import gradio as gr
import pandas as pd
from sentence_transformers import SentenceTransformer, util

# Load FAQ data from a CSV file containing questions and answers
try:
    faq_df = pd.read_csv("lic_faq.csv", encoding="utf-8")
except UnicodeDecodeError:
    faq_df = pd.read_csv("lic_faq.csv", encoding="ISO-8859-1")

# Initialize the SentenceTransformer model for semantic text similarity
model = SentenceTransformer('all-MiniLM-L6-v2')

# Convert FAQ questions into embeddings for similarity comparison
faq_embeddings = model.encode(faq_df['question'].tolist(), convert_to_tensor=True)

# Define policy suggestions to recommend based on keywords in the user query
policy_suggestions = {
    "term": "💡 You might consider LIC Tech Term Plan for pure protection at low cost.",
    "money back": "💡 LIC Money Back Policy is great for periodic returns along with insurance.",
    "endowment": "💡 LIC New Endowment Plan offers savings and insurance benefits together.",
    "ulip": "💡 LIC SIIP and Nivesh Plus are good ULIP options with market-linked returns.",
    "pension": "💡 LIC Jeevan Akshay and PM Vaya Vandana Yojana are best for pension seekers."
}

# Chatbot function to process user queries and generate responses
def chatbot(history, query):
    query_lower = query.lower().strip()

    # Handle common greetings with a friendly response
    if query_lower in ["hi", "hello", "hey", "good morning", "good evening"]:
        response = "👋 Hello! I’m your LIC Assistant. Ask me anything about policies, claims, onboarding, or commission."
    else:
        query_embedding = model.encode(query, convert_to_tensor=True)
        scores = util.pytorch_cos_sim(query_embedding, faq_embeddings)[0]
        best_score = float(scores.max())
        best_idx = int(scores.argmax())

        if best_score < 0.6:
            response = "🤖 I'm not confident I have the right answer for that. Please ask about LIC policies, claims, commissions, onboarding, or KYC."
        else:
            response = faq_df.iloc[best_idx]['answer']

            for keyword, suggestion in policy_suggestions.items():
                if keyword in query_lower:
                    response += f"\n\n{suggestion}"
                    break

    history.append((query, response))
    return history, history

# Custom CSS for a normal chatbot visual design
custom_css = """
/* General container styling */
.gradio-container {
    font-family: 'Arial', sans-serif;
    background: linear-gradient(135deg, #f5f7fa 0%, #c3cfe2 100%); /* Neutral gradient */
    min-height: 100vh;
    padding: 0;
    margin: 0;
    font-size: 14px; /* Base font size */
}

/* Header styling */
h1 {
    color: #2c3e50 !important; /* Dark blue for professionalism */
    font-size: 24px !important;
    text-align: center;
    margin: 15px 0;
    text-shadow: none;
}

p {
    color: #34495e !important; /* Slightly lighter blue */
    font-size: 14px !important;
    text-align: center;
    margin-bottom: 15px;
}

/* Chatbot container */
.chatbot {
    border-radius: 8px !important;
    box-shadow: 0 2px 10px rgba(0,0,0,0.1) !important;
    background: #ffffff !important;
    padding: 10px !important;
    max-width: 800px !important;
    margin: 0 auto !important;
    height: 70vh !important; /* Adjusted for normal view */
    overflow-y: auto !important;
    display: block; /* Reset to default flow */
}

/* Chat bubbles: Normal design */
.chatbot .bubble {
    border-radius: 8px !important;
    padding: 8px 12px !important;
    margin: 5px 0 !important;
    max-width: 70% !important;
    font-size: 14px !important; /* Normal text size */
    min-height: 30px !important; /* Controlled minimum height */
    min-width: 60px !important; /* Controlled minimum width */
    word-wrap: break-word !important;
    line-height: 1.4 !important; /* Readable line spacing */
    color: #2c3e50 !important; /* Consistent text color */
}

/* User bubble: Right-aligned, light blue */
.chatbot .bubble:nth-child(odd) {
    background: #3498db !important; /* Light blue for user */
    color: #ffffff !important; /* White text for contrast */
    margin-left: auto !important;
    align-self: flex-end !important;
}

/* Bot bubble: Left-aligned, light gray */
.chatbot .bubble:nth-child(even) {
    background: #ecf0f1 !important; /* Light gray for bot */
    margin-right: auto !important;
    align-self: flex-start !important;
}

/* Input area */
.gradio-row:last-child {
    width: 100% !important;
    max-width: 800px !important;
    margin: 10px auto !important;
    padding: 10px !important;
    display: flex !important;
    align-items: center !important;
    background: transparent !important;
}

/* Textbox styling */
input[type="text"] {
    border: 1px solid #bdc3c7 !important;
    border-radius: 4px !important;
    padding: 8px 12px !important;
    font-size: 14px !important;
    box-shadow: 0 1px 3px rgba(0,0,0,0.1) !important;
    flex-grow: 1 !important;
    margin-right: 10px !important;
    color: #2c3e50 !important;
}

input[type="text"]::placeholder {
    color: #7f8c8d !important; /* Gray placeholder */
    font-size: 14px !important;
}

input[type="text"]:focus {
    border-color: #3498db !important;
    outline: none !important;
}

/* Send button */
button {
    border-radius: 4px !important;
    background: #3498db !important; /* Matching user bubble */
    color: white !important;
    padding: 8px 20px !important;
    font-size: 14px !important;
    border: none !important;
    min-width: 80px !important;
    transition: background 0.3s ease !important;
    text-align: center !important;
}

button:hover {
    background: #2980b9 !important;
}

/* Clear button */
button[aria-label="Clear Chat"] {
    background: #bdc3c7 !important; /* Silver gray */
    padding: 6px 15px !important;
    font-size: 14px !important;
    min-width: 90px !important;
    margin-top: 10px !important;
    align-self: center !important;
    color: #ffffff !important;
}

button[aria-label="Clear Chat"]:hover {
    background: #95a5a6 !important;
}

/* Scrollbar styling */
.chatbot::-webkit-scrollbar {
    width: 6px;
}

.chatbot::-webkit-scrollbar-track {
    background: #f5f7fa;
}

.chatbot::-webkit-scrollbar-thumb {
    background: #3498db;
    border-radius: 3px;
}

.chatbot::-webkit-scrollbar-thumb:hover {
    background: #2980b9;
}

/* Mobile-specific adjustments */
@media (max-width: 768px) {
    .chatbot {
        max-width: 100% !important;
        height: 65vh !important; /* Adjusted for mobile */
        margin: 0 !important;
    }
    .gradio-row:last-child {
        max-width: 100% !important;
        padding: 5px !important;
    }
    input[type="text"] {
        padding: 6px 10px !important;
    }
    button {
        padding: 6px 15px !important;
    }
}
"""

with gr.Blocks(title="LIC Agent Chatbot", css=custom_css) as demo:
    gr.Markdown(
        "<h1>🧑‍💼 LIC Agent Assistant</h1>"
        "<p>Ask me anything about policies, claims, commissions, onboarding, and KYC.</p>"
    )

    chatbot_ui = gr.Chatbot(
        label="LIC Assistant",
        height=600,
        bubble_full_width=False,
        avatar_images=("🧑", "🤖"),
        show_copy_button=True
    )

    with gr.Row():
        msg = gr.Textbox(
            placeholder="Type your question here...",
            show_label=False,
            scale=8,
            elem_classes="chat-input"
        )
        send = gr.Button("Send", variant="primary", scale=2)

    clear = gr.Button("Clear Chat", variant="secondary")

    state = gr.State([])

    # Send message on Enter key and clear textbox
    def send_message(history, query):
        if query:
            history, _ = chatbot(history, query)
            return history, ""
        return history, query

    send.click(fn=send_message, inputs=[state, msg], outputs=[chatbot_ui, msg])
    msg.submit(fn=send_message, inputs=[state, msg], outputs=[chatbot_ui, msg])
    clear.click(lambda: ([], []), None, [chatbot_ui, state], queue=False)

demo.launch()