app.py

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

# Configure the API
genai.configure(api_key=os.environ["GEMINI_API_KEY"])

# Create the model
generation_config = {
    "temperature": 1,
    "top_p": 0.95,
    "top_k": 40,
    "max_output_tokens": 8192,
}

model = genai.GenerativeModel(
    model_name="gemini-1.5-pro-latest",
    generation_config=generation_config,
)

# Initialize the chat session
chat_session = model.start_chat(history=[])

def process_file(file):
    if file.type.startswith('image'):
        return Image.open(file.path)
    elif file.type.startswith('audio') or file.type.startswith('video'):
        return file.path
    else:
        return None

def respond(message, history):
    files = []
    for file in message.get('files', []):
        processed_file = process_file(file['file'])
        if processed_file:
            files.append(processed_file)

    prompt = message['text']
    
    if files:
        response = chat_session.send_message([prompt, *files])
    else:
        response = chat_session.send_message(prompt)
    
    return {"text": response.text, "files": []}

with gr.Blocks() as demo:
    gr.Markdown("# Gemini Multimodal Chatbot")
    gr.Markdown("Chat with the Gemini 1.5 Pro model. You can send text, images, audio, and video!")
    
    chatbot = MultimodalChatbot(
        height=600,
        bubble_full_width=False,
        avatar_images=(None, "https://lh3.googleusercontent.com/d/1pIo02xepBgqt9gMdFkJHSocJfH_A2dqj"),
        render_markdown=True
    )
    
    chatbot.chat(respond, fill_height=False)

if __name__ == "__main__":
    demo.launch()