import streamlit as st
from transformers import pipeline

# Configure Streamlit page
st.set_page_config(
    page_title="Customer Service Ticket Analyzer",
    page_icon="🎫",
    layout="centered"
)

# Initialize pipelines
@st.cache_resource
def load_models():
    sentiment_analyzer = pipeline(
        "sentiment-analysis",
        model="distilbert/distilbert-base-uncased-finetuned-sst-2-english"
    )
    department_classifier = pipeline(
        "zero-shot-classification",
        model="devronn/Finetuned_bge"
    )
    return sentiment_analyzer, department_classifier

# Load models
try:
    sentiment_analyzer, department_classifier = load_models()
except Exception as e:
    st.error(f"Error loading models: {str(e)}")
    st.stop()

# Define departments and their descriptions
departments = {
    "Customer Support": "Handles customer inquiries and product issues.",
    "Technical Support": "Provides technical assistance and troubleshooting.",
    "Billing and Sales": "Handles payment inquiries, sales questions, and general inquiries.",
    "Product Feedback": "Collects feedback about products and services.",
    "Account Management": "Manages customer accounts and retention."
}

# Define potential responses based on keywords
response_templates = {
    "billing": "For billing inquiries, please check your account or contact our billing department directly.",
    "technical": "For technical support, please provide detailed information about the issue.",
    "product": "For product inquiries, please specify the product name and your question.",
    "general": "Thank you for your inquiry! We will get back to you shortly.",
    "order": "For order-related questions, please provide your order number.",
    "refund": "For refund inquiries, please allow us to assist you with the process.",
}

# Title and description
st.title("Customer Service Ticket Analyzer")
st.markdown("Analyze customer tickets for sentiment and department routing")

# Main input form
customer_name = st.text_input("Customer Name")
ticket_subject = st.text_input("Ticket Subject")
ticket_content = st.text_area("Ticket Content", height=150)

if st.button("Analyze Ticket") and ticket_content.strip():
    try:
        with st.spinner('Analyzing...'):
            # Limit input length for faster processing
            limited_content = ticket_content[:500]  # Limit to 500 characters

            # Sentiment Analysis
            sentiment = sentiment_analyzer(limited_content)[0]

            # Department Classification
            department_result = department_classifier(
                limited_content,
                candidate_labels=list(departments.keys()),
                multi_label=False
            )

            # Keyword Extraction
            keyword_matches = []
            for keyword in response_templates.keys():
                if keyword in limited_content.lower():
                    keyword_matches.append(response_templates[keyword])

            # Display results
            st.markdown("### Analysis Results")

            # Create three columns for results
            col1, col2, col3 = st.columns(3)

            with col1:
                st.markdown("#### Sentiment")
                sentiment_color = "green" if sentiment['label'] == "POSITIVE" else "red"
                st.markdown(
                    f"<p style='color: {sentiment_color};'>{sentiment['label']}</p>",
                    unsafe_allow_html=True
                )
                st.write(f"Confidence: {sentiment['score']:.2%}")

            with col2:
                st.markdown("#### Department")
                suggested_dept = department_result['labels'][0]
                st.write(f"**Suggested:** {suggested_dept}")
                st.write(f"Confidence: {department_result['scores'][0]:.2%}")

            with col3:
                st.markdown("#### Priority")
                priority = "HIGH" if sentiment['label'] == "NEGATIVE" and sentiment['score'] > 0.8 else "MEDIUM"
                priority_color = "red" if priority == "HIGH" else "orange"
                st.markdown(
                    f"<p style='color: {priority_color};'>{priority}</p>",
                    unsafe_allow_html=True
                )

            # Ticket Summary
            st.markdown("### Ticket Details")
            st.write(f"**Customer:** {customer_name}")
            st.write(f"**Subject:** {ticket_subject}")
            st.write(f"**Content:** {ticket_content}")

            # Recommended Responses
            st.markdown("### Recommended Responses")
            if keyword_matches:
                for response in keyword_matches:
                    st.write(f"- {response}")
            else:
                st.write("No specific recommendations available.")

    except Exception as e:
        st.error(f"An error occurred during analysis: {str(e)}")

# Sidebar information
with st.sidebar:
    st.markdown("### About")
    st.write("""
    This tool analyzes customer service tickets by:
    - Determining sentiment
    - Suggesting appropriate department
    - Setting priority level
    - Providing confidence scores
    """)

    st.markdown("### Departments")
    for dept, desc in departments.items():
        st.write(f"**{dept}**")
        st.write(desc)
        st.write("---")