[sites:youtube] Updated extraction regex to allow for }; in data (a…

…long with corresponding tests)
xenova · Jul 4, 2021 · 69e4ecb · 69e4ecb
1 parent a854e80
commit 69e4ecb
Showing 1 changed file with 34 additions and 2 deletions.
diff --git a/chat_downloader/sites/youtube.py b/chat_downloader/sites/youtube.py
@@ -260,11 +260,43 @@ def __init__(self, **kwargs):
             'expected_result': {
                 'error': VideoUnplayable,
             }
+        },
+
+        # Potential parsing errors
+        {
+            'name': "Parsing error with '};' inside yt initial data (1)",
+            'params': {
+                'url': 'https://www.youtube.com/watch?v=CHqg6qOn4no',
+            },
+            'expected_result': {
+                'error': NoChatReplay,
+            }
+        },
+        {
+            'name': "Parsing error with '};' inside yt initial data (2)",
+            'params': {
+                'url': 'https://www.youtube.com/watch?v=gVfgbahppCY',
+            },
+            'expected_result': {
+                'error': NoChatReplay,
+            }
+        },
+        {
+            'name': 'Title with JS-like syntax "};"',
+            'params': {
+                'url': 'https://www.youtube.com/watch?v=lsguqyKfVQg',
+            },
+            'expected_result': {
+                'error': NoChatReplay,
+            }
         }
     ]
 
-    _YT_INITIAL_DATA_RE = r'(?:window\s*\[\s*["\']ytInitialData["\']\s*\]|ytInitialData)\s*=\s*({.+?})\s*;'
-    _YT_INITIAL_PLAYER_RESPONSE_RE = r'ytInitialPlayerResponse\s*=\s*({.+?})\s*;'
+    _YT_INITIAL_BOUNDARY_RE = r'\s*(?:var\s+meta|</script|\n)'
+    _YT_INITIAL_DATA_RE = r'(?:window\s*\[\s*["\']ytInitialData["\']\s*\]|ytInitialData)\s*=\s*({.+?})\s*;' + \
+        _YT_INITIAL_BOUNDARY_RE
+    _YT_INITIAL_PLAYER_RESPONSE_RE = r'ytInitialPlayerResponse\s*=\s*({.+?})\s*;' + \
+        _YT_INITIAL_BOUNDARY_RE
     _YT_CFG_RE = r'ytcfg\.set\s*\(\s*({.+?})\s*\)\s*;'
 
     _YT_HOME = 'https://www.youtube.com'