diff options
author | Ben Rog-Wilhelm <[email protected]> | 2021-05-04 14:14:35 -0500 |
---|---|---|
committer | GitHub <[email protected]> | 2021-05-05 02:14:35 +0700 |
commit | fe05191b8c59538a48b6cbc95f4fe54fc7e6a0ac (patch) | |
tree | 9de69c412b0836b723ba9965d98a4b1c01ee1f88 | |
parent | 0204838163bd4068fe23b40414573d1307d817ab (diff) | |
download | youtube-dl-fe05191b8c59538a48b6cbc95f4fe54fc7e6a0ac.tar.gz youtube-dl-fe05191b8c59538a48b6cbc95f4fe54fc7e6a0ac.zip |
[kaltura] Improve iframe extraction (#28969)
Co-authored-by: Sergey M. <[email protected]>
-rw-r--r-- | youtube_dl/extractor/gdcvault.py | 15 | ||||
-rw-r--r-- | youtube_dl/extractor/kaltura.py | 2 |
2 files changed, 16 insertions, 1 deletions
diff --git a/youtube_dl/extractor/gdcvault.py b/youtube_dl/extractor/gdcvault.py index 2f555c1d4..5ad40ee23 100644 --- a/youtube_dl/extractor/gdcvault.py +++ b/youtube_dl/extractor/gdcvault.py @@ -102,6 +102,21 @@ class GDCVaultIE(InfoExtractor): 'format': 'mp4-408', }, }, + { + # Kaltura embed, whitespace between quote and embedded URL in iframe's src + 'url': 'https://www.gdcvault.com/play/1025699', + 'info_dict': { + 'id': '0_zagynv0a', + 'ext': 'mp4', + 'title': 'Tech Toolbox', + 'upload_date': '20190408', + 'uploader_id': '[email protected]', + 'timestamp': 1554764629, + }, + 'params': { + 'skip_download': True, + }, + }, ] def _login(self, webpage_url, display_id): diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py index 5d0ff0418..c731612c4 100644 --- a/youtube_dl/extractor/kaltura.py +++ b/youtube_dl/extractor/kaltura.py @@ -145,7 +145,7 @@ class KalturaIE(InfoExtractor): ''', webpage)) or list(re.finditer( r'''(?xs) - <(?:iframe[^>]+src|meta[^>]+\bcontent)=(?P<q1>["']) + <(?:iframe[^>]+src|meta[^>]+\bcontent)=(?P<q1>["'])\s* (?:https?:)?//(?:(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P<partner_id>\d+) (?:(?!(?P=q1)).)* [?&;]entry_id=(?P<id>(?:(?!(?P=q1))[^&])+) |