Update app.py

app.py

@@ -8,6 +8,7 @@ from functools import lru_cache
 import hashlib
 from langdetect import detect, LangDetectException
 import html
+
 # =========================
 # Config
 # =========================
@@ -76,13 +77,36 @@ def is_likely_author_bio_or_footer(element, text):
     """Check if an element is likely an author bio, footer, or other non-main content."""
     # Check parent and element classes/ids
     parent = element.parent
-    element_attrs = " ".join([
-        element.get('class', []) if isinstance(element.get('class'), list) else [element.get('class', '')],
-        [element.get('id', '')],
-        parent.get('class', []) if parent and isinstance(parent.get('class'), list) else [parent.get('class', '') if parent else ''],
-        [parent.get('id', '') if parent else '']
-    ])
-    element_attrs = " ".join(filter(None, element_attrs)).lower()
+    
+    # Safely extract and flatten all attributes
+    attrs_list = []
+    
+    # Get element classes
+    elem_classes = element.get('class', [])
+    if isinstance(elem_classes, list):
+        attrs_list.extend(elem_classes)
+    elif elem_classes:
+        attrs_list.append(str(elem_classes))
+    
+    # Get element id
+    elem_id = element.get('id', '')
+    if elem_id:
+        attrs_list.append(str(elem_id))
+    
+    # Get parent classes and id if parent exists
+    if parent:
+        parent_classes = parent.get('class', [])
+        if isinstance(parent_classes, list):
+            attrs_list.extend(parent_classes)
+        elif parent_classes:
+            attrs_list.append(str(parent_classes))
+        
+        parent_id = parent.get('id', '')
+        if parent_id:
+            attrs_list.append(str(parent_id))
+    
+    # Join all attributes into a single string
+    element_attrs = " ".join(str(attr) for attr in attrs_list if attr).lower()
     
     # Common indicators of non-main content
     bio_indicators = ['author', 'bio', 'about', 'profile', 'sidebar', 'footer',