fix: Update Notion sync logic to handle existing records and avoid unique constraint errors

company-explorer/backend/scripts/sync_notion_industries.py

@@ -79,27 +79,20 @@ def sync_categories(token, session):
         
         notion_id = page["id"]
         name = extract_title(props.get("Name"))
-        # In the inspected DB, there was no 'key' or 'description' obvious, checking props again:
-        # Properties: Constrains, Product Category, Text, Product Categories, Name
-        # Wait, the inspection output was:
-        # - Constrains (rich_text)
-        # - Product Category (relation)
-        # - Text (rich_text)
-        # - Product Categories (relation)
-        # - Name (title)
-        
-        # It seems the schema might be slightly different than expected or I looked at the wrong DB.
-        # But 'Name' is there. I'll use Name as Key (lowercase) for now.
-        # And 'Text' as Description?
         
         description = extract_rich_text(props.get("Text"))
         key = name.lower().replace(" ", "_") if name else "unknown"
         
         if not name: continue
 
-        # Upsert
+        # Upsert Logic: Check ID -> Check Key -> Create
         cat = session.query(RoboticsCategory).filter(RoboticsCategory.notion_id == notion_id).first()
         if not cat:
+            cat = session.query(RoboticsCategory).filter(RoboticsCategory.key == key).first()
+            if cat:
+                logger.info(f"Linked existing category '{key}' to Notion ID {notion_id}")
+                cat.notion_id = notion_id
+            else:
                 cat = RoboticsCategory(notion_id=notion_id, key=key)
                 session.add(cat)
         
@@ -125,9 +118,15 @@ def sync_industries(token, session):
         name = extract_title(props.get("Industry"))
         if not name: continue
         
+        # Upsert Logic: Check ID -> Check Name -> Create
         industry = session.query(Industry).filter(Industry.notion_id == notion_id).first()
         if not industry:
-            industry = Industry(notion_id=notion_id)
+            industry = session.query(Industry).filter(Industry.name == name).first()
+            if industry:
+                logger.info(f"Linked existing industry '{name}' to Notion ID {notion_id}")
+                industry.notion_id = notion_id
+            else:
+                industry = Industry(notion_id=notion_id, name=name)
                 session.add(industry)
             
         # Map Fields