From f93abcf1da5aa1ca122896254bdec3a3c831ac24 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Sergey=20M=E2=80=A4?= Date: Sun, 8 Mar 2020 05:09:02 +0700 Subject: [PATCH] [youtube] Improve extraction in 429 error conditions (closes #24283) --- youtube_dl/extractor/youtube.py | 23 +++++++++++++++++------ 1 file changed, 17 insertions(+), 6 deletions(-) diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index d3e18a6ad..9cbdc7ac5 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -1790,11 +1790,19 @@ class YoutubeIE(YoutubeBaseInfoExtractor): query['el'] = el if sts: query['sts'] = sts - video_info_webpage = self._download_webpage( - '%s://www.youtube.com/get_video_info' % proto, - video_id, note=False, - errnote='unable to download video info webpage', - fatal=False, query=query) + try: + video_info_webpage = self._download_webpage( + '%s://www.youtube.com/get_video_info' % proto, + video_id, note=False, + errnote='unable to download video info webpage', + query=query) + except ExtractorError as e: + # Skip further retries if we get 429 since solving + # captcha only unblocks access to website but + # not get_video_info end point + if isinstance(e.cause, compat_HTTPError) and e.cause.code == 429: + break + continue if not video_info_webpage: continue get_video_info = compat_parse_qs(video_info_webpage) @@ -1833,13 +1841,16 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if messages: return '\n'.join(messages) - if not video_info: + if not video_info and not player_response: unavailable_message = extract_unavailable_message() if not unavailable_message: unavailable_message = 'Unable to extract video data' raise ExtractorError( 'YouTube said: %s' % unavailable_message, expected=True, video_id=video_id) + if not isinstance(video_info, dict): + video_info = {} + video_details = try_get( player_response, lambda x: x['videoDetails'], dict) or {} -- 2.30.2