[cnn] Add support for articles with videos (fixes #4541)

This commit is contained in:
Adrian Kretz 2014-12-22 18:40:36 +01:00
parent 1362bbbb4b
commit 9532d72371
2 changed files with 26 additions and 0 deletions

View file

@ -64,6 +64,7 @@ from .cnet import CNETIE
from .cnn import (
CNNIE,
CNNBlogsIE,
CNNArticleIE,
)
from .collegehumor import CollegeHumorIE
from .comedycentral import ComedyCentralIE, ComedyCentralShowsIE

View file

@ -127,3 +127,28 @@ class CNNBlogsIE(InfoExtractor):
'url': cnn_url,
'ie_key': CNNIE.ie_key(),
}
class CNNArticleIE(InfoExtractor):
_VALID_URL = r'https?://((edition|www)\.)?cnn\.com/(?!video/)'
_TEST = {
'url': 'http://www.cnn.com/2014/12/21/politics/obama-north-koreas-hack-not-war-but-cyber-vandalism/',
'md5': '275b326f85d80dff7592a9820f5dc887',
'info_dict': {
'id': 'bestoftv/2014/12/21/sotu-crowley-president-obama-north-korea-not-going-to-be-intimidated.cnn',
'ext': 'mp4',
'title': 'Obama: We\'re not going to be intimidated',
'description': 'md5:e735586f3dc936075fa654a4d91b21f9',
'upload_date': '20141220',
},
'add_ie': ['CNN'],
}
def _real_extract(self, url):
webpage = self._download_webpage(url, url_basename(url))
cnn_url = self._html_search_regex(r"video:\s*'([^']+)'", webpage, 'cnn url')
return {
'_type': 'url',
'url': r'http://cnn.com/video/?/video/' + cnn_url,
'ie_key': CNNIE.ie_key(),
}