Update rag_server.py

rag_server.py

@@ -13,22 +13,22 @@ from transformers import AutoModel
 import streamlit as st
 
 # --- Konfiguration ---
-os.environ["HF_HOME"] = "/app/cache"  # Specify cache path
-HF_API_TOKEN = os.environ.get("HF_API_TOKEN")  # Read token from environment variable
+HF_API_TOKEN = os.environ.get("HF_API_TOKEN")  # Lesen Sie den Token aus der Umgebungsvariable
 MODEL_NAME = "dannyk97/mistral-screenplay-model"
+HF_CACHE_DIR = os.environ.get("HF_CACHE_DIR", "/app/cache") #Falls ein Fehler Auftritt, wird der Ordner auf /app/cache gesetzt.
 
 # --- Hilfsfunktionen ---
 
 def query_huggingface_inference_endpoints(prompt):
     """
-    Sends a request to the Hugging Face Inference API.
+    Stellt eine Anfrage an die Hugging Face Inference API.
     """
     try:
         client = InferenceClient(token=HF_API_TOKEN)
         result = client.text_generation(prompt, model=MODEL_NAME)
         return result
     except Exception as e:
-        return f"Error in query_huggingface_inference_endpoints: {e}"
+        return f"Fehler bei der Anfrage an Hugging Face API: {e}"
 
 # Function to download PDF from Google Drive
 def download_pdf_from_drive(drive_link):
@@ -56,16 +56,16 @@ def chunk_text(text, chunk_size=500, chunk_overlap=50):
     return text_splitter.split_text(text)
 
 # Function to create embeddings and store in FAISS
-def create_embeddings_and_store(chunks):
+def create_embeddings_and_store(chunks, cache_folder=HF_CACHE_DIR):
     try:
-        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
+        embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2", cache_folder=cache_folder)
         vector_db = FAISS.from_texts(chunks, embedding=embeddings)
         return vector_db
     except Exception as e:
-        print(f"Error creating embeddings: {e}")
-        print("Using dummy embeddings to proceed (functionality will be limited).")
-        # Fallback to a simpler embedding model (but this might not work well)
-        vector_db = FAISS.from_texts(["fallback text"], HuggingFaceEmbeddings(model_name="all-mpnet-base-v2")) #Ggf mit "" ersetzen, falls die Implementierung nicht passt.
+        print(f"❌ Fehler beim Erstellen der Embeddings: {e}")
+        print("Verwende Dummy Embeddings, um fortzufahren (Funktionen sind eingeschränkt).")
+        # Verwenden Sie eine einfachere Fallback Lösung
+        vector_db = FAISS.from_texts(["fallback text"], HuggingFaceEmbeddings(model_name="all-mpnet-base-v2", cache_folder=cache_folder))
         return vector_db
 
 # Function to query the vector database and interact with Hugging Face Inference API
@@ -77,7 +77,7 @@ def query_vector_db(query, vector_db):
     # Interact with the Text Generation API
     prompt = f"Nutze diesen Kontext um die Frage zu beantworten: {context}\nFrage: {query}"
     try:
-        output = query_huggingface_inference_endpoints(prompt) #Keine Modelangabe mehr
+        output = query_huggingface_inference_endpoints(prompt)
         return output
     except Exception as e:
         return f"FEHLER: {str(e)}"
@@ -116,7 +116,7 @@ for link in drive_links:
 
 if all_chunks:
     # Generate embeddings and store in FAISS
-    vector_db = create_embeddings_and_store(all_chunks)
+    vector_db = create_embeddings_and_store(all_chunks, cache_folder=HF_CACHE_DIR)
     st.write("Embeddings Generated and Stored Successfully!")
     
     # User query input