From c21692fa94df49ef925c06c00e5db1d8bb0f770d Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Tue, 1 May 2018 03:09:04 +0700 Subject: [PATCH] [kaltura] Improve iframe embeds detection (closes #16337) --- youtube_dl/extractor/generic.py | 17 +++++++++++++++++ youtube_dl/extractor/kaltura.py | 3 ++- 2 files changed, 19 insertions(+), 1 deletion(-) diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index 252f97c261..73980601c5 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -1282,6 +1282,23 @@ class GenericIE(InfoExtractor): }, 'add_ie': ['Kaltura'], }, + { + # Kaltura iframe embed, more sophisticated + 'url': 'http://www.cns.nyu.edu/~eero/math-tools/Videos/lecture-05sep2017.html', + 'info_dict': { + 'id': '1_9gzouybz', + 'ext': 'mp4', + 'title': 'lecture-05sep2017', + 'description': 'md5:40f347d91fd4ba047e511c5321064b49', + 'upload_date': '20170913', + 'uploader_id': 'eps2', + 'timestamp': 1505340777, + }, + 'params': { + 'skip_download': True, + }, + 'add_ie': ['Kaltura'], + }, { # meta twitter:player 'url': 'http://thechive.com/2017/12/08/all-i-want-for-christmas-is-more-twerk/', diff --git a/youtube_dl/extractor/kaltura.py b/youtube_dl/extractor/kaltura.py index 0ea89e4d66..04f68fce41 100644 --- a/youtube_dl/extractor/kaltura.py +++ b/youtube_dl/extractor/kaltura.py @@ -136,9 +136,10 @@ class KalturaIE(InfoExtractor): re.search( r'''(?xs) <(?:iframe[^>]+src|meta[^>]+\bcontent)=(?P["']) - (?:https?:)?//(?:(?:www|cdnapi)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P\d+) + (?:https?:)?//(?:(?:www|cdnapi(?:sec)?)\.)?kaltura\.com/(?:(?!(?P=q1)).)*\b(?:p|partner_id)/(?P\d+) (?:(?!(?P=q1)).)* [?&;]entry_id=(?P(?:(?!(?P=q1))[^&])+) + (?:(?!(?P=q1)).)* (?P=q1) ''', webpage) )