add project files

Taiyi-Stable-Diffusion-1B-Chinese-v0.1

@@ -0,0 +1 @@
+Subproject commit 373a5f530de018c086e5acff83519f40fe3df1ce

app.py

@@ -0,0 +1,140 @@
+import os; os.environ['no_proxy'] = '*' # 避免代理网络产生意外污染
+import gradio as gr
+from predict import predict
+from funtional_picture import infer_text2img
+from toolbox import format_io, find_free_port, get_conf
+import numpy as np
+
+# 建议您复制一个config_private.py放自己的秘密, 如API和代理网址, 避免不小心传github被别人看到
+proxies, WEB_PORT, LLM_MODEL, CONCURRENT_COUNT, AUTHENTICATION, CHATBOT_HEIGHT = \
+    get_conf('proxies', 'WEB_PORT', 'LLM_MODEL', 'CONCURRENT_COUNT', 'AUTHENTICATION', 'CHATBOT_HEIGHT')
+
+# 如果WEB_PORT是-1, 则随机选取WEB端口
+PORT = find_free_port() if WEB_PORT <= 0 else WEB_PORT
+if not AUTHENTICATION: AUTHENTICATION = None
+
+initial_prompt = "Serve me as a writing and programming assistant."
+title_html = "<h1 align=\"center\">展示你的机器学习模型</h1>"
+description =  """"""
+
+# 问询记录, python 版本建议3.9+（越新越好）
+import logging
+os.makedirs("work_log", exist_ok=True)
+try:logging.basicConfig(filename="work_log/chat_secrets.log", level=logging.INFO, encoding="utf-8")
+except:logging.basicConfig(filename="gpt_log/chat_secrets.log", level=logging.INFO)
+print("所有问询记录将自动保存在本地目录./gpt_log/chat_secrets.log, 请注意自我隐私保护哦！")
+
+# 一些普通功能模块
+from functional import get_functionals
+functional = get_functionals()
+
+
+
+# 处理markdown文本格式的转变
+gr.Chatbot.postprocess = format_io
+
+# 做一些外观色彩上的调整
+from theme import adjust_theme, advanced_css
+set_theme = adjust_theme()
+
+cancel_handles = []
+with gr.Blocks(theme=set_theme, analytics_enabled=False, css=advanced_css) as demo:
+    gr.HTML(title_html)
+    with gr.Tab("ChatGPT"):
+        with gr.Row().style(equal_height=True):
+            with gr.Column(scale=2):
+                chatbot = gr.Chatbot()
+                chatbot.style(height=CHATBOT_HEIGHT/2)
+                history = gr.State([])
+                with gr.Row():
+                    txt = gr.Textbox(show_label=False, placeholder="Input question here.").style(container=False)
+                with gr.Row():
+                    submitBtn = gr.Button("提交", variant="primary")
+                with gr.Row():
+                    resetBtn = gr.Button("重置", variant="secondary");
+                    resetBtn.style(size="sm")
+                    stopBtn = gr.Button("停止", variant="secondary");
+                    stopBtn.style(size="sm")
+
+            with gr.Column(scale=1):
+                with gr.Row():
+                    from check_proxy import check_proxy
+                    status = gr.Markdown(f"Tip: 按Enter提交, 按Shift+Enter换行。当前模型: {LLM_MODEL} \n {check_proxy(proxies)}")
+                with gr.Accordion("基础功能区", open=True) as area_basic_fn:
+                    with gr.Row():
+                        for k in functional:
+                            variant = functional[k]["Color"] if "Color" in functional[k] else "secondary"
+                            functional[k]["Button"] = gr.Button(k, variant=variant)
+                with gr.Accordion("展开SysPrompt & 交互界面布局 & Github地址", open=True):
+                    system_prompt = gr.Textbox(show_label=True, placeholder=f"System Prompt", label="System prompt", value=initial_prompt)
+                    top_p = gr.Slider(minimum=-0, maximum=1.0, value=1.0, step=0.01,interactive=True, label="Top-p (nucleus sampling)",)
+                    temperature = gr.Slider(minimum=-0, maximum=2.0, value=1.0, step=0.01, interactive=True, label="Temperature",)
+                    checkboxes = gr.CheckboxGroup(["基础功能区", "函数插件区"], value=["基础功能区", "函数插件区"], label="显示/隐藏功能区")
+                    gr.Markdown(description)
+    with gr.Tab("AI绘画"):
+        examples = [
+            ["铁马冰河入梦来, 梦幻, 插画"],
+            ["东临碣石, 以观沧海, 波涛汹涌, 插画"],
+            ["孤帆远影碧空尽，惟见长江天际流,油画"],
+            ["动漫化，帅气，插画"],
+            ["女孩背影, 日落, 唯美插画"],
+        ]
+        with gr.Row():
+            with gr.Column(scale=1, ):
+                image_out = gr.Image(label='输出(output)')
+            with gr.Column(scale=1, ):
+                image_in = gr.Image(source='upload', elem_id="image_upload", type="pil", label="参考图（非必须）(ref)")
+                prompt = gr.Textbox(label='提示词(prompt)')
+                submit_btn = gr.Button("生成图像(Generate)")
+                with gr.Row(scale=0.5):
+                    guide = gr.Slider(2, 15, value=7, step=0.1, label='文本引导强度(guidance scale)')
+                    steps = gr.Slider(10, 30, value=20, step=1, label='迭代次数(inference steps)')
+                    width = gr.Slider(384, 640, value=512, step=64, label='宽度(width)')
+                    height = gr.Slider(384, 640, value=512, step=64, label='高度(height)')
+                    strength = gr.Slider(0, 1.0, value=0.8, step=0.02, label='参考图改变程度(strength)')
+                    ex = gr.Examples(examples, fn=infer_text2img, inputs=[prompt, guide, steps, width, height],
+                                     outputs=image_out)
+
+            submit_btn.click(fn=infer_text2img, inputs=[prompt, guide, steps, width, height, image_in, strength],
+                             outputs=image_out)
+
+    # demo.queue(concurrency_count=1, max_size=8).launch()
+
+
+    # 功能区显示开关与功能区的互动
+    def fn_area_visibility(a):
+        ret = {}
+        ret.update({area_basic_fn: gr.update(visible=("基础功能区" in a))})
+        return ret
+
+    checkboxes.select(fn_area_visibility, [checkboxes], [area_basic_fn])
+    # 整理反复出现的控件句柄组合
+    input_combo = [txt, top_p, temperature, chatbot, history, system_prompt]
+    output_combo = [chatbot, history, status]
+    predict_args = dict(fn=predict, inputs=input_combo, outputs=output_combo)
+    empty_txt_args = dict(fn=lambda: "", inputs=[], outputs=[txt]) # 用于在提交后清空输入栏
+    # 提交按钮、重置按钮
+    cancel_handles.append(txt.submit(**predict_args)) #; txt.submit(**empty_txt_args) 在提交后清空输入栏
+    cancel_handles.append(submitBtn.click(**predict_args)) #; submitBtn.click(**empty_txt_args) 在提交后清空输入栏
+    resetBtn.click(lambda: ([], [], "已重置"), None, output_combo)
+    # 基础功能区的回调函数注册
+    for k in functional:
+        click_handle = functional[k]["Button"].click(predict, [*input_combo, gr.State(True), gr.State(k)], output_combo)
+        cancel_handles.append(click_handle)
+    cancel_handles.append(click_handle)
+    # 终止按钮的回调函数注册
+    stopBtn.click(fn=None, inputs=None, outputs=None, cancels=cancel_handles)
+
+
+
+# gradio的inbrowser触发不太稳定，回滚代码到原始的浏览器打开函数
+def auto_opentab_delay():
+    import threading, webbrowser, time
+    print(f"如果浏览器没有自动打开，请复制并转到以下URL: http://localhost:{PORT}")
+    def open():
+        time.sleep(2)
+        webbrowser.open_new_tab(f"http://localhost:{PORT}")
+    threading.Thread(target=open, name="open-browser", daemon=True).start()
+auto_opentab_delay()
+demo.title = "展示你的机器学习模型"
+demo.queue(concurrency_count=CONCURRENT_COUNT).launch(server_name="0.0.0.0", share=True, server_port=PORT, auth=AUTHENTICATION)

check_proxy.py

@@ -0,0 +1,27 @@
+
+def check_proxy(proxies):
+    import requests
+    proxies_https = proxies['https'] if proxies is not None else '无'
+    try:
+        response = requests.get("https://ipapi.co/json/", proxies=proxies, timeout=4)
+        data = response.json()
+        print(f'查询代理的地理位置，返回的结果是{data}')
+        if 'country_name' in data:
+            country = data['country_name']
+            result = f"代理配置 {proxies_https}, 代理所在地：{country}"
+        elif 'error' in data:
+            result = f"代理配置 {proxies_https}, 代理所在地：未知，IP查询频率受限"
+        print(result)
+        return result
+    except:
+        result = f"代理配置 {proxies_https}, 代理所在地查询超时，代理可能无效"
+        print(result)
+        return result
+
+
+if __name__ == '__main__':
+    import os; os.environ['no_proxy'] = '*' # 避免代理网络产生意外污染
+    from toolbox import get_conf
+    proxies, = get_conf('proxies')
+    check_proxy(proxies)
+    

config.py

@@ -0,0 +1,46 @@
+# [step 1]>> 例如： API_KEY = "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" （此key无效）
+API_KEY = "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"
+
+# [step 2]>> 改为True应用代理，如果直接在海外服务器部署，此处不修改
+USE_PROXY = True
+if USE_PROXY:
+    # 填写格式是 [协议]://  [地址] :[端口]，填写之前不要忘记把USE_PROXY改成True，如果直接在海外服务器部署，此处不修改
+    # 例如    "socks5h://localhost:11284"
+    # [协议] 常见协议无非socks5h/http; 例如 v2**y 和 ss* 的默认本地协议是socks5h; 而cl**h 的默认本地协议是http
+    # [地址] 懂的都懂，不懂就填localhost或者127.0.0.1肯定错不了（localhost意思是代理软件安装在本机上）
+    # [端口] 在代理软件的设置里找。虽然不同的代理软件界面不一样，但端口号都应该在最显眼的位置上
+
+    # 代理网络的地址，打开你的科学上网软件查看代理的协议(socks5/http)、地址(localhost)和端口(11284)
+    proxies = { 
+        #          [协议]://  [地址]  :[端口]
+        "http":  "socks5h://localhost:7890",
+        "https": "socks5h://localhost:7890",
+    }
+else:
+    proxies = None
+
+
+# [step 3]>> 以下配置可以优化体验，但大部分场合下并不需要修改
+# 对话窗的高度
+CHATBOT_HEIGHT = 1115
+
+# 发送请求到OpenAI后，等待多久判定为超时
+TIMEOUT_SECONDS = 25
+
+# 网页的端口, -1代表随机端口
+WEB_PORT = -1
+
+# 如果OpenAI不响应（网络卡顿、代理失败、KEY失效），重试的次数限制
+MAX_RETRY = 2
+
+# OpenAI模型选择是（gpt4现在只对申请成功的人开放）
+LLM_MODEL = "gpt-3.5-turbo"
+
+# OpenAI的API_URL
+API_URL = "https://api.openai.com/v1/chat/completions"
+
+# 设置并行使用的线程数
+CONCURRENT_COUNT = 100
+
+# 设置用户名和密码（相关功能不稳定，与gradio版本和网络都相关，如果本地使用不建议加这个）
+AUTHENTICATION = [] # [("username", "password"), ("username2", "password2"), ...]

functional.py

@@ -0,0 +1,70 @@
+# 'primary' 颜色对应 theme.py 中的 primary_hue
+# 'secondary' 颜色对应 theme.py 中的 neutral_hue
+# 'stop' 颜色对应 theme.py 中的 color_er
+# 默认按钮颜色是 secondary
+from toolbox import clear_line_break
+
+def get_functionals():
+    return {
+        "英语学术润色": {
+            # 前言
+            "Prefix":   r"Below is a paragraph from an academic paper. Polish the writing to meet the academic style, " +
+                        r"improve the spelling, grammar, clarity, concision and overall readability. When necessary, rewrite the whole sentence. " +
+                        r"Furthermore, list all modification and explain the reasons to do so in markdown table." + "\n\n",
+            # 后语 
+            "Suffix":   r"",
+            "Color":    r"secondary",    # 按钮颜色
+        },
+        "中文学术润色": {
+            "Prefix":   r"作为一名中文学术论文写作改进助理，你的任务是改进所提供文本的拼写、语法、清晰、简洁和整体可读性，" + 
+                        r"同时分解长句，减少重复，并提供改进建议。请只提供文本的更正版本，避免包括解释。请编辑以下文本" + "\n\n",
+            "Suffix":   r"",
+        },
+        "查找语法错误": {
+            "Prefix":   r"Can you help me ensure that the grammar and the spelling is correct? " +
+                        r"Do not try to polish the text, if no mistake is found, tell me that this paragraph is good." +
+                        r"If you find grammar or spelling mistakes, please list mistakes you find in a two-column markdown table, " +
+                        r"put the original text the first column, " +
+                        r"put the corrected text in the second column and highlight the key words you fixed.""\n"
+                        r"Example:""\n"
+                        r"Paragraph: How is you? Do you knows what is it?""\n"
+                        r"| Original sentence | Corrected sentence |""\n"
+                        r"| :--- | :--- |""\n"
+                        r"| How **is** you? | How **are** you? |""\n"
+                        r"| Do you **knows** what **is** **it**? | Do you **know** what **it** **is** ? |""\n"
+                        r"Below is a paragraph from an academic paper. "
+                        r"You need to report all grammar and spelling mistakes as the example before."
+                        + "\n\n",
+            "Suffix":   r"",
+            "PreProcess": clear_line_break,    # 预处理：清除换行符
+        },
+        "中译英": {
+            "Prefix":   r"Please translate following sentence to English:" + "\n\n",
+            "Suffix":   r"",
+        },
+        "学术中英互译": {
+            "Prefix":   r"I want you to act as a scientific English-Chinese translator, " +
+                        r"I will provide you with some paragraphs in one language " +
+                        r"and your task is to accurately and academically translate the paragraphs only into the other language. " +
+                        r"Do not repeat the original provided paragraphs after translation. " +
+                        r"You should use artificial intelligence tools, " +
+                        r"such as natural language processing, and rhetorical knowledge " +
+                        r"and experience about effective writing techniques to reply. " +
+                        r"I'll give you my paragraphs as follows, tell me what language it is written in, and then translate:" + "\n\n",
+            "Suffix": "",
+            "Color": "secondary",
+        },
+        "英译中": {
+            "Prefix":   r"请翻译成中文：" + "\n\n",
+            "Suffix":   r"",
+        },
+        "找图片": {
+            "Prefix":   r"我需要你找一张网络图片。使用Unsplash API(https://source.unsplash.com/960x640/?<英语关键词>)获取图片URL，" +
+                        r"然后请使用Markdown格式封装，并且不要有反斜线，不要用代码块。现在，请按以下描述给我发送图片：" + "\n\n",
+            "Suffix":   r"",
+        },
+        "解释代码": {
+            "Prefix":   r"请解释以下代码：" + "\n```\n",
+            "Suffix":   "\n```\n",
+        },
+    }

funtional_picture.py

@@ -0,0 +1,22 @@
+
+#GAN模型
+from PIL import Image
+import torch
+from diffusers import (
+    StableDiffusionPipeline,
+    StableDiffusionImg2ImgPipeline,
+    StableDiffusionInpaintPipeline,
+)
+device="cuda"
+model_id = "./Taiyi-Stable-Diffusion-1B-Chinese-v0.1"
+
+pipe_text2img = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16).to(device)
+pipe_img2img = StableDiffusionImg2ImgPipeline(**pipe_text2img.components).to(device)
+def infer_text2img(prompt, guide, steps, width, height, image_in, strength):
+    if image_in is not None:
+        init_image = image_in.convert("RGB").resize((width, height))
+        output = pipe_img2img(prompt, image=init_image, strength=strength, guidance_scale=guide, num_inference_steps=steps)
+    else:
+        output = pipe_text2img(prompt, width=width, height=height, guidance_scale=guide, num_inference_steps=steps,)
+    image = output.images[0]
+    return image

predict.py

@@ -0,0 +1,177 @@
+# 借鉴了 https://github.com/GaiZhenbiao/ChuanhuChatGPT 项目
+
+"""
+    该文件中主要包含三个函数
+
+    不具备多线程能力的函数：
+    1. predict: 正常对话时使用，具备完备的交互功能，不可多线程
+
+    具备多线程调用能力的函数
+    2. predict_no_ui：高级实验性功能模块调用，不会实时显示在界面上，参数简单，可以多线程并行，方便实现复杂的功能逻辑
+    3. predict_no_ui_long_connection：在实验过程中发现调用predict_no_ui处理长文档时，和openai的连接容易断掉，这个函数用stream的方式解决这个问题，同样支持多线程
+"""
+
+import json
+import gradio as gr
+import logging
+import traceback
+import requests
+import importlib
+
+# config_private.py放自己的秘密如API和代理网址
+# 读取时首先看是否存在私密的config_private配置文件（不受git管控），如果有，则覆盖原config文件
+from toolbox import get_conf
+proxies, API_URL, API_KEY, TIMEOUT_SECONDS, MAX_RETRY, LLM_MODEL = \
+    get_conf('proxies', 'API_URL', 'API_KEY', 'TIMEOUT_SECONDS', 'MAX_RETRY', 'LLM_MODEL')
+
+timeout_bot_msg = '[Local Message] Request timeout. Network error. Please check proxy settings in config.py.' + \
+                  '网络错误，检查代理服务器是否可用，以及代理设置的格式是否正确，格式须是[协议]://[地址]:[端口]，缺一不可。'
+
+def get_full_error(chunk, stream_response):
+    """
+        获取完整的从Openai返回的报错
+    """
+    while True:
+        try:
+            chunk += next(stream_response)
+        except:
+            break
+    return chunk
+
+#
+
+
+def predict(inputs, top_p, temperature, chatbot=[], history=[], system_prompt='', 
+            stream = True, additional_fn=None):
+    """
+        发送至chatGPT，流式获取输出。
+        用于基础的对话功能。
+        inputs 是本次问询的输入
+        top_p, temperature是chatGPT的内部调优参数
+        history 是之前的对话列表（注意无论是inputs还是history，内容太长了都会触发token数量溢出的错误）
+        chatbot 为WebUI中显示的对话列表，修改它，然后yeild出去，可以直接修改对话界面内容
+        additional_fn代表点击的哪个按钮，按钮见functional.py
+    """
+    if additional_fn is not None:
+        import functional
+        importlib.reload(functional)    # 热更新prompt
+        functional = functional.get_functionals()
+        if "PreProcess" in functional[additional_fn]: inputs = functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
+        inputs = functional[additional_fn]["Prefix"] + inputs + functional[additional_fn]["Suffix"]
+
+    if stream:
+        raw_input = inputs
+        logging.info(f'[raw_input] {raw_input}')
+        chatbot.append((inputs, ""))
+        yield chatbot, history, "等待响应"
+
+    headers, payload = generate_payload(inputs, top_p, temperature, history, system_prompt, stream)
+    history.append(inputs); history.append(" ")
+
+    retry = 0
+    while True:
+        try:
+            # make a POST request to the API endpoint, stream=True
+            response = requests.post(API_URL, headers=headers, proxies=proxies,
+                                    json=payload, stream=True, timeout=TIMEOUT_SECONDS);break
+        except:
+            retry += 1
+            chatbot[-1] = ((chatbot[-1][0], timeout_bot_msg))
+            retry_msg = f"，正在重试 ({retry}/{MAX_RETRY}) ……" if MAX_RETRY > 0 else ""
+            yield chatbot, history, "请求超时"+retry_msg
+            if retry > MAX_RETRY: raise TimeoutError
+
+    gpt_replying_buffer = ""
+    
+    is_head_of_the_stream = True
+    if stream:
+        stream_response =  response.iter_lines()
+        while True:
+            chunk = next(stream_response)
+            # print(chunk.decode()[6:])
+            if is_head_of_the_stream:
+                # 数据流的第一帧不携带content
+                is_head_of_the_stream = False; continue
+            
+            if chunk:
+                try:
+                    if len(json.loads(chunk.decode()[6:])['choices'][0]["delta"]) == 0:
+                        # 判定为数据流的结束，gpt_replying_buffer也写完了
+                        logging.info(f'[response] {gpt_replying_buffer}')
+                        break
+                    # 处理数据流的主体
+                    chunkjson = json.loads(chunk.decode()[6:])
+                    status_text = f"finish_reason: {chunkjson['choices'][0]['finish_reason']}"
+                    # 如果这里抛出异常，一般是文本过长，详情见get_full_error的输出
+                    gpt_replying_buffer = gpt_replying_buffer + json.loads(chunk.decode()[6:])['choices'][0]["delta"]["content"]
+                    history[-1] = gpt_replying_buffer
+                    chatbot[-1] = (history[-2], history[-1])
+                    yield chatbot, history, status_text
+
+                except Exception as e:
+                    traceback.print_exc()
+                    yield chatbot, history, "Json解析不合常规"
+                    chunk = get_full_error(chunk, stream_response)
+                    error_msg = chunk.decode()
+                    if "reduce the length" in error_msg:
+                        chatbot[-1] = (chatbot[-1][0], "[Local Message] Input (or history) is too long, please reduce input or clear history by refreshing this page.")
+                        history = []    # 清除历史
+                    elif "Incorrect API key" in error_msg:
+                        chatbot[-1] = (chatbot[-1][0], "[Local Message] Incorrect API key provided.")
+                    elif "exceeded your current quota" in error_msg:
+                        chatbot[-1] = (chatbot[-1][0], "[Local Message] You exceeded your current quota. OpenAI以账户额度不足为由，拒绝服务.")
+                    else:
+                        from toolbox import regular_txt_to_markdown
+                        tb_str = '```\n' + traceback.format_exc() + '```'
+                        chatbot[-1] = (chatbot[-1][0], f"[Local Message] 异常 \n\n{tb_str} \n\n{regular_txt_to_markdown(chunk.decode()[4:])}")
+                    yield chatbot, history, "Json异常" + error_msg
+                    return
+
+def generate_payload(inputs, top_p, temperature, history, system_prompt, stream):
+    """
+        整合所有信息，选择LLM模型，生成http请求，为发送请求做准备
+    """
+    headers = {
+        "Content-Type": "application/json",
+        "Authorization": f"Bearer {API_KEY}"
+    }
+
+    conversation_cnt = len(history) // 2
+
+    messages = [{"role": "system", "content": system_prompt}]
+    if conversation_cnt:
+        for index in range(0, 2*conversation_cnt, 2):
+            what_i_have_asked = {}
+            what_i_have_asked["role"] = "user"
+            what_i_have_asked["content"] = history[index]
+            what_gpt_answer = {}
+            what_gpt_answer["role"] = "assistant"
+            what_gpt_answer["content"] = history[index+1]
+            if what_i_have_asked["content"] != "":
+                if what_gpt_answer["content"] == "": continue
+                if what_gpt_answer["content"] == timeout_bot_msg: continue
+                messages.append(what_i_have_asked)
+                messages.append(what_gpt_answer)
+            else:
+                messages[-1]['content'] = what_gpt_answer['content']
+
+    what_i_ask_now = {}
+    what_i_ask_now["role"] = "user"
+    what_i_ask_now["content"] = inputs
+    messages.append(what_i_ask_now)
+
+    payload = {
+        "model": LLM_MODEL,
+        "messages": messages, 
+        "temperature": temperature,  # 1.0,
+        "top_p": top_p,  # 1.0,
+        "n": 1,
+        "stream": stream,
+        "presence_penalty": 0,
+        "frequency_penalty": 0,
+    }
+    
+    print(f" {LLM_MODEL} : {conversation_cnt} : {inputs}")
+    return headers,payload
+
+

requirements.txt

@@ -0,0 +1,11 @@
+--extra-index-url https://download.pytorch.org/whl/cu113 
+torch
+torchvision
+git+https://github.com/huggingface/diffusers.git
+transformers
+accelerate
+gradio>=3.23
+requests[socks]
+mdtex2html
+Markdown
+latex2mathml

show_math.py

@@ -0,0 +1,80 @@
+# This program is written by: https://github.com/polarwinkel/mdtex2html
+
+from latex2mathml.converter import convert as tex2mathml
+import re
+
+incomplete = '<font style="color:orange;" class="tooltip">&#9888;<span class="tooltiptext">formula incomplete</span></font>'
+convError = '<font style="color:red" class="tooltip">&#9888;<span class="tooltiptext">LaTeX-convert-error</span></font>'
+
+def convert(mdtex, extensions=[], splitParagraphs=True):
+    ''' converts recursively the Markdown-LaTeX-mixture to HTML with MathML '''
+    found = False
+    # handle all paragraphs separately (prevents aftereffects)
+    if splitParagraphs:
+        parts = re.split("\n\n", mdtex)
+        result = ''
+        for part in parts:
+            result += convert(part, extensions, splitParagraphs=False)
+        return result
+    # find first $$-formula:
+    parts = re.split('\${2}', mdtex, 2)
+    if len(parts)>1:
+        found = True
+        result = convert(parts[0], extensions, splitParagraphs=False)+'\n'
+        try:
+            result += '<div class="blockformula">'+tex2mathml(parts[1])+'</div>\n'
+        except:
+            result += '<div class="blockformula">'+convError+'</div>'
+        if len(parts)==3:
+            result += convert(parts[2], extensions, splitParagraphs=False)
+        else:
+            result += '<div class="blockformula">'+incomplete+'</div>'
+    # else find first $-formulas:
+    else:
+        parts = re.split('\${1}', mdtex, 2)
+    if len(parts)>1 and not found:
+        found = True
+        try:
+            mathml = tex2mathml(parts[1])
+        except:
+            mathml = convError
+        if parts[0].endswith('\n\n') or parts[0]=='': # make sure textblock starts before formula!
+            parts[0]=parts[0]+'&#x200b;'
+        if len(parts)==3:
+            result = convert(parts[0]+mathml+parts[2], extensions, splitParagraphs=False)
+        else:
+            result = convert(parts[0]+mathml+incomplete, extensions, splitParagraphs=False)
+    # else find first \[..\]-equation:
+    else:
+        parts = re.split(r'\\\[', mdtex, 1)
+    if len(parts)>1 and not found:
+        found = True
+        result = convert(parts[0], extensions, splitParagraphs=False)+'\n'
+        parts = re.split(r'\\\]', parts[1], 1)
+        try:
+            result += '<div class="blockformula">'+tex2mathml(parts[0])+'</div>\n'
+        except:
+            result += '<div class="blockformula">'+convError+'</div>'
+        if len(parts)==2:
+            result += convert(parts[1], extensions, splitParagraphs=False)
+        else:
+            result += '<div class="blockformula">'+incomplete+'</div>'
+    # else find first \(..\)-equation:
+    else:
+        parts = re.split(r'\\\(', mdtex, 1)
+    if len(parts)>1 and not found:
+        found = True
+        subp = re.split(r'\\\)', parts[1], 1)
+        try:
+            mathml = tex2mathml(subp[0])
+        except:
+            mathml = convError
+        if parts[0].endswith('\n\n') or parts[0]=='': # make sure textblock starts before formula!
+            parts[0]=parts[0]+'&#x200b;'
+        if len(subp)==2:
+            result = convert(parts[0]+mathml+subp[1], extensions, splitParagraphs=False)
+        else:
+            result = convert(parts[0]+mathml+incomplete, extensions, splitParagraphs=False)
+    if not found:
+        result = mdtex
+    return result

theme.py

@@ -0,0 +1,152 @@
+import gradio as gr 
+
+# gradio可用颜色列表
+# gr.themes.utils.colors.slate (石板色)
+# gr.themes.utils.colors.gray (灰色)
+# gr.themes.utils.colors.zinc (锌色)
+# gr.themes.utils.colors.neutral (中性色)
+# gr.themes.utils.colors.stone (石头色)
+# gr.themes.utils.colors.red (红色)
+# gr.themes.utils.colors.orange (橙色)
+# gr.themes.utils.colors.amber (琥珀色)
+# gr.themes.utils.colors.yellow (黄色)
+# gr.themes.utils.colors.lime (酸橙色)
+# gr.themes.utils.colors.green (绿色)
+# gr.themes.utils.colors.emerald (祖母绿)
+# gr.themes.utils.colors.teal (青蓝色)
+# gr.themes.utils.colors.cyan (青色)
+# gr.themes.utils.colors.sky (天蓝色)
+# gr.themes.utils.colors.blue (蓝色)
+# gr.themes.utils.colors.indigo (靛蓝色)
+# gr.themes.utils.colors.violet (紫罗兰色)
+# gr.themes.utils.colors.purple (紫色)
+# gr.themes.utils.colors.fuchsia (洋红色)
+# gr.themes.utils.colors.pink (粉红色)
+# gr.themes.utils.colors.rose (玫瑰色)
+
+def adjust_theme():
+    try: 
+        color_er = gr.themes.utils.colors.pink
+        set_theme = gr.themes.Default( 
+                        primary_hue=gr.themes.utils.colors.orange,
+                        neutral_hue=gr.themes.utils.colors.gray,
+                        font=["sans-serif", "Microsoft YaHei", "ui-sans-serif", "system-ui", "sans-serif", gr.themes.utils.fonts.GoogleFont("Source Sans Pro")], 
+                        font_mono=["ui-monospace", "Consolas", "monospace", gr.themes.utils.fonts.GoogleFont("IBM Plex Mono")])
+        set_theme.set(
+            # Colors
+            input_background_fill_dark="*neutral_800",
+            # Transition
+            button_transition="none",
+            # Shadows
+            button_shadow="*shadow_drop",
+            button_shadow_hover="*shadow_drop_lg",
+            button_shadow_active="*shadow_inset",
+            input_shadow="0 0 0 *shadow_spread transparent, *shadow_inset",
+            input_shadow_focus="0 0 0 *shadow_spread *secondary_50, *shadow_inset",
+            input_shadow_focus_dark="0 0 0 *shadow_spread *neutral_700, *shadow_inset",
+            checkbox_label_shadow="*shadow_drop",
+            block_shadow="*shadow_drop",
+            form_gap_width="1px",
+            # Button borders
+            input_border_width="1px",
+            input_background_fill="white",
+            # Gradients
+            stat_background_fill="linear-gradient(to right, *primary_400, *primary_200)",
+            stat_background_fill_dark="linear-gradient(to right, *primary_400, *primary_600)",
+            error_background_fill=f"linear-gradient(to right, {color_er.c100}, *background_fill_secondary)",
+            error_background_fill_dark="*background_fill_primary",
+            checkbox_label_background_fill="linear-gradient(to top, *neutral_50, white)",
+            checkbox_label_background_fill_dark="linear-gradient(to top, *neutral_900, *neutral_800)",
+            checkbox_label_background_fill_hover="linear-gradient(to top, *neutral_100, white)",
+            checkbox_label_background_fill_hover_dark="linear-gradient(to top, *neutral_900, *neutral_800)",
+            button_primary_background_fill="linear-gradient(to bottom right, *primary_100, *primary_300)",
+            button_primary_background_fill_dark="linear-gradient(to bottom right, *primary_500, *primary_600)",
+            button_primary_background_fill_hover="linear-gradient(to bottom right, *primary_100, *primary_200)",
+            button_primary_background_fill_hover_dark="linear-gradient(to bottom right, *primary_500, *primary_500)",
+            button_primary_border_color_dark="*primary_500",
+            button_secondary_background_fill="linear-gradient(to bottom right, *neutral_100, *neutral_200)",
+            button_secondary_background_fill_dark="linear-gradient(to bottom right, *neutral_600, *neutral_700)",
+            button_secondary_background_fill_hover="linear-gradient(to bottom right, *neutral_100, *neutral_100)",
+            button_secondary_background_fill_hover_dark="linear-gradient(to bottom right, *neutral_600, *neutral_600)",
+            button_cancel_background_fill=f"linear-gradient(to bottom right, {color_er.c100}, {color_er.c200})",
+            button_cancel_background_fill_dark=f"linear-gradient(to bottom right, {color_er.c600}, {color_er.c700})",
+            button_cancel_background_fill_hover=f"linear-gradient(to bottom right, {color_er.c100}, {color_er.c100})",
+            button_cancel_background_fill_hover_dark=f"linear-gradient(to bottom right, {color_er.c600}, {color_er.c600})",
+            button_cancel_border_color=color_er.c200,
+            button_cancel_border_color_dark=color_er.c600,
+            button_cancel_text_color=color_er.c600,
+            button_cancel_text_color_dark="white",
+        )
+    except: 
+        set_theme = None; print('gradio版本较旧, 不能自定义字体和颜色')
+    return set_theme
+
+advanced_css = """
+/* 设置表格的外边距为1em，内部单元格之间边框合并，空单元格显示. */
+.markdown-body table {
+    margin: 1em 0;
+    border-collapse: collapse;
+    empty-cells: show;
+}
+
+/* 设置表格单元格的内边距为5px，边框粗细为1.2px，颜色为--border-color-primary. */
+.markdown-body th, .markdown-body td {
+    border: 1.2px solid var(--border-color-primary);
+    padding: 5px;
+}
+
+/* 设置表头背景颜色为rgba(175,184,193,0.2)，透明度为0.2. */
+.markdown-body thead {
+    background-color: rgba(175,184,193,0.2);
+}
+
+/* 设置表头单元格的内边距为0.5em和0.2em. */
+.markdown-body thead th {
+    padding: .5em .2em;
+}
+
+/* 去掉列表前缀的默认间距，使其与文本线对齐. */
+.markdown-body ol, .markdown-body ul {
+    padding-inline-start: 2em !important;
+}
+
+/* 设定聊天气泡的样式，包括圆角、最大宽度和阴影等. */
+[class *= "message"] {
+    border-radius: var(--radius-xl) !important;
+    /* padding: var(--spacing-xl) !important; */
+    /* font-size: var(--text-md) !important; */
+    /* line-height: var(--line-md) !important; */
+    /* min-height: calc(var(--text-md)*var(--line-md) + 2*var(--spacing-xl)); */
+    /* min-width: calc(var(--text-md)*var(--line-md) + 2*var(--spacing-xl)); */
+}
+[data-testid = "bot"] {
+    max-width: 95%;
+    /* width: auto !important; */
+    border-bottom-left-radius: 0 !important;
+}
+[data-testid = "user"] {
+    max-width: 100%;
+    /* width: auto !important; */
+    border-bottom-right-radius: 0 !important;
+}
+
+/* 行内代码的背景设为淡灰色，设定圆角和间距. */
+.markdown-body code {
+    display: inline;
+    white-space: break-spaces;
+    border-radius: 6px;
+    margin: 0 2px 0 2px;
+    padding: .2em .4em .1em .4em;
+    background-color: rgba(175,184,193,0.2);
+}
+/* 设定代码块的样式，包括背景颜色、内、外边距、圆角。 */
+.markdown-body pre code {
+    display: block;
+    overflow: auto;
+    white-space: pre;
+    background-color: rgba(175,184,193,0.2);
+    border-radius: 10px;
+    padding: 1em;
+    margin: 1em 2em 1em 0.5em;
+}
+"""

toolbox.py

@@ -0,0 +1,317 @@
+import markdown, mdtex2html, threading, importlib, traceback, importlib, inspect, re
+from show_math import convert as convert_math
+from functools import wraps, lru_cache
+
+def get_reduce_token_percent(text):
+    try:
+        # text = "maximum context length is 4097 tokens. However, your messages resulted in 4870 tokens"
+        pattern = r"(\d+)\s+tokens\b"
+        match = re.findall(pattern, text)
+        EXCEED_ALLO = 500 # 稍微留一点余地，否则在回复时会因余量太少出问题
+        max_limit = float(match[0]) - EXCEED_ALLO
+        current_tokens = float(match[1])
+        ratio = max_limit/current_tokens
+        assert ratio > 0 and ratio < 1
+        return ratio, str(int(current_tokens-max_limit))
+    except:
+        return 0.5, '不详'
+
+def predict_no_ui_but_counting_down(i_say, i_say_show_user, chatbot, top_p, temperature, history=[], sys_prompt='', long_connection=True):
+    """
+        调用简单的predict_no_ui接口，但是依然保留了些许界面心跳功能，当对话太长时，会自动采用二分法截断
+        i_say: 当前输入
+        i_say_show_user: 显示到对话界面上的当前输入，例如，输入整个文件时，你绝对不想把文件的内容都糊到对话界面上
+        chatbot: 对话界面句柄
+        top_p, temperature: gpt参数
+        history: gpt参数 对话历史
+        sys_prompt: gpt参数 sys_prompt
+        long_connection: 是否采用更稳定的连接方式（推荐）
+    """
+    import time
+    from predict import predict_no_ui, predict_no_ui_long_connection
+    from toolbox import get_conf
+    TIMEOUT_SECONDS, MAX_RETRY = get_conf('TIMEOUT_SECONDS', 'MAX_RETRY')
+    # 多线程的时候，需要一个mutable结构在不同线程之间传递信息
+    # list就是最简单的mutable结构，我们第一个位置放gpt输出，第二个位置传递报错信息
+    mutable = [None, '']
+    # multi-threading worker
+    def mt(i_say, history):
+        while True:
+            try:
+                if long_connection:
+                    mutable[0] = predict_no_ui_long_connection(inputs=i_say, top_p=top_p, temperature=temperature, history=history, sys_prompt=sys_prompt)
+                else:
+                    mutable[0] = predict_no_ui(inputs=i_say, top_p=top_p, temperature=temperature, history=history, sys_prompt=sys_prompt)
+                break
+            except ConnectionAbortedError as token_exceeded_error:
+                # 尝试计算比例，尽可能多地保留文本
+                p_ratio, n_exceed = get_reduce_token_percent(str(token_exceeded_error))
+                if len(history) > 0:
+                    history = [his[     int(len(his)    *p_ratio):      ] for his in history if his is not None]
+                else:
+                    i_say = i_say[:     int(len(i_say)  *p_ratio)     ]
+                mutable[1] = f'警告，文本过长将进行截断，Token溢出数：{n_exceed}，截断比例：{(1-p_ratio):.0%}。'
+            except TimeoutError as e:
+                mutable[0] = '[Local Message] 请求超时。'
+                raise TimeoutError
+            except Exception as e:
+                mutable[0] = f'[Local Message] 异常：{str(e)}.'
+                raise RuntimeError(f'[Local Message] 异常：{str(e)}.')
+    # 创建新线程发出http请求
+    thread_name = threading.Thread(target=mt, args=(i_say, history)); thread_name.start()
+    # 原来的线程则负责持续更新UI，实现一个超时倒计时，并等待新线程的任务完成
+    cnt = 0
+    while thread_name.is_alive():
+        cnt += 1
+        chatbot[-1] = (i_say_show_user, f"[Local Message] {mutable[1]}waiting gpt response {cnt}/{TIMEOUT_SECONDS*2*(MAX_RETRY+1)}"+''.join(['.']*(cnt%4)))
+        yield chatbot, history, '正常'
+        time.sleep(1)
+    # 把gpt的输出从mutable中取出来
+    gpt_say = mutable[0]
+    if gpt_say=='[Local Message] Failed with timeout.': raise TimeoutError
+    return gpt_say
+
+def write_results_to_file(history, file_name=None):
+    """
+        将对话记录history以Markdown格式写入文件中。如果没有指定文件名，则使用当前时间生成文件名。
+    """
+    import os, time
+    if file_name is None:
+        # file_name = time.strftime("chatGPT分析报告%Y-%m-%d-%H-%M-%S", time.localtime()) + '.md'
+        file_name = 'chatGPT分析报告' + time.strftime("%Y-%m-%d-%H-%M-%S", time.localtime()) + '.md'
+    os.makedirs('./gpt_log/', exist_ok=True)
+    with open(f'./gpt_log/{file_name}', 'w', encoding = 'utf8') as f:
+        f.write('# chatGPT 分析报告\n')
+        for i, content in enumerate(history):
+            try:    # 这个bug没找到触发条件，暂时先这样顶一下
+                if type(content) != str: content = str(content)
+            except:
+                continue
+            if i%2==0: f.write('## ')
+            f.write(content)
+            f.write('\n\n')
+    res = '以上材料已经被写入' + os.path.abspath(f'./gpt_log/{file_name}')
+    print(res)
+    return res
+
+def regular_txt_to_markdown(text):
+    """
+        将普通文本转换为Markdown格式的文本。
+    """
+    text = text.replace('\n', '\n\n')
+    text = text.replace('\n\n\n', '\n\n')
+    text = text.replace('\n\n\n', '\n\n')
+    return text
+
+def CatchException(f):
+    """
+        装饰器函数，捕捉函数f中的异常并封装到一个生成器中返回，并显示到聊天当中。
+    """
+    @wraps(f)
+    def decorated(txt, top_p, temperature, chatbot, history, systemPromptTxt, WEB_PORT):
+        try:
+            yield from f(txt, top_p, temperature, chatbot, history, systemPromptTxt, WEB_PORT)
+        except Exception as e:
+            from check_proxy import check_proxy
+            from toolbox import get_conf
+            proxies, = get_conf('proxies')
+            tb_str = '```\n' + traceback.format_exc() + '```'
+            if len(chatbot) == 0: chatbot.append(["插件调度异常","异常原因"])
+            chatbot[-1] = (chatbot[-1][0], f"[Local Message] 实验性函数调用出错: \n\n{tb_str} \n\n当前代理可用性: \n\n{check_proxy(proxies)}")
+            yield chatbot, history, f'异常 {e}'
+    return decorated
+
+def HotReload(f):
+    """
+        装饰器函数，实现函数插件热更新
+    """
+    @wraps(f)
+    def decorated(*args, **kwargs):
+        fn_name = f.__name__
+        f_hot_reload = getattr(importlib.reload(inspect.getmodule(f)), fn_name)
+        yield from f_hot_reload(*args, **kwargs)
+    return decorated
+
+def report_execption(chatbot, history, a, b):
+    """
+        向chatbot中添加错误信息
+    """
+    chatbot.append((a, b))
+    history.append(a); history.append(b)
+
+def text_divide_paragraph(text):
+    """
+        将文本按照段落分隔符分割开，生成带有段落标签的HTML代码。
+    """
+    if '```' in text:
+        # careful input
+        return text
+    else:
+        # shit input
+        lines = text.split("\n")
+        for i, line in enumerate(lines):
+            lines[i] = lines[i].replace(" ", "&nbsp;")
+        text = "</br>".join(lines)
+        return text
+
+def markdown_convertion(txt):
+    """
+        将Markdown格式的文本转换为HTML格式。如果包含数学公式，则先将公式转换为HTML格式。
+    """
+    pre = '<div class="markdown-body">'
+    suf = '</div>'
+    if ('$' in txt) and ('```' not in txt):
+        return pre + markdown.markdown(txt,extensions=['fenced_code','tables']) + '<br><br>' + markdown.markdown(convert_math(txt, splitParagraphs=False),extensions=['fenced_code','tables']) + suf
+    else:
+        return pre + markdown.markdown(txt,extensions=['fenced_code','tables']) + suf
+
+def close_up_code_segment_during_stream(gpt_reply):
+    """
+        在gpt输出代码的中途（输出了前面的```，但还没输出完后面的```），补上后面的```
+    """
+    if '```' not in gpt_reply: return gpt_reply
+    if gpt_reply.endswith('```'): return gpt_reply
+
+    # 排除了以上两个情况，我们
+    segments = gpt_reply.split('```')
+    n_mark = len(segments) - 1
+    if n_mark % 2 == 1:
+        # print('输出代码片段中！')
+        return gpt_reply+'\n```'
+    else:
+        return gpt_reply
+ 
+
+
+def format_io(self, y):
+    """
+        将输入和输出解析为HTML格式。将y中最后一项的输入部分段落化，并将输出部分的Markdown和数学公式转换为HTML格式。
+    """
+    if y is None or y == []: return []
+    i_ask, gpt_reply = y[-1]
+    i_ask = text_divide_paragraph(i_ask) # 输入部分太自由，预处理一波
+    gpt_reply = close_up_code_segment_during_stream(gpt_reply)  # 当代码输出半截的时候，试着补上后个```
+    y[-1] = (
+        None if i_ask is None else markdown.markdown(i_ask, extensions=['fenced_code','tables']),
+        None if gpt_reply is None else markdown_convertion(gpt_reply)
+    )
+    return y
+
+
+def find_free_port():
+    """
+        返回当前系统中可用的未使用端口。
+    """
+    import socket
+    from contextlib import closing
+    with closing(socket.socket(socket.AF_INET, socket.SOCK_STREAM)) as s:
+        s.bind(('', 0))
+        s.setsockopt(socket.SOL_SOCKET, socket.SO_REUSEADDR, 1)
+        return s.getsockname()[1]
+
+
+def extract_archive(file_path, dest_dir):
+    import zipfile
+    import tarfile
+    import os
+    # Get the file extension of the input file
+    file_extension = os.path.splitext(file_path)[1]
+
+    # Extract the archive based on its extension
+    if file_extension == '.zip':
+        with zipfile.ZipFile(file_path, 'r') as zipobj:
+            zipobj.extractall(path=dest_dir)
+            print("Successfully extracted zip archive to {}".format(dest_dir))
+
+    elif file_extension in ['.tar', '.gz', '.bz2']:
+        with tarfile.open(file_path, 'r:*') as tarobj:
+            tarobj.extractall(path=dest_dir)
+            print("Successfully extracted tar archive to {}".format(dest_dir))
+
+    # 第三方库，需要预先pip install rarfile
+    # 此外，Windows��还需要安装winrar软件，配置其Path环境变量，如"C:\Program Files\WinRAR"才可以
+    elif file_extension == '.rar':
+        try:
+            import rarfile
+            with rarfile.RarFile(file_path) as rf:
+                rf.extractall(path=dest_dir)
+                print("Successfully extracted rar archive to {}".format(dest_dir))
+        except:
+            print("Rar format requires additional dependencies to install")
+            return '\n\n需要安装pip install rarfile来解压rar文件'
+
+    # 第三方库，需要预先pip install py7zr
+    elif file_extension == '.7z':
+        try:
+            import py7zr
+            with py7zr.SevenZipFile(file_path, mode='r') as f:
+                f.extractall(path=dest_dir)
+                print("Successfully extracted 7z archive to {}".format(dest_dir))
+        except:
+            print("7z format requires additional dependencies to install")
+            return '\n\n需要安装pip install py7zr来解压7z文件'
+    else:
+        return ''
+    return ''
+
+def find_recent_files(directory):
+    """
+        me: find files that is created with in one minutes under a directory with python, write a function
+        gpt: here it is!
+    """
+    import os
+    import time
+    current_time = time.time()
+    one_minute_ago = current_time - 60
+    recent_files = []
+
+    for filename in os.listdir(directory):
+        file_path = os.path.join(directory, filename)
+        if file_path.endswith('.log'): continue
+        created_time = os.path.getctime(file_path)
+        if created_time >= one_minute_ago:
+            if os.path.isdir(file_path): continue
+            recent_files.append(file_path)
+
+    return recent_files
+
+
+
+
+
+
+
+@lru_cache(maxsize=128)
+def read_single_conf_with_lru_cache(arg):
+    try: r = getattr(importlib.import_module('config_private'), arg)
+    except: r = getattr(importlib.import_module('config'), arg)
+    # 在读取API_KEY时，检查一下是不是忘了改config
+    if arg=='API_KEY':
+        # 正确的 API_KEY 是 "sk-" + 48 位大小写字母数字的组合
+        API_MATCH = re.match(r"sk-[a-zA-Z0-9]{48}$", r)
+        if API_MATCH:
+            print(f"[API_KEY] 您的 API_KEY 是: {r[:15]}*** API_KEY 导入成功")
+        else:
+            assert False, "正确的 API_KEY 是 'sk-' + '48 位大小写字母数字' 的组合，请在config文件中修改API密钥, 添加海外代理之后再运行。" + \
+                        "（如果您刚更新过代码，请确保旧版config_private文件中没有遗留任何新增键值）"
+    if arg=='proxies':
+        if r is None: 
+            print('[PROXY] 网络代理状态：未配置。无代理状态下很可能无法访问。建议：检查USE_PROXY选项是否修改。')
+        else: 
+            print('[PROXY] 网络代理状态：已配置。配置信息如下：', r)
+            assert isinstance(r, dict), 'proxies格式错误，请注意proxies选项的格式，不要遗漏括号。'
+    return r
+
+def get_conf(*args):
+    # 建议您复制一个config_private.py放自己的秘密, 如API和代理网址, 避免不小心传github被别人看到
+    res = []
+    for arg in args:
+        r = read_single_conf_with_lru_cache(arg)
+        res.append(r)
+    return res
+
+def clear_line_break(txt):
+    txt = txt.replace('\n', ' ')
+    txt = txt.replace('  ', ' ')
+    txt = txt.replace('  ', ' ')
+    return txt