Spaces:
Sleeping
Sleeping
| # app.py | |
| import streamlit as st | |
| from shopping_agent import ShoppingAgent | |
| from dotenv import load_dotenv | |
| # Load environment variables | |
| load_dotenv() | |
| # --- Page Configuration --- | |
| st.set_page_config( | |
| page_title="🤖 Shopping Agent", | |
| page_icon="🛒", | |
| layout="wide" | |
| ) | |
| # --- State Management --- | |
| if "agent" not in st.session_state: | |
| st.session_state.agent = ShoppingAgent() | |
| if "messages" not in st.session_state: | |
| st.session_state.messages = [] | |
| # Add initial assistant message | |
| st.session_state.messages.append( | |
| {"role": "assistant", "content": "Hello! How can I help you with your shopping today?"} | |
| ) | |
| # --- UI Rendering --- | |
| st.title("🤖🛒 Your Personal Shopping Agent") | |
| st.caption("I can search the web, analyze products, and help you build your shopping list.") | |
| # Sidebar for Shopping List and Agent Internals | |
| with st.sidebar: | |
| st.header("🛒 Shopping List") | |
| # Use the definitive state from the agent instance | |
| shopping_list = st.session_state.agent.state.get("shopping_list", []) | |
| if not shopping_list: | |
| st.info("Your shopping list is empty.") | |
| else: | |
| for i, item in enumerate(shopping_list): | |
| st.markdown(f"**{i+1}. {item.get('name', 'N/A')}**") | |
| if url := item.get('url'): | |
| st.markdown(f" - **URL:** [{url.split('//')[-1]}]({url})") | |
| st.markdown(f" - **Details:** {item.get('details', 'No details provided.')}") | |
| st.divider() | |
| st.header("🕵️ Agent Thoughts") | |
| # This placeholder will be updated as the agent runs | |
| agent_thoughts_placeholder = st.container(height=350, border=True) | |
| # Display chat messages from history | |
| for message in st.session_state.messages: | |
| with st.chat_message(message["role"]): | |
| st.markdown(message["content"]) | |
| # --- Chat Input and Agent Interaction --- | |
| if prompt := st.chat_input("e.g., Find me the best budget wireless earbuds"): | |
| # Add user message to UI and history | |
| st.chat_message("user").markdown(prompt) | |
| st.session_state.messages.append({"role": "user", "content": prompt}) | |
| # Display agent's response | |
| with st.chat_message("assistant"): | |
| response_placeholder = st.empty() | |
| thought_log = [] | |
| with st.spinner("Processing..."): | |
| # The run_agent method is a generator that yields thought strings | |
| for thought in st.session_state.agent.run_agent(prompt): | |
| thought_log.append(thought) | |
| # Update the placeholder with the growing log of thoughts | |
| agent_thoughts_placeholder.markdown("\n\n---\n\n".join(thought_log)) | |
| # After the generator is exhausted, the agent's state is fully updated | |
| # Get the final response from the last message in the agent's state | |
| final_response = st.session_state.agent.state["messages"][-1].content | |
| response_placeholder.markdown(final_response) | |
| # Add final agent response to the UI history | |
| st.session_state.messages.append({"role": "assistant", "content": final_response}) | |
| # Rerun the script to update the sidebar with the new shopping list | |
| st.rerun() |