import hashlib import re from .common import InfoExtractor from ..utils import ( ExtractorError, determine_ext, float_or_none, int_or_none, merge_dicts, unified_strdate, ) class ProSiebenSat1BaseIE(InfoExtractor): _GEO_BYPASS = False _ACCESS_ID = None _SUPPORTED_PROTOCOLS = 'dash:clear,hls:clear,progressive:clear' _V4_BASE_URL = 'https://vas-v4.p7s1video.net/4.0/get' def _extract_video_info(self, url, clip_id): client_location = url video = self._download_json( 'http://vas.sim-technik.de/vas/live/v2/videos', clip_id, 'Downloading videos JSON', query={ 'access_token': self._TOKEN, 'client_location': client_location, 'client_name': self._CLIENT_NAME, 'ids': clip_id, })[0] if not self.get_param('allow_unplayable_formats') and video.get('is_protected') is True: self.report_drm(clip_id) formats = [] if self._ACCESS_ID: raw_ct = self._ENCRYPTION_KEY + clip_id + self._IV + self._ACCESS_ID protocols = self._download_json( self._V4_BASE_URL + 'protocols', clip_id, 'Downloading protocols JSON', headers=self.geo_verification_headers(), query={ 'access_id': self._ACCESS_ID, 'client_token': hashlib.sha1((raw_ct).encode()).hexdigest(), 'video_id': clip_id, }, fatal=False, expected_status=(403,)) or {} error = protocols.get('error') or {} if error.get('title') == 'Geo check failed': self.raise_geo_restricted(countries=['AT', 'CH', 'DE']) server_token = protocols.get('server_token') if server_token: urls = (self._download_json( self._V4_BASE_URL + 'urls', clip_id, 'Downloading urls JSON', query={ 'access_id': self._ACCESS_ID, 'client_token': hashlib.sha1((raw_ct + server_token + self._SUPPORTED_PROTOCOLS).encode()).hexdigest(), 'protocols': self._SUPPORTED_PROTOCOLS, 'server_token': server_token, 'video_id': clip_id, }, fatal=False) or {}).get('urls') or {} for protocol, variant in urls.items(): source_url = variant.get('clear', {}).get('url') if not source_url: continue if protocol == 'dash': formats.extend(self._extract_mpd_formats( source_url, clip_id, mpd_id=protocol, fatal=False)) elif protocol == 'hls': formats.extend(self._extract_m3u8_formats( source_url, clip_id, 'mp4', 'm3u8_native', m3u8_id=protocol, fatal=False)) else: formats.append({ 'url': source_url, 'format_id': protocol, }) if not formats: source_ids = [str(source['id']) for source in video['sources']] client_id = self._SALT[:2] + hashlib.sha1(''.join([clip_id, self._SALT, self._TOKEN, client_location, self._SALT, self._CLIENT_NAME]).encode()).hexdigest() sources = self._download_json( f'http://vas.sim-technik.de/vas/live/v2/videos/{clip_id}/sources', clip_id, 'Downloading sources JSON', query={ 'access_token': self._TOKEN, 'client_id': client_id, 'client_location': client_location, 'client_name': self._CLIENT_NAME, }) server_id = sources['server_id'] def fix_bitrate(bitrate): bitrate = int_or_none(bitrate) if not bitrate: return None return (bitrate // 1000) if bitrate % 1000 == 0 else bitrate for source_id in source_ids: client_id = self._SALT[:2] + hashlib.sha1(''.join([self._SALT, clip_id, self._TOKEN, server_id, client_location, source_id, self._SALT, self._CLIENT_NAME]).encode()).hexdigest() urls = self._download_json( f'http://vas.sim-technik.de/vas/live/v2/videos/{clip_id}/sources/url', clip_id, 'Downloading urls JSON', fatal=False, query={ 'access_token': self._TOKEN, 'client_id': client_id, 'client_location': client_location, 'client_name': self._CLIENT_NAME, 'server_id': server_id, 'source_ids': source_id, }) if not urls: continue if urls.get('status_code') != 0: raise ExtractorError('This video is unavailable', expected=True) urls_sources = urls['sources'] if isinstance(urls_sources, dict): urls_sources = urls_sources.values() for source in urls_sources: source_url = source.get('url') if not source_url: continue protocol = source.get('protocol') mimetype = source.get('mimetype') if mimetype == 'application/f4m+xml' or 'f4mgenerator' in source_url or determine_ext(source_url) == 'f4m': formats.extend(self._extract_f4m_formats( source_url, clip_id, f4m_id='hds', fatal=False)) elif mimetype == 'application/x-mpegURL': formats.extend(self._extract_m3u8_formats( source_url, clip_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)) elif mimetype == 'application/dash+xml': formats.extend(self._extract_mpd_formats( source_url, clip_id, mpd_id='dash', fatal=False)) else: tbr = fix_bitrate(source['bitrate']) if protocol in ('rtmp', 'rtmpe'): mobj = re.search(r'^(?Prtmpe?://[^/]+)/(?P.+)$', source_url) if not mobj: continue path = mobj.group('path') mp4colon_index = path.rfind('mp4:') app = path[:mp4colon_index] play_path = path[mp4colon_index:] formats.append({ 'url': '{}/{}'.format(mobj.group('url'), app), 'app': app, 'play_path': play_path, 'player_url': 'http://livepassdl.conviva.com/hf/ver/2.79.0.17083/LivePassModuleMain.swf', 'page_url': 'http://www.prosieben.de', 'tbr': tbr, 'ext': 'flv', 'format_id': 'rtmp{}'.format(f'-{tbr}' if tbr else ''), }) else: formats.append({ 'url': source_url, 'tbr': tbr, 'format_id': 'http{}'.format(f'-{tbr}' if tbr else ''), }) return { 'duration': float_or_none(video.get('duration')), 'formats': formats, } class ProSiebenSat1IE(ProSiebenSat1BaseIE): IE_NAME = 'prosiebensat1' IE_DESC = 'ProSiebenSat.1 Digital' _VALID_URL = r'''(?x) https?:// (?:www\.)? (?: (?:beta\.)? (?: prosieben(?:maxx)?|sixx|sat1(?:gold)?|kabeleins(?:doku)?|the-voice-of-germany|advopedia )\.(?:de|at|ch)| ran\.de|fem\.com|advopedia\.de|galileo\.tv/video ) /(?P.+) ''' _TESTS = [ { # Tests changes introduced in https://github.com/ytdl-org/youtube-dl/pull/6242 # in response to fixing https://github.com/ytdl-org/youtube-dl/issues/6215: # - malformed f4m manifest support # - proper handling of URLs starting with `https?://` in 2.0 manifests # - recursive child f4m manifests extraction 'url': 'http://www.prosieben.de/tv/circus-halligalli/videos/218-staffel-2-episode-18-jahresrueckblick-ganze-folge', 'info_dict': { 'id': '2104602', 'ext': 'mp4', 'title': 'CIRCUS HALLIGALLI - Episode 18 - Staffel 2', 'description': 'md5:8733c81b702ea472e069bc48bb658fc1', 'upload_date': '20131231', 'duration': 5845.04, 'series': 'CIRCUS HALLIGALLI', 'season_number': 2, 'episode': 'Episode 18 - Staffel 2', 'episode_number': 18, }, }, { 'url': 'http://www.prosieben.de/videokatalog/Gesellschaft/Leben/Trends/video-Lady-Umstyling-f%C3%BCr-Audrina-Rebekka-Audrina-Fergen-billig-aussehen-Battal-Modica-700544.html', 'info_dict': { 'id': '2570327', 'ext': 'mp4', 'title': 'Lady-Umstyling für Audrina', 'description': 'md5:4c16d0c17a3461a0d43ea4084e96319d', 'upload_date': '20131014', 'duration': 606.76, }, 'params': { # rtmp download 'skip_download': True, }, 'skip': 'Seems to be broken', }, { 'url': 'http://www.prosiebenmaxx.de/tv/experience/video/144-countdown-fuer-die-autowerkstatt-ganze-folge', 'info_dict': { 'id': '2429369', 'ext': 'mp4', 'title': 'Countdown für die Autowerkstatt', 'description': 'md5:809fc051a457b5d8666013bc40698817', 'upload_date': '20140223', 'duration': 2595.04, }, 'params': { # rtmp download 'skip_download': True, }, 'skip': 'This video is unavailable', }, { 'url': 'http://www.sixx.de/stars-style/video/sexy-laufen-in-ugg-boots-clip', 'info_dict': { 'id': '2904997', 'ext': 'mp4', 'title': 'Sexy laufen in Ugg Boots', 'description': 'md5:edf42b8bd5bc4e5da4db4222c5acb7d6', 'upload_date': '20140122', 'duration': 245.32, }, 'params': { # rtmp download 'skip_download': True, }, 'skip': 'This video is unavailable', }, { 'url': 'http://www.sat1.de/film/der-ruecktritt/video/im-interview-kai-wiesinger-clip', 'info_dict': { 'id': '2906572', 'ext': 'mp4', 'title': 'Im Interview: Kai Wiesinger', 'description': 'md5:e4e5370652ec63b95023e914190b4eb9', 'upload_date': '20140203', 'duration': 522.56, }, 'params': { # rtmp download 'skip_download': True, }, 'skip': 'This video is unavailable', }, { 'url': 'http://www.kabeleins.de/tv/rosins-restaurants/videos/jagd-auf-fertigkost-im-elsthal-teil-2-ganze-folge', 'info_dict': { 'id': '2992323', 'ext': 'mp4', 'title': 'Jagd auf Fertigkost im Elsthal - Teil 2', 'description': 'md5:2669cde3febe9bce13904f701e774eb6', 'upload_date': '20141014', 'duration': 2410.44, }, 'params': { # rtmp download 'skip_download': True, }, 'skip': 'This video is unavailable', }, { 'url': 'http://www.ran.de/fussball/bundesliga/video/schalke-toennies-moechte-raul-zurueck-ganze-folge', 'info_dict': { 'id': '3004256', 'ext': 'mp4', 'title': 'Schalke: Tönnies möchte Raul zurück', 'description': 'md5:4b5b271d9bcde223b54390754c8ece3f', 'upload_date': '20140226', 'duration': 228.96, }, 'params': { # rtmp download 'skip_download': True, }, 'skip': 'This video is unavailable', }, { 'url': 'http://www.the-voice-of-germany.de/video/31-andreas-kuemmert-rocket-man-clip', 'info_dict': { 'id': '2572814', 'ext': 'mp4', 'title': 'The Voice of Germany - Andreas Kümmert: Rocket Man', 'description': 'md5:6ddb02b0781c6adf778afea606652e38', 'timestamp': 1382041620, 'upload_date': '20131017', 'duration': 469.88, }, 'params': { 'skip_download': True, }, }, { 'url': 'http://www.fem.com/videos/beauty-lifestyle/kurztrips-zum-valentinstag', 'info_dict': { 'id': '2156342', 'ext': 'mp4', 'title': 'Kurztrips zum Valentinstag', 'description': 'Romantischer Kurztrip zum Valentinstag? Nina Heinemann verrät, was sich hier wirklich lohnt.', 'duration': 307.24, }, 'params': { 'skip_download': True, }, }, { 'url': 'http://www.prosieben.de/tv/joko-gegen-klaas/videos/playlists/episode-8-ganze-folge-playlist', 'info_dict': { 'id': '439664', 'title': 'Episode 8 - Ganze Folge - Playlist', 'description': 'md5:63b8963e71f481782aeea877658dec84', }, 'playlist_count': 2, 'skip': 'This video is unavailable', }, { # title in

'url': 'http://www.prosieben.de/stars/oscar-award/videos/jetzt-erst-enthuellt-das-geheimnis-von-emma-stones-oscar-robe-clip', 'info_dict': { 'id': '4895826', 'ext': 'mp4', 'title': 'Jetzt erst enthüllt: Das Geheimnis von Emma Stones Oscar-Robe', 'description': 'md5:e5ace2bc43fadf7b63adc6187e9450b9', 'upload_date': '20170302', }, 'params': { 'skip_download': True, }, 'skip': 'geo restricted to Germany', }, { # geo restricted to Germany 'url': 'http://www.kabeleinsdoku.de/tv/mayday-alarm-im-cockpit/video/102-notlandung-im-hudson-river-ganze-folge', 'only_matching': True, }, { # geo restricted to Germany 'url': 'http://www.sat1gold.de/tv/edel-starck/video/11-staffel-1-episode-1-partner-wider-willen-ganze-folge', 'only_matching': True, }, { # geo restricted to Germany 'url': 'https://www.galileo.tv/video/diese-emojis-werden-oft-missverstanden', 'only_matching': True, }, { 'url': 'http://www.sat1gold.de/tv/edel-starck/playlist/die-gesamte-1-staffel', 'only_matching': True, }, { 'url': 'http://www.advopedia.de/videos/lenssen-klaert-auf/lenssen-klaert-auf-folge-8-staffel-3-feiertage-und-freie-tage', 'only_matching': True, }, ] _TOKEN = 'prosieben' _SALT = '01!8d8F_)r9]4s[qeuXfP%' _CLIENT_NAME = 'kolibri-2.0.19-splec4' _ACCESS_ID = 'x_prosiebenmaxx-de' _ENCRYPTION_KEY = 'Eeyeey9oquahthainoofashoyoikosag' _IV = 'Aeluchoc6aevechuipiexeeboowedaok' _CLIPID_REGEXES = [ r'"clip_id"\s*:\s+"(\d+)"', r'clipid: "(\d+)"', r'clip[iI]d=(\d+)', r'clip[iI][dD]\s*=\s*["\'](\d+)', r"'itemImageUrl'\s*:\s*'/dynamic/thumbnails/full/\d+/(\d+)", r'proMamsId"\s*:\s*"(\d+)', r'proMamsId"\s*:\s*"(\d+)', ] _TITLE_REGEXES = [ r'

\s*(.+?)

', r'
\s*

(.+?)

', r'\s*

(.+?)

', r'

\s*(.+?)

', r'
\s*

([^<]+)

\s*
', r'

\s*(.+?)

', r']+id="veeseoTitle"[^>]*>(.+?)', r']+class="subtitle"[^>]*>([^<]+)', ] _DESCRIPTION_REGEXES = [ r'

\s*(.+?)

', r'
\s*

Beschreibung: (.+?)

', r'
\s*
\s*
\s*(.+?)\s*