2 from __future__ import unicode_literals
6 from .common import InfoExtractor
14 class NovaIE(InfoExtractor):
15 IE_DESC = 'TN.cz, Prásk.tv, Nova.cz, Novaplus.cz, FANDA.tv, Krásná.cz and Doma.cz'
16 _VALID_URL = 'http://(?:[^.]+\.)?(?P<site>tv(?:noviny)?|tn|novaplus|vymena|fanda|krasna|doma|prask)\.nova\.cz/(?:[^/]+/)+(?P<id>[^/]+?)(?:\.html|/|$)'
18 'url': 'http://tvnoviny.nova.cz/clanek/novinky/co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou.html?utm_source=tvnoviny&utm_medium=cpfooter&utm_campaign=novaplus',
21 'display_id': 'co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou',
23 'title': 'Duel: Michal Hrdlička a Petr Suchoň',
24 'description': 'md5:d0cc509858eee1b1374111c588c6f5d5',
25 'thumbnail': 're:^https?://.*\.(?:jpg)',
29 'skip_download': True,
32 'url': 'http://tn.nova.cz/clanek/tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci.html#player_13260',
33 'md5': '1dd7b9d5ea27bc361f110cd855a19bd3',
36 'display_id': 'tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci',
38 'title': 'Podzemní nemocnice v pražské Krči',
39 'description': 'md5:f0a42dd239c26f61c28f19e62d20ef53',
40 'thumbnail': 're:^https?://.*\.(?:jpg)',
43 'url': 'http://novaplus.nova.cz/porad/policie-modrava/video/5591-policie-modrava-15-dil-blondynka-na-hrbitove',
46 'display_id': '5591-policie-modrava-15-dil-blondynka-na-hrbitove',
48 'title': 'Policie Modrava - 15. díl - Blondýnka na hřbitově',
49 'description': 'md5:dc24e50be5908df83348e50d1431295e', # Make sure this description is clean of html tags
50 'thumbnail': 're:^https?://.*\.(?:jpg)',
54 'skip_download': True,
57 'url': 'http://novaplus.nova.cz/porad/televizni-noviny/video/5585-televizni-noviny-30-5-2015/',
61 'title': 'Televizní noviny - 30. 5. 2015',
62 'thumbnail': 're:^https?://.*\.(?:jpg)',
63 'upload_date': '20150530',
67 'skip_download': True,
70 'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
74 'title': 'Zaklínač 3: Divoký hon',
75 'description': 're:.*Pokud se stejně jako my nemůžete.*',
76 'thumbnail': 're:https?://.*\.jpg(\?.*)?',
77 'upload_date': '20150521',
81 'skip_download': True,
84 'url': 'http://sport.tn.nova.cz/clanek/sport/hokej/nhl/zivot-jde-dal-hodnotil-po-vyrazeni-z-playoff-jiri-sekac.html',
85 'only_matching': True,
87 'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
88 'only_matching': True,
90 'url': 'http://doma.nova.cz/clanek/zdravi/prijdte-se-zapsat-do-registru-kostni-drene-jiz-ve-stredu-3-cervna.html',
91 'only_matching': True,
93 'url': 'http://prask.nova.cz/clanek/novinky/co-si-na-sobe-nase-hvezdy-nechaly-pojistit.html',
94 'only_matching': True,
96 'url': 'http://tv.nova.cz/clanek/novinky/zivot-je-zivot-bondovsky-trailer.html',
97 'only_matching': True,
100 def _real_extract(self, url):
101 mobj = re.match(self._VALID_URL, url)
102 display_id = mobj.group('id')
103 site = mobj.group('site')
105 webpage = self._download_webpage(url, display_id)
107 video_id = self._search_regex(
108 [r"(?:media|video_id)\s*:\s*'(\d+)'",
110 r'id="article_video_(\d+)"',
111 r'id="player_(\d+)"'],
114 config_url = self._search_regex(
115 r'src="(http://tn\.nova\.cz/bin/player/videojs/config\.php\?[^"]+)"',
116 webpage, 'config url', default=None)
119 DEFAULT_SITE_ID = '23000'
121 'tvnoviny': DEFAULT_SITE_ID,
122 'novaplus': DEFAULT_SITE_ID,
123 'vymena': DEFAULT_SITE_ID,
124 'krasna': DEFAULT_SITE_ID,
130 site_id = self._search_regex(
131 r'site=(\d+)', webpage, 'site id', default=None) or SITES.get(site, DEFAULT_SITE_ID)
133 config_url = ('http://tn.nova.cz/bin/player/videojs/config.php?site=%s&media=%s&jsVar=vjsconfig'
134 % (site_id, video_id))
136 config = self._download_json(
137 config_url, display_id,
138 'Downloading config JSON',
139 transform_source=lambda s: s[s.index('{'):s.rindex('}') + 1])
141 mediafile = config['mediafile']
142 video_url = mediafile['src']
143 ext = determine_ext(video_url)
144 video_url = video_url.replace('&%s:' % ext, '')
146 title = mediafile.get('meta', {}).get('title') or self._og_search_title(webpage)
147 description = clean_html(self._og_search_description(webpage, default=None))
148 thumbnail = config.get('poster')
150 if site == 'novaplus':
151 upload_date = unified_strdate(self._search_regex(
152 r'(\d{1,2}-\d{1,2}-\d{4})$', display_id, 'upload date', default=None))
153 elif site == 'fanda':
154 upload_date = unified_strdate(self._search_regex(
155 r'<span class="date_time">(\d{1,2}\.\d{1,2}\.\d{4})', webpage, 'upload date', default=None))
161 'display_id': display_id,
163 'description': description,
164 'upload_date': upload_date,
165 'thumbnail': thumbnail,