2 from __future__ import unicode_literals
6 from .common import InfoExtractor
17 class NovaEmbedIE(InfoExtractor):
18 _VALID_URL = r'https?://media\.cms\.nova\.cz/embed/(?P<id>[^/?#&]+)'
20 'url': 'https://media.cms.nova.cz/embed/8o0n0r?autoplay=1',
21 'md5': 'b3834f6de5401baabf31ed57456463f7',
26 'thumbnail': r're:^https?://.*\.jpg',
31 def _real_extract(self, url):
32 video_id = self._match_id(url)
34 webpage = self._download_webpage(url, video_id)
36 bitrates = self._parse_json(
38 r'(?s)(?:src|bitrates)\s*=\s*({.+?})\s*;', webpage, 'formats'),
39 video_id, transform_source=js_to_json)
41 QUALITIES = ('lq', 'mq', 'hq', 'hd')
42 quality_key = qualities(QUALITIES)
45 for format_id, format_list in bitrates.items():
46 if format_id == 'hls':
47 m3u8_url = url_or_none(format_list)
50 formats.extend(self._extract_m3u8_formats(
51 m3u8_url, video_id, ext='mp4', m3u8_id='hls', fatal=False))
53 if not isinstance(format_list, list):
55 for format_url in format_list:
56 format_url = url_or_none(format_url)
63 for quality in QUALITIES:
64 if '%s.mp4' % quality in format_url:
65 f_id += '-%s' % quality
67 'quality': quality_key(quality),
68 'format_note': quality.upper(),
73 self._sort_formats(formats)
75 title = self._og_search_title(
76 webpage, default=None) or self._search_regex(
77 (r'<value>(?P<title>[^<]+)',
78 r'videoTitle\s*:\s*(["\'])(?P<value>(?:(?!\1).)+)\1'), webpage,
79 'title', group='value')
80 thumbnail = self._og_search_thumbnail(
81 webpage, default=None) or self._search_regex(
82 r'poster\s*:\s*(["\'])(?P<value>(?:(?!\1).)+)\1', webpage,
83 'thumbnail', fatal=False, group='value')
84 duration = int_or_none(self._search_regex(
85 r'videoDuration\s*:\s*(\d+)', webpage, 'duration', fatal=False))
90 'thumbnail': thumbnail,
96 class NovaIE(InfoExtractor):
97 IE_DESC = 'TN.cz, Prásk.tv, Nova.cz, Novaplus.cz, FANDA.tv, Krásná.cz and Doma.cz'
98 _VALID_URL = r'https?://(?:[^.]+\.)?(?P<site>tv(?:noviny)?|tn|novaplus|vymena|fanda|krasna|doma|prask)\.nova\.cz/(?:[^/]+/)+(?P<id>[^/]+?)(?:\.html|/|$)'
100 'url': 'http://tn.nova.cz/clanek/tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci.html#player_13260',
101 'md5': '1dd7b9d5ea27bc361f110cd855a19bd3',
104 'display_id': 'tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci',
106 'title': 'Podzemní nemocnice v pražské Krči',
107 'description': 'md5:f0a42dd239c26f61c28f19e62d20ef53',
108 'thumbnail': r're:^https?://.*\.(?:jpg)',
111 'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
115 'title': 'Zaklínač 3: Divoký hon',
116 'description': 're:.*Pokud se stejně jako my nemůžete.*',
117 'thumbnail': r're:https?://.*\.jpg(\?.*)?',
118 'upload_date': '20150521',
122 'skip_download': True,
125 # media.cms.nova.cz embed
126 'url': 'https://novaplus.nova.cz/porad/ulice/epizoda/18760-2180-dil',
130 'title': '2180. díl',
131 'thumbnail': r're:^https?://.*\.jpg',
135 'skip_download': True,
137 'add_ie': [NovaEmbedIE.ie_key()],
139 'url': 'http://sport.tn.nova.cz/clanek/sport/hokej/nhl/zivot-jde-dal-hodnotil-po-vyrazeni-z-playoff-jiri-sekac.html',
140 'only_matching': True,
142 'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
143 'only_matching': True,
145 'url': 'http://doma.nova.cz/clanek/zdravi/prijdte-se-zapsat-do-registru-kostni-drene-jiz-ve-stredu-3-cervna.html',
146 'only_matching': True,
148 'url': 'http://prask.nova.cz/clanek/novinky/co-si-na-sobe-nase-hvezdy-nechaly-pojistit.html',
149 'only_matching': True,
151 'url': 'http://tv.nova.cz/clanek/novinky/zivot-je-zivot-bondovsky-trailer.html',
152 'only_matching': True,
155 def _real_extract(self, url):
156 mobj = re.match(self._VALID_URL, url)
157 display_id = mobj.group('id')
158 site = mobj.group('site')
160 webpage = self._download_webpage(url, display_id)
163 embed_id = self._search_regex(
164 r'<iframe[^>]+\bsrc=["\'](?:https?:)?//media\.cms\.nova\.cz/embed/([^/?#&]+)',
165 webpage, 'embed url', default=None)
167 return self.url_result(
168 'https://media.cms.nova.cz/embed/%s' % embed_id,
169 ie=NovaEmbedIE.ie_key(), video_id=embed_id)
171 video_id = self._search_regex(
172 [r"(?:media|video_id)\s*:\s*'(\d+)'",
174 r'id="article_video_(\d+)"',
175 r'id="player_(\d+)"'],
178 config_url = self._search_regex(
179 r'src="(https?://(?:tn|api)\.nova\.cz/bin/player/videojs/config\.php\?[^"]+)"',
180 webpage, 'config url', default=None)
184 player = self._parse_json(
186 r'(?s)Player\s*\(.+?\s*,\s*({.+?\bmedia\b["\']?\s*:\s*["\']?\d+.+?})\s*\)', webpage,
187 'player', default='{}'),
188 video_id, transform_source=js_to_json, fatal=False)
190 config_url = url_or_none(player.get('configUrl'))
191 params = player.get('configParams')
192 if isinstance(params, dict):
193 config_params = params
196 DEFAULT_SITE_ID = '23000'
198 'tvnoviny': DEFAULT_SITE_ID,
199 'novaplus': DEFAULT_SITE_ID,
200 'vymena': DEFAULT_SITE_ID,
201 'krasna': DEFAULT_SITE_ID,
207 site_id = self._search_regex(
208 r'site=(\d+)', webpage, 'site id', default=None) or SITES.get(
209 site, DEFAULT_SITE_ID)
211 config_url = 'https://api.nova.cz/bin/player/videojs/config.php'
219 config = self._download_json(
220 config_url, display_id,
221 'Downloading config JSON', query=config_params,
222 transform_source=lambda s: s[s.index('{'):s.rindex('}') + 1])
224 mediafile = config['mediafile']
225 video_url = mediafile['src']
227 m = re.search(r'^(?P<url>rtmpe?://[^/]+/(?P<app>[^/]+?))/&*(?P<playpath>.+)$', video_url)
230 'url': m.group('url'),
231 'app': m.group('app'),
232 'play_path': m.group('playpath'),
233 'player_path': 'http://tvnoviny.nova.cz/static/shared/app/videojs/video-js.swf',
240 self._sort_formats(formats)
242 title = mediafile.get('meta', {}).get('title') or self._og_search_title(webpage)
243 description = clean_html(self._og_search_description(webpage, default=None))
244 thumbnail = config.get('poster')
246 if site == 'novaplus':
247 upload_date = unified_strdate(self._search_regex(
248 r'(\d{1,2}-\d{1,2}-\d{4})$', display_id, 'upload date', default=None))
249 elif site == 'fanda':
250 upload_date = unified_strdate(self._search_regex(
251 r'<span class="date_time">(\d{1,2}\.\d{1,2}\.\d{4})', webpage, 'upload date', default=None))
257 'display_id': display_id,
259 'description': description,
260 'upload_date': upload_date,
261 'thumbnail': thumbnail,