[youtube] Improve extraction in 429 error conditions (closes #24283)

This commit is contained in:
Sergey M․ 2020-03-08 05:09:02 +07:00
parent 0ec9d4e565
commit f93abcf1da
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -1790,11 +1790,19 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
query['el'] = el query['el'] = el
if sts: if sts:
query['sts'] = sts query['sts'] = sts
try:
video_info_webpage = self._download_webpage( video_info_webpage = self._download_webpage(
'%s://www.youtube.com/get_video_info' % proto, '%s://www.youtube.com/get_video_info' % proto,
video_id, note=False, video_id, note=False,
errnote='unable to download video info webpage', errnote='unable to download video info webpage',
fatal=False, query=query) 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: if not video_info_webpage:
continue continue
get_video_info = compat_parse_qs(video_info_webpage) get_video_info = compat_parse_qs(video_info_webpage)
@ -1833,13 +1841,16 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
if messages: if messages:
return '\n'.join(messages) return '\n'.join(messages)
if not video_info: if not video_info and not player_response:
unavailable_message = extract_unavailable_message() unavailable_message = extract_unavailable_message()
if not unavailable_message: if not unavailable_message:
unavailable_message = 'Unable to extract video data' unavailable_message = 'Unable to extract video data'
raise ExtractorError( raise ExtractorError(
'YouTube said: %s' % unavailable_message, expected=True, video_id=video_id) 'YouTube said: %s' % unavailable_message, expected=True, video_id=video_id)
if not isinstance(video_info, dict):
video_info = {}
video_details = try_get( video_details = try_get(
player_response, lambda x: x['videoDetails'], dict) or {} player_response, lambda x: x['videoDetails'], dict) or {}