Update rag_engine.py

rag_engine.py

@@ -16,12 +16,6 @@ import gc
 # Force model to CPU for stability
 os.environ["CUDA_VISIBLE_DEVICES"] = ""
 
-# GCS Paths
-metadata_file_gcs = "metadata/metadata.jsonl"
-embeddings_file_gcs = "processed/embeddings/all_embeddings.npy"
-faiss_index_file_gcs = "processed/indices/faiss_index.faiss"
-text_chunks_file_gcs = "processed/chunks/text_chunks.txt"
-
 # Local Paths
 local_embeddings_file = "all_embeddings.npy"
 local_faiss_index_file = "faiss_index.faiss"
@@ -32,9 +26,16 @@ local_metadata_file = "metadata.jsonl"
 def setup_gcp_client():
     try:
         credentials = setup_gcp_auth()
+        
+        # Get bucket name from secrets - required
+        try:
+            bucket_name_gcs = st.secrets["bucket_name_gcs"]
+        except KeyError:
+            print("❌ Error: GCS bucket name not found in secrets")
+            return None
+            
         storage_client = storage.Client(credentials=credentials)
-        bucket_name = "indian_spiritual-1"
-        bucket = storage_client.bucket(bucket_name)
+        bucket = storage_client.bucket(bucket_name_gcs)
         print("✅ GCP client initialized successfully")
         return bucket
     except Exception as e:
@@ -64,10 +65,17 @@ def load_model():
         # Force model to CPU
         device = torch.device("cpu")
         
+        # Get embedding model path from secrets
+        try:
+            embedding_model_path = st.secrets["embedding_model_path"]
+        except KeyError:
+            print("❌ Error: Embedding model path not found in secrets")
+            return None, None
+            
         # Load tokenizer and model
-        tokenizer = AutoTokenizer.from_pretrained("intfloat/e5-large-v2")
+        tokenizer = AutoTokenizer.from_pretrained(embedding_model_path)
         model = AutoModel.from_pretrained(
-            "intfloat/e5-large-v2",
+            embedding_model_path,
             torch_dtype=torch.float16
         )
         
@@ -121,6 +129,16 @@ def load_data_files():
         print("Failed to initialize required services")
         return None, None, None
     
+    # Get GCS paths from secrets - required
+    try:
+        metadata_file_gcs = st.secrets["metadata_file_gcs"]
+        embeddings_file_gcs = st.secrets["embeddings_file_gcs"]
+        faiss_index_file_gcs = st.secrets["faiss_index_file_gcs"]
+        text_chunks_file_gcs = st.secrets["text_chunks_file_gcs"]
+    except KeyError as e:
+        print(f"❌ Error: Required GCS path not found in secrets: {e}")
+        return None, None, None
+    
     # Download necessary files
     success = True
     success &= download_file_from_gcs(bucket, faiss_index_file_gcs, local_faiss_index_file)
@@ -318,9 +336,16 @@ def answer_with_llm(query, context=None, word_limit=100):
         {query}
         """
 
+        # Get LLM model from secrets
+        try:
+            llm_model = st.secrets["llm_model"]
+        except KeyError:
+            print("❌ Error: LLM model not found in secrets")
+            return "I apologize, but I'm unable to answer at the moment."
+            
         # Call OpenAI API
         response = openai.chat.completions.create(
-            model="gpt-3.5-turbo",
+            model=llm_model,
             messages=[
                 {"role": "system", "content": system_message},
                 {"role": "user", "content": user_message}