1 from __future__ import unicode_literals
\r
5 from .common import InfoExtractor
\r
8 class BildIE(InfoExtractor):
\r
11 'url': 'http://www.bild.de/video/clip/apple-ipad-air/das-koennen-die-neuen-ipads-38184146.bild.html',
\r
14 'title': 'BILD hat sie getestet',
\r
15 'thumbnail': 'http://bilder.bild.de/fotos/stand-das-koennen-die-neuen-ipads-38184138/Bild/1.bild.jpg',
\r
20 #http://www.bild.de/video/clip/apple-ipad-air/das-koennen-die-neuen-ipads-38184146.bild.html
\r
21 _VALID_URL = r'http?://(?:www\.)?bild\.de/(?:[^/]+/)+(?P<display_id>[^/]+)-(?P<id>\d+)(?:,auto=true)?\.bild\.html'
\r
23 def _real_extract(self, url):
\r
24 m = re.match(self._VALID_URL, url)
\r
25 video_id = m.group('id')
\r
27 #webpage = self._download_webpage(url, video_id)
\r
29 xml_url = url.split(".bild.html")[0]+",view=xml.bild.xml"
\r
31 doc = self._download_xml(xml_url, video_id)
\r
33 video_url = doc.attrib['src']
\r
34 title = doc.attrib['ueberschrift']
\r
35 description = doc.attrib['text']
\r
36 thumbnail = doc.attrib['img']
\r
37 duration = int(doc.attrib['duration'])/1000
\r
42 'description': description,
\r
44 'thumbnail': thumbnail,
\r
45 'duration': duration,
\r