updated app.py

app.py

@@ -1,133 +1,71 @@
-import os
 import gradio as gr
-from langchain.embeddings import HuggingFaceEmbeddings
-from langchain.vectorstores import Chroma
-from langchain.llms import HuggingFacePipeline
-from langchain.chains import RetrievalQA
-from transformers import AutoConfig, AutoTokenizer, pipeline, AutoModelForCausalLM
-from langchain_community.document_loaders import DirectoryLoader
-import torch
-import re
-import transformers
-from urllib.parse import urlencode
-import spaces
-
-# Initialize embeddings and ChromaDB
-model_name = "sentence-transformers/all-mpnet-base-v2"
-device = "cuda" if torch.cuda.is_available() else "cpu"
-model_kwargs = {"device": device}
-embeddings = HuggingFaceEmbeddings(model_name=model_name, model_kwargs=model_kwargs)
-
-loader = DirectoryLoader('./example', glob="**/*.pdf", recursive=True, use_multithreading=True)
-docs = loader.load()
-vectordb = Chroma.from_documents(documents=docs, embedding=embeddings, persist_directory="companies_db")
-books_db = Chroma(persist_directory="./companies_db", embedding_function=embeddings)
-books_db_client = books_db.as_retriever()
-
-# Initialize the model and tokenizer
-model_name = "stabilityai/stablelm-zephyr-3b"
-model_config = transformers.AutoConfig.from_pretrained(model_name, max_new_tokens=1024)
-model = transformers.AutoModelForCausalLM.from_pretrained(
-    model_name,
-    trust_remote_code=True,
-    config=model_config,
-    device_map=device,
-)
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-
-query_pipeline = transformers.pipeline(
-    "text-generation",
-    model=model,
-    tokenizer=tokenizer,
-    return_full_text=True,
-    torch_dtype=torch.float16,
-    device_map=device,
-    do_sample=True,
-    temperature=0.7,
-    top_p=0.9,
-    top_k=50,
-    max_new_tokens=256
-)
-
-llm = HuggingFacePipeline(pipeline=query_pipeline)
-
-books_db_client_retriever = RetrievalQA.from_chain_type(
-    llm=llm,
-    chain_type="stuff",
-    retriever=books_db_client,
-    verbose=True
-)
-
-# Function to retrieve answer using the RAG system
-@spaces.GPU(duration=60)
-def test_rag(query):
-    books_retriever = books_db_client_retriever.run(query)
-    corrected_text_match = re.search(r"Helpful Answer:(.*)", books_retriever, re.DOTALL)
-    
-    if corrected_text_match:
-        corrected_text_books = corrected_text_match.group(1).strip()
-    else:
-        corrected_text_books = "No helpful answer found."
-    
-    return corrected_text_books
+import requests
+import webbrowser
+from http.server import BaseHTTPRequestHandler, HTTPServer
+import threading
 
 # OAuth Configuration
-TENANT_ID = os.getenv("TENANT_ID")
-CLIENT_ID = os.getenv("OAUTH_CLIENT_ID")
-CLIENT_SECRET = os.getenv("CLIENT_SECRET")
-REDIRECT_URI = os.getenv("SPACE_HOST")  # Make sure this is the correct redirect URI
-AUTH_URL = os.getenv("AUTH_URL")
-TOKEN_URL = os.getenv("TOKEN_URL")
-SCOPE = os.getenv("SCOPE")
+TENANT_ID = '2b093ced-2571-463f-bc3e-b4f8bcb427ee'
+CLIENT_ID = '2a7c884c-942d-49e2-9e5d-7a29d8a0d3e5'
+CLIENT_SECRET = 'EOF8Q~kKHCRgx8tnlLM-H8e93ifetxI6x7sU6bGW'
+REDIRECT_URI = 'https://sanjeevbora-chatbot.hf.space/' 
+AUTH_URL = f"https://login.microsoftonline.com/{TENANT_ID}/oauth2/v2.0/authorize"
+TOKEN_URL = f"https://login.microsoftonline.com/{TENANT_ID}/oauth2/v2.0/token"
+SCOPE = 'User.Read'
 access_token = None
 
-# OAuth Login Functionality
-def oauth_login():
-    params = {
-        'client_id': CLIENT_ID,
-        'response_type': 'code',
-        'redirect_uri': REDIRECT_URI,
-        'response_mode': 'query',
-        'scope': SCOPE,
-        'state': 'random_state_string'  # Optional: Use for security
-    }
-    login_url = f"{AUTH_URL}?{urlencode(params)}"
-    return login_url
-
-# Define the Gradio interface
-def chat(query, history=None):
-    if history is None:
-        history = []
-    if query:
-        answer = test_rag(query)
-        history.append((query, answer))
-    return history, ""  # Clear input after submission
-
-# Function to clear input text
-def clear_input():
-    return "",  # Return empty string to clear input field
-
-with gr.Blocks() as interface:
-    gr.Markdown("## RAG Chatbot")
-    gr.Markdown("Ask a question and get answers based on retrieved documents.")
-
-    # Sign-In Button
-    login_btn = gr.Button("Sign In with HF")
+class RequestHandler(BaseHTTPRequestHandler):
+    def do_GET(self):
+        global access_token
+        if self.path.startswith("/callback"):
+            # Capture the authorization code
+            code = self.path.split("code=")[1]
+            token_url = f"{AUTHORITY_URL}/token"
+            response = requests.post(token_url, data={
+                'client_id': CLIENT_ID,
+                'client_secret': CLIENT_SECRET,
+                'grant_type': 'authorization_code',
+                'code': code,
+                'redirect_uri': REDIRECT_URI
+            })
+            token_data = response.json()
+            access_token = token_data.get('access_token')
+            self.send_response(200)
+            self.end_headers()
+            self.wfile.write(b"Login successful! You can close this window.")
+            return
+
+        self.send_response(404)
+        self.end_headers()
+
+def start_http_server():
+    server_address = ('', 8080)
+    httpd = HTTPServer(server_address, RequestHandler)
+    httpd.serve_forever()
+
+def login():
+    auth_url = f"{AUTHORITY_URL}/authorize?client_id={CLIENT_ID}&response_type=code&redirect_uri={REDIRECT_URI}&scope={SCOPE}"
+    webbrowser.open(auth_url)
+
+def gradio_interface():
+    with gr.Blocks() as demo:
+        gr.Markdown("### Welcome to the App")
+        btn_login = gr.Button("Login with Microsoft")
+        output = gr.Textbox(label="Status")
+
+        def check_login():
+            if access_token:
+                return "You are logged in!"
+            return "You are not logged in."
+
+        btn_login.click(lambda: (login(), check_login()), None, output)
+
+    return demo
+
+if __name__ == "__main__":
+    # Start the HTTP server in a separate thread
+    threading.Thread(target=start_http_server, daemon=True).start()
     
-    # Redirect to OAuth login
-    login_btn.click(lambda: f"window.open('{oauth_login()}')", outputs=None)
-
-    # Hidden components initially
-    input_box = gr.Textbox(label="Enter your question", placeholder="Type your question here...", visible=False)
-    submit_btn = gr.Button("Submit", visible=False)
-    chat_history = gr.Chatbot(label="Chat History", visible=False)
-
-    # Show components after login
-    def show_components():
-        return gr.update(visible=True), gr.update(visible=True), gr.update(visible=True)
-
-    # After a successful login, show the input box and buttons
-    submit_btn.click(show_components, outputs=[input_box, submit_btn, chat_history])
-    submit_btn.click(chat, inputs=[input_box, chat_history], outputs=[chat_history, input_box])
-
-interface.launch()
+    # Launch Gradio app
+    demo = gradio_interface()
+    demo.launch()