': 13,
273 'skip_download': True,
276 'url': 'https://tv.nrk.no/serie/nytt-paa-nytt/MUHH46000317/27-01-2017',
278 'id': 'MUHH46000317AA',
280 'title': 'Nytt på Nytt 27.01.2017',
281 'description': 'md5:5358d6388fba0ea6f0b6d11c48b9eb4b',
283 'series': 'Nytt på nytt',
284 'episode': '27.01.2017',
287 'skip_download': True,
290 'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#',
291 'only_matching': True,
293 'url': 'https://tv.nrk.no/serie/lindmo/2018/MUHU11006318/avspiller',
294 'only_matching': True,
296 'url': 'https://radio.nrk.no/serie/dagsnytt/sesong/201507/NPUB21019315',
297 'only_matching': True,
302 def _extract_from_mediaelement(self, video_id):
303 api_hosts = (self._api_host, ) if self._api_host else self._API_HOSTS
305 for api_host in api_hosts:
306 data = self._download_json(
307 'http://%s/mediaelement/%s' % (api_host, video_id),
308 video_id, 'Downloading mediaelement JSON',
309 fatal=api_host == api_hosts[-1])
312 self._api_host = api_host
315 title = data.get('fullTitle') or data.get('mainTitle') or data['title']
316 video_id = data.get('id') or video_id
321 conviva = data.get('convivaStatistics') or {}
322 live = (data.get('mediaElementType') == 'Live'
323 or data.get('isLive') is True or conviva.get('isLive'))
326 return self._live_title(t) if live else t
328 media_assets = data.get('mediaAssets')
329 if media_assets and isinstance(media_assets, list):
330 def video_id_and_title(idx):
331 return ((video_id, title) if len(media_assets) == 1
332 else ('%s-%d' % (video_id, idx), '%s (Part %d)' % (title, idx)))
333 for num, asset in enumerate(media_assets, 1):
334 asset_url = asset.get('url')
335 if not asset_url or asset_url in urls:
337 formats = self._extract_nrk_formats(asset_url, video_id)
340 self._sort_formats(formats)
342 entry_id, entry_title = video_id_and_title(num)
343 duration = parse_duration(asset.get('duration'))
345 for subtitle in ('webVtt', 'timedText'):
346 subtitle_url = asset.get('%sSubtitlesUrl' % subtitle)
348 subtitles.setdefault('no', []).append({
349 'url': compat_urllib_parse_unquote(subtitle_url)
352 'id': asset.get('carrierId') or entry_id,
353 'title': make_title(entry_title),
354 'duration': duration,
355 'subtitles': subtitles,
360 media_url = data.get('mediaUrl')
361 if media_url and media_url not in urls:
362 formats = self._extract_nrk_formats(media_url, video_id)
364 self._sort_formats(formats)
365 duration = parse_duration(data.get('duration'))
368 'title': make_title(title),
369 'duration': duration,
374 self._raise_error(data)
376 series = conviva.get('seriesName') or data.get('seriesTitle')
377 episode = conviva.get('episodeName') or data.get('episodeNumberOrDate')
380 episode_number = None
381 if data.get('mediaElementType') == 'Episode':
382 _season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \
383 data.get('relativeOriginUrl', '')
385 r'/s(?P<season>\d{,2})e(?P<episode>\d{,2})\.',
386 r'/sesong-(?P<season>\d{,2})/episode-(?P<episode>\d{,2})',
388 season_number = int_or_none(self._search_regex(
389 EPISODENUM_RE, _season_episode, 'season number',
390 default=None, group='season'))
391 episode_number = int_or_none(self._search_regex(
392 EPISODENUM_RE, _season_episode, 'episode number',
393 default=None, group='episode'))
396 images = data.get('images')
397 if images and isinstance(images, dict):
398 web_images = images.get('webImages')
399 if isinstance(web_images, list):
401 'url': image['imageUrl'],
402 'width': int_or_none(image.get('width')),
403 'height': int_or_none(image.get('height')),
404 } for image in web_images if image.get('imageUrl')]
406 description = data.get('description')
407 category = data.get('mediaAnalytics', {}).get('category')
410 'description': description,
413 'season_number': season_number,
414 'episode_number': episode_number,
415 'categories': [category] if category else None,
416 'age_limit': parse_age_limit(data.get('legalAge')),
417 'thumbnails': thumbnails,
420 vcodec = 'none' if data.get('mediaType') == 'Audio' else None
422 for entry in entries:
423 entry.update(common_info)
424 for f in entry['formats']:
427 points = data.get('shortIndexPoints')
428 if isinstance(points, list):
430 for next_num, point in enumerate(points, start=1):
431 if not isinstance(point, dict):
433 start_time = parse_duration(point.get('startPoint'))
434 if start_time is None:
436 end_time = parse_duration(
438 if next_num == len(points)
439 else points[next_num].get('startPoint'))
443 'start_time': start_time,
444 'end_time': end_time,
445 'title': point.get('title'),
447 if chapters and len(entries) == 1:
448 entries[0]['chapters'] = chapters
450 return self.playlist_result(entries, video_id, title, description)
452 def _real_extract(self, url):
453 video_id = self._match_id(url)
454 return self._extract_from_mediaelement(video_id)
457 class NRKTVEpisodeIE(InfoExtractor):
458 _VALID_URL = r'https?://tv\.nrk\.no/serie/(?P<id>[^/]+/sesong/\d+/episode/\d+)'
460 'url': 'https://tv.nrk.no/serie/hellums-kro/sesong/1/episode/2',
462 'id': 'MUHH36005220BA',
464 'title': 'Kro, krig og kjærlighet 2:6',
465 'description': 'md5:b32a7dc0b1ed27c8064f58b97bda4350',
467 'series': 'Hellums kro',
474 'skip_download': True,
477 'url': 'https://tv.nrk.no/serie/backstage/sesong/1/episode/8',
479 'id': 'MSUI14000816AA',
481 'title': 'Backstage 8:30',
482 'description': 'md5:de6ca5d5a2d56849e4021f2bf2850df4',
484 'series': 'Backstage',
490 'skip_download': True,
492 'skip': 'ProgramRightsHasExpired',
495 def _real_extract(self, url):
496 display_id = self._match_id(url)
498 webpage = self._download_webpage(url, display_id)
500 info = self._search_json_ld(webpage, display_id, default={})
501 nrk_id = info.get('@id') or self._html_search_meta(
502 'nrk:program-id', webpage, default=None) or self._search_regex(
503 r'data-program-id=["\'](%s)' % NRKTVIE._EPISODE_RE, webpage,
505 assert re.match(NRKTVIE._EPISODE_RE, nrk_id)
508 '_type': 'url_transparent',
510 'url': 'nrk:%s' % nrk_id,
511 'ie_key': NRKIE.ie_key(),
516 class NRKTVSerieBaseIE(InfoExtractor):
517 def _extract_series(self, webpage, display_id, fatal=True):
518 config = self._parse_json(
520 (r'INITIAL_DATA(?:_V\d)?_*\s*=\s*({.+?})\s*;',
521 r'({.+?})\s*,\s*"[^"]+"\s*\)\s*</script>',
522 r'PRELOADED_STATE_*\s*=\s*({.+?})\s*\n'),
523 webpage, 'config', default='{}' if not fatal else NO_DEFAULT),
524 display_id, fatal=False, transform_source=js_to_json)
529 (lambda x: x['initialState']['series'], lambda x: x['series']),
532 def _extract_seasons(self, domain, series_id, seasons):
533 if isinstance(seasons, dict):
534 seasons = seasons.get('seasons')
535 if not isinstance(seasons, list):
538 for season in seasons:
539 if not isinstance(season, dict):
541 episodes = self._extract_episodes(season)
543 entries.extend(episodes)
545 season_name = season.get('name')
546 if season_name and isinstance(season_name, compat_str):
547 entries.append(self.url_result(
548 'https://%s.nrk.no/serie/%s/sesong/%s'
549 % (domain, series_id, season_name),
550 ie=NRKTVSeasonIE.ie_key(),
551 video_title=season.get('title')))
554 def _extract_episodes(self, season):
555 if not isinstance(season, dict):
557 return self._extract_entries(season.get('episodes'))
559 def _extract_entries(self, entry_list):
560 if not isinstance(entry_list, list):
563 for episode in entry_list:
564 nrk_id = episode.get('prfId') or episode.get('episodeId')
565 if not nrk_id or not isinstance(nrk_id, compat_str):
567 if not re.match(NRKTVIE._EPISODE_RE, nrk_id):
569 entries.append(self.url_result(
570 'nrk:%s' % nrk_id, ie=NRKIE.ie_key(), video_id=nrk_id))
573 _ASSETS_KEYS = ('episodes', 'instalments',)
575 def _extract_assets_key(self, embedded):
576 for asset_key in self._ASSETS_KEYS:
577 if embedded.get(asset_key):
580 def _entries(self, data, display_id):
581 for page_num in itertools.count(1):
582 embedded = data.get('_embedded')
583 if not isinstance(embedded, dict):
585 assets_key = self._extract_assets_key(embedded)
591 (lambda x: x[assets_key]['_embedded'][assets_key],
592 lambda x: x[assets_key]),
594 for e in self._extract_entries(entries):
598 'https://psapi.nrk.no/',
601 (lambda x: x['_links']['next']['href'],
602 lambda x: x['_embedded'][assets_key]['_links']['next']['href']),
606 data = self._download_json(
607 next_url, display_id,
608 'Downloading %s JSON page %d' % (assets_key, page_num),
614 class NRKTVSeasonIE(NRKTVSerieBaseIE):
615 _VALID_URL = r'https?://(?P<domain>tv|radio)\.nrk\.no/serie/(?P<serie>[^/]+)/(?:sesong/)?(?P<id>\d+)'
617 'url': 'https://tv.nrk.no/serie/backstage/sesong/1',
622 'playlist_mincount': 30,
624 # no /sesong/ in path
625 'url': 'https://tv.nrk.no/serie/lindmo/2016',
630 'playlist_mincount': 29,
632 # weird nested _embedded in catalog JSON response
633 'url': 'https://radio.nrk.no/serie/dickie-dick-dickens/sesong/1',
635 'id': 'dickie-dick-dickens/1',
638 'playlist_mincount': 11,
640 # 841 entries, multi page
641 'url': 'https://radio.nrk.no/serie/dagsnytt/sesong/201509',
643 'id': 'dagsnytt/201509',
644 'title': 'September 2015',
646 'playlist_mincount': 841,
648 # 180 entries, single page
649 'url': 'https://tv.nrk.no/serie/spangas/sesong/1',
650 'only_matching': True,
654 def suitable(cls, url):
655 return (False if NRKTVIE.suitable(url) or NRKTVEpisodeIE.suitable(url)
656 else super(NRKTVSeasonIE, cls).suitable(url))
658 def _real_extract(self, url):
659 mobj = re.match(self._VALID_URL, url)
660 domain = mobj.group('domain')
661 serie = mobj.group('serie')
662 season_id = mobj.group('id')
663 display_id = '%s/%s' % (serie, season_id)
665 data = self._download_json(
666 'https://psapi.nrk.no/%s/catalog/series/%s/seasons/%s'
667 % (domain, serie, season_id), display_id, query={'pageSize': 50})
669 title = try_get(data, lambda x: x['titles']['title'], compat_str) or display_id
670 return self.playlist_result(
671 self._entries(data, display_id),
675 class NRKTVSeriesIE(NRKTVSerieBaseIE):
676 _VALID_URL = r'https?://(?P<domain>tv|radio)\.nrk(?:super)?\.no/serie/(?P<id>[^/]+)'
677 _ITEM_RE = r'(?:data-season=["\']|id=["\']season-)(?P<id>\d+)'
679 # new layout, instalments
680 'url': 'https://tv.nrk.no/serie/groenn-glede',
682 'id': 'groenn-glede',
683 'title': 'Grønn glede',
684 'description': 'md5:7576e92ae7f65da6993cf90ee29e4608',
686 'playlist_mincount': 90,
688 # new layout, instalments, more entries
689 'url': 'https://tv.nrk.no/serie/lindmo',
690 'only_matching': True,
692 'url': 'https://tv.nrk.no/serie/blank',
696 'description': 'md5:7664b4e7e77dc6810cd3bca367c25b6e',
698 'playlist_mincount': 30,
699 'expected_warnings': ['HTTP Error 404: Not Found'],
701 # new layout, seasons
702 'url': 'https://tv.nrk.no/serie/backstage',
705 'title': 'Backstage',
706 'description': 'md5:63692ceb96813d9a207e9910483d948b',
708 'playlist_mincount': 60,
709 'expected_warnings': ['HTTP Error 404: Not Found'],
712 'url': 'https://tv.nrksuper.no/serie/labyrint',
716 'description': 'md5:318b597330fdac5959247c9b69fdb1ec',
718 'playlist_mincount': 3,
720 'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene',
721 'only_matching': True,
723 'url': 'https://tv.nrk.no/serie/saving-the-human-race',
724 'only_matching': True,
726 'url': 'https://tv.nrk.no/serie/postmann-pat',
727 'only_matching': True,
729 'url': 'https://radio.nrk.no/serie/dickie-dick-dickens',
731 'id': 'dickie-dick-dickens',
733 'playlist_mincount': 8,
734 'expected_warnings': ['HTTP Error 404: Not Found'],
738 def suitable(cls, url):
740 False if any(ie.suitable(url)
741 for ie in (NRKTVIE, NRKTVEpisodeIE, NRKTVSeasonIE))
742 else super(NRKTVSeriesIE, cls).suitable(url))
744 def _real_extract(self, url):
745 mobj = re.match(self._VALID_URL, url)
746 domain = mobj.group('domain')
747 series_id = mobj.group('id')
749 title = description = None
751 webpage = self._download_webpage(url, series_id)
753 series = self._extract_series(webpage, series_id, fatal=False)
755 title = try_get(series, lambda x: x['titles']['title'], compat_str)
756 description = try_get(
757 series, lambda x: x['titles']['subtitle'], compat_str)
759 data = self._download_json(
760 'https://psapi.nrk.no/%s/catalog/series/%s/instalments'
761 % (domain, series_id), series_id, query={'pageSize': 50},
764 return self.playlist_result(
765 self._entries(data, series_id), series_id, title, description)
767 # New layout (e.g. https://tv.nrk.no/serie/backstage)
770 entries.extend(self._extract_seasons(domain, series_id, series.get('seasons')))
771 entries.extend(self._extract_entries(series.get('instalments')))
772 entries.extend(self._extract_episodes(series.get('extraMaterial')))
773 return self.playlist_result(entries, series_id, title, description)
775 # Old layout (e.g. https://tv.nrksuper.no/serie/labyrint)
778 'https://tv.nrk.no/program/Episodes/{series}/{season}'.format(
779 series=series_id, season=season_id))
780 for season_id in re.findall(self._ITEM_RE, webpage)
783 title = self._html_search_meta(
784 'seriestitle', webpage,
785 'title', default=None) or self._og_search_title(
786 webpage, fatal=False)
788 title = self._search_regex(
789 r'NRK (?:Super )?TV\s*[-–]\s*(.+)', title, 'title', default=title)
791 description = self._html_search_meta(
792 'series_description', webpage,
793 'description', default=None) or self._og_search_description(webpage)
795 return self.playlist_result(entries, series_id, title, description)
798 class NRKTVDirekteIE(NRKTVIE):
799 IE_DESC = 'NRK TV Direkte and NRK Radio Direkte'
800 _VALID_URL = r'https?://(?:tv|radio)\.nrk\.no/direkte/(?P<id>[^/?#&]+)'
803 'url': 'https://tv.nrk.no/direkte/nrk1',
804 'only_matching': True,
806 'url': 'https://radio.nrk.no/direkte/p1_oslo_akershus',
807 'only_matching': True,
811 class NRKPlaylistBaseIE(InfoExtractor):
812 def _extract_description(self, webpage):
815 def _real_extract(self, url):
816 playlist_id = self._match_id(url)
818 webpage = self._download_webpage(url, playlist_id)
821 self.url_result('nrk:%s' % video_id, NRKIE.ie_key())
822 for video_id in re.findall(self._ITEM_RE, webpage)
825 playlist_title = self. _extract_title(webpage)
826 playlist_description = self._extract_description(webpage)
828 return self.playlist_result(
829 entries, playlist_id, playlist_title, playlist_description)
832 class NRKPlaylistIE(NRKPlaylistBaseIE):
833 _VALID_URL = r'https?://(?:www\.)?nrk\.no/(?!video|skole)(?:[^/]+/)+(?P<id>[^/]+)'
834 _ITEM_RE = r'class="[^"]*\brich\b[^"]*"[^>]+data-video-id="([^"]+)"'
836 'url': 'http://www.nrk.no/troms/gjenopplev-den-historiske-solformorkelsen-1.12270763',
838 'id': 'gjenopplev-den-historiske-solformorkelsen-1.12270763',
839 'title': 'Gjenopplev den historiske solformørkelsen',
840 'description': 'md5:c2df8ea3bac5654a26fc2834a542feed',
844 'url': 'http://www.nrk.no/kultur/bok/rivertonprisen-til-karin-fossum-1.12266449',
846 'id': 'rivertonprisen-til-karin-fossum-1.12266449',
847 'title': 'Rivertonprisen til Karin Fossum',
848 'description': 'Første kvinne på 15 år til å vinne krimlitteraturprisen.',
853 def _extract_title(self, webpage):
854 return self._og_search_title(webpage, fatal=False)
856 def _extract_description(self, webpage):
857 return self._og_search_description(webpage)
860 class NRKTVEpisodesIE(NRKPlaylistBaseIE):
861 _VALID_URL = r'https?://tv\.nrk\.no/program/[Ee]pisodes/[^/]+/(?P<id>\d+)'
862 _ITEM_RE = r'data-episode=["\']%s' % NRKTVIE._EPISODE_RE
864 'url': 'https://tv.nrk.no/program/episodes/nytt-paa-nytt/69031',
867 'title': 'Nytt på nytt, sesong: 201210',
872 def _extract_title(self, webpage):
873 return self._html_search_regex(
874 r'<h1>([^<]+)</h1>', webpage, 'title', fatal=False)
877 class NRKSkoleIE(InfoExtractor):
878 IE_DESC = 'NRK Skole'
879 _VALID_URL = r'https?://(?:www\.)?nrk\.no/skole/?\?.*\bmediaId=(?P<id>\d+)'
882 'url': 'https://www.nrk.no/skole/?page=search&q=&mediaId=14099',
883 'md5': '18c12c3d071953c3bf8d54ef6b2587b7',
887 'title': 'Genetikk og eneggede tvillinger',
888 'description': 'md5:3aca25dcf38ec30f0363428d2b265f8d',
892 'url': 'https://www.nrk.no/skole/?page=objectives&subject=naturfag&objective=K15114&mediaId=19355',
893 'only_matching': True,
896 def _real_extract(self, url):
897 video_id = self._match_id(url)
899 webpage = self._download_webpage(
900 'https://mimir.nrk.no/plugin/1.0/static?mediaId=%s' % video_id,
903 nrk_id = self._parse_json(
905 r'<script[^>]+type=["\']application/json["\'][^>]*>({.+?})</script>',
906 webpage, 'application json'),
907 video_id)['activeMedia']['psId']
909 return self.url_result('nrk:%s' % nrk_id)