commit: 31dbd054c801ec14c1ea29a2167b70c980f1d782
parent 66d04c74e097c03e4d644d7292546884cbee3d2b
Author: Sergey M․ <dstftw@gmail.com>
Date: Tue, 3 Sep 2019 01:24:20 +0700
[platzi] Improve client data extraction (closes #22290)
Diffstat:
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/youtube_dl/extractor/platzi.py b/youtube_dl/extractor/platzi.py
@@ -107,7 +107,11 @@ class PlatziIE(PlatziBaseIE):
data = self._parse_json(
self._search_regex(
- r'client_data\s*=\s*({.+?})\s*;', webpage, 'client data'),
+ # client_data may contain "};" so that we have to try more
+ # strict regex first
+ (r'client_data\s*=\s*({.+?})\s*;\s*\n',
+ r'client_data\s*=\s*({.+?})\s*;'),
+ webpage, 'client data'),
lecture_id)
material = data['initialState']['material']