Update app.py

app.py

@@ -1,14 +1,14 @@
 import os
 import gradio as gr
+import time
 from langchain.chains import RetrievalQA
 from langchain_community.vectorstores import Chroma
-from langchain_community.llms import OpenAI  # ✅ Fixed Import
-from langchain_huggingface import HuggingFaceEndpoint, HuggingFaceEmbeddings  # ✅ Fixed Import
-from langchain_community.document_loaders import PyPDFLoader  # ✅ Fixed Import
-import time
+from langchain_community.document_loaders import PyPDFLoader
+from langchain_huggingface import HuggingFaceEmbeddings  # ✅ Fixed Import
+from huggingface_hub import InferenceClient  # ✅ New method for querying Hugging Face LLM
 
-# Ensure required packages are up-to-date
-os.system("pip install -U huggingface_hub langchain_huggingface langchain_core langchain")
+# Install required dependencies (ensure latest versions)
+os.system("pip install -U huggingface_hub langchain_huggingface langchain_core langchain gradio")
 
 # Define paths for cybersecurity training PDFs
 PDF_FILES = [
@@ -20,7 +20,7 @@ PDF_FILES = [
 
 # Fetch Hugging Face API token securely from environment variables
 HUGGINGFACE_API_KEY = os.getenv("HUGGINGFACEHUB_API_TOKEN")
-if not HUGGINGFACE_API_KEY:
+if HUGGINGFACE_API_KEY is None:
     raise ValueError("❌ Hugging Face API token is missing! Set it in Hugging Face Spaces Secrets.")
 
 # Load PDFs into ChromaDB
@@ -28,37 +28,37 @@ def load_data():
     """Loads multiple PDFs and stores embeddings in ChromaDB"""
     all_docs = []
     for pdf in PDF_FILES:
-        if os.path.exists(pdf):  # Ensure the PDF exists in the Hugging Face Space
+        if os.path.exists(pdf):  # Ensure the PDF exists
             loader = PyPDFLoader(pdf)
             all_docs.extend(loader.load())
 
-    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")  # ✅ Updated import
+    # Use updated embedding model
+    embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
 
     return Chroma.from_documents(all_docs, embeddings)
 
 # Load the knowledge base
 vector_db = load_data()
 
-# Load LLM from Hugging Face securely
-llm = HuggingFaceEndpoint(
-    repo_id="google/flan-t5-large",
-    temperature=0.5,  # ✅ Ensure temperature is explicit
-    max_new_tokens=250,  # ✅ Ensure API limit is followed
-    huggingfacehub_api_token=HUGGINGFACE_API_KEY
+# Initialize Hugging Face Inference Client (new recommended method)
+client = InferenceClient(
+    "https://api-inference.huggingface.co/models/google/flan-t5-large",
+    token=HUGGINGFACE_API_KEY
 )
 
+# Function to interact with the Hugging Face model
+def query_llm(prompt):
+    """Send query to Hugging Face API and return response"""
+    response = client.chat_completion(messages=[{"role": "user", "content": prompt}])
+    return response["choices"][0]["message"]["content"]
+
 # Create Retrieval QA chain
-qa_chain = RetrievalQA.from_chain_type(llm=llm, retriever=vector_db.as_retriever())
+qa_chain = RetrievalQA.from_chain_type(llm=query_llm, retriever=vector_db.as_retriever())
 
 # Function to simulate futuristic typing effect
 def chatbot_response(question):
     """Handles chatbot queries with a typing effect"""
     response = qa_chain.invoke(question)  # ✅ Use `invoke` instead of deprecated `run`
-    
-    # Ensure response is a string to avoid errors
-    if not isinstance(response, str):
-        response = str(response)
-
     displayed_response = ""
     for char in response:
         displayed_response += char
@@ -86,5 +86,5 @@ iface = gr.Interface(
     live=True,  # Enables real-time updates for typing effect
 )
 
-# Launch chatbot
-iface.launch()
+# Launch chatbot with public link
+iface.launch(share=True)  # ✅ Now launches with a public link