fix ytInitialData parsing

pull/2/head
siikamiika 4 years ago
parent f0f76a33dc
commit eaedbfd97e

@ -28,8 +28,14 @@ class YoutubeLiveChatReplayFD(FragmentFD):
return self._download_fragment(ctx, url, info_dict, headers)
def parse_yt_initial_data(data):
raw_json = re.search(b'window\\["ytInitialData"\\]\\s*=\\s*(.*);', data).group(1)
return json.loads(raw_json)
window_patt = b'window\\["ytInitialData"\\]\\s*=\\s*(.*?);'
var_patt = b'var\\s+ytInitialData\\s*=\\s*(.*?);'
for patt in window_patt, var_patt:
try:
raw_json = re.search(patt, data).group(1)
return json.loads(raw_json)
except AttributeError:
continue
self._prepare_and_start_frag_download(ctx)

@ -1495,7 +1495,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
def _get_yt_initial_data(self, video_id, webpage):
config = self._search_regex(
r'window\["ytInitialData"\]\s*=\s*(.*);',
(r'window\["ytInitialData"\]\s*=\s*(.*);',
r'var\s+ytInitialData\s*=\s*(.*?);'),
webpage, 'ytInitialData', default=None)
if config:
return self._parse_json(

Loading…
Cancel
Save