2 from __future__ import unicode_literals
7 from .common import InfoExtractor
8 from ..compat import compat_urllib_parse_unquote
17 class NRKBaseIE(InfoExtractor):
20 def _download_webpage_handle(self, *args, **kwargs):
21 # NRK checks X-Forwarded-For HTTP header in order to figure out the
22 # origin of the client behind proxy. This allows to bypass geo
23 # restriction by faking this header's value to some Norway IP.
24 # We will do so once we encounter any geo restriction error.
26 # NB: str is intentional
27 kwargs.setdefault(str('headers'), {})['X-Forwarded-For'] = self._faked_ip
28 return super(NRKBaseIE, self)._download_webpage_handle(*args, **kwargs)
31 # Use fake IP from 37.191.128.0/17 in order to workaround geo
33 def octet(lb=0, ub=255):
34 return random.randint(lb, ub)
35 self._faked_ip = '37.191.%d.%d' % (octet(128), octet())
37 def _real_extract(self, url):
38 video_id = self._match_id(url)
40 data = self._download_json(
41 'http://%s/mediaelement/%s' % (self._API_HOST, video_id),
42 video_id, 'Downloading mediaelement JSON')
44 title = data.get('fullTitle') or data.get('mainTitle') or data['title']
45 video_id = data.get('id') or video_id
47 http_headers = {'X-Forwarded-For': self._faked_ip} if self._faked_ip else {}
51 conviva = data.get('convivaStatistics') or {}
52 live = (data.get('mediaElementType') == 'Live' or
53 data.get('isLive') is True or conviva.get('isLive'))
56 return self._live_title(t) if live else t
58 media_assets = data.get('mediaAssets')
59 if media_assets and isinstance(media_assets, list):
60 def video_id_and_title(idx):
61 return ((video_id, title) if len(media_assets) == 1
62 else ('%s-%d' % (video_id, idx), '%s (Part %d)' % (title, idx)))
63 for num, asset in enumerate(media_assets, 1):
64 asset_url = asset.get('url')
67 formats = self._extract_akamai_formats(asset_url, video_id)
70 self._sort_formats(formats)
72 # Some f4m streams may not work with hdcore in fragments' URLs
74 extra_param = f.get('extra_param_to_segment_url')
75 if extra_param and 'hdcore' in extra_param:
76 del f['extra_param_to_segment_url']
78 entry_id, entry_title = video_id_and_title(num)
79 duration = parse_duration(asset.get('duration'))
81 for subtitle in ('webVtt', 'timedText'):
82 subtitle_url = asset.get('%sSubtitlesUrl' % subtitle)
84 subtitles.setdefault('no', []).append({
85 'url': compat_urllib_parse_unquote(subtitle_url)
88 'id': asset.get('carrierId') or entry_id,
89 'title': make_title(entry_title),
91 'subtitles': subtitles,
93 'http_headers': http_headers,
97 media_url = data.get('mediaUrl')
99 formats = self._extract_akamai_formats(media_url, video_id)
100 self._sort_formats(formats)
101 duration = parse_duration(data.get('duration'))
104 'title': make_title(title),
105 'duration': duration,
110 message_type = data.get('messageType', '')
111 # Can be ProgramIsGeoBlocked or ChannelIsGeoBlocked*
112 if 'IsGeoBlocked' in message_type and not self._faked_ip:
114 'Video is geo restricted, trying to fake IP')
116 return self._real_extract(url)
119 'ProgramRightsAreNotReady': 'Du kan dessverre ikke se eller høre programmet',
120 'ProgramRightsHasExpired': 'Programmet har gått ut',
121 'ProgramIsGeoBlocked': 'NRK har ikke rettigheter til å vise dette programmet utenfor Norge',
123 raise ExtractorError(
124 '%s said: %s' % (self.IE_NAME, MESSAGES.get(
125 message_type, message_type)),
128 series = conviva.get('seriesName') or data.get('seriesTitle')
129 episode = conviva.get('episodeName') or data.get('episodeNumberOrDate')
132 episode_number = None
133 if data.get('mediaElementType') == 'Episode':
134 _season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \
135 data.get('relativeOriginUrl', '')
137 r'/s(?P<season>\d+)e(?P<episode>\d+)\.',
138 r'/sesong-(?P<season>\d+)/episode-(?P<episode>\d+)',
140 season_number = int_or_none(self._search_regex(EPISODENUM_RE, _season_episode, "S##E##", fatal=False, group='season'))
141 episode_number = int_or_none(self._search_regex(EPISODENUM_RE, _season_episode, "S##E##", fatal=False, group='episode'))
144 images = data.get('images')
145 if images and isinstance(images, dict):
146 web_images = images.get('webImages')
147 if isinstance(web_images, list):
149 'url': image['imageUrl'],
150 'width': int_or_none(image.get('width')),
151 'height': int_or_none(image.get('height')),
152 } for image in web_images if image.get('imageUrl')]
154 description = data.get('description')
155 category = data.get('mediaAnalytics', {}).get('category')
158 'description': description,
161 'season_number': season_number,
162 'episode_number': episode_number,
163 'categories': [category] if category else None,
164 'age_limit': parse_age_limit(data.get('legalAge')),
165 'thumbnails': thumbnails,
168 vcodec = 'none' if data.get('mediaType') == 'Audio' else None
170 # TODO: extract chapters when https://github.com/rg3/youtube-dl/pull/9409 is merged
172 for entry in entries:
173 entry.update(common_info)
174 for f in entry['formats']:
177 return self.playlist_result(entries, video_id, title, description)
180 class NRKIE(NRKBaseIE):
181 _VALID_URL = r'''(?x)
186 (?:www\.)?nrk\.no/video/PS\*|
187 v8-psapi\.nrk\.no/mediaelement/
192 _API_HOST = 'v8.psapi.nrk.no'
195 'url': 'http://www.nrk.no/video/PS*150533',
196 'md5': '2f7f6eeb2aacdd99885f355428715cfa',
200 'title': 'Dompap og andre fugler i Piip-Show',
201 'description': 'md5:d9261ba34c43b61c812cb6b0269a5c8f',
206 'url': 'http://www.nrk.no/video/PS*154915',
211 'title': 'Slik høres internett ut når du er blind',
212 'description': 'md5:a621f5cc1bd75c8d5104cb048c6b8568',
216 'url': 'nrk:ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
217 'only_matching': True,
219 'url': 'https://v8-psapi.nrk.no/mediaelement/ecc1b952-96dc-4a98-81b9-5296dc7a98d9',
220 'only_matching': True,
224 class NRKTVIE(NRKBaseIE):
225 IE_DESC = 'NRK TV and NRK Radio'
226 _EPISODE_RE = r'(?P<id>[a-zA-Z]{4}\d{8})'
227 _VALID_URL = r'''(?x)
229 (?:tv|radio)\.nrk(?:super)?\.no/
230 (?:serie/[^/]+|program)/
232 (?:/\d{2}-\d{2}-\d{4})?
233 (?:\#del=(?P<part_id>\d+))?
235 _API_HOST = 'psapi-we.nrk.no'
238 'url': 'https://tv.nrk.no/serie/20-spoersmaal-tv/MUHH48000314/23-05-2014',
239 'md5': '4e9ca6629f09e588ed240fb11619922a',
241 'id': 'MUHH48000314AA',
243 'title': '20 spørsmål 23.05.2014',
244 'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
248 'url': 'https://tv.nrk.no/program/mdfp15000514',
249 'md5': '43d0be26663d380603a9cf0c24366531',
251 'id': 'MDFP15000514CA',
253 'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014',
254 'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db',
258 # single playlist video
259 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2',
260 'md5': 'adbd1dbd813edaf532b0a253780719c2',
262 'id': 'MSPO40010515-part2',
264 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
265 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
267 'skip': 'Only works from Norway',
269 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015',
271 'md5': '9480285eff92d64f06e02a5367970a7a',
273 'id': 'MSPO40010515-part1',
275 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 1:2)',
276 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
279 'md5': 'adbd1dbd813edaf532b0a253780719c2',
281 'id': 'MSPO40010515-part2',
283 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
284 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
288 'id': 'MSPO40010515',
289 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn',
290 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
293 'skip': 'Only works from Norway',
295 'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#',
296 'only_matching': True,
300 class NRKTVDirekteIE(NRKTVIE):
301 IE_DESC = 'NRK TV Direkte and NRK Radio Direkte'
302 _VALID_URL = r'https?://(?:tv|radio)\.nrk\.no/direkte/(?P<id>[^/?#&]+)'
305 'url': 'https://tv.nrk.no/direkte/nrk1',
306 'only_matching': True,
308 'url': 'https://radio.nrk.no/direkte/p1_oslo_akershus',
309 'only_matching': True,
313 class NRKPlaylistBaseIE(InfoExtractor):
314 def _extract_description(self, webpage):
317 def _real_extract(self, url):
318 playlist_id = self._match_id(url)
320 webpage = self._download_webpage(url, playlist_id)
323 self.url_result('nrk:%s' % video_id, NRKIE.ie_key())
324 for video_id in re.findall(self._ITEM_RE, webpage)
327 playlist_title = self. _extract_title(webpage)
328 playlist_description = self._extract_description(webpage)
330 return self.playlist_result(
331 entries, playlist_id, playlist_title, playlist_description)
334 class NRKPlaylistIE(NRKPlaylistBaseIE):
335 _VALID_URL = r'https?://(?:www\.)?nrk\.no/(?!video|skole)(?:[^/]+/)+(?P<id>[^/]+)'
336 _ITEM_RE = r'class="[^"]*\brich\b[^"]*"[^>]+data-video-id="([^"]+)"'
338 'url': 'http://www.nrk.no/troms/gjenopplev-den-historiske-solformorkelsen-1.12270763',
340 'id': 'gjenopplev-den-historiske-solformorkelsen-1.12270763',
341 'title': 'Gjenopplev den historiske solformørkelsen',
342 'description': 'md5:c2df8ea3bac5654a26fc2834a542feed',
346 'url': 'http://www.nrk.no/kultur/bok/rivertonprisen-til-karin-fossum-1.12266449',
348 'id': 'rivertonprisen-til-karin-fossum-1.12266449',
349 'title': 'Rivertonprisen til Karin Fossum',
350 'description': 'Første kvinne på 15 år til å vinne krimlitteraturprisen.',
355 def _extract_title(self, webpage):
356 return self._og_search_title(webpage, fatal=False)
358 def _extract_description(self, webpage):
359 return self._og_search_description(webpage)
362 class NRKTVEpisodesIE(NRKPlaylistBaseIE):
363 _VALID_URL = r'https?://tv\.nrk\.no/program/[Ee]pisodes/[^/]+/(?P<id>\d+)'
364 _ITEM_RE = r'data-episode=["\']%s' % NRKTVIE._EPISODE_RE
366 'url': 'https://tv.nrk.no/program/episodes/nytt-paa-nytt/69031',
369 'title': 'Nytt på nytt, sesong: 201210',
374 def _extract_title(self, webpage):
375 return self._html_search_regex(
376 r'<h1>([^<]+)</h1>', webpage, 'title', fatal=False)
379 class NRKTVSeriesIE(InfoExtractor):
380 _VALID_URL = r'https?://tv\.nrk\.no/serie/(?P<id>[^/]+)/?'
381 _ITEM_RE = r'data-season=["\'](?P<id>\d+)["\']'
383 'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene',
386 'url': 'https://tv.nrk.no/serie/saving-the-human-race',
389 'url': 'https://tv.nrk.no/serie/postmann-pat',
392 'url': 'https://tv.nrk.no/serie/groenn-glede',
396 def _real_extract(self, url):
397 series_id = self._match_id(url)
399 webpage = self._download_webpage(url, series_id)
402 self.url_result('https://tv.nrk.no/program/Episodes/{series}/{season}'.format(
406 for season_id in re.findall(self._ITEM_RE, webpage)
409 return self.playlist_result(entries)
412 class NRKSkoleIE(InfoExtractor):
413 IE_DESC = 'NRK Skole'
414 _VALID_URL = r'https?://(?:www\.)?nrk\.no/skole/?\?.*\bmediaId=(?P<id>\d+)'
417 'url': 'https://www.nrk.no/skole/?page=search&q=&mediaId=14099',
418 'md5': '6bc936b01f9dd8ed45bc58b252b2d9b6',
422 'title': 'Genetikk og eneggede tvillinger',
423 'description': 'md5:3aca25dcf38ec30f0363428d2b265f8d',
427 'url': 'https://www.nrk.no/skole/?page=objectives&subject=naturfag&objective=K15114&mediaId=19355',
428 'only_matching': True,
431 def _real_extract(self, url):
432 video_id = self._match_id(url)
434 webpage = self._download_webpage(
435 'https://mimir.nrk.no/plugin/1.0/static?mediaId=%s' % video_id,
438 nrk_id = self._parse_json(
440 r'<script[^>]+type=["\']application/json["\'][^>]*>({.+?})</script>',
441 webpage, 'application json'),
442 video_id)['activeMedia']['psId']
444 return self.url_result('nrk:%s' % nrk_id)