[youtube:entrylistbase] Retry on 5xx HTTP errors (#20965)

This commit is contained in:
Sergey M․ 2019-05-06 01:12:32 +07:00
parent 876fed6bf3
commit f8c55c6664
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D
1 changed files with 20 additions and 4 deletions

View File

@ -16,6 +16,7 @@ from ..jsinterp import JSInterpreter
from ..swfinterp import SWFInterpreter from ..swfinterp import SWFInterpreter
from ..compat import ( from ..compat import (
compat_chr, compat_chr,
compat_HTTPError,
compat_kwargs, compat_kwargs,
compat_parse_qs, compat_parse_qs,
compat_urllib_parse_unquote, compat_urllib_parse_unquote,
@ -288,10 +289,25 @@ class YoutubeEntryListBaseInfoExtractor(YoutubeBaseInfoExtractor):
if not mobj: if not mobj:
break break
count = 0
retries = 3
while count <= retries:
try:
# Downloading page may result in intermittent 5xx HTTP error
# that is usually worked around with a retry
more = self._download_json( more = self._download_json(
'https://youtube.com/%s' % mobj.group('more'), playlist_id, 'https://youtube.com/%s' % mobj.group('more'), playlist_id,
'Downloading page #%s' % page_num, 'Downloading page #%s%s'
% (page_num, ' (retry #%d)' % count if count else ''),
transform_source=uppercase_escape) transform_source=uppercase_escape)
break
except ExtractorError as e:
if isinstance(e.cause, compat_HTTPError) and e.cause.code in (500, 503):
count += 1
if count <= retries:
continue
raise
content_html = more['content_html'] content_html = more['content_html']
if not content_html.strip(): if not content_html.strip():
# Some webpages show a "Load more" button but they don't # Some webpages show a "Load more" button but they don't