import xml.etree.ElementTree
from .common import InfoExtractor
+from ..utils import ExtractorError
class ClipfishIE(InfoExtractor):
_VALID_URL = r'^https?://(?:www\.)?clipfish\.de/.*?/video/(?P<id>[0-9]+)/'
_TEST = {
- u'url': u'http://www.clipfish.de/special/supertalent/video/4028320/supertalent-2013-ivana-opacak-singt-nobodys-perfect/',
- u'file': u'4028320.f4v',
- u'md5': u'5e38bda8c329fbfb42be0386a3f5a382',
+ u'url': u'http://www.clipfish.de/special/game-trailer/video/3966754/fifa-14-e3-2013-trailer/',
+ u'file': u'3966754.mp4',
+ u'md5': u'2521cd644e862936cf2e698206e47385',
u'info_dict': {
- u'title': u'Supertalent 2013: Ivana Opacak singt Nobody\'s Perfect',
- u'duration': 399,
- }
+ u'title': u'FIFA 14 - E3 2013 Trailer',
+ u'duration': 82,
+ },
+ u'skip': 'Blocked in the US'
}
def _real_extract(self, url):
info_url = ('http://www.clipfish.de/devxml/videoinfo/%s?ts=%d' %
(video_id, int(time.time())))
- info_xml = self._download_webpage(
+ doc = self._download_xml(
info_url, video_id, note=u'Downloading info page')
- doc = xml.etree.ElementTree.fromstring(info_xml)
title = doc.find('title').text
video_url = doc.find('filename').text
+ if video_url is None:
+ xml_bytes = xml.etree.ElementTree.tostring(doc)
+ raise ExtractorError(u'Cannot find video URL in document %r' %
+ xml_bytes)
thumbnail = doc.find('imageurl').text
duration_str = doc.find('duration').text
m = re.match(