2 from __future__ import unicode_literals
15 class MediasetIE(InfoExtractor):
16 _VALID_URL = r'https?://www\.video\.mediaset\.it/(?:(?:video|on-demand)/(?:.+)_|player/playerIFrame(?:Twitter)?\.shtml\?id=)(?P<id>[0-9]+)(?:.html|&.+)'
19 'url': 'http://www.video.mediaset.it/video/hello_goodbye/full/quarta-puntata_661824.html',
20 'md5': '9b75534d42c44ecef7bf1ffeacb7f85d',
24 'title': 'Quarta puntata',
25 'thumbnail': 're:^https?://.*\.jpg$',
26 'description': 'md5:7183696d6df570e3412a5ef74b27c5e2',
27 'uploader': 'mediaset'
31 'url': 'http://www.video.mediaset.it/video/domenica_live/interviste/il-fenomeno-elettra-lamborghini_716283.html',
32 'md5': '81c57566bf2ee02e995f5342f079ca25',
36 'title': 'Il fenomeno Elettra Lamborghini',
37 'thumbnail': 're:^https?://.*\.jpg$',
38 'description': 'md5:dabf0e7cf48fc6d0a3417b989028748a',
39 'uploader': 'mediaset'
43 'url': 'http://www.video.mediaset.it/video/gogglebox/clip/un-grande-classico-della-commedia-sexy_661680.html',
44 'md5': '189ca72fe399db80dbfa595a4abf42d0',
48 'title': 'Un grande classico della commedia sexy',
49 'thumbnail': 're:^https?://.*\.jpg$',
50 'description': 'Un film che riesce a risvegliare i sensi di Gigi.',
51 'uploader': 'mediaset'
55 'url': 'http://www.video.mediaset.it/player/playerIFrame.shtml?id=665924&autoplay=true',
56 'md5': '308430901e55e1ad83dddb4be2a4454a',
60 'title': 'Gianna Nannini incontra i fan a Milano',
61 'thumbnail': 're:^https?://.*\.jpg$',
62 'description': 'La cantante parla del nuovo libro',
63 'uploader': 'mediaset'
66 # iframe twitter (from http://www.wittytv.it/se-prima-mi-fidavo-zero/)
67 'url': 'https://www.video.mediaset.it/player/playerIFrameTwitter.shtml?id=665104&playrelated=false&autoplay=false&related=true&hidesocial=true',
68 'md5': '6f53a834b3b5eac1ebc2037ccf7194d0',
72 'title': '\"Se prima mi fidavo zero...\"',
73 'thumbnail': 're:^https?://.*\.jpg$',
74 'description': 'Una piccola anteprima della prossima puntata del Trono Classico',
75 'uploader': 'mediaset'
79 def _real_extract(self, url):
80 video_id = self._match_id(url)
85 mediainfo = self._download_json(
86 'http://plr.video.mediaset.it/html/metainfo.sjson?id=%s' % video_id,
87 video_id, 'Downloading video info JSON').get('video')
89 if 'brand-info' in mediainfo:
90 uploader = mediainfo.get('brand-info').get('publisher')
91 categories = [mediainfo.get('brand-info').get('category')]
93 cnd = self._download_json(
94 'http://cdnsel01.mediaset.net/GetCdn.aspx?streamid=%s&format=json' % video_id,
95 video_id, 'Downloading video CND JSON')
97 if not cnd.get('videoList'):
98 raise ExtractorError('Video not found')
100 for media_url in cnd.get('videoList'):
103 'ext': determine_ext(media_url)
105 self._sort_formats(formats)
109 'title': mediainfo.get('title'),
111 'description': mediainfo.get('short-description'),
112 'uploader': uploader,
113 'thumbnail': mediainfo.get('thumbnail'),
114 'duration': parse_duration(mediainfo.get('duration')),
115 'release_date': unified_strdate(mediainfo.get('production-date')),
116 'webpage_url': mediainfo.get('url'),
117 'categories': categories