projects
/
youtube-dl
/ blobdiff
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
raw
|
inline
| side by side
Merge branch 'karrierevideos' of https://github.com/misterhat/youtube-dl into misterh...
[youtube-dl]
/
youtube_dl
/
extractor
/
nbc.py
diff --git
a/youtube_dl/extractor/nbc.py
b/youtube_dl/extractor/nbc.py
index 6cbe03d0f634138cf4b6f651590be6d700c54c08..dc2091be0d0c8706b2f3b6d78d88fa22fcb8b6d1 100644
(file)
--- a/
youtube_dl/extractor/nbc.py
+++ b/
youtube_dl/extractor/nbc.py
@@
-10,6
+10,8
@@
from ..compat import (
from ..utils import (
ExtractorError,
find_xpath_attr,
from ..utils import (
ExtractorError,
find_xpath_attr,
+ lowercase_escape,
+ unescapeHTML,
)
)
@@
-46,18
+48,23
@@
class NBCIE(InfoExtractor):
'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442',
},
'skip': 'Only works from US',
'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442',
},
'skip': 'Only works from US',
+ },
+ {
+ # This video has expired but with an escaped embedURL
+ 'url': 'http://www.nbc.com/parenthood/episode-guide/season-5/just-like-at-home/515',
+ 'skip': 'Expired'
}
]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
}
]
def _real_extract(self, url):
video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
- theplatform_url =
self.
_search_regex(
+ theplatform_url =
unescapeHTML(lowercase_escape(self._html
_search_regex(
[
r'(?:class="video-player video-player-full" data-mpx-url|class="player" src)="(.*?)"',
r'"embedURL"\s*:\s*"([^"]+)"'
],
[
r'(?:class="video-player video-player-full" data-mpx-url|class="player" src)="(.*?)"',
r'"embedURL"\s*:\s*"([^"]+)"'
],
- webpage, 'theplatform url').replace('_no_endcard', '')
+ webpage, 'theplatform url').replace('_no_endcard', '')
.replace('\\/', '/')))
if theplatform_url.startswith('//'):
theplatform_url = 'http:' + theplatform_url
return self.url_result(theplatform_url)
if theplatform_url.startswith('//'):
theplatform_url = 'http:' + theplatform_url
return self.url_result(theplatform_url)