[genric] Eliminate duplicated video URLs (closes #6562)

This commit is contained in:
Yen Chi Hsuan 2016-05-22 22:22:27 +08:00
parent a4a7c44bd3
commit 4a12077855
No known key found for this signature in database
GPG Key ID: 3FDDD575826C5C30
1 changed files with 11 additions and 1 deletions

View File

@ -1194,6 +1194,16 @@ class GenericIE(InfoExtractor):
'uploader': 'Lake8737', 'uploader': 'Lake8737',
} }
}, },
# Duplicated embedded video URLs
{
'url': 'http://www.hudl.com/athlete/2538180/highlights/149298443',
'info_dict': {
'id': '149298443_480_16c25b74_2',
'ext': 'mp4',
'title': 'vs. Blue Orange Spring Game',
'uploader': 'www.hudl.com',
},
},
] ]
def report_following_redirect(self, new_url): def report_following_redirect(self, new_url):
@ -2111,7 +2121,7 @@ class GenericIE(InfoExtractor):
raise UnsupportedError(url) raise UnsupportedError(url)
entries = [] entries = []
for video_url in found: for video_url in orderedSet(found):
video_url = unescapeHTML(video_url) video_url = unescapeHTML(video_url)
video_url = video_url.replace('\\/', '/') video_url = video_url.replace('\\/', '/')
video_url = compat_urlparse.urljoin(url, video_url) video_url = compat_urlparse.urljoin(url, video_url)