aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorSergey M. <dstftw@gmail.com>2014-12-29 20:21:39 +0600
committerSergey M. <dstftw@gmail.com>2014-12-29 20:21:39 +0600
commit974739aab5935ddf19c005ea084c83a45c855348 (patch)
tree5b051cd41cc22743401363c871a8db75c9f3dac8
parentae6986fb895ceb72e084b587bc9d070c86848347 (diff)
parent9532d723715f1d1a08c126ba8af6940113b88d3b (diff)
downloadyoutube-dl-974739aab5935ddf19c005ea084c83a45c855348.tar.xz
Merge pull request #4543 from akretz/cnn_fix
[cnn] Add support for articles with videos (fixes #4541)
-rw-r--r--youtube_dl/extractor/__init__.py1
-rw-r--r--youtube_dl/extractor/cnn.py25
2 files changed, 26 insertions, 0 deletions
diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
index 32d616397..ab0f76862 100644
--- a/youtube_dl/extractor/__init__.py
+++ b/youtube_dl/extractor/__init__.py
@@ -65,6 +65,7 @@ from .cnet import CNETIE
from .cnn import (
CNNIE,
CNNBlogsIE,
+ CNNArticleIE,
)
from .collegehumor import CollegeHumorIE
from .comedycentral import ComedyCentralIE, ComedyCentralShowsIE
diff --git a/youtube_dl/extractor/cnn.py b/youtube_dl/extractor/cnn.py
index 81142ee41..cddcacdf1 100644
--- a/youtube_dl/extractor/cnn.py
+++ b/youtube_dl/extractor/cnn.py
@@ -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(),
+ }