1
0
mirror of https://github.com/l1ving/youtube-dl synced 2024-12-04 02:02:52 +08:00

[vice] Remove extractor

The generic ooyala detection works fine.
This commit is contained in:
Jaime Marquínez Ferrándiz 2014-03-26 15:03:34 +01:00
parent 6a66ae96ed
commit 9a90636805
2 changed files with 0 additions and 39 deletions

View File

@ -263,7 +263,6 @@ from .veehd import VeeHDIE
from .veoh import VeohIE from .veoh import VeohIE
from .vesti import VestiIE from .vesti import VestiIE
from .vevo import VevoIE from .vevo import VevoIE
from .vice import ViceIE
from .viddler import ViddlerIE from .viddler import ViddlerIE
from .videobam import VideoBamIE from .videobam import VideoBamIE
from .videodetective import VideoDetectiveIE from .videodetective import VideoDetectiveIE

View File

@ -1,38 +0,0 @@
import re
from .common import InfoExtractor
from .ooyala import OoyalaIE
from ..utils import ExtractorError
class ViceIE(InfoExtractor):
_VALID_URL = r'http://www\.vice\.com/.*?/(?P<name>.+)'
_TEST = {
u'url': u'http://www.vice.com/Fringes/cowboy-capitalists-part-1',
u'file': u'43cW1mYzpia9IlestBjVpd23Yu3afAfp.mp4',
u'info_dict': {
u'title': u'VICE_COWBOYCAPITALISTS_PART01_v1_VICE_WM_1080p.mov',
},
u'params': {
# Requires ffmpeg (m3u8 manifest)
u'skip_download': True,
},
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
name = mobj.group('name')
webpage = self._download_webpage(url, name)
try:
ooyala_url = self._og_search_video_url(webpage)
except ExtractorError:
try:
embed_code = self._search_regex(
r'OO.Player.create\(\'ooyalaplayer\', \'(.+?)\'', webpage,
u'ooyala embed code')
ooyala_url = OoyalaIE._url_for_embed_code(embed_code)
except ExtractorError:
raise ExtractorError(u'The page doesn\'t contain a video', expected=True)
return self.url_result(ooyala_url, ie='Ooyala')