app.py

import gradio as gr
from huggingface_hub import InferenceClient
from pydantic import BaseModel
from typing import List, Dict

# Inisialisasi client model
client = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct")

# Model untuk data input API
class ChatMessage(BaseModel):
    role: str
    content: str

def chat_llama(chat_history: List[Dict[str, str]]):
    # Mengirim chat_history ke model dan mendapatkan respons
    chat_completion = client.chat_completion(
        messages=chat_history,
        max_tokens=500
    )
    # Menambahkan respons model ke chat_history
    chat_history.append({"role": "assistant", "content": chat_completion.choices[0].message.content})

    return chat_history

def chat_mem(message, chat_history):
    # Membuat chat_history_role untuk pengolahan model
    chat_history_role = [{"role": "system", "content": "You are a helpful assistant."}]
    
    # Menambahkan pesan dari chat_history ke chat_history_role
    if chat_history:
        for user_message, assistant_response in chat_history:
            chat_history_role.append({"role": "user", "content": user_message})
            chat_history_role.append({"role": "assistant", "content": assistant_response})
    
    # Menambahkan pesan pengguna terbaru
    chat_history_role.append({"role": "user", "content": message})
    
    # Mendapatkan respons dari model
    chat_completion = client.chat_completion(
        messages=chat_history_role,
        max_tokens=500
    )
    
    # Menambahkan respons model ke chat_history_role
    chat_history_role.append({"role": "assistant", "content": chat_completion.choices[0].message.content})
    
    # Format ulang chat_history
    modified = [entry["content"] for entry in chat_history_role]
    chat_history = [(modified[i*2], modified[i*2+1]) for i in range(len(modified)//2)]

    return "", chat_history  # Kembalikan pesan kosong dan chat_history yang diperbarui

def api_chat(chat_history: List[Dict[str, str]]):
    # Memanggil chat_llama untuk mendapatkan respons
    updated_history = chat_llama(chat_history)
    # Mengambil respons terakhir sebagai output
    return updated_history[-1] if updated_history else {}

# Mengatur antarmuka Gradio
with gr.Blocks() as demo:
    gr.Markdown("## Chat Demo")
    with gr.Row():
        with gr.Column():
            # Bagian Antarmuka Pengguna
            chatbot = gr.Chatbot()
            msg = gr.Textbox(placeholder="Type your message here...", interactive=True)
            with gr.Row():
                clear = gr.ClearButton([msg, chatbot], icon="https://img.icons8.com/?size=100&id=Xnx8cxDef16O&format=png&color=000000")
                send_btn = gr.Button("Send", variant='primary', icon="https://img.icons8.com/?size=100&id=g8ltXTwIfJ1n&format=png&color=000000")
            msg.submit(fn=chat_mem, inputs=[msg, chatbot], outputs=[msg, chatbot])
            send_btn.click(fn=chat_mem, inputs=[msg, chatbot], outputs=[msg, chatbot])

    gr.Markdown("## API Endpoint for Testing")
    gr.Markdown("### Send a POST request to `/api/chat` with the following JSON body:")
    gr.Markdown("```json\n[ { \"role\": \"user\", \"content\": \"Hello, how are you?\" }, { \"role\": \"assistant\", \"content\": \"I'm fine, thank you! How can I assist you today?\" }, { \"role\": \"user\", \"content\": \"Can you tell me a joke?\" } ]\n```")
    gr.Markdown("### API Response:")
    gr.Interface(fn=api_chat, inputs="json", outputs="json").launch()

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