From 9edf47df7bfbcdd67bce68ea04865aa1f51df2f6 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Mon, 1 May 2017 01:03:47 +0700 Subject: [PATCH] [brightcove] Allow whitespace around attribute names in embedded code --- youtube_dl/extractor/brightcove.py | 2 +- youtube_dl/extractor/generic.py | 16 ++++++++++++++++ 2 files changed, 17 insertions(+), 1 deletion(-) diff --git a/youtube_dl/extractor/brightcove.py b/youtube_dl/extractor/brightcove.py index 97602ca30..3f017a2b1 100644 --- a/youtube_dl/extractor/brightcove.py +++ b/youtube_dl/extractor/brightcove.py @@ -522,7 +522,7 @@ def _extract_urls(ie, webpage): # [2] looks like: for video, script_tag, account_id, player_id, embed in re.findall( r'''(?isx) - (]*data-video-id=['"]?[^>]+>) + (]*\bdata-video-id\s*=\s*['"]?[^>]+>) (?:.*? (]+ src=["\'](?:https?:)?//players\.brightcove\.net/ diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py index bd9d9aa13..b06f43446 100644 --- a/youtube_dl/extractor/generic.py +++ b/youtube_dl/extractor/generic.py @@ -1429,6 +1429,22 @@ class GenericIE(InfoExtractor): 'skip_download': True, }, }, + { + # Brightcove embed with whitespace around attribute names + 'url': 'http://www.stack.com/video/3167554373001/learn-to-hit-open-three-pointers-with-damian-lillard-s-baseline-drift-drill', + 'info_dict': { + 'id': '3167554373001', + 'ext': 'mp4', + 'title': "Learn to Hit Open Three-Pointers With Damian Lillard's Baseline Drift Drill", + 'description': 'md5:57bacb0e0f29349de4972bfda3191713', + 'uploader_id': '1079349493', + 'upload_date': '20140207', + 'timestamp': 1391810548, + }, + 'params': { + 'skip_download': True, + }, + }, # Another form of arte.tv embed { 'url': 'http://www.tv-replay.fr/redirection/09-04-16/arte-reportage-arte-11508975.html',