2 from __future__ import unicode_literals
6 from .common import InfoExtractor
15 class NovaIE(InfoExtractor):
16 IE_DESC = 'TN.cz, Prásk.tv, Nova.cz, Novaplus.cz, FANDA.tv, Krásná.cz and Doma.cz'
17 _VALID_URL = r'https?://(?:[^.]+\.)?(?P<site>tv(?:noviny)?|tn|novaplus|vymena|fanda|krasna|doma|prask)\.nova\.cz/(?:[^/]+/)+(?P<id>[^/]+?)(?:\.html|/|$)'
19 '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',
22 'display_id': 'co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou',
24 'title': 'Duel: Michal Hrdlička a Petr Suchoň',
25 'description': 'md5:d0cc509858eee1b1374111c588c6f5d5',
26 'thumbnail': r're:^https?://.*\.(?:jpg)',
30 'skip_download': True,
33 'url': 'http://tn.nova.cz/clanek/tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci.html#player_13260',
34 'md5': '1dd7b9d5ea27bc361f110cd855a19bd3',
37 'display_id': 'tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci',
39 'title': 'Podzemní nemocnice v pražské Krči',
40 'description': 'md5:f0a42dd239c26f61c28f19e62d20ef53',
41 'thumbnail': r're:^https?://.*\.(?:jpg)',
44 'url': 'http://novaplus.nova.cz/porad/policie-modrava/video/5591-policie-modrava-15-dil-blondynka-na-hrbitove',
47 'display_id': '5591-policie-modrava-15-dil-blondynka-na-hrbitove',
49 'title': 'Policie Modrava - 15. díl - Blondýnka na hřbitově',
50 'description': 'md5:dc24e50be5908df83348e50d1431295e', # Make sure this description is clean of html tags
51 'thumbnail': r're:^https?://.*\.(?:jpg)',
55 'skip_download': True,
58 'url': 'http://novaplus.nova.cz/porad/televizni-noviny/video/5585-televizni-noviny-30-5-2015/',
62 'title': 'Televizní noviny - 30. 5. 2015',
63 'thumbnail': r're:^https?://.*\.(?:jpg)',
64 'upload_date': '20150530',
68 'skip_download': True,
71 'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
75 'title': 'Zaklínač 3: Divoký hon',
76 'description': 're:.*Pokud se stejně jako my nemůžete.*',
77 'thumbnail': r're:https?://.*\.jpg(\?.*)?',
78 'upload_date': '20150521',
82 'skip_download': True,
85 'url': 'http://sport.tn.nova.cz/clanek/sport/hokej/nhl/zivot-jde-dal-hodnotil-po-vyrazeni-z-playoff-jiri-sekac.html',
86 'only_matching': True,
88 'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
89 'only_matching': True,
91 'url': 'http://doma.nova.cz/clanek/zdravi/prijdte-se-zapsat-do-registru-kostni-drene-jiz-ve-stredu-3-cervna.html',
92 'only_matching': True,
94 'url': 'http://prask.nova.cz/clanek/novinky/co-si-na-sobe-nase-hvezdy-nechaly-pojistit.html',
95 'only_matching': True,
97 'url': 'http://tv.nova.cz/clanek/novinky/zivot-je-zivot-bondovsky-trailer.html',
98 'only_matching': True,
101 def _real_extract(self, url):
102 mobj = re.match(self._VALID_URL, url)
103 display_id = mobj.group('id')
104 site = mobj.group('site')
106 webpage = self._download_webpage(url, display_id)
108 video_id = self._search_regex(
109 [r"(?:media|video_id)\s*:\s*'(\d+)'",
111 r'id="article_video_(\d+)"',
112 r'id="player_(\d+)"'],
115 config_url = self._search_regex(
116 r'src="(https?://(?:tn|api)\.nova\.cz/bin/player/videojs/config\.php\?[^"]+)"',
117 webpage, 'config url', default=None)
121 player = self._parse_json(
123 r'(?s)Player\s*\(.+?\s*,\s*({.+?\bmedia\b["\']?\s*:\s*["\']?\d+.+?})\s*\)', webpage,
124 'player', default='{}'),
125 video_id, transform_source=js_to_json, fatal=False)
127 config_url = url_or_none(player.get('configUrl'))
128 params = player.get('configParams')
129 if isinstance(params, dict):
130 config_params = params
133 DEFAULT_SITE_ID = '23000'
135 'tvnoviny': DEFAULT_SITE_ID,
136 'novaplus': DEFAULT_SITE_ID,
137 'vymena': DEFAULT_SITE_ID,
138 'krasna': DEFAULT_SITE_ID,
144 site_id = self._search_regex(
145 r'site=(\d+)', webpage, 'site id', default=None) or SITES.get(
146 site, DEFAULT_SITE_ID)
148 config_url = 'https://api.nova.cz/bin/player/videojs/config.php'
156 config = self._download_json(
157 config_url, display_id,
158 'Downloading config JSON', query=config_params,
159 transform_source=lambda s: s[s.index('{'):s.rindex('}') + 1])
161 mediafile = config['mediafile']
162 video_url = mediafile['src']
164 m = re.search(r'^(?P<url>rtmpe?://[^/]+/(?P<app>[^/]+?))/&*(?P<playpath>.+)$', video_url)
167 'url': m.group('url'),
168 'app': m.group('app'),
169 'play_path': m.group('playpath'),
170 'player_path': 'http://tvnoviny.nova.cz/static/shared/app/videojs/video-js.swf',
177 self._sort_formats(formats)
179 title = mediafile.get('meta', {}).get('title') or self._og_search_title(webpage)
180 description = clean_html(self._og_search_description(webpage, default=None))
181 thumbnail = config.get('poster')
183 if site == 'novaplus':
184 upload_date = unified_strdate(self._search_regex(
185 r'(\d{1,2}-\d{1,2}-\d{4})$', display_id, 'upload date', default=None))
186 elif site == 'fanda':
187 upload_date = unified_strdate(self._search_regex(
188 r'<span class="date_time">(\d{1,2}\.\d{1,2}\.\d{4})', webpage, 'upload date', default=None))
194 'display_id': display_id,
196 'description': description,
197 'upload_date': upload_date,
198 'thumbnail': thumbnail,