# streamlit_app.py

import streamlit as st
from Customer_Support.workflow import run_customer_support

# Set page configuration
st.set_page_config(
    page_title="Customer Support Assistant",
    page_icon="💬",
    layout="centered"
)

# Title and description
st.title("Customer Support Assistant")
st.markdown("Ask a question about technical issues, billing, or general inquiries.")

# API endpoint (change this if your FastAPI app is running on a different address)
#API_URL = "http://127.0.0.1:8000/process"
# "https://amna2024-customer-support-assistant.hf.space/process"


# Create a function to call the API
#def process_query(query):
#    try:
#        response = requests.post(
#            API_URL,
#            json={"query": query},
#            headers={"Content-Type": "application/json"},
#            timeout=10
#        )
#        return response.json()
#    except requests.exceptions.RequestException as e:
#        return {"error": f"API request failed: {str(e)}"}


# Create the main interface
query = st.text_area("What can I help you with today?", height=100)
submit_button = st.button("Submit")

# Display previous conversation if available
if "conversation" not in st.session_state:
    st.session_state.conversation = []

# Process input when the button is clicked
if submit_button and query:
    with st.spinner("Processing your request..."):
        # Call the API
        #result = process_query(query)
        result = run_customer_support(query)
        st.write("🔧 Raw response:", result)  # 👈 Add this line for debugging
        if "error" in result:
            st.error(result["error"])
        else:
            # Add the query and response to the conversation history
            st.session_state.conversation.append({
                "query": query,
                "category": result.get("category", "Unknown"),
                "sentiment": result.get("sentiment", "Unknown"),
                "response": result.get("response", "Sorry, I couldn't process your request.")
            })

    # Clear the input area after submission
    st.rerun()

# Display conversation history in reverse order (newest first)
if st.session_state.conversation:
    st.markdown("### Conversation History")

    for i, interaction in enumerate(reversed(st.session_state.conversation)):
        # Create an expandable section for each interaction
        with st.expander(f"Query: {interaction['query'][:50]}...", expanded=(i == 0)):
            st.markdown(f"**Category**: {interaction['category']}")
            st.markdown(f"**Sentiment**: {interaction['sentiment']}")

            # Display the response in a colored box based on sentiment
            sentiment_color = {
                "Positive": "rgba(0, 180, 0, 0.1)",
                "Neutral": "rgba(100, 100, 100, 0.1)",
                "Negative": "rgba(255, 0, 0, 0.1)"
            }.get(interaction['sentiment'], "rgba(100, 100, 100, 0.1)")

            st.markdown(
                f"""
                <div style="background-color: {sentiment_color}; padding: 15px; border-radius: 5px;">
                    {interaction['response']}
                </div>
                """,
                unsafe_allow_html=True
            )

# Add a sidebar with some helpful information
with st.sidebar:
    st.header("About")
    st.markdown("""
    This Customer Support Assistant uses AI to:
    - Categorize your query (Technical, Billing, General)
    - Analyze sentiment
    - Provide relevant support responses

    Try asking questions about:
    - Technical issues with products or services
    - Billing inquiries or problems
    - General information requests
    """)

    # Add a health check indicator
    #try:
    #    health_response = requests.get("http://localhost:8000/health")
    #    if health_response.status_code == 200:
    #        st.success("✅ API is online")
    #    else:
    #        st.error("❌ API is having issues")
    #except:
    #    st.error("❌ Could not connect to API")