app.py

import os
import nltk 
import openai
import time
import gradio as gr
import tiktoken
from threading import Thread #线程  用于定时器

from assets.char_poses_base64 import ( #角色动作
    CHAR_IDLE_HTML, CHAR_THINKING_HTML, CHAR_TALKING_HTML)

from app_utils import (
    get_chat_history, initialize_knowledge_base, 
    text_to_speech_gen, logging, buzz_user)

global max_response_tokens
global token_limit
max_response_tokens = 500
token_limit= 15000


global FUNC_CALL #全局变量 用于判断角色动作
FUNC_CALL = 0

global BUZZ_TIMEOUT #全局变量 用于定时器
BUZZ_TIMEOUT = 60

global MESSAGES

GENERAL_RSPONSE_TRIGGERS = ["I don't understand the question.", "I don't know", "Hello, my name is", "mentioned in the context provided","I don't know the answer"]
MESSAGES = [{"role": "system", "content": "你现在是一个优秀的展览馆讲解员，你可以通过文字或语音与客户交流，你可以讲述上海老建筑和历史人物之间的关系。"}]
LOGGER = logging.getLogger('voice_agent') #日志
AUDIO_HTML = ''

# Uncomment If this is your first Run: 
nltk.download('averaged_perceptron_tagger')  #下载语料库
conv_model, voice_model = initialize_knowledge_base()  #初始化知识库

def num_tokens_from_messages(messages, model="gpt-3.5-turbo-16k"):
    encoding = tiktoken.encoding_for_model(model)
    num_tokens = 0
    for message in messages:
        num_tokens += 4  # every message follows <im_start>{role/name}\n{content}<im_end>\n
        for key, value in message.items():
            num_tokens += len(encoding.encode(value))
            if key == "name":  # if there's a name, the role is omitted
                num_tokens += -1  # role is always required and always 1 token
    num_tokens += 2  # every reply is primed with <im_start>assistant
    return num_tokens

def idle_timer():
    global BUZZ_TIMEOUT

    while True:
        time.sleep(BUZZ_TIMEOUT)
        buzz_user()

        if BUZZ_TIMEOUT == 80:
            time.sleep(BUZZ_TIMEOUT)
            BUZZ_TIMEOUT = 60


def update_img():
    global FUNC_CALL
    FUNC_CALL += 1

    if FUNC_CALL % 2== 0:
        return CHAR_TALKING_HTML
    else:
        return CHAR_THINKING_HTML


def get_response(history, audio_input):

    query_type = 'text'
    question =history[-1][0]
    conv_history_tokens = 0

    global BUZZ_TIMEOUT
    BUZZ_TIMEOUT = 80

    if not question:
        if audio_input:
            query_type = 'audio'
            os.rename(audio_input, audio_input + '.wav')
            audio_file = open(audio_input + '.wav', "rb")
            transcript = openai.Audio.transcribe("whisper-1", audio_file)
            question = transcript['text']
        else:
            return None, None

    LOGGER.info("\nquery_type: %s", query_type)
    LOGGER.info("query_text: %s", question)
    print('\nquery_type:', query_type)
    print('\nquery_text:', question)

    if question.lower().strip() == 'hi':
        question = 'hello'
    
    answer = conv_model.run(question)
    LOGGER.info("\ndocument_response: %s", answer)
    print('\ndocument_response:', answer)

    conv_history_tokens = num_tokens_from_messages(MESSAGES)
    print("conv_history_tokens: ", conv_history_tokens)
    print("MESSAGES", MESSAGES)
    while (conv_history_tokens + max_response_tokens >= token_limit):
        del MESSAGES[1] 
        conv_history_tokens = num_tokens_from_messages(MESSAGES)
    print("conv_history_tokens_ajust: ", conv_history_tokens)
    MESSAGES.append({"role": "user", "content": question})
    MESSAGES.append({"role": "assistant", "content": answer})            
    for trigger in GENERAL_RSPONSE_TRIGGERS:
        if trigger in answer:    
            MESSAGES.append({"role": "user", "content": question})
            chat = openai.ChatCompletion.create(
                    model="gpt-3.5-turbo-16k", 
                    messages=MESSAGES,
                    max_tokens=500,
                    temperature=0.7,
                    n=128,
                    stop="\n"
                    )
            answer = chat.choices[0].message.content
            MESSAGES.append({"role": "assistant", "content": answer})
            LOGGER.info("general_response: %s", answer)
            print('\ngeneral_response:', answer)
    AUDIO_HTML = text_to_speech_gen(answer)
    history[-1][1] = answer

    return history, AUDIO_HTML

# buzz_usr_proc = Thread(target=idle_timer)

with gr.Blocks(css = """#col_image{width:800px; height:800px; margin-left: auto; margin-right: auto;}""") as demo: 
    with gr.Row(scale=0.7):
        output_html = gr.HTML(label="Felix's Voice", value=AUDIO_HTML)
        output_html.visible = False
        image1= gr.Image("assets/NPCtest1.png").style(height=700) #elem_id = "col_image"
        #assistant_character = gr.HTML(label=None, value=CHAR_IDLE_HTML, show_label=False)
        with gr.Column(scale=0.3):
            chatbot = gr.Chatbot(label='Send a text or a voice input').style(height=285)
            with gr.Column():
                msg = gr.Textbox(placeholder='Write a chat & press Enter.', show_label=False).style(container=False)
                with gr.Column(scale=0.5):
                    audio_input = gr.Audio(source="microphone", type='filepath', show_label=False).style(container=False)
                    button = gr.Button(value="Send")

    msg.submit(get_chat_history, [msg, chatbot], [msg, chatbot]
                ).then(get_response, [chatbot, audio_input], [chatbot, output_html]
                )

    button.click(get_chat_history, [msg, chatbot], [msg, chatbot]
                ).then(get_response, [chatbot, audio_input], [chatbot, output_html]
                )
    
    # buzz_usr_proc.start()
    
demo.launch(debug=False, favicon_path='assets/favicon.png', show_api=False, share=False)