from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration
import torch
import gradio as gr

device = "cuda" if torch.cuda.is_available() else "cpu"
model_name = "./blenderbot-1B-distill" 
tokenizer = BlenderbotTokenizer.from_pretrained(model_name)
model = BlenderbotForConditionalGeneration.from_pretrained(model_name)
model.to(device)

def get_reply(response, history = []):
    response = input("You: ")
    history.append(response)
    if response.endswith(("bye", "Bye", "bye.", "Bye.")):
        return "<div class='chatbot'>Chatbot restarted</div>", []
    if len(history) > 4: history = history[-4:]
    inputs = tokenizer("    ".join(history), return_tensors="pt")
    inputs.to(device)
    outputs = model.generate(**inputs)
    reply = tokenizer.decode(outputs[0][1:-1]).strip()
    history.append(reply)
    
    html = "<div class='chatbot'>"
    for m, msg in enumerate(history):
        cls = "user" if m%2 == 0 else "bot"
        html += "<div class='msg {}'> {}</div>".format(cls, msg)
    html += "</div>"
    
    return html, history

css = """
.chatbox {display:flex;flex-direction:column}
.msg {padding:4px;margin-bottom:4px;border-radius:4px;width:80%}
.msg.user {background-color:cornflowerblue;color:white}
.msg.bot {background-color:lightgray;align-self:self-end}
.footer {display:none !important}
"""

gr.Interface(fn=get_reply,
             theme="default",
             inputs=[gr.inputs.Textbox(placeholder="How are you?"), 
                    "state"],
             outputs=["html", "state"],
             enable_queue=True,
             css=css).launch(debug=True)