2 from __future__ import unicode_literals
6 from .common import InfoExtractor
9 compat_urllib_parse_unquote,
24 class NRKBaseIE(InfoExtractor):
25 _GEO_COUNTRIES = ['NO']
27 def _extract_nrk_formats(self, asset_url, video_id):
28 return self._extract_m3u8_formats(
29 re.sub(r'(?:bw_(?:low|high)=\d+|no_audio_only)&?', '', asset_url),
30 video_id, 'mp4', 'm3u8_native', fatal=False)
33 class NRKIE(NRKBaseIE):
39 (?:www\.)?nrk\.no/video/(?:PS\*|[^_]+_)|
40 v8[-.]psapi\.nrk\.no/mediaelement/
48 'url': 'http://www.nrk.no/video/PS*150533',
49 'md5': '706f34cdf1322577589e369e522b50ef',
53 'title': 'Dompap og andre fugler i Piip-Show',
54 'description': 'md5:d9261ba34c43b61c812cb6b0269a5c8f',
59 'url': 'http://www.nrk.no/video/PS*154915',
64 'title': 'Slik høres internett ut når du er blind',
65 'description': 'md5:a621f5cc1bd75c8d5104cb048c6b8568',
69 'url': 'nrk:ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
70 'only_matching': True,
72 'url': 'nrk:clip/7707d5a3-ebe7-434a-87d5-a3ebe7a34a70',
73 'only_matching': True,
75 'url': 'https://v8-psapi.nrk.no/mediaelement/ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
76 'only_matching': True,
78 'url': 'https://www.nrk.no/video/dompap-og-andre-fugler-i-piip-show_150533',
79 'only_matching': True,
81 'url': 'https://www.nrk.no/video/humor/kommentatorboksen-reiser-til-sjos_d1fda11f-a4ad-437a-a374-0398bc84e999',
82 'only_matching': True,
85 def _extract_from_playback(self, video_id):
86 manifest = self._download_json(
87 'http://psapi.nrk.no/playback/manifest/%s' % video_id,
88 video_id, 'Downloading manifest JSON')
90 playable = manifest['playable']
93 for asset in playable['assets']:
94 if not isinstance(asset, dict):
96 if asset.get('encrypted'):
98 format_url = url_or_none(asset.get('url'))
101 if asset.get('format') == 'HLS' or determine_ext(format_url) == 'm3u8':
102 formats.extend(self._extract_nrk_formats(format_url, video_id))
103 self._sort_formats(formats)
105 data = self._download_json(
106 'http://psapi.nrk.no/playback/metadata/%s' % video_id,
107 video_id, 'Downloading metadata JSON')
109 preplay = data['preplay']
110 titles = preplay['titles']
111 title = titles['title']
112 alt_title = titles.get('subtitle')
114 description = preplay.get('description')
115 duration = parse_duration(playable.get('duration')) or parse_duration(data.get('duration'))
118 for image in try_get(
119 preplay, lambda x: x['poster']['images'], list) or []:
120 if not isinstance(image, dict):
122 image_url = url_or_none(image.get('url'))
127 'width': int_or_none(image.get('pixelWidth')),
128 'height': int_or_none(image.get('pixelHeight')),
134 'alt_title': alt_title,
135 'description': description,
136 'duration': duration,
137 'thumbnails': thumbnails,
141 def _real_extract(self, url):
142 video_id = self._match_id(url)
143 return self._extract_from_playback(video_id)
146 class NRKTVIE(NRKBaseIE):
147 IE_DESC = 'NRK TV and NRK Radio'
148 _EPISODE_RE = r'(?P<id>[a-zA-Z]{4}\d{8})'
149 _VALID_URL = r'''(?x)
151 (?:tv|radio)\.nrk(?:super)?\.no/
152 (?:serie(?:/[^/]+){1,}|program)/
154 (?:/\d{2}-\d{2}-\d{4})?
155 (?:\#del=(?P<part_id>\d+))?
157 _API_HOSTS = ('psapi-ne.nrk.no', 'psapi-we.nrk.no')
159 'url': 'https://tv.nrk.no/program/MDDP12000117',
160 'md5': '8270824df46ec629b66aeaa5796b36fb',
162 'id': 'MDDP12000117AA',
164 'title': 'Alarm Trolltunga',
165 'description': 'md5:46923a6e6510eefcce23d5ef2a58f2ce',
170 'url': 'https://tv.nrk.no/serie/20-spoersmaal-tv/MUHH48000314/23-05-2014',
171 'md5': '9a167e54d04671eb6317a37b7bc8a280',
173 'id': 'MUHH48000314AA',
175 'title': '20 spørsmål 23.05.2014',
176 'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
178 'series': '20 spørsmål',
179 'episode': '23.05.2014',
181 'skip': 'NoProgramRights',
183 'url': 'https://tv.nrk.no/program/mdfp15000514',
185 'id': 'MDFP15000514CA',
187 'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014',
188 'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db',
190 'series': 'Kunnskapskanalen',
191 'episode': '24.05.2014',
194 'skip_download': True,
197 # single playlist video
198 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2',
200 'id': 'MSPO40010515-part2',
202 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
203 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
206 'skip_download': True,
208 'expected_warnings': ['Video is geo restricted'],
209 'skip': 'particular part is not supported currently',
211 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015',
214 'id': 'MSPO40010515AH',
216 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 1)',
217 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
219 'series': 'Tour de Ski',
220 'episode': '06.01.2015',
223 'skip_download': True,
227 'id': 'MSPO40010515BH',
229 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 2)',
230 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
232 'series': 'Tour de Ski',
233 'episode': '06.01.2015',
236 'skip_download': True,
240 'id': 'MSPO40010515',
241 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015',
242 'description': 'md5:1f97a41f05a9486ee00c56f35f82993d',
244 'expected_warnings': ['Video is geo restricted'],
246 'url': 'https://tv.nrk.no/serie/anno/KMTE50001317/sesong-3/episode-13',
248 'id': 'KMTE50001317AA',
250 'title': 'Anno 13:30',
251 'description': 'md5:11d9613661a8dbe6f9bef54e3a4cbbfa',
256 'episode_number': 13,
259 'skip_download': True,
262 'url': 'https://tv.nrk.no/serie/nytt-paa-nytt/MUHH46000317/27-01-2017',
264 'id': 'MUHH46000317AA',
266 'title': 'Nytt på Nytt 27.01.2017',
267 'description': 'md5:5358d6388fba0ea6f0b6d11c48b9eb4b',
269 'series': 'Nytt på nytt',
270 'episode': '27.01.2017',
273 'skip_download': True,
276 'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#',
277 'only_matching': True,
279 'url': 'https://tv.nrk.no/serie/lindmo/2018/MUHU11006318/avspiller',
280 'only_matching': True,
282 'url': 'https://radio.nrk.no/serie/dagsnytt/sesong/201507/NPUB21019315',
283 'only_matching': True,
288 def _extract_from_mediaelement(self, video_id):
289 api_hosts = (self._api_host, ) if self._api_host else self._API_HOSTS
291 for api_host in api_hosts:
292 data = self._download_json(
293 'http://%s/mediaelement/%s' % (api_host, video_id),
294 video_id, 'Downloading mediaelement JSON',
295 fatal=api_host == api_hosts[-1])
298 self._api_host = api_host
301 title = data.get('fullTitle') or data.get('mainTitle') or data['title']
302 video_id = data.get('id') or video_id
307 conviva = data.get('convivaStatistics') or {}
308 live = (data.get('mediaElementType') == 'Live'
309 or data.get('isLive') is True or conviva.get('isLive'))
312 return self._live_title(t) if live else t
314 media_assets = data.get('mediaAssets')
315 if media_assets and isinstance(media_assets, list):
316 def video_id_and_title(idx):
317 return ((video_id, title) if len(media_assets) == 1
318 else ('%s-%d' % (video_id, idx), '%s (Part %d)' % (title, idx)))
319 for num, asset in enumerate(media_assets, 1):
320 asset_url = asset.get('url')
321 if not asset_url or asset_url in urls:
323 formats = self._extract_nrk_formats(asset_url, video_id)
326 self._sort_formats(formats)
328 entry_id, entry_title = video_id_and_title(num)
329 duration = parse_duration(asset.get('duration'))
331 for subtitle in ('webVtt', 'timedText'):
332 subtitle_url = asset.get('%sSubtitlesUrl' % subtitle)
334 subtitles.setdefault('no', []).append({
335 'url': compat_urllib_parse_unquote(subtitle_url)
338 'id': asset.get('carrierId') or entry_id,
339 'title': make_title(entry_title),
340 'duration': duration,
341 'subtitles': subtitles,
346 media_url = data.get('mediaUrl')
347 if media_url and media_url not in urls:
348 formats = self._extract_nrk_formats(media_url, video_id)
350 self._sort_formats(formats)
351 duration = parse_duration(data.get('duration'))
354 'title': make_title(title),
355 'duration': duration,
361 'ProgramRightsAreNotReady': 'Du kan dessverre ikke se eller høre programmet',
362 'ProgramRightsHasExpired': 'Programmet har gått ut',
363 'NoProgramRights': 'Ikke tilgjengelig',
364 'ProgramIsGeoBlocked': 'NRK har ikke rettigheter til å vise dette programmet utenfor Norge',
366 message_type = data.get('messageType', '')
367 # Can be ProgramIsGeoBlocked or ChannelIsGeoBlocked*
368 if 'IsGeoBlocked' in message_type or try_get(data, lambda x: x['usageRights']['isGeoBlocked']) is True:
369 self.raise_geo_restricted(
370 msg=MESSAGES.get('ProgramIsGeoBlocked'),
371 countries=self._GEO_COUNTRIES)
372 raise ExtractorError(
373 '%s said: %s' % (self.IE_NAME, MESSAGES.get(
374 message_type, message_type)),
377 series = conviva.get('seriesName') or data.get('seriesTitle')
378 episode = conviva.get('episodeName') or data.get('episodeNumberOrDate')
381 episode_number = None
382 if data.get('mediaElementType') == 'Episode':
383 _season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \
384 data.get('relativeOriginUrl', '')
386 r'/s(?P<season>\d{,2})e(?P<episode>\d{,2})\.',
387 r'/sesong-(?P<season>\d{,2})/episode-(?P<episode>\d{,2})',
389 season_number = int_or_none(self._search_regex(
390 EPISODENUM_RE, _season_episode, 'season number',
391 default=None, group='season'))
392 episode_number = int_or_none(self._search_regex(
393 EPISODENUM_RE, _season_episode, 'episode number',
394 default=None, group='episode'))
397 images = data.get('images')
398 if images and isinstance(images, dict):
399 web_images = images.get('webImages')
400 if isinstance(web_images, list):
402 'url': image['imageUrl'],
403 'width': int_or_none(image.get('width')),
404 'height': int_or_none(image.get('height')),
405 } for image in web_images if image.get('imageUrl')]
407 description = data.get('description')
408 category = data.get('mediaAnalytics', {}).get('category')
411 'description': description,
414 'season_number': season_number,
415 'episode_number': episode_number,
416 'categories': [category] if category else None,
417 'age_limit': parse_age_limit(data.get('legalAge')),
418 'thumbnails': thumbnails,
421 vcodec = 'none' if data.get('mediaType') == 'Audio' else None
423 for entry in entries:
424 entry.update(common_info)
425 for f in entry['formats']:
428 points = data.get('shortIndexPoints')
429 if isinstance(points, list):
431 for next_num, point in enumerate(points, start=1):
432 if not isinstance(point, dict):
434 start_time = parse_duration(point.get('startPoint'))
435 if start_time is None:
437 end_time = parse_duration(
439 if next_num == len(points)
440 else points[next_num].get('startPoint'))
444 'start_time': start_time,
445 'end_time': end_time,
446 'title': point.get('title'),
448 if chapters and len(entries) == 1:
449 entries[0]['chapters'] = chapters
451 return self.playlist_result(entries, video_id, title, description)
453 def _real_extract(self, url):
454 video_id = self._match_id(url)
455 return self._extract_from_mediaelement(video_id)
458 class NRKTVEpisodeIE(InfoExtractor):
459 _VALID_URL = r'https?://tv\.nrk\.no/serie/(?P<id>[^/]+/sesong/\d+/episode/\d+)'
461 'url': 'https://tv.nrk.no/serie/hellums-kro/sesong/1/episode/2',
463 'id': 'MUHH36005220BA',
465 'title': 'Kro, krig og kjærlighet 2:6',
466 'description': 'md5:b32a7dc0b1ed27c8064f58b97bda4350',
468 'series': 'Hellums kro',
475 'skip_download': True,
478 'url': 'https://tv.nrk.no/serie/backstage/sesong/1/episode/8',
480 'id': 'MSUI14000816AA',
482 'title': 'Backstage 8:30',
483 'description': 'md5:de6ca5d5a2d56849e4021f2bf2850df4',
485 'series': 'Backstage',
491 'skip_download': True,
493 'skip': 'ProgramRightsHasExpired',
496 def _real_extract(self, url):
497 display_id = self._match_id(url)
499 webpage = self._download_webpage(url, display_id)
501 info = self._search_json_ld(webpage, display_id, default={})
502 nrk_id = info.get('@id') or self._html_search_meta(
503 'nrk:program-id', webpage, default=None) or self._search_regex(
504 r'data-program-id=["\'](%s)' % NRKTVIE._EPISODE_RE, webpage,
506 assert re.match(NRKTVIE._EPISODE_RE, nrk_id)
509 '_type': 'url_transparent',
511 'url': 'nrk:%s' % nrk_id,
512 'ie_key': NRKIE.ie_key(),
517 class NRKTVSerieBaseIE(InfoExtractor):
518 def _extract_series(self, webpage, display_id, fatal=True):
519 config = self._parse_json(
521 (r'INITIAL_DATA(?:_V\d)?_*\s*=\s*({.+?})\s*;',
522 r'({.+?})\s*,\s*"[^"]+"\s*\)\s*</script>'),
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, seasons):
533 if not isinstance(seasons, list):
536 for season in seasons:
537 entries.extend(self._extract_episodes(season))
540 def _extract_episodes(self, season):
541 if not isinstance(season, dict):
543 return self._extract_entries(season.get('episodes'))
545 def _extract_entries(self, entry_list):
546 if not isinstance(entry_list, list):
549 for episode in entry_list:
550 nrk_id = episode.get('prfId')
551 if not nrk_id or not isinstance(nrk_id, compat_str):
553 entries.append(self.url_result(
554 'nrk:%s' % nrk_id, ie=NRKIE.ie_key(), video_id=nrk_id))
558 class NRKTVSeasonIE(NRKTVSerieBaseIE):
559 _VALID_URL = r'https?://tv\.nrk\.no/serie/[^/]+/sesong/(?P<id>\d+)'
561 'url': 'https://tv.nrk.no/serie/backstage/sesong/1',
566 'playlist_mincount': 30,
570 def suitable(cls, url):
571 return (False if NRKTVIE.suitable(url) or NRKTVEpisodeIE.suitable(url)
572 else super(NRKTVSeasonIE, cls).suitable(url))
574 def _real_extract(self, url):
575 display_id = self._match_id(url)
577 webpage = self._download_webpage(url, display_id)
579 series = self._extract_series(webpage, display_id)
582 s for s in series['seasons']
583 if int(display_id) == s.get('seasonNumber'))
585 title = try_get(season, lambda x: x['titles']['title'], compat_str)
586 return self.playlist_result(
587 self._extract_episodes(season), display_id, title)
590 class NRKTVSeriesIE(NRKTVSerieBaseIE):
591 _VALID_URL = r'https?://(?:tv|radio)\.nrk(?:super)?\.no/serie/(?P<id>[^/]+)'
592 _ITEM_RE = r'(?:data-season=["\']|id=["\']season-)(?P<id>\d+)'
594 'url': 'https://tv.nrk.no/serie/blank',
598 'description': 'md5:7664b4e7e77dc6810cd3bca367c25b6e',
600 'playlist_mincount': 30,
602 # new layout, seasons
603 'url': 'https://tv.nrk.no/serie/backstage',
606 'title': 'Backstage',
607 'description': 'md5:c3ec3a35736fca0f9e1207b5511143d3',
609 'playlist_mincount': 60,
611 # new layout, instalments
612 'url': 'https://tv.nrk.no/serie/groenn-glede',
614 'id': 'groenn-glede',
615 'title': 'Grønn glede',
616 'description': 'md5:7576e92ae7f65da6993cf90ee29e4608',
618 'playlist_mincount': 10,
621 'url': 'https://tv.nrksuper.no/serie/labyrint',
625 'description': 'md5:318b597330fdac5959247c9b69fdb1ec',
627 'playlist_mincount': 3,
629 'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene',
630 'only_matching': True,
632 'url': 'https://tv.nrk.no/serie/saving-the-human-race',
633 'only_matching': True,
635 'url': 'https://tv.nrk.no/serie/postmann-pat',
636 'only_matching': True,
640 def suitable(cls, url):
642 False if any(ie.suitable(url)
643 for ie in (NRKTVIE, NRKTVEpisodeIE, NRKTVSeasonIE))
644 else super(NRKTVSeriesIE, cls).suitable(url))
646 def _real_extract(self, url):
647 series_id = self._match_id(url)
649 webpage = self._download_webpage(url, series_id)
651 # New layout (e.g. https://tv.nrk.no/serie/backstage)
652 series = self._extract_series(webpage, series_id, fatal=False)
654 title = try_get(series, lambda x: x['titles']['title'], compat_str)
655 description = try_get(
656 series, lambda x: x['titles']['subtitle'], compat_str)
658 entries.extend(self._extract_seasons(series.get('seasons')))
659 entries.extend(self._extract_entries(series.get('instalments')))
660 entries.extend(self._extract_episodes(series.get('extraMaterial')))
661 return self.playlist_result(entries, series_id, title, description)
663 # Old layout (e.g. https://tv.nrksuper.no/serie/labyrint)
666 'https://tv.nrk.no/program/Episodes/{series}/{season}'.format(
667 series=series_id, season=season_id))
668 for season_id in re.findall(self._ITEM_RE, webpage)
671 title = self._html_search_meta(
672 'seriestitle', webpage,
673 'title', default=None) or self._og_search_title(
674 webpage, fatal=False)
676 title = self._search_regex(
677 r'NRK (?:Super )?TV\s*[-–]\s*(.+)', title, 'title', default=title)
679 description = self._html_search_meta(
680 'series_description', webpage,
681 'description', default=None) or self._og_search_description(webpage)
683 return self.playlist_result(entries, series_id, title, description)
686 class NRKTVDirekteIE(NRKTVIE):
687 IE_DESC = 'NRK TV Direkte and NRK Radio Direkte'
688 _VALID_URL = r'https?://(?:tv|radio)\.nrk\.no/direkte/(?P<id>[^/?#&]+)'
691 'url': 'https://tv.nrk.no/direkte/nrk1',
692 'only_matching': True,
694 'url': 'https://radio.nrk.no/direkte/p1_oslo_akershus',
695 'only_matching': True,
699 class NRKPlaylistBaseIE(InfoExtractor):
700 def _extract_description(self, webpage):
703 def _real_extract(self, url):
704 playlist_id = self._match_id(url)
706 webpage = self._download_webpage(url, playlist_id)
709 self.url_result('nrk:%s' % video_id, NRKIE.ie_key())
710 for video_id in re.findall(self._ITEM_RE, webpage)
713 playlist_title = self. _extract_title(webpage)
714 playlist_description = self._extract_description(webpage)
716 return self.playlist_result(
717 entries, playlist_id, playlist_title, playlist_description)
720 class NRKPlaylistIE(NRKPlaylistBaseIE):
721 _VALID_URL = r'https?://(?:www\.)?nrk\.no/(?!video|skole)(?:[^/]+/)+(?P<id>[^/]+)'
722 _ITEM_RE = r'class="[^"]*\brich\b[^"]*"[^>]+data-video-id="([^"]+)"'
724 'url': 'http://www.nrk.no/troms/gjenopplev-den-historiske-solformorkelsen-1.12270763',
726 'id': 'gjenopplev-den-historiske-solformorkelsen-1.12270763',
727 'title': 'Gjenopplev den historiske solformørkelsen',
728 'description': 'md5:c2df8ea3bac5654a26fc2834a542feed',
732 'url': 'http://www.nrk.no/kultur/bok/rivertonprisen-til-karin-fossum-1.12266449',
734 'id': 'rivertonprisen-til-karin-fossum-1.12266449',
735 'title': 'Rivertonprisen til Karin Fossum',
736 'description': 'Første kvinne på 15 år til å vinne krimlitteraturprisen.',
741 def _extract_title(self, webpage):
742 return self._og_search_title(webpage, fatal=False)
744 def _extract_description(self, webpage):
745 return self._og_search_description(webpage)
748 class NRKTVEpisodesIE(NRKPlaylistBaseIE):
749 _VALID_URL = r'https?://tv\.nrk\.no/program/[Ee]pisodes/[^/]+/(?P<id>\d+)'
750 _ITEM_RE = r'data-episode=["\']%s' % NRKTVIE._EPISODE_RE
752 'url': 'https://tv.nrk.no/program/episodes/nytt-paa-nytt/69031',
755 'title': 'Nytt på nytt, sesong: 201210',
760 def _extract_title(self, webpage):
761 return self._html_search_regex(
762 r'<h1>([^<]+)</h1>', webpage, 'title', fatal=False)
765 class NRKSkoleIE(InfoExtractor):
766 IE_DESC = 'NRK Skole'
767 _VALID_URL = r'https?://(?:www\.)?nrk\.no/skole/?\?.*\bmediaId=(?P<id>\d+)'
770 'url': 'https://www.nrk.no/skole/?page=search&q=&mediaId=14099',
771 'md5': '18c12c3d071953c3bf8d54ef6b2587b7',
775 'title': 'Genetikk og eneggede tvillinger',
776 'description': 'md5:3aca25dcf38ec30f0363428d2b265f8d',
780 'url': 'https://www.nrk.no/skole/?page=objectives&subject=naturfag&objective=K15114&mediaId=19355',
781 'only_matching': True,
784 def _real_extract(self, url):
785 video_id = self._match_id(url)
787 webpage = self._download_webpage(
788 'https://mimir.nrk.no/plugin/1.0/static?mediaId=%s' % video_id,
791 nrk_id = self._parse_json(
793 r'<script[^>]+type=["\']application/json["\'][^>]*>({.+?})</script>',
794 webpage, 'application json'),
795 video_id)['activeMedia']['psId']
797 return self.url_result('nrk:%s' % nrk_id)