2015-06-01 18:53:04 +00:00
|
|
|
# encoding: utf-8
|
|
|
|
from __future__ import unicode_literals
|
|
|
|
|
|
|
|
import re
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
2015-06-02 09:56:36 +00:00
|
|
|
from ..utils import clean_html, determine_ext
|
2015-06-01 18:53:04 +00:00
|
|
|
|
|
|
|
|
|
|
|
class NovaIE(InfoExtractor):
|
|
|
|
IE_DESC = 'TN.cz, Prásk.tv, Nova.cz, Novaplus.cz, FANDA.tv, Krásná.cz and Doma.cz'
|
2015-06-02 09:47:52 +00:00
|
|
|
_VALID_URL = 'http://(?:[^.]+\.)?(?P<site>tv(?:noviny)?|tn|novaplus|vymena|fanda|krasna|doma|prask)\.nova\.cz/(?:[^/]+/)+(?P<id>[^/]+?)(?:\.html|/?)$'
|
2015-06-01 18:53:04 +00:00
|
|
|
_TESTS = [{
|
|
|
|
'url': 'http://tvnoviny.nova.cz/clanek/novinky/co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou.html',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '1608920',
|
|
|
|
'display_id': 'co-na-sebe-sportaci-praskli-vime-jestli-pujde-hrdlicka-na-materskou',
|
|
|
|
'ext': 'flv',
|
|
|
|
'title': 'Duel: Michal Hrdlička a Petr Suchoň',
|
|
|
|
'description': 'md5:d0cc509858eee1b1374111c588c6f5d5',
|
|
|
|
'thumbnail': 're:^https?://.*\.(?:jpg)',
|
|
|
|
},
|
|
|
|
'params': {
|
|
|
|
# rtmp download
|
|
|
|
'skip_download': True,
|
|
|
|
}
|
|
|
|
}, {
|
|
|
|
'url': 'http://tn.nova.cz/clanek/tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci.html',
|
|
|
|
'md5': '1dd7b9d5ea27bc361f110cd855a19bd3',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '1757139',
|
|
|
|
'display_id': 'tajemstvi-ukryte-v-podzemi-specialni-nemocnice-v-prazske-krci',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': 'Podzemní nemocnice v pražské Krči',
|
|
|
|
'description': 'md5:f0a42dd239c26f61c28f19e62d20ef53',
|
|
|
|
'thumbnail': 're:^https?://.*\.(?:jpg)',
|
|
|
|
}
|
|
|
|
}, {
|
|
|
|
'url': 'http://novaplus.nova.cz/porad/policie-modrava/video/5591-policie-modrava-15-dil-blondynka-na-hrbitove/',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '1756825',
|
|
|
|
'display_id': '5591-policie-modrava-15-dil-blondynka-na-hrbitove',
|
2015-06-02 09:52:39 +00:00
|
|
|
'ext': 'mp4',
|
2015-06-01 18:53:04 +00:00
|
|
|
'title': 'Policie Modrava - 15. díl - Blondýnka na hřbitově',
|
2015-06-02 09:56:36 +00:00
|
|
|
'description': 'md5:dc24e50be5908df83348e50d1431295e',
|
2015-06-01 18:53:04 +00:00
|
|
|
'thumbnail': 're:^https?://.*\.(?:jpg)',
|
|
|
|
},
|
|
|
|
'params': {
|
|
|
|
# rtmp download
|
|
|
|
'skip_download': True,
|
|
|
|
}
|
2015-06-02 09:55:41 +00:00
|
|
|
}, {
|
|
|
|
'url': 'http://novaplus.nova.cz/porad/televizni-noviny/video/5585-televizni-noviny-30-5-2015/',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '1756858',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': 'Televizní noviny - 30. 5. 2015',
|
|
|
|
'thumbnail': 're:^https?://.*\.(?:jpg)',
|
|
|
|
'upload_date': '20150530',
|
|
|
|
},
|
|
|
|
'params': {
|
|
|
|
# rtmp download
|
|
|
|
'skip_download': True,
|
|
|
|
}
|
|
|
|
}, {
|
|
|
|
'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
|
|
|
|
'info_dict': {
|
|
|
|
'id': '1753621',
|
|
|
|
'ext': 'mp4',
|
|
|
|
'title': 'Zaklínač 3: Divoký hon',
|
|
|
|
'description': 're:.*Pokud se stejně jako my nemůžete.*',
|
|
|
|
'thumbnail': 're:https?://.*\.jpg(\?.*)?',
|
|
|
|
'upload_date': '20150521',
|
|
|
|
},
|
|
|
|
'params': {
|
|
|
|
# rtmp download
|
|
|
|
'skip_download': True,
|
|
|
|
}
|
2015-06-01 18:53:04 +00:00
|
|
|
}, {
|
|
|
|
'url': 'http://sport.tn.nova.cz/clanek/sport/hokej/nhl/zivot-jde-dal-hodnotil-po-vyrazeni-z-playoff-jiri-sekac.html',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'http://fanda.nova.cz/clanek/fun-and-games/krvavy-epos-zaklinac-3-divoky-hon-vychazi-vyhrajte-ho-pro-sebe.html',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'http://doma.nova.cz/clanek/zdravi/prijdte-se-zapsat-do-registru-kostni-drene-jiz-ve-stredu-3-cervna.html',
|
|
|
|
'only_matching': True,
|
|
|
|
}, {
|
|
|
|
'url': 'http://prask.nova.cz/clanek/novinky/co-si-na-sobe-nase-hvezdy-nechaly-pojistit.html',
|
|
|
|
'only_matching': True,
|
2015-06-01 18:57:08 +00:00
|
|
|
}, {
|
|
|
|
'url': 'http://tv.nova.cz/clanek/novinky/zivot-je-zivot-bondovsky-trailer.html',
|
|
|
|
'only_matching': True,
|
2015-06-01 18:53:04 +00:00
|
|
|
}]
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
|
|
|
display_id = mobj.group('id')
|
|
|
|
site = mobj.group('site')
|
|
|
|
|
|
|
|
webpage = self._download_webpage(url, display_id)
|
|
|
|
|
|
|
|
video_id = self._search_regex(
|
|
|
|
[r"(?:media|video_id)\s*:\s*'(\d+)'",
|
|
|
|
r'media=(\d+)',
|
|
|
|
r'id="article_video_(\d+)"',
|
|
|
|
r'id="player_(\d+)"'],
|
|
|
|
webpage, 'video id')
|
|
|
|
|
|
|
|
config_url = self._search_regex(
|
|
|
|
r'src="(http://tn\.nova\.cz/bin/player/videojs/config\.php\?[^"]+)"',
|
|
|
|
webpage, 'config url', default=None)
|
|
|
|
|
|
|
|
if not config_url:
|
|
|
|
DEFAULT_SITE_ID = '23000'
|
|
|
|
SITES = {
|
|
|
|
'tvnoviny': DEFAULT_SITE_ID,
|
|
|
|
'novaplus': DEFAULT_SITE_ID,
|
|
|
|
'vymena': DEFAULT_SITE_ID,
|
|
|
|
'krasna': DEFAULT_SITE_ID,
|
|
|
|
'fanda': '30',
|
|
|
|
'tn': '30',
|
|
|
|
'doma': '30',
|
|
|
|
}
|
|
|
|
|
|
|
|
site_id = self._search_regex(
|
|
|
|
r'site=(\d+)', webpage, 'site id', default=None) or SITES.get(site, DEFAULT_SITE_ID)
|
|
|
|
|
|
|
|
config_url = ('http://tn.nova.cz/bin/player/videojs/config.php?site=%s&media=%s&jsVar=vjsconfig'
|
|
|
|
% (site_id, video_id))
|
|
|
|
|
|
|
|
config = self._download_json(
|
|
|
|
config_url, display_id,
|
|
|
|
'Downloading config JSON',
|
2015-06-02 09:57:03 +00:00
|
|
|
transform_source=lambda s: s[s.index('{'):s.rindex('}') + 1])
|
2015-06-01 18:53:04 +00:00
|
|
|
|
|
|
|
mediafile = config['mediafile']
|
|
|
|
video_url = mediafile['src']
|
2015-06-02 09:52:39 +00:00
|
|
|
ext = determine_ext(video_url)
|
|
|
|
video_url = video_url.replace('&{}:'.format(ext), '')
|
2015-06-01 18:53:04 +00:00
|
|
|
|
|
|
|
title = mediafile.get('meta', {}).get('title') or self._og_search_title(webpage)
|
2015-06-02 09:56:36 +00:00
|
|
|
description = clean_html(self._og_search_description(webpage, default=None))
|
2015-06-01 18:53:04 +00:00
|
|
|
thumbnail = config.get('poster')
|
|
|
|
|
2015-06-02 09:55:41 +00:00
|
|
|
mobj = None
|
|
|
|
if site == 'novaplus':
|
|
|
|
mobj = re.search(r'(?P<day>\d{1,2})-(?P<month>\d{1,2})-(?P<year>\d{4})$', display_id)
|
|
|
|
if site == 'fanda':
|
|
|
|
mobj = re.search(
|
|
|
|
r'<span class="date_time">(?P<day>\d{1,2})\.(?P<month>\d{1,2})\.(?P<year>\d{4})\b', webpage)
|
|
|
|
if mobj:
|
|
|
|
upload_date = '{}{:02d}{:02d}'.format(mobj.group('year'), int(mobj.group('month')), int(mobj.group('day')))
|
|
|
|
else:
|
|
|
|
upload_date = None
|
|
|
|
|
2015-06-01 18:53:04 +00:00
|
|
|
return {
|
|
|
|
'id': video_id,
|
|
|
|
'display_id': display_id,
|
|
|
|
'title': title,
|
|
|
|
'description': description,
|
2015-06-02 09:55:41 +00:00
|
|
|
'upload_date': upload_date,
|
2015-06-01 18:53:04 +00:00
|
|
|
'thumbnail': thumbnail,
|
2015-06-02 09:52:39 +00:00
|
|
|
'url': video_url,
|
|
|
|
'ext': ext,
|
2015-06-01 18:53:04 +00:00
|
|
|
}
|