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)
35 class NRKIE(NRKBaseIE):
41 (?:www\.)?nrk\.no/video/(?:PS\*|[^_]+_)|
42 v8[-.]psapi\.nrk\.no/mediaelement/
50 'url': 'http://www.nrk.no/video/PS*150533',
51 'md5': '706f34cdf1322577589e369e522b50ef',
55 'title': 'Dompap og andre fugler i Piip-Show',
56 'description': 'md5:d9261ba34c43b61c812cb6b0269a5c8f',
61 'url': 'http://www.nrk.no/video/PS*154915',
66 'title': 'Slik høres internett ut når du er blind',
67 'description': 'md5:a621f5cc1bd75c8d5104cb048c6b8568',
71 'url': 'nrk:ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
72 'only_matching': True,
74 'url': 'nrk:clip/7707d5a3-ebe7-434a-87d5-a3ebe7a34a70',
75 'only_matching': True,
77 'url': 'https://v8-psapi.nrk.no/mediaelement/ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
78 'only_matching': True,
80 'url': 'https://www.nrk.no/video/dompap-og-andre-fugler-i-piip-show_150533',
81 'only_matching': True,
83 'url': 'https://www.nrk.no/video/humor/kommentatorboksen-reiser-til-sjos_d1fda11f-a4ad-437a-a374-0398bc84e999',
84 'only_matching': True,
87 def _extract_from_playback(self, video_id):
88 manifest = self._download_json(
89 'http://psapi.nrk.no/playback/manifest/%s' % video_id,
90 video_id, 'Downloading manifest JSON')
92 playable = manifest['playable']
95 for asset in playable['assets']:
96 if not isinstance(asset, dict):
98 if asset.get('encrypted'):
100 format_url = url_or_none(asset.get('url'))
103 if asset.get('format') == 'HLS' or determine_ext(format_url) == 'm3u8':
104 formats.extend(self._extract_nrk_formats(format_url, video_id))
105 self._sort_formats(formats)
107 data = self._download_json(
108 'http://psapi.nrk.no/playback/metadata/%s' % video_id,
109 video_id, 'Downloading metadata JSON')
111 preplay = data['preplay']
112 titles = preplay['titles']
113 title = titles['title']
114 alt_title = titles.get('subtitle')
116 description = preplay.get('description')
117 duration = parse_duration(playable.get('duration')) or parse_duration(data.get('duration'))
120 for image in try_get(
121 preplay, lambda x: x['poster']['images'], list) or []:
122 if not isinstance(image, dict):
124 image_url = url_or_none(image.get('url'))
129 'width': int_or_none(image.get('pixelWidth')),
130 'height': int_or_none(image.get('pixelHeight')),
136 'alt_title': alt_title,
137 'description': description,
138 'duration': duration,
139 'thumbnails': thumbnails,
143 def _real_extract(self, url):
144 video_id = self._match_id(url)
145 return self._extract_from_playback(video_id)
148 class NRKTVIE(NRKBaseIE):
149 IE_DESC = 'NRK TV and NRK Radio'
150 _EPISODE_RE = r'(?P<id>[a-zA-Z]{4}\d{8})'
151 _VALID_URL = r'''(?x)
153 (?:tv|radio)\.nrk(?:super)?\.no/
154 (?:serie(?:/[^/]+){1,}|program)/
156 (?:/\d{2}-\d{2}-\d{4})?
157 (?:\#del=(?P<part_id>\d+))?
159 _API_HOSTS = ('psapi-ne.nrk.no', 'psapi-we.nrk.no')
161 'url': 'https://tv.nrk.no/program/MDDP12000117',
162 'md5': '8270824df46ec629b66aeaa5796b36fb',
164 'id': 'MDDP12000117AA',
166 'title': 'Alarm Trolltunga',
167 'description': 'md5:46923a6e6510eefcce23d5ef2a58f2ce',
172 'url': 'https://tv.nrk.no/serie/20-spoersmaal-tv/MUHH48000314/23-05-2014',
173 'md5': '9a167e54d04671eb6317a37b7bc8a280',
175 'id': 'MUHH48000314AA',
177 'title': '20 spørsmål 23.05.2014',
178 'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
180 'series': '20 spørsmål',
181 'episode': '23.05.2014',
183 'skip': 'NoProgramRights',
185 'url': 'https://tv.nrk.no/program/mdfp15000514',
187 'id': 'MDFP15000514CA',
189 'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014',
190 'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db',
192 'series': 'Kunnskapskanalen',
193 'episode': '24.05.2014',
196 'skip_download': True,
199 # single playlist video
200 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2',
202 'id': 'MSPO40010515-part2',
204 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
205 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
208 'skip_download': True,
210 'expected_warnings': ['Video is geo restricted'],
211 'skip': 'particular part is not supported currently',
213 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015',
216 'id': 'MSPO40010515AH',
218 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 1)',
219 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
221 'series': 'Tour de Ski',
222 'episode': '06.01.2015',
225 'skip_download': True,
229 'id': 'MSPO40010515BH',
231 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 2)',
232 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
234 'series': 'Tour de Ski',
235 'episode': '06.01.2015',
238 'skip_download': True,
242 'id': 'MSPO40010515',
243 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015',
244 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
246 'expected_warnings': ['Video is geo restricted'],
248 'url': 'https://tv.nrk.no/serie/anno/KMTE50001317/sesong-3/episode-13',
250 'id': 'KMTE50001317AA',
252 'title': 'Anno 13:30',
253 'description': 'md5:11d9613661a8dbe6f9bef54e3a4cbbfa',
258 'episode_number': 13,
261 'skip_download': True,
264 'url': 'https://tv.nrk.no/serie/nytt-paa-nytt/MUHH46000317/27-01-2017',
266 'id': 'MUHH46000317AA',
268 'title': 'Nytt på Nytt 27.01.2017',
269 'description': 'md5:5358d6388fba0ea6f0b6d11c48b9eb4b',
271 'series': 'Nytt på nytt',
272 'episode': '27.01.2017',
275 'skip_download': True,
278 'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#',
279 'only_matching': True,
281 'url': 'https://tv.nrk.no/serie/lindmo/2018/MUHU11006318/avspiller',
282 'only_matching': True,
284 'url': 'https://radio.nrk.no/serie/dagsnytt/sesong/201507/NPUB21019315',
285 'only_matching': True,
290 def _extract_from_mediaelement(self, video_id):
291 api_hosts = (self._api_host, ) if self._api_host else self._API_HOSTS
293 for api_host in api_hosts:
294 data = self._download_json(
295 'http://%s/mediaelement/%s' % (api_host, video_id),
296 video_id, 'Downloading mediaelement JSON',
297 fatal=api_host == api_hosts[-1])
300 self._api_host = api_host
303 title = data.get('fullTitle') or data.get('mainTitle') or data['title']
304 video_id = data.get('id') or video_id
309 conviva = data.get('convivaStatistics') or {}
310 live = (data.get('mediaElementType') == 'Live'
311 or data.get('isLive') is True or conviva.get('isLive'))
314 return self._live_title(t) if live else t
316 media_assets = data.get('mediaAssets')
317 if media_assets and isinstance(media_assets, list):
318 def video_id_and_title(idx):
319 return ((video_id, title) if len(media_assets) == 1
320 else ('%s-%d' % (video_id, idx), '%s (Part %d)' % (title, idx)))
321 for num, asset in enumerate(media_assets, 1):
322 asset_url = asset.get('url')
323 if not asset_url or asset_url in urls:
325 formats = self._extract_nrk_formats(asset_url, video_id)
328 self._sort_formats(formats)
330 entry_id, entry_title = video_id_and_title(num)
331 duration = parse_duration(asset.get('duration'))
333 for subtitle in ('webVtt', 'timedText'):
334 subtitle_url = asset.get('%sSubtitlesUrl' % subtitle)
336 subtitles.setdefault('no', []).append({
337 'url': compat_urllib_parse_unquote(subtitle_url)
340 'id': asset.get('carrierId') or entry_id,
341 'title': make_title(entry_title),
342 'duration': duration,
343 'subtitles': subtitles,
348 media_url = data.get('mediaUrl')
349 if media_url and media_url not in urls:
350 formats = self._extract_nrk_formats(media_url, video_id)
352 self._sort_formats(formats)
353 duration = parse_duration(data.get('duration'))
356 'title': make_title(title),
357 'duration': duration,
363 'ProgramRightsAreNotReady': 'Du kan dessverre ikke se eller høre programmet',
364 'ProgramRightsHasExpired': 'Programmet har gått ut',
365 'NoProgramRights': 'Ikke tilgjengelig',
366 'ProgramIsGeoBlocked': 'NRK har ikke rettigheter til å vise dette programmet utenfor Norge',
368 message_type = data.get('messageType', '')
369 # Can be ProgramIsGeoBlocked or ChannelIsGeoBlocked*
370 if 'IsGeoBlocked' in message_type or try_get(data, lambda x: x['usageRights']['isGeoBlocked']) is True:
371 self.raise_geo_restricted(
372 msg=MESSAGES.get('ProgramIsGeoBlocked'),
373 countries=self._GEO_COUNTRIES)
374 raise ExtractorError(
375 '%s said: %s' % (self.IE_NAME, MESSAGES.get(
376 message_type, message_type)),
379 series = conviva.get('seriesName') or data.get('seriesTitle')
380 episode = conviva.get('episodeName') or data.get('episodeNumberOrDate')
383 episode_number = None
384 if data.get('mediaElementType') == 'Episode':
385 _season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \
386 data.get('relativeOriginUrl', '')
388 r'/s(?P<season>\d{,2})e(?P<episode>\d{,2})\.',
389 r'/sesong-(?P<season>\d{,2})/episode-(?P<episode>\d{,2})',
391 season_number = int_or_none(self._search_regex(
392 EPISODENUM_RE, _season_episode, 'season number',
393 default=None, group='season'))
394 episode_number = int_or_none(self._search_regex(
395 EPISODENUM_RE, _season_episode, 'episode number',
396 default=None, group='episode'))
399 images = data.get('images')
400 if images and isinstance(images, dict):
401 web_images = images.get('webImages')
402 if isinstance(web_images, list):
404 'url': image['imageUrl'],
405 'width': int_or_none(image.get('width')),
406 'height': int_or_none(image.get('height')),
407 } for image in web_images if image.get('imageUrl')]
409 description = data.get('description')
410 category = data.get('mediaAnalytics', {}).get('category')
413 'description': description,
416 'season_number': season_number,
417 'episode_number': episode_number,
418 'categories': [category] if category else None,
419 'age_limit': parse_age_limit(data.get('legalAge')),
420 'thumbnails': thumbnails,
423 vcodec = 'none' if data.get('mediaType') == 'Audio' else None
425 for entry in entries:
426 entry.update(common_info)
427 for f in entry['formats']:
430 points = data.get('shortIndexPoints')
431 if isinstance(points, list):
433 for next_num, point in enumerate(points, start=1):
434 if not isinstance(point, dict):
436 start_time = parse_duration(point.get('startPoint'))
437 if start_time is None:
439 end_time = parse_duration(
441 if next_num == len(points)
442 else points[next_num].get('startPoint'))
446 'start_time': start_time,
447 'end_time': end_time,
448 'title': point.get('title'),
450 if chapters and len(entries) == 1:
451 entries[0]['chapters'] = chapters
453 return self.playlist_result(entries, video_id, title, description)
455 def _real_extract(self, url):
456 video_id = self._match_id(url)
457 return self._extract_from_mediaelement(video_id)
460 class NRKTVEpisodeIE(InfoExtractor):
461 _VALID_URL = r'https?://tv\.nrk\.no/serie/(?P<id>[^/]+/sesong/\d+/episode/\d+)'
463 'url': 'https://tv.nrk.no/serie/hellums-kro/sesong/1/episode/2',
465 'id': 'MUHH36005220BA',
467 'title': 'Kro, krig og kjærlighet 2:6',
468 'description': 'md5:b32a7dc0b1ed27c8064f58b97bda4350',
470 'series': 'Hellums kro',
477 'skip_download': True,
480 'url': 'https://tv.nrk.no/serie/backstage/sesong/1/episode/8',
482 'id': 'MSUI14000816AA',
484 'title': 'Backstage 8:30',
485 'description': 'md5:de6ca5d5a2d56849e4021f2bf2850df4',
487 'series': 'Backstage',
493 'skip_download': True,
495 'skip': 'ProgramRightsHasExpired',
498 def _real_extract(self, url):
499 display_id = self._match_id(url)
501 webpage = self._download_webpage(url, display_id)
503 info = self._search_json_ld(webpage, display_id, default={})
504 nrk_id = info.get('@id') or self._html_search_meta(
505 'nrk:program-id', webpage, default=None) or self._search_regex(
506 r'data-program-id=["\'](%s)' % NRKTVIE._EPISODE_RE, webpage,
508 assert re.match(NRKTVIE._EPISODE_RE, nrk_id)
511 '_type': 'url_transparent',
513 'url': 'nrk:%s' % nrk_id,
514 'ie_key': NRKIE.ie_key(),
519 class NRKTVSerieBaseIE(InfoExtractor):
520 def _extract_series(self, webpage, display_id, fatal=True):
521 config = self._parse_json(
523 (r'INITIAL_DATA(?:_V\d)?_*\s*=\s*({.+?})\s*;',
524 r'({.+?})\s*,\s*"[^"]+"\s*\)\s*</script>'),
525 webpage, 'config', default='{}' if not fatal else NO_DEFAULT),
526 display_id, fatal=False, transform_source=js_to_json)
531 (lambda x: x['initialState']['series'], lambda x: x['series']),
534 def _extract_seasons(self, seasons):
535 if not isinstance(seasons, list):
538 for season in seasons:
539 entries.extend(self._extract_episodes(season))
542 def _extract_episodes(self, season):
543 if not isinstance(season, dict):
545 return self._extract_entries(season.get('episodes'))
547 def _extract_entries(self, entry_list):
548 if not isinstance(entry_list, list):
551 for episode in entry_list:
552 nrk_id = episode.get('prfId') or episode.get('episodeId')
553 if not nrk_id or not isinstance(nrk_id, compat_str):
555 if not re.match(NRKTVIE._EPISODE_RE, nrk_id):
557 entries.append(self.url_result(
558 'nrk:%s' % nrk_id, ie=NRKIE.ie_key(), video_id=nrk_id))
562 class NRKTVSeasonIE(NRKTVSerieBaseIE):
563 _VALID_URL = r'https?://(?P<domain>tv|radio)\.nrk\.no/serie/(?P<serie>[^/]+)/(?:sesong/)?(?P<id>\d+)'
565 'url': 'https://tv.nrk.no/serie/backstage/sesong/1',
570 'playlist_mincount': 30,
572 # no /sesong/ in path
573 'url': 'https://tv.nrk.no/serie/lindmo/2016',
578 'playlist_mincount': 29,
580 # weird nested _embedded in catalog JSON response
581 'url': 'https://radio.nrk.no/serie/dickie-dick-dickens/sesong/1',
583 'id': 'dickie-dick-dickens/1',
586 'playlist_mincount': 11,
588 # 841 entries, multi page
589 'url': 'https://radio.nrk.no/serie/dagsnytt/sesong/201509',
591 'id': 'dagsnytt/201509',
592 'title': 'September 2015',
594 'playlist_mincount': 841,
596 # 180 entries, single page
597 'url': 'https://tv.nrk.no/serie/spangas/sesong/1',
598 'only_matching': True,
602 def suitable(cls, url):
603 return (False if NRKTVIE.suitable(url) or NRKTVEpisodeIE.suitable(url)
604 else super(NRKTVSeasonIE, cls).suitable(url))
606 _ASSETS_KEYS = ('episodes', 'instalments',)
608 def _entries(self, data, display_id):
609 for page_num in itertools.count(1):
610 embedded = data.get('_embedded')
611 if not isinstance(embedded, dict):
614 for asset_key in self._ASSETS_KEYS:
617 (lambda x: x[asset_key]['_embedded'][asset_key],
618 lambda x: x[asset_key]),
620 for e in self._extract_entries(entries):
623 for asset_key in self._ASSETS_KEYS:
625 'https://psapi.nrk.no/',
628 (lambda x: x['_links']['next']['href'],
629 lambda x: x['_embedded'][asset_key]['_links']['next']['href']),
635 data = self._download_json(
636 next_url, display_id,
637 'Downloading season JSON page %d' % page_num, fatal=False)
641 def _real_extract(self, url):
642 mobj = re.match(self._VALID_URL, url)
643 domain = mobj.group('domain')
644 serie = mobj.group('serie')
645 season_id = mobj.group('id')
646 display_id = '%s/%s' % (serie, season_id)
648 data = self._download_json(
649 'https://psapi.nrk.no/%s/catalog/series/%s/seasons/%s'
650 % (domain, serie, season_id), display_id, query={'pageSize': 50})
651 title = try_get(data, lambda x: x['titles']['title'], compat_str) or display_id
652 return self.playlist_result(
653 self._entries(data, display_id),
657 class NRKTVSeriesIE(NRKTVSerieBaseIE):
658 _VALID_URL = r'https?://(?:tv|radio)\.nrk(?:super)?\.no/serie/(?P<id>[^/]+)'
659 _ITEM_RE = r'(?:data-season=["\']|id=["\']season-)(?P<id>\d+)'
661 'url': 'https://tv.nrk.no/serie/blank',
665 'description': 'md5:7664b4e7e77dc6810cd3bca367c25b6e',
667 'playlist_mincount': 30,
669 # new layout, seasons
670 'url': 'https://tv.nrk.no/serie/backstage',
673 'title': 'Backstage',
674 'description': 'md5:c3ec3a35736fca0f9e1207b5511143d3',
676 'playlist_mincount': 60,
678 # new layout, instalments
679 'url': 'https://tv.nrk.no/serie/groenn-glede',
681 'id': 'groenn-glede',
682 'title': 'Grønn glede',
683 'description': 'md5:7576e92ae7f65da6993cf90ee29e4608',
685 'playlist_mincount': 10,
688 'url': 'https://tv.nrksuper.no/serie/labyrint',
692 'description': 'md5:318b597330fdac5959247c9b69fdb1ec',
694 'playlist_mincount': 3,
696 'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene',
697 'only_matching': True,
699 'url': 'https://tv.nrk.no/serie/saving-the-human-race',
700 'only_matching': True,
702 'url': 'https://tv.nrk.no/serie/postmann-pat',
703 'only_matching': True,
707 def suitable(cls, url):
709 False if any(ie.suitable(url)
710 for ie in (NRKTVIE, NRKTVEpisodeIE, NRKTVSeasonIE))
711 else super(NRKTVSeriesIE, cls).suitable(url))
713 def _real_extract(self, url):
714 series_id = self._match_id(url)
716 webpage = self._download_webpage(url, series_id)
718 # New layout (e.g. https://tv.nrk.no/serie/backstage)
719 series = self._extract_series(webpage, series_id, fatal=False)
721 title = try_get(series, lambda x: x['titles']['title'], compat_str)
722 description = try_get(
723 series, lambda x: x['titles']['subtitle'], compat_str)
725 entries.extend(self._extract_seasons(series.get('seasons')))
726 entries.extend(self._extract_entries(series.get('instalments')))
727 entries.extend(self._extract_episodes(series.get('extraMaterial')))
728 return self.playlist_result(entries, series_id, title, description)
730 # Old layout (e.g. https://tv.nrksuper.no/serie/labyrint)
733 'https://tv.nrk.no/program/Episodes/{series}/{season}'.format(
734 series=series_id, season=season_id))
735 for season_id in re.findall(self._ITEM_RE, webpage)
738 title = self._html_search_meta(
739 'seriestitle', webpage,
740 'title', default=None) or self._og_search_title(
741 webpage, fatal=False)
743 title = self._search_regex(
744 r'NRK (?:Super )?TV\s*[-–]\s*(.+)', title, 'title', default=title)
746 description = self._html_search_meta(
747 'series_description', webpage,
748 'description', default=None) or self._og_search_description(webpage)
750 return self.playlist_result(entries, series_id, title, description)
753 class NRKTVDirekteIE(NRKTVIE):
754 IE_DESC = 'NRK TV Direkte and NRK Radio Direkte'
755 _VALID_URL = r'https?://(?:tv|radio)\.nrk\.no/direkte/(?P<id>[^/?#&]+)'
758 'url': 'https://tv.nrk.no/direkte/nrk1',
759 'only_matching': True,
761 'url': 'https://radio.nrk.no/direkte/p1_oslo_akershus',
762 'only_matching': True,
766 class NRKPlaylistBaseIE(InfoExtractor):
767 def _extract_description(self, webpage):
770 def _real_extract(self, url):
771 playlist_id = self._match_id(url)
773 webpage = self._download_webpage(url, playlist_id)
776 self.url_result('nrk:%s' % video_id, NRKIE.ie_key())
777 for video_id in re.findall(self._ITEM_RE, webpage)
780 playlist_title = self. _extract_title(webpage)
781 playlist_description = self._extract_description(webpage)
783 return self.playlist_result(
784 entries, playlist_id, playlist_title, playlist_description)
787 class NRKPlaylistIE(NRKPlaylistBaseIE):
788 _VALID_URL = r'https?://(?:www\.)?nrk\.no/(?!video|skole)(?:[^/]+/)+(?P<id>[^/]+)'
789 _ITEM_RE = r'class="[^"]*\brich\b[^"]*"[^>]+data-video-id="([^"]+)"'
791 'url': 'http://www.nrk.no/troms/gjenopplev-den-historiske-solformorkelsen-1.12270763',
793 'id': 'gjenopplev-den-historiske-solformorkelsen-1.12270763',
794 'title': 'Gjenopplev den historiske solformørkelsen',
795 'description': 'md5:c2df8ea3bac5654a26fc2834a542feed',
799 'url': 'http://www.nrk.no/kultur/bok/rivertonprisen-til-karin-fossum-1.12266449',
801 'id': 'rivertonprisen-til-karin-fossum-1.12266449',
802 'title': 'Rivertonprisen til Karin Fossum',
803 'description': 'Første kvinne på 15 år til å vinne krimlitteraturprisen.',
808 def _extract_title(self, webpage):
809 return self._og_search_title(webpage, fatal=False)
811 def _extract_description(self, webpage):
812 return self._og_search_description(webpage)
815 class NRKTVEpisodesIE(NRKPlaylistBaseIE):
816 _VALID_URL = r'https?://tv\.nrk\.no/program/[Ee]pisodes/[^/]+/(?P<id>\d+)'
817 _ITEM_RE = r'data-episode=["\']%s' % NRKTVIE._EPISODE_RE
819 'url': 'https://tv.nrk.no/program/episodes/nytt-paa-nytt/69031',
822 'title': 'Nytt på nytt, sesong: 201210',
827 def _extract_title(self, webpage):
828 return self._html_search_regex(
829 r'<h1>([^<]+)</h1>', webpage, 'title', fatal=False)
832 class NRKSkoleIE(InfoExtractor):
833 IE_DESC = 'NRK Skole'
834 _VALID_URL = r'https?://(?:www\.)?nrk\.no/skole/?\?.*\bmediaId=(?P<id>\d+)'
837 'url': 'https://www.nrk.no/skole/?page=search&q=&mediaId=14099',
838 'md5': '18c12c3d071953c3bf8d54ef6b2587b7',
842 'title': 'Genetikk og eneggede tvillinger',
843 'description': 'md5:3aca25dcf38ec30f0363428d2b265f8d',
847 'url': 'https://www.nrk.no/skole/?page=objectives&subject=naturfag&objective=K15114&mediaId=19355',
848 'only_matching': True,
851 def _real_extract(self, url):
852 video_id = self._match_id(url)
854 webpage = self._download_webpage(
855 'https://mimir.nrk.no/plugin/1.0/static?mediaId=%s' % video_id,
858 nrk_id = self._parse_json(
860 r'<script[^>]+type=["\']application/json["\'][^>]*>({.+?})</script>',
861 webpage, 'application json'),
862 video_id)['activeMedia']['psId']
864 return self.url_result('nrk:%s' % nrk_id)