import re
from .common import InfoExtractor
-from ..utils import ExtractorError
+from ..utils import (
+ clean_html,
+ ExtractorError,
+)
class RTLnowIE(InfoExtractor):
- """Information Extractor for RTL NOW, RTL2 NOW, SUPER RTL NOW and VOX NOW"""
- _VALID_URL = r'(?:http://)?(?P<url>(?P<base_url>rtl-now\.rtl\.de/|rtl2now\.rtl2\.de/|(?:www\.)?voxnow\.de/|(?:www\.)?superrtlnow\.de/)[a-zA-Z0-9-]+/[a-zA-Z0-9-]+\.php\?(?:container_id|film_id)=(?P<video_id>[0-9]+)&player=1(?:&season=[0-9]+)?(?:&.*)?)'
+ """Information Extractor for RTL NOW, RTL2 NOW, RTL NITRO, SUPER RTL NOW and VOX NOW"""
+ _VALID_URL = r'(?:http://)?(?P<url>(?P<base_url>rtl-now\.rtl\.de/|rtl2now\.rtl2\.de/|(?:www\.)?voxnow\.de/|(?:www\.)?rtlnitronow\.de/|(?:www\.)?superrtlnow\.de/)[a-zA-Z0-9-]+/[a-zA-Z0-9-]+\.php\?(?:container_id|film_id)=(?P<video_id>[0-9]+)&player=1(?:&season=[0-9]+)?(?:&.*)?)'
_TESTS = [{
u'url': u'http://rtl-now.rtl.de/ahornallee/folge-1.php?film_id=90419&player=1&season=1',
u'file': u'90419.flv',
u'params': {
u'skip_download': True,
},
+ u'skip': u'Only works from Germany',
},
{
u'url': u'http://rtl2now.rtl2.de/aerger-im-revier/episode-15-teil-1.php?film_id=69756&player=1&season=2&index=5',
u'params': {
u'skip_download': True,
},
+ u'skip': u'Only works from Germany',
},
{
u'url': u'www.voxnow.de/voxtours/suedafrika-reporter-ii.php?film_id=13883&player=1&season=17',
u'params': {
u'skip_download': True,
},
+ },
+ {
+ u'url': u'http://www.rtlnitronow.de/recht-ordnung/lebensmittelkontrolle-erlangenordnungsamt-berlin.php?film_id=127367&player=1&season=1',
+ u'file': u'127367.flv',
+ u'info_dict': {
+ u'upload_date': u'20130926',
+ u'title': u'Recht & Ordnung - Lebensmittelkontrolle Erlangen/Ordnungsamt...',
+ u'description': u'Lebensmittelkontrolle Erlangen/Ordnungsamt Berlin',
+ u'thumbnail': u'http://autoimg.static-fra.de/nitronow/344787/1500x1500/image2.jpg',
+ },
+ u'params': {
+ u'skip_download': True,
+ },
}]
def _real_extract(self,url):
video_id = mobj.group(u'video_id')
webpage = self._download_webpage(webpage_url, video_id)
- video_title = self._html_search_regex(r'<title>(?P<title>[^<]+)</title>',
+
+ note_m = re.search(r'''(?sx)
+ <div[ ]style="margin-left:[ ]20px;[ ]font-size:[ ]13px;">(.*?)
+ <div[ ]id="playerteaser">''', webpage)
+ if note_m:
+ msg = clean_html(note_m.group(1))
+ raise ExtractorError(msg)
+
+ video_title = self._html_search_regex(r'<title>(?P<title>[^<]+?)( \| [^<]*)?</title>',
webpage, u'title')
playerdata_url = self._html_search_regex(r'\'playerdata\': \'(?P<playerdata_url>[^\']+)\'',
webpage, u'playerdata_url')