2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..compat import compat_urllib_parse_unquote
17 class NRKBaseIE(InfoExtractor):
18 _GEO_COUNTRIES = ['NO']
22 def _real_extract(self, url):
23 video_id = self._match_id(url)
25 api_hosts = (self._api_host, ) if self._api_host else self._API_HOSTS
27 for api_host in api_hosts:
28 data = self._download_json(
29 'http://%s/mediaelement/%s' % (api_host, video_id),
30 video_id, 'Downloading mediaelement JSON',
31 fatal=api_host == api_hosts[-1])
34 self._api_host = api_host
37 title = data.get('fullTitle') or data.get('mainTitle') or data['title']
38 video_id = data.get('id') or video_id
42 conviva = data.get('convivaStatistics') or {}
43 live = (data.get('mediaElementType') == 'Live' or
44 data.get('isLive') is True or conviva.get('isLive'))
47 return self._live_title(t) if live else t
49 media_assets = data.get('mediaAssets')
50 if media_assets and isinstance(media_assets, list):
51 def video_id_and_title(idx):
52 return ((video_id, title) if len(media_assets) == 1
53 else ('%s-%d' % (video_id, idx), '%s (Part %d)' % (title, idx)))
54 for num, asset in enumerate(media_assets, 1):
55 asset_url = asset.get('url')
58 formats = self._extract_akamai_formats(asset_url, video_id)
61 self._sort_formats(formats)
63 # Some f4m streams may not work with hdcore in fragments' URLs
65 extra_param = f.get('extra_param_to_segment_url')
66 if extra_param and 'hdcore' in extra_param:
67 del f['extra_param_to_segment_url']
69 entry_id, entry_title = video_id_and_title(num)
70 duration = parse_duration(asset.get('duration'))
72 for subtitle in ('webVtt', 'timedText'):
73 subtitle_url = asset.get('%sSubtitlesUrl' % subtitle)
75 subtitles.setdefault('no', []).append({
76 'url': compat_urllib_parse_unquote(subtitle_url)
79 'id': asset.get('carrierId') or entry_id,
80 'title': make_title(entry_title),
82 'subtitles': subtitles,
87 media_url = data.get('mediaUrl')
89 formats = self._extract_akamai_formats(media_url, video_id)
90 self._sort_formats(formats)
91 duration = parse_duration(data.get('duration'))
94 'title': make_title(title),
101 'ProgramRightsAreNotReady': 'Du kan dessverre ikke se eller høre programmet',
102 'ProgramRightsHasExpired': 'Programmet har gått ut',
103 'ProgramIsGeoBlocked': 'NRK har ikke rettigheter til å vise dette programmet utenfor Norge',
105 message_type = data.get('messageType', '')
106 # Can be ProgramIsGeoBlocked or ChannelIsGeoBlocked*
107 if 'IsGeoBlocked' in message_type:
108 self.raise_geo_restricted(
109 msg=MESSAGES.get('ProgramIsGeoBlocked'),
110 countries=self._GEO_COUNTRIES)
111 raise ExtractorError(
112 '%s said: %s' % (self.IE_NAME, MESSAGES.get(
113 message_type, message_type)),
116 series = conviva.get('seriesName') or data.get('seriesTitle')
117 episode = conviva.get('episodeName') or data.get('episodeNumberOrDate')
120 episode_number = None
121 if data.get('mediaElementType') == 'Episode':
122 _season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \
123 data.get('relativeOriginUrl', '')
125 r'/s(?P<season>\d{,2})e(?P<episode>\d{,2})\.',
126 r'/sesong-(?P<season>\d{,2})/episode-(?P<episode>\d{,2})',
128 season_number = int_or_none(self._search_regex(
129 EPISODENUM_RE, _season_episode, 'season number',
130 default=None, group='season'))
131 episode_number = int_or_none(self._search_regex(
132 EPISODENUM_RE, _season_episode, 'episode number',
133 default=None, group='episode'))
136 images = data.get('images')
137 if images and isinstance(images, dict):
138 web_images = images.get('webImages')
139 if isinstance(web_images, list):
141 'url': image['imageUrl'],
142 'width': int_or_none(image.get('width')),
143 'height': int_or_none(image.get('height')),
144 } for image in web_images if image.get('imageUrl')]
146 description = data.get('description')
147 category = data.get('mediaAnalytics', {}).get('category')
150 'description': description,
153 'season_number': season_number,
154 'episode_number': episode_number,
155 'categories': [category] if category else None,
156 'age_limit': parse_age_limit(data.get('legalAge')),
157 'thumbnails': thumbnails,
160 vcodec = 'none' if data.get('mediaType') == 'Audio' else None
162 for entry in entries:
163 entry.update(common_info)
164 for f in entry['formats']:
167 points = data.get('shortIndexPoints')
168 if isinstance(points, list):
170 for next_num, point in enumerate(points, start=1):
171 if not isinstance(point, dict):
173 start_time = parse_duration(point.get('startPoint'))
174 if start_time is None:
176 end_time = parse_duration(
178 if next_num == len(points)
179 else points[next_num].get('startPoint'))
183 'start_time': start_time,
184 'end_time': end_time,
185 'title': point.get('title'),
187 if chapters and len(entries) == 1:
188 entries[0]['chapters'] = chapters
190 return self.playlist_result(entries, video_id, title, description)
193 class NRKIE(NRKBaseIE):
194 _VALID_URL = r'''(?x)
199 (?:www\.)?nrk\.no/video/PS\*|
200 v8[-.]psapi\.nrk\.no/mediaelement/
205 _API_HOSTS = ('psapi.nrk.no', 'v8-psapi.nrk.no')
208 'url': 'http://www.nrk.no/video/PS*150533',
209 'md5': '2f7f6eeb2aacdd99885f355428715cfa',
213 'title': 'Dompap og andre fugler i Piip-Show',
214 'description': 'md5:d9261ba34c43b61c812cb6b0269a5c8f',
219 'url': 'http://www.nrk.no/video/PS*154915',
224 'title': 'Slik høres internett ut når du er blind',
225 'description': 'md5:a621f5cc1bd75c8d5104cb048c6b8568',
229 'url': 'nrk:ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
230 'only_matching': True,
232 'url': 'nrk:clip/7707d5a3-ebe7-434a-87d5-a3ebe7a34a70',
233 'only_matching': True,
235 'url': 'https://v8-psapi.nrk.no/mediaelement/ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
236 'only_matching': True,
240 class NRKTVIE(NRKBaseIE):
241 IE_DESC = 'NRK TV and NRK Radio'
242 _EPISODE_RE = r'(?P<id>[a-zA-Z]{4}\d{8})'
243 _VALID_URL = r'''(?x)
245 (?:tv|radio)\.nrk(?:super)?\.no/
246 (?:serie/[^/]+|program)/
248 (?:/\d{2}-\d{2}-\d{4})?
249 (?:\#del=(?P<part_id>\d+))?
251 _API_HOSTS = ('psapi-ne.nrk.no', 'psapi-we.nrk.no')
253 'url': 'https://tv.nrk.no/serie/20-spoersmaal-tv/MUHH48000314/23-05-2014',
254 'md5': '4e9ca6629f09e588ed240fb11619922a',
256 'id': 'MUHH48000314AA',
258 'title': '20 spørsmål 23.05.2014',
259 'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
261 'series': '20 spørsmål - TV',
262 'episode': '23.05.2014',
265 'url': 'https://tv.nrk.no/program/mdfp15000514',
267 'id': 'MDFP15000514CA',
269 'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014',
270 'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db',
272 'series': 'Kunnskapskanalen',
273 'episode': '24.05.2014',
276 'skip_download': True,
279 # single playlist video
280 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2',
282 'id': 'MSPO40010515-part2',
284 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
285 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
288 'skip_download': True,
290 'expected_warnings': ['Video is geo restricted'],
291 'skip': 'particular part is not supported currently',
293 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015',
296 'id': 'MSPO40010515AH',
298 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 1)',
299 'description': 'md5:c03aba1e917561eface5214020551b7a',
301 'series': 'Tour de Ski',
302 'episode': '06.01.2015',
305 'skip_download': True,
309 'id': 'MSPO40010515BH',
311 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 2)',
312 'description': 'md5:c03aba1e917561eface5214020551b7a',
314 'series': 'Tour de Ski',
315 'episode': '06.01.2015',
318 'skip_download': True,
322 'id': 'MSPO40010515',
323 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015',
324 'description': 'md5:c03aba1e917561eface5214020551b7a',
326 'expected_warnings': ['Video is geo restricted'],
328 'url': 'https://tv.nrk.no/serie/anno/KMTE50001317/sesong-3/episode-13',
330 'id': 'KMTE50001317AA',
332 'title': 'Anno 13:30',
333 'description': 'md5:11d9613661a8dbe6f9bef54e3a4cbbfa',
338 'episode_number': 13,
341 'skip_download': True,
344 'url': 'https://tv.nrk.no/serie/nytt-paa-nytt/MUHH46000317/27-01-2017',
346 'id': 'MUHH46000317AA',
348 'title': 'Nytt på Nytt 27.01.2017',
349 'description': 'md5:5358d6388fba0ea6f0b6d11c48b9eb4b',
351 'series': 'Nytt på nytt',
352 'episode': '27.01.2017',
355 'skip_download': True,
358 'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#',
359 'only_matching': True,
363 class NRKTVEpisodeIE(InfoExtractor):
364 _VALID_URL = r'https?://tv\.nrk\.no/serie/(?P<id>[^/]+/sesong/\d+/episode/\d+)'
366 'url': 'https://tv.nrk.no/serie/backstage/sesong/1/episode/8',
368 'id': 'MSUI14000816AA',
370 'title': 'Backstage 8:30',
371 'description': 'md5:de6ca5d5a2d56849e4021f2bf2850df4',
373 'series': 'Backstage',
379 'skip_download': True,
383 def _real_extract(self, url):
384 display_id = self._match_id(url)
386 webpage = self._download_webpage(url, display_id)
388 nrk_id = self._parse_json(
389 self._search_regex(JSON_LD_RE, webpage, 'JSON-LD', group='json_ld'),
392 assert re.match(NRKTVIE._EPISODE_RE, nrk_id)
393 return self.url_result(
394 'nrk:%s' % nrk_id, ie=NRKIE.ie_key(), video_id=nrk_id)
397 class NRKTVDirekteIE(NRKTVIE):
398 IE_DESC = 'NRK TV Direkte and NRK Radio Direkte'
399 _VALID_URL = r'https?://(?:tv|radio)\.nrk\.no/direkte/(?P<id>[^/?#&]+)'
402 'url': 'https://tv.nrk.no/direkte/nrk1',
403 'only_matching': True,
405 'url': 'https://radio.nrk.no/direkte/p1_oslo_akershus',
406 'only_matching': True,
410 class NRKPlaylistBaseIE(InfoExtractor):
411 def _extract_description(self, webpage):
414 def _real_extract(self, url):
415 playlist_id = self._match_id(url)
417 webpage = self._download_webpage(url, playlist_id)
420 self.url_result('nrk:%s' % video_id, NRKIE.ie_key())
421 for video_id in re.findall(self._ITEM_RE, webpage)
424 playlist_title = self. _extract_title(webpage)
425 playlist_description = self._extract_description(webpage)
427 return self.playlist_result(
428 entries, playlist_id, playlist_title, playlist_description)
431 class NRKPlaylistIE(NRKPlaylistBaseIE):
432 _VALID_URL = r'https?://(?:www\.)?nrk\.no/(?!video|skole)(?:[^/]+/)+(?P<id>[^/]+)'
433 _ITEM_RE = r'class="[^"]*\brich\b[^"]*"[^>]+data-video-id="([^"]+)"'
435 'url': 'http://www.nrk.no/troms/gjenopplev-den-historiske-solformorkelsen-1.12270763',
437 'id': 'gjenopplev-den-historiske-solformorkelsen-1.12270763',
438 'title': 'Gjenopplev den historiske solformørkelsen',
439 'description': 'md5:c2df8ea3bac5654a26fc2834a542feed',
443 'url': 'http://www.nrk.no/kultur/bok/rivertonprisen-til-karin-fossum-1.12266449',
445 'id': 'rivertonprisen-til-karin-fossum-1.12266449',
446 'title': 'Rivertonprisen til Karin Fossum',
447 'description': 'Første kvinne på 15 år til å vinne krimlitteraturprisen.',
452 def _extract_title(self, webpage):
453 return self._og_search_title(webpage, fatal=False)
455 def _extract_description(self, webpage):
456 return self._og_search_description(webpage)
459 class NRKTVEpisodesIE(NRKPlaylistBaseIE):
460 _VALID_URL = r'https?://tv\.nrk\.no/program/[Ee]pisodes/[^/]+/(?P<id>\d+)'
461 _ITEM_RE = r'data-episode=["\']%s' % NRKTVIE._EPISODE_RE
463 'url': 'https://tv.nrk.no/program/episodes/nytt-paa-nytt/69031',
466 'title': 'Nytt på nytt, sesong: 201210',
471 def _extract_title(self, webpage):
472 return self._html_search_regex(
473 r'<h1>([^<]+)</h1>', webpage, 'title', fatal=False)
476 class NRKTVSeriesIE(InfoExtractor):
477 _VALID_URL = r'https?://(?:tv|radio)\.nrk(?:super)?\.no/serie/(?P<id>[^/]+)'
478 _ITEM_RE = r'(?:data-season=["\']|id=["\']season-)(?P<id>\d+)'
480 'url': 'https://tv.nrk.no/serie/groenn-glede',
482 'id': 'groenn-glede',
483 'title': 'Grønn glede',
484 'description': 'md5:7576e92ae7f65da6993cf90ee29e4608',
486 'playlist_mincount': 9,
488 'url': 'http://tv.nrksuper.no/serie/labyrint',
492 'description': 'md5:58afd450974c89e27d5a19212eee7115',
494 'playlist_mincount': 3,
496 'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene',
497 'only_matching': True,
499 'url': 'https://tv.nrk.no/serie/saving-the-human-race',
500 'only_matching': True,
502 'url': 'https://tv.nrk.no/serie/postmann-pat',
503 'only_matching': True,
507 def suitable(cls, url):
508 return (False if NRKTVIE.suitable(url) or NRKTVEpisodeIE.suitable(url)
509 else super(NRKTVSeriesIE, cls).suitable(url))
511 def _real_extract(self, url):
512 series_id = self._match_id(url)
514 webpage = self._download_webpage(url, series_id)
518 'https://tv.nrk.no/program/Episodes/{series}/{season}'.format(
519 series=series_id, season=season_id))
520 for season_id in re.findall(self._ITEM_RE, webpage)
523 title = self._html_search_meta(
524 'seriestitle', webpage,
525 'title', default=None) or self._og_search_title(
526 webpage, fatal=False)
528 description = self._html_search_meta(
529 'series_description', webpage,
530 'description', default=None) or self._og_search_description(webpage)
532 return self.playlist_result(entries, series_id, title, description)
535 class NRKSkoleIE(InfoExtractor):
536 IE_DESC = 'NRK Skole'
537 _VALID_URL = r'https?://(?:www\.)?nrk\.no/skole/?\?.*\bmediaId=(?P<id>\d+)'
540 'url': 'https://www.nrk.no/skole/?page=search&q=&mediaId=14099',
541 'md5': '6bc936b01f9dd8ed45bc58b252b2d9b6',
545 'title': 'Genetikk og eneggede tvillinger',
546 'description': 'md5:3aca25dcf38ec30f0363428d2b265f8d',
550 'url': 'https://www.nrk.no/skole/?page=objectives&subject=naturfag&objective=K15114&mediaId=19355',
551 'only_matching': True,
554 def _real_extract(self, url):
555 video_id = self._match_id(url)
557 webpage = self._download_webpage(
558 'https://mimir.nrk.no/plugin/1.0/static?mediaId=%s' % video_id,
561 nrk_id = self._parse_json(
563 r'<script[^>]+type=["\']application/json["\'][^>]*>({.+?})</script>',
564 webpage, 'application json'),
565 video_id)['activeMedia']['psId']
567 return self.url_result('nrk:%s' % nrk_id)