import gradio as gr
import requests, json

public_ip = '71.202.66.108'

model = 'llama3.1:latest'  # You can replace the model name if needed
context = [] 

ollama_serve = f"http://{public_ip}:11434/api/generate"

# Call Ollama API
def generate(prompt, context, top_k, top_p, temp):
    r = requests.post(ollama_serve,
                      json={
                          'model': model,
                          'prompt': prompt,
                          'context': context,
                          'options': {
                              'top_k': top_k,
                              'temperature': top_p,
                              'top_p': temp
                          }
                      },
                      stream=True)
    r.raise_for_status()

    response = ""  

    for line in r.iter_lines():
        body = json.loads(line)
        response_part = body.get('response', '')
        
        if 'error' in body:
            yield f"Error: {body['error']}"
            return

        # Append token to the growing response and yield the entire response so far
        if response_part:
            response += response_part
            yield response  # Yield the growing response incrementally

        if body.get('done', False):
            context = body.get('context', [])
            return  # End the generator once done

def chat(input, chat_history, top_k, top_p, temp):
    chat_history = chat_history or []
    global context
    
    # Initialize the user input as part of the chat history
    chat_history.append((input, ""))  # Add user input first
    response = ""  # Initialize empty response

    # Stream each part of the response as it's received
    response_stream = generate(input, context, top_k, top_p, temp)

    for response_part in response_stream:
        response = response_part  # Keep updating with the new part of the response
        # Update the latest assistant response (the second part of the tuple)
        chat_history[-1] = (input, response)
        yield chat_history, chat_history  # Yield the updated chat history


######################### Gradio Code ##########################
# background-image: url('https://cdn.shoplightspeed.com/shops/631940/files/45845092/800x800x3/apple-apple-macpro-trashcan-12-core-27ghz-64gb-1tb.jpg');
block = gr.Blocks(css="""
    .chatbox {
        background-image: url('https://cdn.shoplightspeed.com/shops/631940/files/45845092/800x800x3/apple-apple-macpro-trashcan-12-core-27ghz-64gb-1tb.jpg');
        background-size: contain;  /* Ensure the image fits the height */
        background-repeat: no-repeat;
        background-position: center;
        height: 100%;  /* Make the chatbox fill the available height */
    }
""")



with block:

    gr.Markdown("""<h1><center> Trashcan AI </center></h1>""")
    gr.Markdown("""<h3><center> LLama3.1 hosted on a 2013 "Trashcan" Mac Pro with ollama </center></h3>""")

    # Add a custom class 'chatbox' to apply the background image
    chatbot = gr.Chatbot(elem_classes="chatbox")
    message = gr.Textbox(placeholder="Type here")

    state = gr.State()
    with gr.Row():
        top_k = gr.Slider(0.0, 100.0, label="top_k", value=40, info="Reduces the probability of generating nonsense. A higher value (e.g. 100) will give more diverse answers, while a lower value (e.g. 10) will be more conservative. (Default: 40)")
        top_p = gr.Slider(0.0, 1.0, label="top_p", value=0.9, info="Works together with top-k. A higher value (e.g., 0.95) will lead to more diverse text, while a lower value (e.g., 0.5) will generate more focused and conservative text. (Default: 0.9)")
        temp = gr.Slider(0.0, 2.0, label="temperature", value=0.8, info="The temperature of the model. Increasing the temperature will make the model answer more creatively. (Default: 0.8)")

    submit = gr.Button("SEND")
    
    # Use .click() to trigger the response streaming
    submit.click(chat, inputs=[message, state, top_k, top_p, temp], outputs=[chatbot, state])

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