import google.generativeai as genai
import os
import PIL.Image
import gradio as gr
from gradio_multimodalchatbot import MultimodalChatbot
from gradio.data_classes import FileData

# For better security practices, retrieve sensitive information like API keys from environment variables.

# Fetch an environment variable.
GOOGLE_API_KEY = os.environ.get('GOOGLE_API_KEY')
genai.configure(api_key=GOOGLE_API_KEY)


# Initialize genai models
model = genai.GenerativeModel('gemini-pro')

def gemini(input, file, chatbot=[]):
    """
    Function to handle gemini model and gemini vision model interactions.
    Parameters:
    input (str): The input text.
    file (File): An optional file object for image processing.
    chatbot (list): A list to keep track of chatbot interactions.
    Returns:
    tuple: Updated chatbot interaction list, an empty string, and None.
    """

    messages = []
    print(chatbot)

    # Process previous chatbot messages if present
    if len(chatbot) != 0:
        for messages_dict in chatbot:
            user_text = messages_dict[0]['text']
            bot_text = messages_dict[1]['text']
            messages.extend([
                {'role': 'user', 'parts': [user_text]},
                {'role': 'model', 'parts': [bot_text]}
            ])
        messages.append({'role': 'user', 'parts': [input]})
    else:
        messages.append({'role': 'user', 'parts': [input]})

    try:
        response = model.generate_content(messages,stream=True)
        gemini_resp = response.text
        # Construct list of messages in the required format
        user_msg = {"text": input, "files": []}
        bot_msg = {"text": gemini_resp, "files": []}
        chatbot.append([user_msg, bot_msg])
   
    except Exception as e:
        # Handling exceptions and raising error to the modal
        print(f"An error occurred: {e}")
        raise gr.Error(e)

    return chatbot, "", None

# Define the Gradio Blocks interface
with gr.Blocks() as demo:
    # Add a centered header using HTML
    gr.HTML("<center><h1>Gemini Chat PRO API</h1></center>")

    # Initialize the MultimodalChatbot component
    multi = MultimodalChatbot(value=[], height=800)

    with gr.Row():
        # Textbox for user input with increased scale for better visibility
        tb = gr.Textbox(scale=4, placeholder='Input text and press Enter')

       

    # Define the behavior on text submission
    tb.submit(gemini, [tb, multi], [multi, tb])


# Launch the demo with a queue to handle multiple users
demo.queue().launch()