Notion Fix Nested Properties (#2877)

backend/danswer/connectors/notion/connector.py

@@ -241,24 +241,29 @@ def _recurse_properties(inner_dict: dict[str, Any]) -> str | None:
                     )
 
             # TODO there may be more types to handle here
-            if "name" in inner_dict:
-                return inner_dict["name"]
-            if "content" in inner_dict:
-                return inner_dict["content"]
-            start = inner_dict.get("start")
-            end = inner_dict.get("end")
-            if start is not None:
-                if end is not None:
-                    return f"{start} - {end}"
-                return start
-            elif end is not None:
-                return f"Until {end}"
-
-            if "id" in inner_dict:
-                # This is not useful to index, it's a reference to another Notion object
-                # and this ID value in plaintext is useless outside of the Notion context
-                logger.debug("Skipping Notion object id field property")
-                return None
+            if isinstance(inner_dict, str):
+                # For some objects the innermost value could just be a string, not sure what causes this
+                return inner_dict
+
+            elif isinstance(inner_dict, dict):
+                if "name" in inner_dict:
+                    return inner_dict["name"]
+                if "content" in inner_dict:
+                    return inner_dict["content"]
+                start = inner_dict.get("start")
+                end = inner_dict.get("end")
+                if start is not None:
+                    if end is not None:
+                        return f"{start} - {end}"
+                    return start
+                elif end is not None:
+                    return f"Until {end}"
+
+                if "id" in inner_dict:
+                    # This is not useful to index, it's a reference to another Notion object
+                    # and this ID value in plaintext is useless outside of the Notion context
+                    logger.debug("Skipping Notion object id field property")
+                    return None
 
             logger.debug(f"Unreadable property from innermost prop: {inner_dict}")
             return None
@@ -268,7 +273,13 @@ def _recurse_properties(inner_dict: dict[str, Any]) -> str | None:
             if not prop:
                 continue
 
-            inner_value = _recurse_properties(prop)
+            try:
+                inner_value = _recurse_properties(prop)
+            except Exception as e:
+                # This is not a critical failure, these properties are not the actual contents of the page
+                # more similar to metadata
+                logger.warning(f"Error recursing properties for {prop_name}: {e}")
+                continue
             # Not a perfect way to format Notion database tables but there's no perfect representation
             # since this must be represented as plaintext
             if inner_value:

backend/scripts/dev_run_background_jobs.py

@@ -1,4 +1,3 @@
-import argparse
 import subprocess
 import threading
 
@@ -135,7 +134,4 @@ def run_jobs() -> None:
 
 
 if __name__ == "__main__":
-    parser = argparse.ArgumentParser(description="Run background jobs.")
-    args = parser.parse_args()
-
     run_jobs()