Create fix_collection.py

app/fix_collection.py

@@ -0,0 +1,50 @@
+import asyncio
+from app.services.vector_store import vector_store
+from app.services.document_ingestion import initialize_knowledge_base_async
+
+async def fix_collection():
+    print("=" * 60)
+    print("🔧 FIXING QDRANT COLLECTION DIMENSIONS")
+    print("=" * 60)
+    
+    try:
+        # Step 1: Delete the old collection
+        print("\n1️⃣ Deleting old collection (1536 dimensions)...")
+        vector_store.delete_collection()
+        print("   ✓ Old collection deleted")
+    except Exception as e:
+        print(f"   ℹ Collection delete: {e}")
+    
+    try:
+        # Step 2: Create new collection
+        print("\n2️⃣ Creating new collection (768 dimensions for Gemini)...")
+        vector_store.create_collection(force_recreate=True)
+        print("   ✓ New collection created")
+    except Exception as e:
+        print(f"   ✗ Error creating collection: {e}")
+        return
+    
+    try:
+        # Step 3: Re-ingest all documents
+        print("\n3️⃣ Re-ingesting documents with Gemini embeddings...")
+        print("   (This may take a few minutes...)")
+        doc_ids = await initialize_knowledge_base_async()
+        print(f"   ✓ Successfully ingested {len(doc_ids)} documents")
+    except Exception as e:
+        print(f"   ✗ Error ingesting documents: {e}")
+        import traceback
+        traceback.print_exc()
+        return
+    
+    print("\n" + "=" * 60)
+    print("✅ MIGRATION COMPLETE!")
+    print("=" * 60)
+    print(f"✓ Collection: book_documents (768 dimensions)")
+    print(f"✓ Documents: {len(doc_ids)} chunks")
+    print(f"✓ Embedding model: Gemini text-embedding-004")
+    print(f"✓ LLM model: Gemini 2.0 Flash")
+    print("\nYou can now start your application! 🚀")
+    print("=" * 60)
+
+if __name__ == "__main__":
+    asyncio.run(fix_collection())