Recognize a third format of the upload_date in the 'watch-uploader-info' element

This commit is contained in:
Anton Novosyolov 2014-06-02 13:30:23 +04:00
parent b7e8b6e37a
commit 263bd4ec50

View File

@ -1140,7 +1140,7 @@ def _real_extract(self, url):
mobj = re.search(r'(?s)id="eow-date.*?>(.*?)</span>', video_webpage) mobj = re.search(r'(?s)id="eow-date.*?>(.*?)</span>', video_webpage)
if mobj is None: if mobj is None:
mobj = re.search( mobj = re.search(
r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded) on (.*?)</strong>', r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live) on (.*?)</strong>',
video_webpage) video_webpage)
if mobj is not None: if mobj is not None:
upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split()) upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split())