[platzi] Improve client data extraction (closes #22290)

5 years ago · 31dbd054c8
parent 66d04c74e0
commit 31dbd054c8
1 changed files with 5 additions and 1 deletions
--- 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']