app.py

import gradio as gr
import os 
import json 
import requests

#Streaming endpoint 
API_URL = "https://api.openai.com/v1/chat/completions" #os.getenv("API_URL") + "/generate_stream"

#Testing with my Open AI Key 
#OPENAI_API_KEY = os.getenv("OPENAI_API_KEY") 





#def chatbot_interface_restaurant_chinese(inputs, openai_api_key, chat_counter,chatbot=[], history=[]):









def predict1(inputs, openai_api_key, chat_counter,chatbot=[], history=[]):  #repetition_penalty, top_k

    payload = {
    "model": "gpt-3.5-turbo-16k",
    "messages": [{"role": "user", "content": f"{inputs}"}],
    "temperature" : 1.0,
    "top_p":1.0,
    "n" : 1,
    "stream": True,
    "presence_penalty":0,
    "frequency_penalty":0,
    }

    headers = {
    "Content-Type": "application/json",
    "Authorization": f"Bearer {openai_api_key}"
    }

    print(f"chat_counter - {chat_counter}")
    if chat_counter != 0 :
        messages=[{'role':'system', 'content':"""
                    你是订餐机器人，为瓦力快餐厅自动收集订单信息。\
                    注意如果客户来了，发送了空消息，请主动与客户说你好。
                    你要首先问候顾客，并主动告知客户今天的菜单，询问客户今天要什么食物。
                    然后等待用户回复收集订单信息。
                    收集完信息需确认顾客是否还需要添加其他内容。
                    最后需要询问是否自取或外送，如果是外送，你要询问地址。
                    最后告诉顾客订单总金额，并送上祝福。
                    你的回应应该以简短、助人和友好的风格呈现。
                    请确保明确所有选项、附加项和尺寸，以便从菜单中识别出该项唯一的内容。
                    请确保用中文回复。\
                    
                    如果客户询问的产品名称或者品类不在以下四个品类中，可以试着询问客户是以下哪个菜品 \
                    今日菜的品类是：
                    冷菜：
                    热菜：
                    饮料：
                    火锅：
                    """} ]
        for data in chatbot:
          temp1 = {}
          temp1["role"] = "user" 
          temp1["content"] = data[0] 
          temp2 = {}
          temp2["role"] = "assistant" 
          temp2["content"] = data[1]
          messages.append(temp1)
          messages.append(temp2)
        temp3 = {}
        temp3["role"] = "user" 
        temp3["content"] = inputs
        messages.append(temp3)
        #messages
        payload = {
        "model": "gpt-3.5-turbo-16k",
        "messages": messages, #[{"role": "user", "content": f"{inputs}"}],
        "temperature" : 1.0, #1.0,
        "top_p": 1.0, #1.0,
        "n" : 1,
        "stream": True,
        "presence_penalty":0,
        "frequency_penalty":0,
        }

    chat_counter+=1

    history.append(inputs)
    print(f"payload is - {payload}")
    # make a POST request to the API endpoint using the requests.post method, passing in stream=True
    response = requests.post(API_URL, headers=headers, json=payload, stream=True)
    #response = requests.post(API_URL, headers=headers, json=payload, stream=True)
    token_counter = 0 
    partial_words = "" 

    counter=0
    for chunk in response.iter_lines():
        #Skipping first chunk
        if counter == 0:
          counter+=1
          continue
        #counter+=1
        # check whether each line is non-empty
        if chunk.decode() :
          chunk = chunk.decode()
          # decode each line as response data is in bytes
          if len(chunk) > 12 and "content" in json.loads(chunk[6:])['choices'][0]['delta']:
              #if len(json.loads(chunk.decode()[6:])['choices'][0]["delta"]) == 0:
              #  break
              partial_words = partial_words + json.loads(chunk[6:])['choices'][0]["delta"]["content"]
              if token_counter == 0:
                history.append(" " + partial_words)
              else:
                history[-1] = partial_words
              chat = [(history[i], history[i + 1]) for i in range(0, len(history) - 1, 2) ]  # convert to tuples of list
              token_counter+=1
              yield chat, history, chat_counter  # resembles {chatbot: chat, state: history}  
                   

def reset_textbox():
    return gr.update(value='')

title = """<h1 align="center">🔥瓦力西餐外卖ChatBot🚀</h1>"""
title1 = """<h3 align="center">输入你的API，接着机器人就可以为你进行点餐服务了</h3>"""
description = """Language models can be conditioned to act like dialogue agents through a conversational prompt that typically takes the form:
```
User: <utterance>
Assistant: <utterance>
User: <utterance>
Assistant: <utterance>
...
```
In this app, you can explore the outputs of a gpt-3.5-turbo LLM.
"""
                
with gr.Blocks(css = """#col_container {width: 1000px; margin-left: auto; margin-right: auto;}
                #chatbot {height: 520px; overflow: auto;}""") as demo:
    gr.HTML(title)
    gr.HTML(title1)
    #gr.HTML('''<center><a href="https://huggingface.co/spaces/ysharma/ChatGPTwithAPI?duplicate=true"><img src="https://bit.ly/3gLdBN6" alt="Duplicate Space"></a>Duplicate the Space and run securely with your OpenAI API Key</center>''')
    with gr.Column(elem_id = "col_container"):
        openai_api_key = gr.Textbox(type='password', label="Enter your OpenAI API key here")
        chatbot = gr.Chatbot(elem_id='chatbot') #c
        inputs = gr.Textbox(placeholder= "Hi there!", label= "Type an input and press Enter") #t
        state = gr.State([]) #s
        b1 = gr.Button()
        chat_counter = gr.Number(value=0, visible=False, precision=0)

    #inputs.submit( predict, [inputs, top_p, temperature, openai_api_key, chat_counter, chatbot, state], [chatbot, state, chat_counter],)
    b1.click( predict, [inputs, openai_api_key, chat_counter,chatbot, state], [chatbot, state,chat_counter],)
    #b1.click(reset_textbox, [], [inputs])
    #inputs.submit(reset_textbox, [], [inputs])
                    
    #gr.Markdown(description)
    demo.queue().launch(debug=True)