2 from __future__ import unicode_literals
6 from .common import InfoExtractor
9 compat_urllib_parse_unquote,
24 class NRKBaseIE(InfoExtractor):
25 _GEO_COUNTRIES = ['NO']
28 class NRKIE(NRKBaseIE):
34 (?:www\.)?nrk\.no/video/(?:PS\*|[^_]+_)|
35 v8[-.]psapi\.nrk\.no/mediaelement/
43 'url': 'http://www.nrk.no/video/PS*150533',
44 'md5': '706f34cdf1322577589e369e522b50ef',
48 'title': 'Dompap og andre fugler i Piip-Show',
49 'description': 'md5:d9261ba34c43b61c812cb6b0269a5c8f',
54 'url': 'http://www.nrk.no/video/PS*154915',
59 'title': 'Slik høres internett ut når du er blind',
60 'description': 'md5:a621f5cc1bd75c8d5104cb048c6b8568',
64 'url': 'nrk:ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
65 'only_matching': True,
67 'url': 'nrk:clip/7707d5a3-ebe7-434a-87d5-a3ebe7a34a70',
68 'only_matching': True,
70 'url': 'https://v8-psapi.nrk.no/mediaelement/ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
71 'only_matching': True,
73 'url': 'https://www.nrk.no/video/dompap-og-andre-fugler-i-piip-show_150533',
74 'only_matching': True,
76 'url': 'https://www.nrk.no/video/humor/kommentatorboksen-reiser-til-sjos_d1fda11f-a4ad-437a-a374-0398bc84e999',
77 'only_matching': True,
80 def _extract_from_playback(self, video_id):
81 manifest = self._download_json(
82 'http://psapi.nrk.no/playback/manifest/%s' % video_id,
83 video_id, 'Downloading manifest JSON')
85 playable = manifest['playable']
88 for asset in playable['assets']:
89 if not isinstance(asset, dict):
91 if asset.get('encrypted'):
93 format_url = url_or_none(asset.get('url'))
96 if asset.get('format') == 'HLS' or determine_ext(format_url) == 'm3u8':
97 formats.extend(self._extract_m3u8_formats(
98 format_url, video_id, 'mp4', entry_protocol='m3u8_native',
99 m3u8_id='hls', fatal=False))
100 self._sort_formats(formats)
102 data = self._download_json(
103 'http://psapi.nrk.no/playback/metadata/%s' % video_id,
104 video_id, 'Downloading metadata JSON')
106 preplay = data['preplay']
107 titles = preplay['titles']
108 title = titles['title']
109 alt_title = titles.get('subtitle')
111 description = preplay.get('description')
112 duration = parse_duration(playable.get('duration')) or parse_duration(data.get('duration'))
115 for image in try_get(
116 preplay, lambda x: x['poster']['images'], list) or []:
117 if not isinstance(image, dict):
119 image_url = url_or_none(image.get('url'))
124 'width': int_or_none(image.get('pixelWidth')),
125 'height': int_or_none(image.get('pixelHeight')),
131 'alt_title': alt_title,
132 'description': description,
133 'duration': duration,
134 'thumbnails': thumbnails,
138 def _real_extract(self, url):
139 video_id = self._match_id(url)
140 return self._extract_from_playback(video_id)
143 class NRKTVIE(NRKBaseIE):
144 IE_DESC = 'NRK TV and NRK Radio'
145 _EPISODE_RE = r'(?P<id>[a-zA-Z]{4}\d{8})'
146 _VALID_URL = r'''(?x)
148 (?:tv|radio)\.nrk(?:super)?\.no/
149 (?:serie(?:/[^/]+){1,}|program)/
151 (?:/\d{2}-\d{2}-\d{4})?
152 (?:\#del=(?P<part_id>\d+))?
154 _API_HOSTS = ('psapi-ne.nrk.no', 'psapi-we.nrk.no')
156 'url': 'https://tv.nrk.no/program/MDDP12000117',
157 'md5': '8270824df46ec629b66aeaa5796b36fb',
159 'id': 'MDDP12000117AA',
161 'title': 'Alarm Trolltunga',
162 'description': 'md5:46923a6e6510eefcce23d5ef2a58f2ce',
167 'url': 'https://tv.nrk.no/serie/20-spoersmaal-tv/MUHH48000314/23-05-2014',
168 'md5': '9a167e54d04671eb6317a37b7bc8a280',
170 'id': 'MUHH48000314AA',
172 'title': '20 spørsmål 23.05.2014',
173 'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
175 'series': '20 spørsmål',
176 'episode': '23.05.2014',
178 'skip': 'NoProgramRights',
180 'url': 'https://tv.nrk.no/program/mdfp15000514',
182 'id': 'MDFP15000514CA',
184 'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014',
185 'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db',
187 'series': 'Kunnskapskanalen',
188 'episode': '24.05.2014',
191 'skip_download': True,
194 # single playlist video
195 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2',
197 'id': 'MSPO40010515-part2',
199 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
200 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
203 'skip_download': True,
205 'expected_warnings': ['Video is geo restricted'],
206 'skip': 'particular part is not supported currently',
208 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015',
211 'id': 'MSPO40010515AH',
213 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 1)',
214 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
216 'series': 'Tour de Ski',
217 'episode': '06.01.2015',
220 'skip_download': True,
224 'id': 'MSPO40010515BH',
226 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 2)',
227 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
229 'series': 'Tour de Ski',
230 'episode': '06.01.2015',
233 'skip_download': True,
237 'id': 'MSPO40010515',
238 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015',
239 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
241 'expected_warnings': ['Video is geo restricted'],
243 'url': 'https://tv.nrk.no/serie/anno/KMTE50001317/sesong-3/episode-13',
245 'id': 'KMTE50001317AA',
247 'title': 'Anno 13:30',
248 'description': 'md5:11d9613661a8dbe6f9bef54e3a4cbbfa',
253 'episode_number': 13,
256 'skip_download': True,
259 'url': 'https://tv.nrk.no/serie/nytt-paa-nytt/MUHH46000317/27-01-2017',
261 'id': 'MUHH46000317AA',
263 'title': 'Nytt på Nytt 27.01.2017',
264 'description': 'md5:5358d6388fba0ea6f0b6d11c48b9eb4b',
266 'series': 'Nytt på nytt',
267 'episode': '27.01.2017',
270 'skip_download': True,
273 'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#',
274 'only_matching': True,
276 'url': 'https://tv.nrk.no/serie/lindmo/2018/MUHU11006318/avspiller',
277 'only_matching': True,
279 'url': 'https://radio.nrk.no/serie/dagsnytt/sesong/201507/NPUB21019315',
280 'only_matching': True,
285 def _extract_from_mediaelement(self, video_id):
286 api_hosts = (self._api_host, ) if self._api_host else self._API_HOSTS
288 for api_host in api_hosts:
289 data = self._download_json(
290 'http://%s/mediaelement/%s' % (api_host, video_id),
291 video_id, 'Downloading mediaelement JSON',
292 fatal=api_host == api_hosts[-1])
295 self._api_host = api_host
298 title = data.get('fullTitle') or data.get('mainTitle') or data['title']
299 video_id = data.get('id') or video_id
303 conviva = data.get('convivaStatistics') or {}
304 live = (data.get('mediaElementType') == 'Live'
305 or data.get('isLive') is True or conviva.get('isLive'))
308 return self._live_title(t) if live else t
310 media_assets = data.get('mediaAssets')
311 if media_assets and isinstance(media_assets, list):
312 def video_id_and_title(idx):
313 return ((video_id, title) if len(media_assets) == 1
314 else ('%s-%d' % (video_id, idx), '%s (Part %d)' % (title, idx)))
315 for num, asset in enumerate(media_assets, 1):
316 asset_url = asset.get('url')
319 formats = self._extract_akamai_formats(asset_url, video_id)
322 self._sort_formats(formats)
324 # Some f4m streams may not work with hdcore in fragments' URLs
326 extra_param = f.get('extra_param_to_segment_url')
327 if extra_param and 'hdcore' in extra_param:
328 del f['extra_param_to_segment_url']
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')
350 formats = self._extract_akamai_formats(media_url, video_id)
351 self._sort_formats(formats)
352 duration = parse_duration(data.get('duration'))
355 'title': make_title(title),
356 'duration': duration,
362 'ProgramRightsAreNotReady': 'Du kan dessverre ikke se eller høre programmet',
363 'ProgramRightsHasExpired': 'Programmet har gått ut',
364 'NoProgramRights': 'Ikke tilgjengelig',
365 'ProgramIsGeoBlocked': 'NRK har ikke rettigheter til å vise dette programmet utenfor Norge',
367 message_type = data.get('messageType', '')
368 # Can be ProgramIsGeoBlocked or ChannelIsGeoBlocked*
369 if 'IsGeoBlocked' in message_type:
370 self.raise_geo_restricted(
371 msg=MESSAGES.get('ProgramIsGeoBlocked'),
372 countries=self._GEO_COUNTRIES)
373 raise ExtractorError(
374 '%s said: %s' % (self.IE_NAME, MESSAGES.get(
375 message_type, message_type)),
378 series = conviva.get('seriesName') or data.get('seriesTitle')
379 episode = conviva.get('episodeName') or data.get('episodeNumberOrDate')
382 episode_number = None
383 if data.get('mediaElementType') == 'Episode':
384 _season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \
385 data.get('relativeOriginUrl', '')
387 r'/s(?P<season>\d{,2})e(?P<episode>\d{,2})\.',
388 r'/sesong-(?P<season>\d{,2})/episode-(?P<episode>\d{,2})',
390 season_number = int_or_none(self._search_regex(
391 EPISODENUM_RE, _season_episode, 'season number',
392 default=None, group='season'))
393 episode_number = int_or_none(self._search_regex(
394 EPISODENUM_RE, _season_episode, 'episode number',
395 default=None, group='episode'))
398 images = data.get('images')
399 if images and isinstance(images, dict):
400 web_images = images.get('webImages')
401 if isinstance(web_images, list):
403 'url': image['imageUrl'],
404 'width': int_or_none(image.get('width')),
405 'height': int_or_none(image.get('height')),
406 } for image in web_images if image.get('imageUrl')]
408 description = data.get('description')
409 category = data.get('mediaAnalytics', {}).get('category')
412 'description': description,
415 'season_number': season_number,
416 'episode_number': episode_number,
417 'categories': [category] if category else None,
418 'age_limit': parse_age_limit(data.get('legalAge')),
419 'thumbnails': thumbnails,
422 vcodec = 'none' if data.get('mediaType') == 'Audio' else None
424 for entry in entries:
425 entry.update(common_info)
426 for f in entry['formats']:
429 points = data.get('shortIndexPoints')
430 if isinstance(points, list):
432 for next_num, point in enumerate(points, start=1):
433 if not isinstance(point, dict):
435 start_time = parse_duration(point.get('startPoint'))
436 if start_time is None:
438 end_time = parse_duration(
440 if next_num == len(points)
441 else points[next_num].get('startPoint'))
445 'start_time': start_time,
446 'end_time': end_time,
447 'title': point.get('title'),
449 if chapters and len(entries) == 1:
450 entries[0]['chapters'] = chapters
452 return self.playlist_result(entries, video_id, title, description)
454 def _real_extract(self, url):
455 video_id = self._match_id(url)
456 return self._extract_from_mediaelement(video_id)
459 class NRKTVEpisodeIE(InfoExtractor):
460 _VALID_URL = r'https?://tv\.nrk\.no/serie/(?P<id>[^/]+/sesong/\d+/episode/\d+)'
462 'url': 'https://tv.nrk.no/serie/hellums-kro/sesong/1/episode/2',
464 'id': 'MUHH36005220BA',
466 'title': 'Kro, krig og kjærlighet 2:6',
467 'description': 'md5:b32a7dc0b1ed27c8064f58b97bda4350',
469 'series': 'Hellums kro',
476 'skip_download': True,
479 'url': 'https://tv.nrk.no/serie/backstage/sesong/1/episode/8',
481 'id': 'MSUI14000816AA',
483 'title': 'Backstage 8:30',
484 'description': 'md5:de6ca5d5a2d56849e4021f2bf2850df4',
486 'series': 'Backstage',
492 'skip_download': True,
494 'skip': 'ProgramRightsHasExpired',
497 def _real_extract(self, url):
498 display_id = self._match_id(url)
500 webpage = self._download_webpage(url, display_id)
502 info = self._search_json_ld(webpage, display_id, default={})
503 nrk_id = info.get('@id') or self._html_search_meta(
504 'nrk:program-id', webpage, default=None) or self._search_regex(
505 r'data-program-id=["\'](%s)' % NRKTVIE._EPISODE_RE, webpage,
507 assert re.match(NRKTVIE._EPISODE_RE, nrk_id)
510 '_type': 'url_transparent',
512 'url': 'nrk:%s' % nrk_id,
513 'ie_key': NRKIE.ie_key(),
518 class NRKTVSerieBaseIE(InfoExtractor):
519 def _extract_series(self, webpage, display_id, fatal=True):
520 config = self._parse_json(
522 (r'INITIAL_DATA(?:_V\d)?_*\s*=\s*({.+?})\s*;',
523 r'({.+?})\s*,\s*"[^"]+"\s*\)\s*</script>'),
524 webpage, 'config', default='{}' if not fatal else NO_DEFAULT),
525 display_id, fatal=False, transform_source=js_to_json)
530 (lambda x: x['initialState']['series'], lambda x: x['series']),
533 def _extract_seasons(self, seasons):
534 if not isinstance(seasons, list):
537 for season in seasons:
538 entries.extend(self._extract_episodes(season))
541 def _extract_episodes(self, season):
542 if not isinstance(season, dict):
544 return self._extract_entries(season.get('episodes'))
546 def _extract_entries(self, entry_list):
547 if not isinstance(entry_list, list):
550 for episode in entry_list:
551 nrk_id = episode.get('prfId')
552 if not nrk_id or not isinstance(nrk_id, compat_str):
554 entries.append(self.url_result(
555 'nrk:%s' % nrk_id, ie=NRKIE.ie_key(), video_id=nrk_id))
559 class NRKTVSeasonIE(NRKTVSerieBaseIE):
560 _VALID_URL = r'https?://tv\.nrk\.no/serie/[^/]+/sesong/(?P<id>\d+)'
562 'url': 'https://tv.nrk.no/serie/backstage/sesong/1',
567 'playlist_mincount': 30,
571 def suitable(cls, url):
572 return (False if NRKTVIE.suitable(url) or NRKTVEpisodeIE.suitable(url)
573 else super(NRKTVSeasonIE, cls).suitable(url))
575 def _real_extract(self, url):
576 display_id = self._match_id(url)
578 webpage = self._download_webpage(url, display_id)
580 series = self._extract_series(webpage, display_id)
583 s for s in series['seasons']
584 if int(display_id) == s.get('seasonNumber'))
586 title = try_get(season, lambda x: x['titles']['title'], compat_str)
587 return self.playlist_result(
588 self._extract_episodes(season), display_id, title)
591 class NRKTVSeriesIE(NRKTVSerieBaseIE):
592 _VALID_URL = r'https?://(?:tv|radio)\.nrk(?:super)?\.no/serie/(?P<id>[^/]+)'
593 _ITEM_RE = r'(?:data-season=["\']|id=["\']season-)(?P<id>\d+)'
595 'url': 'https://tv.nrk.no/serie/blank',
599 'description': 'md5:7664b4e7e77dc6810cd3bca367c25b6e',
601 'playlist_mincount': 30,
603 # new layout, seasons
604 'url': 'https://tv.nrk.no/serie/backstage',
607 'title': 'Backstage',
608 'description': 'md5:c3ec3a35736fca0f9e1207b5511143d3',
610 'playlist_mincount': 60,
612 # new layout, instalments
613 'url': 'https://tv.nrk.no/serie/groenn-glede',
615 'id': 'groenn-glede',
616 'title': 'Grønn glede',
617 'description': 'md5:7576e92ae7f65da6993cf90ee29e4608',
619 'playlist_mincount': 10,
622 'url': 'https://tv.nrksuper.no/serie/labyrint',
626 'description': 'md5:318b597330fdac5959247c9b69fdb1ec',
628 'playlist_mincount': 3,
630 'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene',
631 'only_matching': True,
633 'url': 'https://tv.nrk.no/serie/saving-the-human-race',
634 'only_matching': True,
636 'url': 'https://tv.nrk.no/serie/postmann-pat',
637 'only_matching': True,
641 def suitable(cls, url):
643 False if any(ie.suitable(url)
644 for ie in (NRKTVIE, NRKTVEpisodeIE, NRKTVSeasonIE))
645 else super(NRKTVSeriesIE, cls).suitable(url))
647 def _real_extract(self, url):
648 series_id = self._match_id(url)
650 webpage = self._download_webpage(url, series_id)
652 # New layout (e.g. https://tv.nrk.no/serie/backstage)
653 series = self._extract_series(webpage, series_id, fatal=False)
655 title = try_get(series, lambda x: x['titles']['title'], compat_str)
656 description = try_get(
657 series, lambda x: x['titles']['subtitle'], compat_str)
659 entries.extend(self._extract_seasons(series.get('seasons')))
660 entries.extend(self._extract_entries(series.get('instalments')))
661 entries.extend(self._extract_episodes(series.get('extraMaterial')))
662 return self.playlist_result(entries, series_id, title, description)
664 # Old layout (e.g. https://tv.nrksuper.no/serie/labyrint)
667 'https://tv.nrk.no/program/Episodes/{series}/{season}'.format(
668 series=series_id, season=season_id))
669 for season_id in re.findall(self._ITEM_RE, webpage)
672 title = self._html_search_meta(
673 'seriestitle', webpage,
674 'title', default=None) or self._og_search_title(
675 webpage, fatal=False)
677 title = self._search_regex(
678 r'NRK (?:Super )?TV\s*[-–]\s*(.+)', title, 'title', default=title)
680 description = self._html_search_meta(
681 'series_description', webpage,
682 'description', default=None) or self._og_search_description(webpage)
684 return self.playlist_result(entries, series_id, title, description)
687 class NRKTVDirekteIE(NRKTVIE):
688 IE_DESC = 'NRK TV Direkte and NRK Radio Direkte'
689 _VALID_URL = r'https?://(?:tv|radio)\.nrk\.no/direkte/(?P<id>[^/?#&]+)'
692 'url': 'https://tv.nrk.no/direkte/nrk1',
693 'only_matching': True,
695 'url': 'https://radio.nrk.no/direkte/p1_oslo_akershus',
696 'only_matching': True,
700 class NRKPlaylistBaseIE(InfoExtractor):
701 def _extract_description(self, webpage):
704 def _real_extract(self, url):
705 playlist_id = self._match_id(url)
707 webpage = self._download_webpage(url, playlist_id)
710 self.url_result('nrk:%s' % video_id, NRKIE.ie_key())
711 for video_id in re.findall(self._ITEM_RE, webpage)
714 playlist_title = self. _extract_title(webpage)
715 playlist_description = self._extract_description(webpage)
717 return self.playlist_result(
718 entries, playlist_id, playlist_title, playlist_description)
721 class NRKPlaylistIE(NRKPlaylistBaseIE):
722 _VALID_URL = r'https?://(?:www\.)?nrk\.no/(?!video|skole)(?:[^/]+/)+(?P<id>[^/]+)'
723 _ITEM_RE = r'class="[^"]*\brich\b[^"]*"[^>]+data-video-id="([^"]+)"'
725 'url': 'http://www.nrk.no/troms/gjenopplev-den-historiske-solformorkelsen-1.12270763',
727 'id': 'gjenopplev-den-historiske-solformorkelsen-1.12270763',
728 'title': 'Gjenopplev den historiske solformørkelsen',
729 'description': 'md5:c2df8ea3bac5654a26fc2834a542feed',
733 'url': 'http://www.nrk.no/kultur/bok/rivertonprisen-til-karin-fossum-1.12266449',
735 'id': 'rivertonprisen-til-karin-fossum-1.12266449',
736 'title': 'Rivertonprisen til Karin Fossum',
737 'description': 'Første kvinne på 15 år til å vinne krimlitteraturprisen.',
742 def _extract_title(self, webpage):
743 return self._og_search_title(webpage, fatal=False)
745 def _extract_description(self, webpage):
746 return self._og_search_description(webpage)
749 class NRKTVEpisodesIE(NRKPlaylistBaseIE):
750 _VALID_URL = r'https?://tv\.nrk\.no/program/[Ee]pisodes/[^/]+/(?P<id>\d+)'
751 _ITEM_RE = r'data-episode=["\']%s' % NRKTVIE._EPISODE_RE
753 'url': 'https://tv.nrk.no/program/episodes/nytt-paa-nytt/69031',
756 'title': 'Nytt på nytt, sesong: 201210',
761 def _extract_title(self, webpage):
762 return self._html_search_regex(
763 r'<h1>([^<]+)</h1>', webpage, 'title', fatal=False)
766 class NRKSkoleIE(InfoExtractor):
767 IE_DESC = 'NRK Skole'
768 _VALID_URL = r'https?://(?:www\.)?nrk\.no/skole/?\?.*\bmediaId=(?P<id>\d+)'
771 'url': 'https://www.nrk.no/skole/?page=search&q=&mediaId=14099',
772 'md5': '18c12c3d071953c3bf8d54ef6b2587b7',
776 'title': 'Genetikk og eneggede tvillinger',
777 'description': 'md5:3aca25dcf38ec30f0363428d2b265f8d',
781 'url': 'https://www.nrk.no/skole/?page=objectives&subject=naturfag&objective=K15114&mediaId=19355',
782 'only_matching': True,
785 def _real_extract(self, url):
786 video_id = self._match_id(url)
788 webpage = self._download_webpage(
789 'https://mimir.nrk.no/plugin/1.0/static?mediaId=%s' % video_id,
792 nrk_id = self._parse_json(
794 r'<script[^>]+type=["\']application/json["\'][^>]*>({.+?})</script>',
795 webpage, 'application json'),
796 video_id)['activeMedia']['psId']
798 return self.url_result('nrk:%s' % nrk_id)