from groq import Groq
import gradio as gr
import os
import speech_recognition as sr

#  Securely retrieve the Groq API key from Hugging Face Secrets
groqkey = os.getenv("groqkey")

#  Ensure the API key exists
if not groqkey:
    raise ValueError("🚨 Error: 'groqkey' is missing! Make sure it is set in Hugging Face Secrets.")

#  Initialize the Groq client
client = Groq(api_key=groqkey)

#  Define chatbot response function
def get_chatbot_response(user_message, insurance_type, country, claim_number, image, language, conversation_history):
    """Fetch insurance-related responses from the AI."""

    system_message = (
        f"You are an insurance expert providing accurate information on {insurance_type} insurance in {country}. "
        f"Respond in {language}. If the user asks about a claim, provide a placeholder response."
        "Your responses should be clear and informative but must not be considered official legal or financial advice. "
        "Always use factual information about policies, coverage, and common practices."
    )

    # Maintain conversation history
    if conversation_history:
        if conversation_history[0]["role"] == "system":
            conversation_history[0]["content"] = system_message
        else:
            conversation_history.insert(0, {"role": "system", "content": system_message})
    else:
        conversation_history.append({"role": "system", "content": system_message})

    # Handle claim status check
    if claim_number:
        claim_response = f"Your claim with number {claim_number} is currently being processed. Please check back later."
        conversation_history.append({"role": "assistant", "content": claim_response})
        return conversation_history, claim_response

    conversation_history.append({"role": "user", "content": user_message})

    if image:
        conversation_history.append({"role": "user", "content": "[User uploaded an image]"})

    # API call to Groq
    completion = client.chat.completions.create(
        model="mixtral-8x7b",
        messages=conversation_history,
        temperature=0.3,
        top_p=0.95
    )

    response = ""
    for chunk in completion:
        response += chunk.choices[0].delta.content or ""

    conversation_history.append({"role": "assistant", "content": response})

    chat_display = [
        (msg["content"], conversation_history[i + 1]["content"])
        for i, msg in enumerate(conversation_history[:-1]) if msg["role"] == "user"
    ]

    return conversation_history, chat_display

# Convert voice input to text
def transcribe_audio(audio):
    recognizer = sr.Recognizer()
    with sr.AudioFile(audio) as source:
        audio_data = recognizer.record(source)
        try:
            text = recognizer.recognize_google(audio)
            return text
        except sr.UnknownValueError:
            return "Sorry, I could not understand the audio."
        except sr.RequestError:
            return "Error processing the audio."

#  Styling and Theme
theme = gr.themes.Base().set(
    body_background_fill="linear-gradient(to right, #001F3F, #0052CC)",  # Dark navy to royal blue
    button_primary_background_fill="linear-gradient(135deg, #FFD700, #FFAA00)",  # Golden gradient
    button_primary_text_color="black"
)

custom_css = """
/* Title text with gold gradient */
.title-text {
    background: linear-gradient(90deg, #FFD700, #FFAA00);
    -webkit-background-clip: text;
    background-clip: text;
    color: transparent;
    -webkit-text-fill-color: transparent;
    font-weight: 700;
    font-size: 42px;
    text-align: center;
    text-transform: uppercase;
    letter-spacing: 1px;
    text-shadow: 1px 1px 5px rgba(255, 215, 0, 0.3);
}

/* Body text - Gold on Dark Background */
body, .gradio-container {
    color: #FFD700 !important;
    font-family: 'Inter', sans-serif;
}

/* Chat messages - Minimalist design */
.chat-message {
    background: rgba(255, 255, 255, 0.05);
    border-radius: 12px;
    padding: 12px;
    backdrop-filter: blur(8px);
    color: white !important;
}

/* Buttons */
button.primary {
    background: linear-gradient(135deg, #FFD700, #FFAA00);
    border: none;
    color: black !important;
    font-weight: bold;
    text-transform: uppercase;
    border-radius: 20px;
    padding: 10px 20px;
    transition: all 0.3s ease-in-out;
}

button.primary:hover {
    transform: scale(1.1);
    box-shadow: 0 0 15px rgba(255, 215, 0, 0.7);
}
"""

#  Build Gradio Interface
with gr.Blocks(theme=theme, css=custom_css) as demo:
    gr.HTML("<h2 class='title-text'>🛡️ AI Insurance Chatbot</h2>")
    gr.Markdown("### Select your insurance type, country, and ask your question.")

    with gr.Row():
        insurance_type_input = gr.Dropdown(
            ["Health", "Car", "Home", "Travel", "Life", "Disability", "Business", "Pet"],
            label="Select Insurance Type",
            interactive=True
        )
        country_input = gr.Dropdown(
            ["USA", "Canada", "UK", "Germany", "France", "India", "Australia", "Other"],
            label="Select Country",
            interactive=True
        )
        language_input = gr.Dropdown(
            ["English", "Spanish", "French"],
            label="Select Language",
            interactive=True
        )

    custom_country_input = gr.Textbox(label="Enter Country (if not listed)", visible=False)

    claim_input = gr.Textbox(label="Enter Claim Number (Optional)", placeholder="e.g. 12345")

    conversation_state = gr.State([])
    
    chatbot = gr.Chatbot(label="Chat History", type="messages")  #  Fix: Set type to "messages"

    clear_button = gr.Button("Clear Chat History")
    clear_button.click(lambda: [], outputs=[conversation_state, chatbot])

    question_input = gr.Textbox(label="Ask your question...", placeholder="Describe your insurance question...", interactive=True)
    image_input = gr.File(label="Upload an image (optional)", type="filepath")  #  Fix: Change type to "filepath"
    voice_input = gr.Audio(label="Speak your question (optional)", type="filepath")

    submit_btn = gr.Button("Send")

    def process_voice(voice_file):
        transcribed_text = transcribe_audio(voice_file)
        return transcribed_text

    voice_input.change(process_voice, inputs=[voice_input], outputs=[question_input])

    submit_btn.click(
        get_chatbot_response,
        inputs=[question_input, insurance_type_input, country_input, claim_input, image_input, language_input, conversation_state],
        outputs=[conversation_state, chatbot]
    )

demo.launch()