Basic MCP Application

""" MCP Scientific Paper Analyzer - Gradio Frontend using latest version(Gradio 5.23.0) This application provides a frontend for Scientific Paper Analyzer using Gradio. It connects to the backend API to process users input and display responses from tools and LLM To run: 1. Backend should be running(i.e in seperate terminal cd fontend running 2. Run: python frontend/app.py """ import os import re import requests import gradio as gr from dotenv import load_dotenv # Load environment variables load_dotenv() # Configuration BACKEND_URL = os.getenv("BACKEND_URL", "http://localhost:8000") API_KEY = os.getenv("GEMINI_API_KEY", "") # Start state conversation_history = [] # same as the backend def clean_text(text): """Clean response text removing unwanted html tags""" if not text: return "" if not text.startswith('\n'): text = '\n' + text # Remove HTML tags text = re.sub(r'</?[a-zA-Z][^>]*>', '', text) return text # This checks the backend status def check_backend_status(): try: response = requests.get(f"{BACKEND_URL}/health", timeout=3) if response.status_code == 200: return "Backend is running ✓" return "Backend not connected ✗" except Exception: return "Backend not connected ✗" # Function to save and verify api key def save_api_key(api_key): global API_KEY if not api_key: return "Please enter an API key ✗" try: response = requests.post( f"{BACKEND_URL}/api/set-api-key", json={"api_key": api_key}, timeout=5 ) data = response.json() if data["success"]: API_KEY = api_key return "API key saved successfully ✓" else: return f"Invalid API key: {data['message']} ✗" except Exception as e: return f"Error connecting to backend: {str(e)} ✗" # Function to send message to backend and get response def send_message(message, chatbot, api_key_input): global conversation_history, API_KEY if not message or message.strip() == "": return chatbot # Use API key from input if provided, else use stored key (.env file) current_api_key = api_key_input if api_key_input else API_KEY # Check if backend is available backend_status = check_backend_status() if "not connected" in backend_status: return chatbot + [[message, f"Error: Backend is not connected. Make sure the backend server is running at {BACKEND_URL}."]] # Send request to backend try: request_data = { "message": message, "conversation_history": conversation_history, "api_key": current_api_key } response = requests.post( f"{BACKEND_URL}/api/chat", json=request_data, timeout=60 ) if response.status_code == 200: data = response.json() conversation_history = data["conversation_history"] # Clean response text response_text = clean_text(data["response"]) return chatbot + [[message, response_text]] else: error_message = f"Error: {response.text}" return chatbot + [[message, error_message]] except requests.exceptions.ConnectionError: connection_error = f"Connection error: Could not connect to {BACKEND_URL}. Is the backend server running?" return chatbot + [[message, connection_error]] except Exception as e: general_error = f"Error: {str(e)}" return chatbot + [[message, general_error]] # Function to handle example selection def use_example(example, chatbot): if example != "Select an example...": return example, chatbot return "", chatbot # Function to clear conversation def clear_conversation(): global conversation_history conversation_history = [] return None, [] # Gradio interface with gr.Blocks() as demo: with gr.Row(): gr.Markdown( """ # Scientific Paper Analyzer ### Analyze, search, and explore scientific papers using AI with Model Context Protocol (MCP)(demo has bugs) """ ) with gr.Row(): # Main chat column with greeting message with gr.Column(scale=7): chatbot = gr.Chatbot( value=[[None, "👋 Hello! I'm your research assistant. Ask me to search for papers, analyze citations, or get paper details!"]], height=500, show_copy_button=True, ) # Input box with gr.Group(): message = gr.Textbox( placeholder="Ask about scientific papers, request analysis, or explore citations(not working)...", lines=2, show_label=False ) # Buttons with gr.Row(): submit = gr.Button("Send", variant="primary") clear = gr.Button("Clear Conversation") # Few Example queries user can try with gr.Accordion("Example Queries", open=False): example_dropdown = gr.Dropdown( choices=[ "Select an example...", "Search for papers about machine learning", "Search for papers about climate change", "Search for papers about quantum computing", "Analyze the citation graph for paper P1", "Get information about paper P2", "Compare papers P1 and P2", "What are the latest research trends in artificial intelligence?" ], label="Try one of these examples", value="Select an example..." ) # Settings and info column with gr.Column(scale=3): # Settings section with gr.Group(): gr.Markdown("## Settings") backend_status_text = gr.Textbox( value=check_backend_status(), label="Backend Status", interactive=False ) api_key_input = gr.Textbox( value=API_KEY, label="Gemini API Key", placeholder="Enter your Gemini API key", type="password" ) save_key_button = gr.Button("Save API Key") # About section with gr.Group(): gr.Markdown("## About") gr.Markdown( """ This application is a demo for the use of Model Context Protocol (MCP) with Google's Gemini API to create a scientific paper analysis tool. As this is a demonstartion, there are few bugs. New features will be added soon. Thanks **Features:** - Search for scientific papers - Analyze citation graphs - Get paper details **How to use:** 1. Enter your Gemini API key 2. Ask questions about scientific papers 3. The AI will use MCP tools to find information """ ) # Event handlers submit.click(send_message, [message, chatbot, api_key_input], [chatbot]).then( lambda: "", None, [message] ) message.submit(send_message, [message, chatbot, api_key_input], [chatbot]).then( lambda: "", None, [message] ) clear.click(clear_conversation, None, [message, chatbot]) example_dropdown.change( use_example, [example_dropdown, chatbot], [message, chatbot] ) save_key_button.click( save_api_key, [api_key_input], [backend_status_text] ) # Launch the app if __name__ == "__main__": print("=" * 70) print(" MCP Scientific Paper Analyzer - Gradio Frontend") print("=" * 70) print(f"Backend URL: {BACKEND_URL}") print(f"API Key configured: {'Yes' if API_KEY else 'No'}") print("\nStarting Gradio interface...") # Launch demo.launch(server_name="0.0.0.0", server_port=8501, share=False)