[googledrive] Use redirect URLs for source format (closes #18877, closes #23919, closes #24689, closes #26565)

This commit is contained in:
Sergey M․ 2020-09-13 20:43:50 +07:00
parent 95c9810015
commit da2069fb22
No known key found for this signature in database
GPG Key ID: 2C393E0F18A9236D

View File

@ -220,19 +220,27 @@ class GoogleDriveIE(InfoExtractor):
'id': video_id, 'id': video_id,
'export': 'download', 'export': 'download',
}) })
urlh = self._request_webpage(
source_url, video_id, note='Requesting source file', def request_source_file(source_url, kind):
errnote='Unable to request source file', fatal=False) return self._request_webpage(
source_url, video_id, note='Requesting %s file' % kind,
errnote='Unable to request %s file' % kind, fatal=False)
urlh = request_source_file(source_url, 'source')
if urlh: if urlh:
def add_source_format(src_url): def add_source_format(urlh):
formats.append({ formats.append({
'url': src_url, # Use redirect URLs as download URLs in order to calculate
# correct cookies in _calc_cookies.
# Using original URLs may result in redirect loop due to
# google.com's cookies mistakenly used for googleusercontent.com
# redirect URLs (see #23919).
'url': urlh.geturl(),
'ext': determine_ext(title, 'mp4').lower(), 'ext': determine_ext(title, 'mp4').lower(),
'format_id': 'source', 'format_id': 'source',
'quality': 1, 'quality': 1,
}) })
if urlh.headers.get('Content-Disposition'): if urlh.headers.get('Content-Disposition'):
add_source_format(source_url) add_source_format(urlh)
else: else:
confirmation_webpage = self._webpage_read_content( confirmation_webpage = self._webpage_read_content(
urlh, url, video_id, note='Downloading confirmation page', urlh, url, video_id, note='Downloading confirmation page',
@ -242,9 +250,12 @@ class GoogleDriveIE(InfoExtractor):
r'confirm=([^&"\']+)', confirmation_webpage, r'confirm=([^&"\']+)', confirmation_webpage,
'confirmation code', fatal=False) 'confirmation code', fatal=False)
if confirm: if confirm:
add_source_format(update_url_query(source_url, { confirmed_source_url = update_url_query(source_url, {
'confirm': confirm, 'confirm': confirm,
})) })
urlh = request_source_file(confirmed_source_url, 'confirmed source')
if urlh and urlh.headers.get('Content-Disposition'):
add_source_format(urlh)
if not formats: if not formats:
reason = self._search_regex( reason = self._search_regex(