pip

Merge branch 'master' of github.com:binary-husky/chatgpt_academic
Update README.md
2023-07-28 12:28:04 +08:00 · 2023-07-28 10:35:50 +08:00 · 2023-07-28 09:54:03 +08:00 · 2023-07-28 09:53:17 +08:00 · 2023-07-27 22:30:55 +08:00 · 2023-07-26 22:46:43 +08:00
30 changed files with 856 additions and 544 deletions
--- a/.github/workflows/build-with-audio-assistant.yml
+++ b/.github/workflows/build-with-audio-assistant.yml
@ -0,0 +1,44 @@
+# https://docs.github.com/en/actions/publishing-packages/publishing-docker-images#publishing-images-to-github-packages
+name: build-with-audio-assistant
+
+on:
+  push:
+    branches:
+      - 'master'
+
+env:
+  REGISTRY: ghcr.io
+  IMAGE_NAME: ${{ github.repository }}_audio_assistant
+
+jobs:
+  build-and-push-image:
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      packages: write
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+
+      - name: Log in to the Container registry
+        uses: docker/login-action@v2
+        with:
+          registry: ${{ env.REGISTRY }}
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Extract metadata (tags, labels) for Docker
+        id: meta
+        uses: docker/metadata-action@v4
+        with:
+          images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
+
+      - name: Build and push Docker image
+        uses: docker/build-push-action@v4
+        with:
+          context: .
+          push: true
+          file: docs/GithubAction+NoLocal+AudioAssistant
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
--- a/README.md
+++ b/README.md
@ -93,7 +93,7 @@ Latex论文一键校对 | [函数插件] 仿Grammarly对Latex文章进行语法

 1. 下载项目
 ```sh
-git clone https://github.com/binary-husky/gpt_academic.git
+git clone --depth=1 https://github.com/binary-husky/gpt_academic.git
 cd gpt_academic
 ```

@ -116,7 +116,7 @@ python -m pip install -r requirements.txt # 这个步骤和pip安装一样的步
 ```


-<details><summary>如果需要支持清华ChatGLM2/复旦MOSS作为后端，请点击展开此处</summary>
+<details><summary>如果需要支持清华ChatGLM2/复旦MOSS/RWKV作为后端，请点击展开此处</summary>
 <p>

 【可选步骤】如果需要支持清华ChatGLM2/复旦MOSS作为后端，需要额外安装更多依赖（前提条件：熟悉Python + 用过Pytorch + 电脑配置够强）：
@ -126,9 +126,12 @@ python -m pip install -r request_llm/requirements_chatglm.txt

 # 【可选步骤II】支持复旦MOSS
 python -m pip install -r request_llm/requirements_moss.txt
-git clone https://github.com/OpenLMLab/MOSS.git request_llm/moss  # 注意执行此行代码时，必须处于项目根路径
+git clone --depth=1 https://github.com/OpenLMLab/MOSS.git request_llm/moss  # 注意执行此行代码时，必须处于项目根路径

-# 【可选步骤III】确保config.py配置文件的AVAIL_LLM_MODELS包含了期望的模型，目前支持的全部模型如下(jittorllms系列目前仅支持docker方案)：
+# 【可选步骤III】支持RWKV Runner
+参考wiki：https://github.com/binary-husky/gpt_academic/wiki/%E9%80%82%E9%85%8DRWKV-Runner
+
+# 【可选步骤IV】确保config.py配置文件的AVAIL_LLM_MODELS包含了期望的模型，目前支持的全部模型如下(jittorllms系列目前仅支持docker方案)：
 AVAIL_LLM_MODELS = ["gpt-3.5-turbo", "api2d-gpt-3.5-turbo", "gpt-4", "api2d-gpt-4", "chatglm", "newbing", "moss"] # + ["jittorllms_rwkv", "jittorllms_pangualpha", "jittorllms_llama"]
 ```

@ -147,9 +150,10 @@ python main.py
 1. 仅ChatGPT（推荐大多数人选择，等价于docker-compose方案1）
 [![basic](https://github.com/binary-husky/gpt_academic/actions/workflows/build-without-local-llms.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-without-local-llms.yml)
 [![basiclatex](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-latex.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-latex.yml)
+[![basicaudio](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml/badge.svg?branch=master)](https://github.com/binary-husky/gpt_academic/actions/workflows/build-with-audio-assistant.yml)

 ``` sh
-git clone https://github.com/binary-husky/gpt_academic.git  # 下载项目
+git clone --depth=1 https://github.com/binary-husky/gpt_academic.git  # 下载项目
 cd gpt_academic                                 # 进入路径
 nano config.py                                      # 用任意文本编辑器编辑config.py, 配置 “Proxy”， “API_KEY” 以及 “WEB_PORT” (例如50923) 等
 docker build -t gpt-academic .                      # 安装
@ -195,10 +199,12 @@ docker-compose up
 5. 远程云服务器部署（需要云服务器知识与经验）。
 请访问[部署wiki-1](https://github.com/binary-husky/gpt_academic/wiki/%E4%BA%91%E6%9C%8D%E5%8A%A1%E5%99%A8%E8%BF%9C%E7%A8%8B%E9%83%A8%E7%BD%B2%E6%8C%87%E5%8D%97)

-6. 使用WSL2（Windows Subsystem for Linux 子系统）。
+6. 使用Sealos[一键部署](https://github.com/binary-husky/gpt_academic/issues/993)。
+
+7. 使用WSL2（Windows Subsystem for Linux 子系统）。
 请访问[部署wiki-2](https://github.com/binary-husky/gpt_academic/wiki/%E4%BD%BF%E7%94%A8WSL2%EF%BC%88Windows-Subsystem-for-Linux-%E5%AD%90%E7%B3%BB%E7%BB%9F%EF%BC%89%E9%83%A8%E7%BD%B2)

-7. 如何在二级网址（如`http://localhost/subpath`）下运行。
+8. 如何在二级网址（如`http://localhost/subpath`）下运行。
 请访问[FastAPI运行说明](docs/WithFastapi.md)


--- a/config.py
+++ b/config.py
@ -80,6 +80,7 @@ ChatGLM_PTUNING_CHECKPOINT = "" # 例如"/home/hmp/ChatGLM2-6B/ptuning/output/6b

 # 本地LLM模型如ChatGLM的执行方式 CPU/GPU
 LOCAL_MODEL_DEVICE = "cpu" # 可选 "cuda"
+LOCAL_MODEL_QUANT = "FP16" # 默认 "FP16" "INT4" 启用量化INT4版本 "INT8" 启用量化INT8版本


 # 设置gradio的并行线程数（不需要修改）
@ -131,9 +132,14 @@ put your new bing cookies here

 # 阿里云实时语音识别 配置难度较高 仅建议高手用户使用 参考 https://github.com/binary-husky/gpt_academic/blob/master/docs/use_audio.md
 ENABLE_AUDIO = False
-ALIYUN_TOKEN=""    # 例如 f37f30e0f9934c34a992f6f64f7eba4f
-ALIYUN_APPKEY=""   # 例如 RoPlZrM88DnAFkZK
-
+ALIYUN_TOKEN=""     # 例如 f37f30e0f9934c34a992f6f64f7eba4f
+ALIYUN_APPKEY=""    # 例如 RoPlZrM88DnAFkZK
+ALIYUN_ACCESSKEY="" # （无需填写）
+ALIYUN_SECRET=""    # （无需填写）

 # Claude API KEY
-ANTHROPIC_API_KEY = ""
+ANTHROPIC_API_KEY = ""
+
+
+# 自定义API KEY格式
+CUSTOM_API_KEY_PATTERN = ""
--- a/core_functional.py
+++ b/core_functional.py
@ -1,7 +1,7 @@
 # '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
+import importlib
 from toolbox import clear_line_break


@ -14,7 +14,12 @@ def get_core_functions():
                        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
+            # 按钮颜色 (默认 secondary)
+            "Color":    r"secondary",
+            # 按钮是否可见 (默认 True，即可见)
+            "Visible": True,
+            # 是否在触发时清除历史 (默认 False，即不处理之前的对话历史)
+            "AutoClearHistory": True
        },
        "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，" +
@ -76,3 +81,13 @@ def get_core_functions():
            "Suffix":   r"",
        }
    }
+
+
+def handle_core_functionality(additional_fn, inputs, history):
+    import core_functional
+    importlib.reload(core_functional)    # 热更新prompt
+    core_functional = core_functional.get_core_functions()
+    if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
+    inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+    history = [] if core_functional[additional_fn].get("AutoClearHistory", False) else history
+    return inputs, history
--- a/crazy_functions/Latex输出PDF结果.py
+++ b/crazy_functions/Latex输出PDF结果.py
@ -157,7 +157,7 @@ def Latex英文纠错加PDF对比(txt, llm_kwargs, plugin_kwargs, chatbot, histo
    try:
        import glob, os, time, subprocess
        subprocess.Popen(['pdflatex', '-version'])
-        from .latex_utils import Latex精细分解与转化, 编译Latex
+        from .latex_fns.latex_actions import Latex精细分解与转化, 编译Latex
    except Exception as e:
        chatbot.append([ f"解析项目: {txt}",
            f"尝试执行Latex指令失败。Latex没有安装, 或者不在环境变量PATH中。安装方法https://tug.org/texlive/。报错信息\n\n```\n\n{trimmed_format_exc()}\n\n```\n\n"])
@ -234,7 +234,7 @@ def Latex翻译中文并重新编译PDF(txt, llm_kwargs, plugin_kwargs, chatbot,
    try:
        import glob, os, time, subprocess
        subprocess.Popen(['pdflatex', '-version'])
-        from .latex_utils import Latex精细分解与转化, 编译Latex
+        from .latex_fns.latex_actions import Latex精细分解与转化, 编译Latex
    except Exception as e:
        chatbot.append([ f"解析项目: {txt}",
            f"尝试执行Latex指令失败。Latex没有安装, 或者不在环境变量PATH中。安装方法https://tug.org/texlive/。报错信息\n\n```\n\n{trimmed_format_exc()}\n\n```\n\n"])
--- a/crazy_functions/crazy_functions_test.py
+++ b/crazy_functions/crazy_functions_test.py
@ -195,9 +195,12 @@ def test_Latex():
    # txt = r"https://arxiv.org/abs/2303.08774"
    # txt = r"https://arxiv.org/abs/2303.12712"
    # txt = r"C:\Users\fuqingxu\arxiv_cache\2303.12712\workfolder"
-    txt = r"2306.17157" # 这个paper有个input命令文件名大小写错误！
-    
-
+    # txt = r"2306.17157" # 这个paper有个input命令文件名大小写错误！
+    # txt = "https://arxiv.org/abs/2205.14135"
+    # txt = r"C:\Users\fuqingxu\arxiv_cache\2205.14135\workfolder"
+    # txt = r"C:\Users\fuqingxu\arxiv_cache\2205.14135\workfolder"
+    txt = r"2210.03629"
+    txt = r"2307.04964"
    for cookies, cb, hist, msg in (Latex翻译中文并重新编译PDF)(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
        cli_printer.print(cb)   #  print(cb)

@ -240,7 +243,7 @@ if __name__ == "__main__":
    # test_数学动画生成manim()
    # test_Langchain知识库()
    # test_Langchain知识库读取()
-    # test_Latex()
-    test_chatglm_finetune()
+    test_Latex()
+    # test_chatglm_finetune()
    input("程序完成，回车退出。")
    print("退出。")
--- a/crazy_functions/latex_fns/latex_actions.py
+++ b/crazy_functions/latex_fns/latex_actions.py
@ -1,320 +1,16 @@
 from toolbox import update_ui, update_ui_lastest_msg    # 刷新Gradio前端界面
 from toolbox import zip_folder, objdump, objload, promote_file_to_downloadzone
+from .latex_toolbox import PRESERVE, TRANSFORM
+from .latex_toolbox import set_forbidden_text, set_forbidden_text_begin_end, set_forbidden_text_careful_brace
+from .latex_toolbox import reverse_forbidden_text_careful_brace, reverse_forbidden_text, convert_to_linklist, post_process
+from .latex_toolbox import fix_content, find_main_tex_file, merge_tex_files, compile_latex_with_timeout
+
 import os, shutil
 import re
 import numpy as np
+
 pj = os.path.join

-"""
-========================================================================
-Part One
-Latex segmentation with a binary mask (PRESERVE=0, TRANSFORM=1)
-========================================================================
-"""
-PRESERVE = 0
-TRANSFORM = 1
-
-def set_forbidden_text(text, mask, pattern, flags=0):
-    """
-    Add a preserve text area in this paper
-    e.g. with pattern = r"\\begin\{algorithm\}(.*?)\\end\{algorithm\}"
-    you can mask out (mask = PRESERVE so that text become untouchable for GPT) 
-    everything between "\begin{equation}" and "\end{equation}"
-    """
-    if isinstance(pattern, list): pattern = '|'.join(pattern)
-    pattern_compile = re.compile(pattern, flags)
-    for res in pattern_compile.finditer(text):
-        mask[res.span()[0]:res.span()[1]] = PRESERVE
-    return text, mask
-
-def reverse_forbidden_text(text, mask, pattern, flags=0, forbid_wrapper=True):
-    """
-    Move area out of preserve area (make text editable for GPT)
-    count the number of the braces so as to catch compelete text area. 
-    e.g.
-    \begin{abstract} blablablablablabla. \end{abstract} 
-    """
-    if isinstance(pattern, list): pattern = '|'.join(pattern)
-    pattern_compile = re.compile(pattern, flags)
-    for res in pattern_compile.finditer(text):
-        if not forbid_wrapper:
-            mask[res.span()[0]:res.span()[1]] = TRANSFORM
-        else:
-            mask[res.regs[0][0]: res.regs[1][0]] = PRESERVE   # '\\begin{abstract}'
-            mask[res.regs[1][0]: res.regs[1][1]] = TRANSFORM   # abstract
-            mask[res.regs[1][1]: res.regs[0][1]] = PRESERVE   # abstract
-    return text, mask
-
-def set_forbidden_text_careful_brace(text, mask, pattern, flags=0):
-    """
-    Add a preserve text area in this paper (text become untouchable for GPT).
-    count the number of the braces so as to catch compelete text area. 
-    e.g.
-    \caption{blablablablabla\texbf{blablabla}blablabla.} 
-    """
-    pattern_compile = re.compile(pattern, flags)
-    for res in pattern_compile.finditer(text):
-        brace_level = -1
-        p = begin = end = res.regs[0][0]
-        for _ in range(1024*16):
-            if text[p] == '}' and brace_level == 0: break
-            elif text[p] == '}':  brace_level -= 1
-            elif text[p] == '{':  brace_level += 1
-            p += 1
-        end = p+1
-        mask[begin:end] = PRESERVE
-    return text, mask
-
-def reverse_forbidden_text_careful_brace(text, mask, pattern, flags=0, forbid_wrapper=True):
-    """
-    Move area out of preserve area (make text editable for GPT)
-    count the number of the braces so as to catch compelete text area. 
-    e.g.
-    \caption{blablablablabla\texbf{blablabla}blablabla.} 
-    """
-    pattern_compile = re.compile(pattern, flags)
-    for res in pattern_compile.finditer(text):
-        brace_level = 0
-        p = begin = end = res.regs[1][0]
-        for _ in range(1024*16):
-            if text[p] == '}' and brace_level == 0: break
-            elif text[p] == '}':  brace_level -= 1
-            elif text[p] == '{':  brace_level += 1
-            p += 1
-        end = p
-        mask[begin:end] = TRANSFORM
-        if forbid_wrapper:
-            mask[res.regs[0][0]:begin] = PRESERVE
-            mask[end:res.regs[0][1]] = PRESERVE
-    return text, mask
-
-def set_forbidden_text_begin_end(text, mask, pattern, flags=0, limit_n_lines=42):
-    """
-    Find all \begin{} ... \end{} text block that with less than limit_n_lines lines.
-    Add it to preserve area
-    """
-    pattern_compile = re.compile(pattern, flags)
-    def search_with_line_limit(text, mask):
-        for res in pattern_compile.finditer(text):
-            cmd = res.group(1)  # begin{what}
-            this = res.group(2) # content between begin and end
-            this_mask = mask[res.regs[2][0]:res.regs[2][1]]
-            white_list = ['document', 'abstract', 'lemma', 'definition', 'sproof', 
-                          'em', 'emph', 'textit', 'textbf', 'itemize', 'enumerate']
-            if (cmd in white_list) or this.count('\n') >= limit_n_lines: # use a magical number 42
-                this, this_mask = search_with_line_limit(this, this_mask)
-                mask[res.regs[2][0]:res.regs[2][1]] = this_mask
-            else:
-                mask[res.regs[0][0]:res.regs[0][1]] = PRESERVE
-        return text, mask
-    return search_with_line_limit(text, mask) 
-
-class LinkedListNode():
-    """
-    Linked List Node
-    """
-    def __init__(self, string, preserve=True) -> None:
-        self.string = string
-        self.preserve = preserve
-        self.next = None
-        # self.begin_line = 0
-        # self.begin_char = 0
-
-def convert_to_linklist(text, mask):
-    root = LinkedListNode("", preserve=True)
-    current_node = root
-    for c, m, i in zip(text, mask, range(len(text))):
-        if (m==PRESERVE and current_node.preserve) \
-            or (m==TRANSFORM and not current_node.preserve):
-            # add
-            current_node.string += c
-        else:
-            current_node.next = LinkedListNode(c, preserve=(m==PRESERVE))
-            current_node = current_node.next
-    return root
-"""
-========================================================================
-Latex Merge File
-========================================================================
-"""
-
-def 寻找Latex主文件(file_manifest, mode):
-    """
-    在多Tex文档中，寻找主文件，必须包含documentclass，返回找到的第一个。
-    P.S. 但愿没人把latex模板放在里面传进来 (6.25 加入判定latex模板的代码)
-    """
-    canidates = []
-    for texf in file_manifest:
-        if os.path.basename(texf).startswith('merge'):
-            continue
-        with open(texf, 'r', encoding='utf8', errors='ignore') as f:
-            file_content = f.read()
-        if r'\documentclass' in file_content:
-            canidates.append(texf)
-        else:
-            continue
-
-    if len(canidates) == 0:
-        raise RuntimeError('无法找到一个主Tex文件（包含documentclass关键字）')
-    elif len(canidates) == 1:
-        return canidates[0]
-    else: # if len(canidates) >= 2 通过一些Latex模板中常见（但通常不会出现在正文）的单词，对不同latex源文件扣分，取评分最高者返回
-        canidates_score = []
-        # 给出一些判定模板文档的词作为扣分项
-        unexpected_words = ['\LaTeX', 'manuscript', 'Guidelines', 'font', 'citations', 'rejected', 'blind review', 'reviewers']
-        expected_words = ['\input', '\ref', '\cite']
-        for texf in canidates:
-            canidates_score.append(0)
-            with open(texf, 'r', encoding='utf8', errors='ignore') as f:
-                file_content = f.read()
-            for uw in unexpected_words:
-                if uw in file_content:
-                    canidates_score[-1] -= 1
-            for uw in expected_words:
-                if uw in file_content:
-                    canidates_score[-1] += 1
-        select = np.argmax(canidates_score) # 取评分最高者返回
-        return canidates[select]
-    
-def rm_comments(main_file):
-    new_file_remove_comment_lines = []
-    for l in main_file.splitlines():
-        # 删除整行的空注释
-        if l.lstrip().startswith("%"):
-            pass
-        else:
-            new_file_remove_comment_lines.append(l)
-    main_file = '\n'.join(new_file_remove_comment_lines)
-    # main_file = re.sub(r"\\include{(.*?)}", r"\\input{\1}", main_file)  # 将 \include 命令转换为 \input 命令
-    main_file = re.sub(r'(?<!\\)%.*', '', main_file)  # 使用正则表达式查找半行注释, 并替换为空字符串
-    return main_file
-
-def find_tex_file_ignore_case(fp):
-    dir_name = os.path.dirname(fp)
-    base_name = os.path.basename(fp)
-    if not base_name.endswith('.tex'): base_name+='.tex'
-    if os.path.exists(pj(dir_name, base_name)): return pj(dir_name, base_name)
-    # go case in-sensitive
-    import glob
-    for f in glob.glob(dir_name+'/*.tex'):
-        base_name_s = os.path.basename(fp)
-        if base_name_s.lower() == base_name.lower(): return f
-    return None
-
-def merge_tex_files_(project_foler, main_file, mode):
-    """
-    Merge Tex project recrusively
-    """
-    main_file = rm_comments(main_file)
-    for s in reversed([q for q in re.finditer(r"\\input\{(.*?)\}", main_file, re.M)]):
-        f = s.group(1)
-        fp = os.path.join(project_foler, f)
-        fp = find_tex_file_ignore_case(fp)
-        if fp:
-            with open(fp, 'r', encoding='utf-8', errors='replace') as fx: c = fx.read()
-        else:
-            raise RuntimeError(f'找不到{fp}，Tex源文件缺失！')
-        c = merge_tex_files_(project_foler, c, mode)
-        main_file = main_file[:s.span()[0]] + c + main_file[s.span()[1]:]
-    return main_file
-
-def merge_tex_files(project_foler, main_file, mode):
-    """
-    Merge Tex project recrusively
-    P.S. 顺便把CTEX塞进去以支持中文
-    P.S. 顺便把Latex的注释去除
-    """
-    main_file = merge_tex_files_(project_foler, main_file, mode)
-    main_file = rm_comments(main_file)
-
-    if mode == 'translate_zh':
-        # find paper documentclass
-        pattern = re.compile(r'\\documentclass.*\n')
-        match = pattern.search(main_file)
-        assert match is not None, "Cannot find documentclass statement!"
-        position = match.end()
-        add_ctex = '\\usepackage{ctex}\n'
-        add_url = '\\usepackage{url}\n' if '{url}' not in main_file else ''
-        main_file = main_file[:position] + add_ctex + add_url + main_file[position:]
-        # fontset=windows
-        import platform
-        main_file = re.sub(r"\\documentclass\[(.*?)\]{(.*?)}", r"\\documentclass[\1,fontset=windows,UTF8]{\2}",main_file)
-        main_file = re.sub(r"\\documentclass{(.*?)}", r"\\documentclass[fontset=windows,UTF8]{\1}",main_file)
-        # find paper abstract
-        pattern_opt1 = re.compile(r'\\begin\{abstract\}.*\n')
-        pattern_opt2 = re.compile(r"\\abstract\{(.*?)\}", flags=re.DOTALL)
-        match_opt1 = pattern_opt1.search(main_file)
-        match_opt2 = pattern_opt2.search(main_file)
-        assert (match_opt1 is not None) or (match_opt2 is not None), "Cannot find paper abstract section!"
-    return main_file
-
-
-
-"""
-========================================================================
-Post process
-========================================================================
-"""
-def mod_inbraket(match):
-    """
-    为啥chatgpt会把cite里面的逗号换成中文逗号呀 
-    """
-    # get the matched string
-    cmd = match.group(1)
-    str_to_modify = match.group(2)
-    # modify the matched string
-    str_to_modify = str_to_modify.replace('：', ':')    # 前面是中文冒号，后面是英文冒号
-    str_to_modify = str_to_modify.replace('，', ',')    # 前面是中文逗号，后面是英文逗号
-    # str_to_modify = 'BOOM'
-    return "\\" + cmd + "{" + str_to_modify + "}"
-
-def fix_content(final_tex, node_string):
-    """
-    Fix common GPT errors to increase success rate
-    """
-    final_tex = re.sub(r"(?<!\\)%", "\\%", final_tex)
-    final_tex = re.sub(r"\\([a-z]{2,10})\ \{", r"\\\1{", string=final_tex)
-    final_tex = re.sub(r"\\\ ([a-z]{2,10})\{", r"\\\1{", string=final_tex)
-    final_tex = re.sub(r"\\([a-z]{2,10})\{([^\}]*?)\}", mod_inbraket, string=final_tex)
-
-    if "Traceback" in final_tex and "[Local Message]" in final_tex:
-        final_tex = node_string # 出问题了，还原原文
-    if node_string.count('\\begin') != final_tex.count('\\begin'):
-        final_tex = node_string # 出问题了，还原原文
-    if node_string.count('\_') > 0 and node_string.count('\_') > final_tex.count('\_'):
-        # walk and replace any _ without \
-        final_tex = re.sub(r"(?<!\\)_", "\\_", final_tex)
-
-    def compute_brace_level(string):
-        # this function count the number of { and }
-        brace_level = 0
-        for c in string:
-            if c == "{": brace_level += 1
-            elif c == "}": brace_level -= 1
-        return brace_level
-    def join_most(tex_t, tex_o):
-        # this function join translated string and original string when something goes wrong
-        p_t = 0
-        p_o = 0
-        def find_next(string, chars, begin):
-            p = begin
-            while p < len(string):
-                if string[p] in chars: return p, string[p]
-                p += 1
-            return None, None
-        while True:
-            res1, char = find_next(tex_o, ['{','}'], p_o)
-            if res1 is None: break
-            res2, char = find_next(tex_t, [char], p_t)
-            if res2 is None: break
-            p_o = res1 + 1
-            p_t = res2 + 1
-        return tex_t[:p_t] + tex_o[p_o:]
-
-    if compute_brace_level(final_tex) != compute_brace_level(node_string):
-        # 出问题了，还原部分原文，保证括号正确
-        final_tex = join_most(final_tex, node_string)
-    return final_tex

 def split_subprocess(txt, project_folder, return_dict, opts):
    """
@ -326,7 +22,8 @@ def split_subprocess(txt, project_folder, return_dict, opts):
    mask = np.zeros(len(txt), dtype=np.uint8) + TRANSFORM

    # 吸收title与作者以上的部分
-    text, mask = set_forbidden_text(text, mask, r"(.*?)\\maketitle", re.DOTALL)
+    text, mask = set_forbidden_text(text, mask, r"^(.*?)\\maketitle", re.DOTALL)
+    text, mask = set_forbidden_text(text, mask, r"^(.*?)\\begin{document}", re.DOTALL)
    # 吸收iffalse注释
    text, mask = set_forbidden_text(text, mask, r"\\iffalse(.*?)\\fi", re.DOTALL)
    # 吸收在42行以内的begin-end组合
@ -356,77 +53,9 @@ def split_subprocess(txt, project_folder, return_dict, opts):
    text, mask = reverse_forbidden_text(text, mask, r"\\begin\{abstract\}(.*?)\\end\{abstract\}", re.DOTALL, forbid_wrapper=True)
    root = convert_to_linklist(text, mask)

-    # 修复括号
-    node = root
-    while True:
-        string = node.string
-        if node.preserve: 
-            node = node.next
-            if node is None: break
-            continue
-        def break_check(string):
-            str_stack = [""] # (lv, index)
-            for i, c in enumerate(string):
-                if c == '{':
-                    str_stack.append('{')
-                elif c == '}':
-                    if len(str_stack) == 1:
-                        print('stack fix')
-                        return i
-                    str_stack.pop(-1)
-                else:
-                    str_stack[-1] += c
-            return -1
-        bp = break_check(string)
+    # 最后一步处理，增强稳健性
+    root = post_process(root)

-        if bp == -1:
-            pass
-        elif bp == 0:
-            node.string = string[:1]
-            q = LinkedListNode(string[1:], False)
-            q.next = node.next
-            node.next = q
-        else:
-            node.string = string[:bp]
-            q = LinkedListNode(string[bp:], False)
-            q.next = node.next
-            node.next = q
-
-        node = node.next
-        if node is None: break
-
-    # 屏蔽空行和太短的句子
-    node = root
-    while True:
-        if len(node.string.strip('\n').strip(''))==0: node.preserve = True
-        if len(node.string.strip('\n').strip(''))<42: node.preserve = True
-        node = node.next
-        if node is None: break
-    node = root
-    while True:
-        if node.next and node.preserve and node.next.preserve:
-            node.string += node.next.string
-            node.next = node.next.next
-        node = node.next
-        if node is None: break
-
-    # 将前后断行符脱离
-    node = root
-    prev_node = None
-    while True:
-        if not node.preserve:
-            lstriped_ = node.string.lstrip().lstrip('\n')
-            if (prev_node is not None) and (prev_node.preserve) and (len(lstriped_)!=len(node.string)):
-                prev_node.string += node.string[:-len(lstriped_)]
-                node.string = lstriped_
-            rstriped_ = node.string.rstrip().rstrip('\n')
-            if (node.next is not None) and (node.next.preserve) and (len(rstriped_)!=len(node.string)):
-                node.next.string = node.string[len(rstriped_):] + node.next.string
-                node.string = rstriped_
-        # =====
-        prev_node = node
-        node = node.next
-        if node is None: break
    # 输出html调试文件，用红色标注处保留区（PRESERVE），用黑色标注转换区（TRANSFORM）
    with open(pj(project_folder, 'debug_log.html'), 'w', encoding='utf8') as f:
        segment_parts_for_gpt = []
@ -437,7 +66,7 @@ def split_subprocess(txt, project_folder, return_dict, opts):
            show_html = node.string.replace('\n','<br/>')
            if not node.preserve:
                segment_parts_for_gpt.append(node.string)
-                f.write(f'<p style="color:black;">#{show_html}#</p>')
+                f.write(f'<p style="color:black;">#{node.range}{show_html}#</p>')
            else:
                f.write(f'<p style="color:red;">{show_html}</p>')
            node = node.next
@ -448,8 +77,6 @@ def split_subprocess(txt, project_folder, return_dict, opts):
    return_dict['segment_parts_for_gpt'] = segment_parts_for_gpt
    return return_dict

-
-
 class LatexPaperSplit():
    """
    break down latex file to a linked list,
@ -464,18 +91,32 @@ class LatexPaperSplit():
        # 请您不要删除或修改这行警告，除非您是论文的原作者（如果您是论文原作者，欢迎加REAME中的QQ联系开发者）
        self.msg_declare = "为了防止大语言模型的意外谬误产生扩散影响，禁止移除或修改此警告。}}\\\\" 

-    def merge_result(self, arr, mode, msg):
+
+    def merge_result(self, arr, mode, msg, buggy_lines=[], buggy_line_surgery_n_lines=10):
        """
        Merge the result after the GPT process completed
        """
        result_string = ""
-        p = 0
+        node_cnt = 0
+        line_cnt = 0
+        
        for node in self.nodes:
            if node.preserve:
+                line_cnt += node.string.count('\n')
                result_string += node.string
            else:
-                result_string += fix_content(arr[p], node.string)
-                p += 1
+                translated_txt = fix_content(arr[node_cnt], node.string)
+                begin_line = line_cnt
+                end_line = line_cnt + translated_txt.count('\n')
+
+                # reverse translation if any error
+                if any([begin_line-buggy_line_surgery_n_lines <= b_line <= end_line+buggy_line_surgery_n_lines for b_line in buggy_lines]):
+                    translated_txt = node.string
+
+                result_string += translated_txt
+                node_cnt += 1
+                line_cnt += translated_txt.count('\n')
+
        if mode == 'translate_zh':
            pattern = re.compile(r'\\begin\{abstract\}.*\n')
            match = pattern.search(result_string)
@ -490,6 +131,7 @@ class LatexPaperSplit():
            result_string = result_string[:position] + self.msg + msg + self.msg_declare + result_string[position:]
        return result_string

+
    def split(self, txt, project_folder, opts): 
        """
        break down latex file to a linked list,
@ -511,7 +153,6 @@ class LatexPaperSplit():
        return self.sp


-
 class LatexPaperFileGroup():
    """
    use tokenizer to break down text according to max_token_limit
@ -539,7 +180,7 @@ class LatexPaperFileGroup():
                self.sp_file_index.append(index)
                self.sp_file_tag.append(self.file_paths[index])
            else:
-                from .crazy_utils import breakdown_txt_to_satisfy_token_limit_for_pdf
+                from ..crazy_utils import breakdown_txt_to_satisfy_token_limit_for_pdf
                segments = breakdown_txt_to_satisfy_token_limit_for_pdf(file_content, self.get_token_num, max_token_limit)
                for j, segment in enumerate(segments):
                    self.sp_file_contents.append(segment)
@ -560,41 +201,14 @@ class LatexPaperFileGroup():
                f.write(res)
        return manifest

-def write_html(sp_file_contents, sp_file_result, chatbot, project_folder):
-
-    # write html
-    try:
-        import shutil
-        from .crazy_utils import construct_html
-        from toolbox import gen_time_str
-        ch = construct_html() 
-        orig = ""
-        trans = ""
-        final = []
-        for c,r in zip(sp_file_contents, sp_file_result): 
-            final.append(c)
-            final.append(r)
-        for i, k in enumerate(final): 
-            if i%2==0:
-                orig = k
-            if i%2==1:
-                trans = k
-                ch.add_row(a=orig, b=trans)
-        create_report_file_name = f"{gen_time_str()}.trans.html"
-        ch.save_file(create_report_file_name)
-        shutil.copyfile(pj('./gpt_log/', create_report_file_name), pj(project_folder, create_report_file_name))
-        promote_file_to_downloadzone(file=f'./gpt_log/{create_report_file_name}', chatbot=chatbot)
-    except:
-        from toolbox import trimmed_format_exc
-        print('writing html result failed:', trimmed_format_exc())

 def Latex精细分解与转化(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, mode='proofread', switch_prompt=None, opts=[]):
    import time, os, re
-    from .crazy_utils import request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
-    from .latex_utils import LatexPaperFileGroup, merge_tex_files, LatexPaperSplit, 寻找Latex主文件
+    from ..crazy_utils import request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
+    from .latex_actions import LatexPaperFileGroup, LatexPaperSplit

    #  <-------- 寻找主tex文件 ----------> 
-    maintex = 寻找Latex主文件(file_manifest, mode)
+    maintex = find_main_tex_file(file_manifest, mode)
    chatbot.append((f"定位主Latex文件", f'[Local Message] 分析结果：该项目的Latex主文件是{maintex}, 如果分析错误, 请立即终止程序, 删除或修改歧义文件, 然后重试。主程序即将开始, 请稍候。'))
    yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
    time.sleep(3)
@ -668,54 +282,51 @@ def Latex精细分解与转化(file_manifest, project_folder, llm_kwargs, plugin
    #  <-------- 写出文件 ----------> 
    msg = f"当前大语言模型: {llm_kwargs['llm_model']}，当前语言模型温度设定: {llm_kwargs['temperature']}。"
    final_tex = lps.merge_result(pfg.file_result, mode, msg)
+    objdump((lps, pfg.file_result, mode, msg), file=pj(project_folder,'merge_result.pkl'))
+
    with open(project_folder + f'/merge_{mode}.tex', 'w', encoding='utf-8', errors='replace') as f:
        if mode != 'translate_zh' or "binary" in final_tex: f.write(final_tex)
        

    #  <-------- 整理结果, 退出 ----------> 
-    chatbot.append((f"完成了吗？", 'GPT结果已输出, 正在编译PDF'))
+    chatbot.append((f"完成了吗？", 'GPT结果已输出, 即将编译PDF'))
    yield from update_ui(chatbot=chatbot, history=history) # 刷新界面

    #  <-------- 返回 ----------> 
    return project_folder + f'/merge_{mode}.tex'


-
-def remove_buggy_lines(file_path, log_path, tex_name, tex_name_pure, n_fix, work_folder_modified):
+def remove_buggy_lines(file_path, log_path, tex_name, tex_name_pure, n_fix, work_folder_modified, fixed_line=[]):
    try:
        with open(log_path, 'r', encoding='utf-8', errors='replace') as f:
            log = f.read()
-        with open(file_path, 'r', encoding='utf-8', errors='replace') as f:
-            file_lines = f.readlines()
        import re
        buggy_lines = re.findall(tex_name+':([0-9]{1,5}):', log)
        buggy_lines = [int(l) for l in buggy_lines]
        buggy_lines = sorted(buggy_lines)
-        print("removing lines that has errors", buggy_lines)
-        file_lines.pop(buggy_lines[0]-1)
+        buggy_line = buggy_lines[0]-1
+        print("reversing tex line that has errors", buggy_line)
+
+        # 重组，逆转出错的段落
+        if buggy_line not in fixed_line:
+            fixed_line.append(buggy_line)
+
+        lps, file_result, mode, msg = objload(file=pj(work_folder_modified,'merge_result.pkl'))
+        final_tex = lps.merge_result(file_result, mode, msg, buggy_lines=fixed_line, buggy_line_surgery_n_lines=5*n_fix)
+
        with open(pj(work_folder_modified, f"{tex_name_pure}_fix_{n_fix}.tex"), 'w', encoding='utf-8', errors='replace') as f:
-            f.writelines(file_lines)
+            f.write(final_tex)
+
        return True, f"{tex_name_pure}_fix_{n_fix}", buggy_lines
    except:
        print("Fatal error occurred, but we cannot identify error, please download zip, read latex log, and compile manually.")
        return False, -1, [-1]
-    
-def compile_latex_with_timeout(command, cwd, timeout=60):
-    import subprocess
-    process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, cwd=cwd)
-    try:
-        stdout, stderr = process.communicate(timeout=timeout)
-    except subprocess.TimeoutExpired:
-        process.kill()
-        stdout, stderr = process.communicate()
-        print("Process timed out!")
-        return False
-    return True
+

 def 编译Latex(chatbot, history, main_file_original, main_file_modified, work_folder_original, work_folder_modified, work_folder, mode='default'):
    import os, time
-    current_dir = os.getcwd()
    n_fix = 1
+    fixed_line = []
    max_try = 32
    chatbot.append([f"正在编译PDF文档", f'编译已经开始。当前工作路径为{work_folder}，如果程序停顿5分钟以上，请直接去该路径下取回翻译结果，或者重启之后再度尝试 ...']); yield from update_ui(chatbot=chatbot, history=history)
    chatbot.append([f"正在编译PDF文档", '...']); yield from update_ui(chatbot=chatbot, history=history); time.sleep(1); chatbot[-1] = list(chatbot[-1]) # 刷新界面
@ -723,6 +334,10 @@ def 编译Latex(chatbot, history, main_file_original, main_file_modified, work_f

    while True:
        import os
+        may_exist_bbl = pj(work_folder_modified, f'merge.bbl')
+        target_bbl = pj(work_folder_modified, f'{main_file_modified}.bbl')
+        if os.path.exists(may_exist_bbl) and not os.path.exists(target_bbl):
+            shutil.copyfile(may_exist_bbl, target_bbl)

        # https://stackoverflow.com/questions/738755/dont-make-me-manually-abort-a-latex-compile-when-theres-an-error
        yield from update_ui_lastest_msg(f'尝试第 {n_fix}/{max_try} 次编译, 编译原始PDF ...', chatbot, history)   # 刷新Gradio前端界面
@ -756,7 +371,6 @@ def 编译Latex(chatbot, history, main_file_original, main_file_modified, work_f
                ok = compile_latex_with_timeout(f'pdflatex  -interaction=batchmode -file-line-error merge_diff.tex', work_folder)
                ok = compile_latex_with_timeout(f'pdflatex  -interaction=batchmode -file-line-error merge_diff.tex', work_folder)

-
        # <---------- 检查结果 ----------->
        results_ = ""
        original_pdf_success = os.path.exists(pj(work_folder_original, f'{main_file_original}.pdf'))
@ -773,9 +387,19 @@ def 编译Latex(chatbot, history, main_file_original, main_file_modified, work_f
        if modified_pdf_success:
            yield from update_ui_lastest_msg(f'转化PDF编译已经成功, 即将退出 ...', chatbot, history)    # 刷新Gradio前端界面
            result_pdf = pj(work_folder_modified, f'{main_file_modified}.pdf') # get pdf path
+            origin_pdf = pj(work_folder_original, f'{main_file_original}.pdf') # get pdf path
            if os.path.exists(pj(work_folder, '..', 'translation')):
                shutil.copyfile(result_pdf, pj(work_folder, '..', 'translation', 'translate_zh.pdf'))
            promote_file_to_downloadzone(result_pdf, rename_file=None, chatbot=chatbot)  # promote file to web UI
+            # 将两个PDF拼接
+            if original_pdf_success: 
+                try:
+                    from .latex_toolbox import merge_pdfs
+                    concat_pdf = pj(work_folder_modified, f'comparison.pdf')
+                    merge_pdfs(origin_pdf, result_pdf, concat_pdf)
+                    promote_file_to_downloadzone(concat_pdf, rename_file=None, chatbot=chatbot)  # promote file to web UI
+                except Exception as e:
+                    pass
            return True # 成功啦
        else:
            if n_fix>=max_try: break
@ -787,6 +411,7 @@ def 编译Latex(chatbot, history, main_file_original, main_file_modified, work_f
                tex_name_pure=f'{main_file_modified}',
                n_fix=n_fix,
                work_folder_modified=work_folder_modified,
+                fixed_line=fixed_line
            )
            yield from update_ui_lastest_msg(f'由于最为关键的转化PDF编译失败, 将根据报错信息修正tex源文件并重试, 当前报错的latex代码处于第{buggy_lines}行 ...', chatbot, history)   # 刷新Gradio前端界面
            if not can_retry: break
@ -794,4 +419,29 @@ def 编译Latex(chatbot, history, main_file_original, main_file_modified, work_f
    return False # 失败啦


-
+def write_html(sp_file_contents, sp_file_result, chatbot, project_folder):
+    # write html
+    try:
+        import shutil
+        from ..crazy_utils import construct_html
+        from toolbox import gen_time_str
+        ch = construct_html() 
+        orig = ""
+        trans = ""
+        final = []
+        for c,r in zip(sp_file_contents, sp_file_result): 
+            final.append(c)
+            final.append(r)
+        for i, k in enumerate(final): 
+            if i%2==0:
+                orig = k
+            if i%2==1:
+                trans = k
+                ch.add_row(a=orig, b=trans)
+        create_report_file_name = f"{gen_time_str()}.trans.html"
+        ch.save_file(create_report_file_name)
+        shutil.copyfile(pj('./gpt_log/', create_report_file_name), pj(project_folder, create_report_file_name))
+        promote_file_to_downloadzone(file=f'./gpt_log/{create_report_file_name}', chatbot=chatbot)
+    except:
+        from toolbox import trimmed_format_exc
+        print('writing html result failed:', trimmed_format_exc())
--- a/crazy_functions/latex_fns/latex_toolbox.py
+++ b/crazy_functions/latex_fns/latex_toolbox.py
@ -0,0 +1,456 @@
+import os, shutil
+import re
+import numpy as np
+PRESERVE = 0
+TRANSFORM = 1
+
+pj = os.path.join
+
+class LinkedListNode():
+    """
+    Linked List Node
+    """
+    def __init__(self, string, preserve=True) -> None:
+        self.string = string
+        self.preserve = preserve
+        self.next = None
+        self.range = None
+        # self.begin_line = 0
+        # self.begin_char = 0
+
+def convert_to_linklist(text, mask):
+    root = LinkedListNode("", preserve=True)
+    current_node = root
+    for c, m, i in zip(text, mask, range(len(text))):
+        if (m==PRESERVE and current_node.preserve) \
+            or (m==TRANSFORM and not current_node.preserve):
+            # add
+            current_node.string += c
+        else:
+            current_node.next = LinkedListNode(c, preserve=(m==PRESERVE))
+            current_node = current_node.next
+    return root
+
+def post_process(root):
+    # 修复括号
+    node = root
+    while True:
+        string = node.string
+        if node.preserve: 
+            node = node.next
+            if node is None: break
+            continue
+        def break_check(string):
+            str_stack = [""] # (lv, index)
+            for i, c in enumerate(string):
+                if c == '{':
+                    str_stack.append('{')
+                elif c == '}':
+                    if len(str_stack) == 1:
+                        print('stack fix')
+                        return i
+                    str_stack.pop(-1)
+                else:
+                    str_stack[-1] += c
+            return -1
+        bp = break_check(string)
+
+        if bp == -1:
+            pass
+        elif bp == 0:
+            node.string = string[:1]
+            q = LinkedListNode(string[1:], False)
+            q.next = node.next
+            node.next = q
+        else:
+            node.string = string[:bp]
+            q = LinkedListNode(string[bp:], False)
+            q.next = node.next
+            node.next = q
+
+        node = node.next
+        if node is None: break
+
+    # 屏蔽空行和太短的句子
+    node = root
+    while True:
+        if len(node.string.strip('\n').strip(''))==0: node.preserve = True
+        if len(node.string.strip('\n').strip(''))<42: node.preserve = True
+        node = node.next
+        if node is None: break
+    node = root
+    while True:
+        if node.next and node.preserve and node.next.preserve:
+            node.string += node.next.string
+            node.next = node.next.next
+        node = node.next
+        if node is None: break
+
+    # 将前后断行符脱离
+    node = root
+    prev_node = None
+    while True:
+        if not node.preserve:
+            lstriped_ = node.string.lstrip().lstrip('\n')
+            if (prev_node is not None) and (prev_node.preserve) and (len(lstriped_)!=len(node.string)):
+                prev_node.string += node.string[:-len(lstriped_)]
+                node.string = lstriped_
+            rstriped_ = node.string.rstrip().rstrip('\n')
+            if (node.next is not None) and (node.next.preserve) and (len(rstriped_)!=len(node.string)):
+                node.next.string = node.string[len(rstriped_):] + node.next.string
+                node.string = rstriped_
+        # =====
+        prev_node = node
+        node = node.next
+        if node is None: break
+
+    # 标注节点的行数范围
+    node = root
+    n_line = 0
+    expansion = 2
+    while True:
+        n_l = node.string.count('\n')
+        node.range = [n_line-expansion, n_line+n_l+expansion]   # 失败时，扭转的范围
+        n_line = n_line+n_l
+        node = node.next
+        if node is None: break
+    return root
+
+
+"""
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+Latex segmentation with a binary mask (PRESERVE=0, TRANSFORM=1)
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+"""
+
+
+def set_forbidden_text(text, mask, pattern, flags=0):
+    """
+    Add a preserve text area in this paper
+    e.g. with pattern = r"\\begin\{algorithm\}(.*?)\\end\{algorithm\}"
+    you can mask out (mask = PRESERVE so that text become untouchable for GPT) 
+    everything between "\begin{equation}" and "\end{equation}"
+    """
+    if isinstance(pattern, list): pattern = '|'.join(pattern)
+    pattern_compile = re.compile(pattern, flags)
+    for res in pattern_compile.finditer(text):
+        mask[res.span()[0]:res.span()[1]] = PRESERVE
+    return text, mask
+
+def reverse_forbidden_text(text, mask, pattern, flags=0, forbid_wrapper=True):
+    """
+    Move area out of preserve area (make text editable for GPT)
+    count the number of the braces so as to catch compelete text area. 
+    e.g.
+    \begin{abstract} blablablablablabla. \end{abstract} 
+    """
+    if isinstance(pattern, list): pattern = '|'.join(pattern)
+    pattern_compile = re.compile(pattern, flags)
+    for res in pattern_compile.finditer(text):
+        if not forbid_wrapper:
+            mask[res.span()[0]:res.span()[1]] = TRANSFORM
+        else:
+            mask[res.regs[0][0]: res.regs[1][0]] = PRESERVE   # '\\begin{abstract}'
+            mask[res.regs[1][0]: res.regs[1][1]] = TRANSFORM   # abstract
+            mask[res.regs[1][1]: res.regs[0][1]] = PRESERVE   # abstract
+    return text, mask
+
+def set_forbidden_text_careful_brace(text, mask, pattern, flags=0):
+    """
+    Add a preserve text area in this paper (text become untouchable for GPT).
+    count the number of the braces so as to catch compelete text area. 
+    e.g.
+    \caption{blablablablabla\texbf{blablabla}blablabla.} 
+    """
+    pattern_compile = re.compile(pattern, flags)
+    for res in pattern_compile.finditer(text):
+        brace_level = -1
+        p = begin = end = res.regs[0][0]
+        for _ in range(1024*16):
+            if text[p] == '}' and brace_level == 0: break
+            elif text[p] == '}':  brace_level -= 1
+            elif text[p] == '{':  brace_level += 1
+            p += 1
+        end = p+1
+        mask[begin:end] = PRESERVE
+    return text, mask
+
+def reverse_forbidden_text_careful_brace(text, mask, pattern, flags=0, forbid_wrapper=True):
+    """
+    Move area out of preserve area (make text editable for GPT)
+    count the number of the braces so as to catch compelete text area. 
+    e.g.
+    \caption{blablablablabla\texbf{blablabla}blablabla.} 
+    """
+    pattern_compile = re.compile(pattern, flags)
+    for res in pattern_compile.finditer(text):
+        brace_level = 0
+        p = begin = end = res.regs[1][0]
+        for _ in range(1024*16):
+            if text[p] == '}' and brace_level == 0: break
+            elif text[p] == '}':  brace_level -= 1
+            elif text[p] == '{':  brace_level += 1
+            p += 1
+        end = p
+        mask[begin:end] = TRANSFORM
+        if forbid_wrapper:
+            mask[res.regs[0][0]:begin] = PRESERVE
+            mask[end:res.regs[0][1]] = PRESERVE
+    return text, mask
+
+def set_forbidden_text_begin_end(text, mask, pattern, flags=0, limit_n_lines=42):
+    """
+    Find all \begin{} ... \end{} text block that with less than limit_n_lines lines.
+    Add it to preserve area
+    """
+    pattern_compile = re.compile(pattern, flags)
+    def search_with_line_limit(text, mask):
+        for res in pattern_compile.finditer(text):
+            cmd = res.group(1)  # begin{what}
+            this = res.group(2) # content between begin and end
+            this_mask = mask[res.regs[2][0]:res.regs[2][1]]
+            white_list = ['document', 'abstract', 'lemma', 'definition', 'sproof', 
+                          'em', 'emph', 'textit', 'textbf', 'itemize', 'enumerate']
+            if (cmd in white_list) or this.count('\n') >= limit_n_lines: # use a magical number 42
+                this, this_mask = search_with_line_limit(this, this_mask)
+                mask[res.regs[2][0]:res.regs[2][1]] = this_mask
+            else:
+                mask[res.regs[0][0]:res.regs[0][1]] = PRESERVE
+        return text, mask
+    return search_with_line_limit(text, mask) 
+
+
+
+"""
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+Latex Merge File
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+"""
+
+def find_main_tex_file(file_manifest, mode):
+    """
+    在多Tex文档中，寻找主文件，必须包含documentclass，返回找到的第一个。
+    P.S. 但愿没人把latex模板放在里面传进来 (6.25 加入判定latex模板的代码)
+    """
+    canidates = []
+    for texf in file_manifest:
+        if os.path.basename(texf).startswith('merge'):
+            continue
+        with open(texf, 'r', encoding='utf8', errors='ignore') as f:
+            file_content = f.read()
+        if r'\documentclass' in file_content:
+            canidates.append(texf)
+        else:
+            continue
+
+    if len(canidates) == 0:
+        raise RuntimeError('无法找到一个主Tex文件（包含documentclass关键字）')
+    elif len(canidates) == 1:
+        return canidates[0]
+    else: # if len(canidates) >= 2 通过一些Latex模板中常见（但通常不会出现在正文）的单词，对不同latex源文件扣分，取评分最高者返回
+        canidates_score = []
+        # 给出一些判定模板文档的词作为扣分项
+        unexpected_words = ['\LaTeX', 'manuscript', 'Guidelines', 'font', 'citations', 'rejected', 'blind review', 'reviewers']
+        expected_words = ['\input', '\ref', '\cite']
+        for texf in canidates:
+            canidates_score.append(0)
+            with open(texf, 'r', encoding='utf8', errors='ignore') as f:
+                file_content = f.read()
+            for uw in unexpected_words:
+                if uw in file_content:
+                    canidates_score[-1] -= 1
+            for uw in expected_words:
+                if uw in file_content:
+                    canidates_score[-1] += 1
+        select = np.argmax(canidates_score) # 取评分最高者返回
+        return canidates[select]
+    
+def rm_comments(main_file):
+    new_file_remove_comment_lines = []
+    for l in main_file.splitlines():
+        # 删除整行的空注释
+        if l.lstrip().startswith("%"):
+            pass
+        else:
+            new_file_remove_comment_lines.append(l)
+    main_file = '\n'.join(new_file_remove_comment_lines)
+    # main_file = re.sub(r"\\include{(.*?)}", r"\\input{\1}", main_file)  # 将 \include 命令转换为 \input 命令
+    main_file = re.sub(r'(?<!\\)%.*', '', main_file)  # 使用正则表达式查找半行注释, 并替换为空字符串
+    return main_file
+
+def find_tex_file_ignore_case(fp):
+    dir_name = os.path.dirname(fp)
+    base_name = os.path.basename(fp)
+    if not base_name.endswith('.tex'): base_name+='.tex'
+    if os.path.exists(pj(dir_name, base_name)): return pj(dir_name, base_name)
+    # go case in-sensitive
+    import glob
+    for f in glob.glob(dir_name+'/*.tex'):
+        base_name_s = os.path.basename(fp)
+        if base_name_s.lower() == base_name.lower(): return f
+    return None
+
+def merge_tex_files_(project_foler, main_file, mode):
+    """
+    Merge Tex project recrusively
+    """
+    main_file = rm_comments(main_file)
+    for s in reversed([q for q in re.finditer(r"\\input\{(.*?)\}", main_file, re.M)]):
+        f = s.group(1)
+        fp = os.path.join(project_foler, f)
+        fp = find_tex_file_ignore_case(fp)
+        if fp:
+            with open(fp, 'r', encoding='utf-8', errors='replace') as fx: c = fx.read()
+        else:
+            raise RuntimeError(f'找不到{fp}，Tex源文件缺失！')
+        c = merge_tex_files_(project_foler, c, mode)
+        main_file = main_file[:s.span()[0]] + c + main_file[s.span()[1]:]
+    return main_file
+
+def merge_tex_files(project_foler, main_file, mode):
+    """
+    Merge Tex project recrusively
+    P.S. 顺便把CTEX塞进去以支持中文
+    P.S. 顺便把Latex的注释去除
+    """
+    main_file = merge_tex_files_(project_foler, main_file, mode)
+    main_file = rm_comments(main_file)
+
+    if mode == 'translate_zh':
+        # find paper documentclass
+        pattern = re.compile(r'\\documentclass.*\n')
+        match = pattern.search(main_file)
+        assert match is not None, "Cannot find documentclass statement!"
+        position = match.end()
+        add_ctex = '\\usepackage{ctex}\n'
+        add_url = '\\usepackage{url}\n' if '{url}' not in main_file else ''
+        main_file = main_file[:position] + add_ctex + add_url + main_file[position:]
+        # fontset=windows
+        import platform
+        main_file = re.sub(r"\\documentclass\[(.*?)\]{(.*?)}", r"\\documentclass[\1,fontset=windows,UTF8]{\2}",main_file)
+        main_file = re.sub(r"\\documentclass{(.*?)}", r"\\documentclass[fontset=windows,UTF8]{\1}",main_file)
+        # find paper abstract
+        pattern_opt1 = re.compile(r'\\begin\{abstract\}.*\n')
+        pattern_opt2 = re.compile(r"\\abstract\{(.*?)\}", flags=re.DOTALL)
+        match_opt1 = pattern_opt1.search(main_file)
+        match_opt2 = pattern_opt2.search(main_file)
+        assert (match_opt1 is not None) or (match_opt2 is not None), "Cannot find paper abstract section!"
+    return main_file
+
+
+"""
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+Post process
+=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
+"""
+def mod_inbraket(match):
+    """
+    为啥chatgpt会把cite里面的逗号换成中文逗号呀 
+    """
+    # get the matched string
+    cmd = match.group(1)
+    str_to_modify = match.group(2)
+    # modify the matched string
+    str_to_modify = str_to_modify.replace('：', ':')    # 前面是中文冒号，后面是英文冒号
+    str_to_modify = str_to_modify.replace('，', ',')    # 前面是中文逗号，后面是英文逗号
+    # str_to_modify = 'BOOM'
+    return "\\" + cmd + "{" + str_to_modify + "}"
+
+def fix_content(final_tex, node_string):
+    """
+    Fix common GPT errors to increase success rate
+    """
+    final_tex = re.sub(r"(?<!\\)%", "\\%", final_tex)
+    final_tex = re.sub(r"\\([a-z]{2,10})\ \{", r"\\\1{", string=final_tex)
+    final_tex = re.sub(r"\\\ ([a-z]{2,10})\{", r"\\\1{", string=final_tex)
+    final_tex = re.sub(r"\\([a-z]{2,10})\{([^\}]*?)\}", mod_inbraket, string=final_tex)
+
+    if "Traceback" in final_tex and "[Local Message]" in final_tex:
+        final_tex = node_string # 出问题了，还原原文
+    if node_string.count('\\begin') != final_tex.count('\\begin'):
+        final_tex = node_string # 出问题了，还原原文
+    if node_string.count('\_') > 0 and node_string.count('\_') > final_tex.count('\_'):
+        # walk and replace any _ without \
+        final_tex = re.sub(r"(?<!\\)_", "\\_", final_tex)
+
+    def compute_brace_level(string):
+        # this function count the number of { and }
+        brace_level = 0
+        for c in string:
+            if c == "{": brace_level += 1
+            elif c == "}": brace_level -= 1
+        return brace_level
+    def join_most(tex_t, tex_o):
+        # this function join translated string and original string when something goes wrong
+        p_t = 0
+        p_o = 0
+        def find_next(string, chars, begin):
+            p = begin
+            while p < len(string):
+                if string[p] in chars: return p, string[p]
+                p += 1
+            return None, None
+        while True:
+            res1, char = find_next(tex_o, ['{','}'], p_o)
+            if res1 is None: break
+            res2, char = find_next(tex_t, [char], p_t)
+            if res2 is None: break
+            p_o = res1 + 1
+            p_t = res2 + 1
+        return tex_t[:p_t] + tex_o[p_o:]
+
+    if compute_brace_level(final_tex) != compute_brace_level(node_string):
+        # 出问题了，还原部分原文，保证括号正确
+        final_tex = join_most(final_tex, node_string)
+    return final_tex
+    
+def compile_latex_with_timeout(command, cwd, timeout=60):
+    import subprocess
+    process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE, cwd=cwd)
+    try:
+        stdout, stderr = process.communicate(timeout=timeout)
+    except subprocess.TimeoutExpired:
+        process.kill()
+        stdout, stderr = process.communicate()
+        print("Process timed out!")
+        return False
+    return True
+
+
+
+def merge_pdfs(pdf1_path, pdf2_path, output_path):
+    import PyPDF2
+    Percent = 0.8
+    # Open the first PDF file
+    with open(pdf1_path, 'rb') as pdf1_file:
+        pdf1_reader = PyPDF2.PdfFileReader(pdf1_file)
+        # Open the second PDF file
+        with open(pdf2_path, 'rb') as pdf2_file:
+            pdf2_reader = PyPDF2.PdfFileReader(pdf2_file)
+            # Create a new PDF file to store the merged pages
+            output_writer = PyPDF2.PdfFileWriter()
+            # Determine the number of pages in each PDF file
+            num_pages = max(pdf1_reader.numPages, pdf2_reader.numPages)
+            # Merge the pages from the two PDF files
+            for page_num in range(num_pages):
+                # Add the page from the first PDF file
+                if page_num < pdf1_reader.numPages:
+                    page1 = pdf1_reader.getPage(page_num)
+                else:
+                    page1 = PyPDF2.PageObject.createBlankPage(pdf1_reader)
+                # Add the page from the second PDF file
+                if page_num < pdf2_reader.numPages:
+                    page2 = pdf2_reader.getPage(page_num)
+                else:
+                    page2 = PyPDF2.PageObject.createBlankPage(pdf1_reader)
+                # Create a new empty page with double width
+                new_page = PyPDF2.PageObject.createBlankPage(
+                    width = int(int(page1.mediaBox.getWidth()) + int(page2.mediaBox.getWidth()) * Percent),
+                    height = max(page1.mediaBox.getHeight(), page2.mediaBox.getHeight())
+                )
+                new_page.mergeTranslatedPage(page1, 0, 0)
+                new_page.mergeTranslatedPage(page2, int(int(page1.mediaBox.getWidth())-int(page2.mediaBox.getWidth())* (1-Percent)), 0)
+                output_writer.addPage(new_page)
+            # Save the merged PDF file
+            with open(output_path, 'wb') as output_file:
+                output_writer.write(output_file)
--- a/crazy_functions/live_audio/aliyunASR.py
+++ b/crazy_functions/live_audio/aliyunASR.py
@ -19,7 +19,7 @@ class AliyunASR():
        pass

    def test_on_error(self, message, *args):
-        # print("on_error args=>{}".format(args))
+        print("on_error args=>{}".format(args))
        pass

    def test_on_close(self, *args):
@ -50,6 +50,8 @@ class AliyunASR():
        rad.clean_up()
        temp_folder = tempfile.gettempdir()
        TOKEN, APPKEY = get_conf('ALIYUN_TOKEN', 'ALIYUN_APPKEY')
+        if len(TOKEN) == 0:
+            TOKEN = self.get_token()
        self.aliyun_service_ok = True
        URL="wss://nls-gateway.aliyuncs.com/ws/v1"
        sr = nls.NlsSpeechTranscriber(
@ -91,3 +93,38 @@ class AliyunASR():
                self.stop = True
                self.stop_msg = 'Aliyun音频服务异常，请检查ALIYUN_TOKEN和ALIYUN_APPKEY是否过期。'
        r = sr.stop()
+
+    def get_token(self):
+        from toolbox import get_conf
+        import json
+        from aliyunsdkcore.request import CommonRequest
+        from aliyunsdkcore.client import AcsClient
+        AccessKey_ID, AccessKey_secret = get_conf('ALIYUN_ACCESSKEY', 'ALIYUN_SECRET')
+
+        # 创建AcsClient实例
+        client = AcsClient(
+            AccessKey_ID,
+            AccessKey_secret,
+            "cn-shanghai"
+        )
+
+        # 创建request，并设置参数。
+        request = CommonRequest()
+        request.set_method('POST')
+        request.set_domain('nls-meta.cn-shanghai.aliyuncs.com')
+        request.set_version('2019-02-28')
+        request.set_action_name('CreateToken')
+
+        try:
+            response = client.do_action_with_exception(request)
+            print(response)
+            jss = json.loads(response)
+            if 'Token' in jss and 'Id' in jss['Token']:
+                token = jss['Token']['Id']
+                expireTime = jss['Token']['ExpireTime']
+                print("token = " + token)
+                print("expireTime = " + str(expireTime))
+        except Exception as e:
+            print(e)
+
+        return token
--- a/crazy_functions/语音助手.py
+++ b/crazy_functions/语音助手.py
@ -179,12 +179,12 @@ def 语音助手(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt
        import nls
        from scipy import io
    except:
-        chatbot.append(["导入依赖失败", "使用该模块需要额外依赖, 安装方法:```pip install --upgrade pyOpenSSL scipy git+https://github.com/aliyun/alibabacloud-nls-python-sdk.git```"])
+        chatbot.append(["导入依赖失败", "使用该模块需要额外依赖, 安装方法:```pip install --upgrade aliyun-python-sdk-core==2.13.3 pyOpenSSL scipy git+https://github.com/aliyun/alibabacloud-nls-python-sdk.git```"])
        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
        return

-    TOKEN, APPKEY = get_conf('ALIYUN_TOKEN', 'ALIYUN_APPKEY')
-    if TOKEN == "" or APPKEY == "":
+    APPKEY = get_conf('ALIYUN_APPKEY')
+    if APPKEY == "":
        chatbot.append(["导入依赖失败", "没有阿里云语音识别APPKEY和TOKEN, 详情见https://help.aliyun.com/document_detail/450255.html"])
        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
        return
--- a/docker-compose.yml
+++ b/docker-compose.yml
@ -115,3 +115,36 @@ services:
    command: >
      bash -c "python3 -u main.py"

+
+## ===================================================
+## 【方案五】 ChatGPT + 语音助手 （请先阅读 docs/use_audio.md）
+## ===================================================
+version: '3'
+services:
+  gpt_academic_with_audio:
+    image: ghcr.io/binary-husky/gpt_academic_audio_assistant:master
+    environment:
+      # 请查阅 `config.py` 以查看所有的配置信息
+      API_KEY:                  '    fk195831-IdP0Pb3W6DCMUIbQwVX6MsSiyxwqybyS                        '
+      USE_PROXY:                '    False                                                            '
+      proxies:                  '    None                                                             '
+      LLM_MODEL:                '    gpt-3.5-turbo                                                    '
+      AVAIL_LLM_MODELS:         '    ["gpt-3.5-turbo", "gpt-4"]                                       '
+      ENABLE_AUDIO:             '    True                                                             '
+      LOCAL_MODEL_DEVICE:       '    cuda                                                             '
+      DEFAULT_WORKER_NUM:       '    20                                                               '
+      WEB_PORT:                 '    12343                                                            '
+      ADD_WAIFU:                '    True                                                             '
+      THEME:                    '    Chuanhu-Small-and-Beautiful                                      '
+      ALIYUN_APPKEY:            '    RoP1ZrM84DnAFkZK                                                 '
+      ALIYUN_TOKEN:             '    f37f30e0f9934c34a992f6f64f7eba4f                                 '
+      # (无需填写) ALIYUN_ACCESSKEY:         '    LTAI5q6BrFUzoRXVGUWnekh1                                         '
+      # (无需填写) ALIYUN_SECRET:            '    eHmI20AVWIaQZ0CiTD2bGQVsaP9i68                                   '
+
+    # 与宿主的网络融合
+    network_mode: "host"
+
+    # 不使用代理网络拉取最新代码
+    command: >
+      bash -c "python3 -u main.py"
+
--- a/docs/GithubAction+NoLocal+AudioAssistant
+++ b/docs/GithubAction+NoLocal+AudioAssistant
@ -0,0 +1,22 @@
+# 此Dockerfile适用于“无本地模型”的环境构建，如果需要使用chatglm等本地模型，请参考 docs/Dockerfile+ChatGLM
+# 如何构建: 先修改 `config.py`， 然后 docker build -t gpt-academic-nolocal -f docs/Dockerfile+NoLocal .
+# 如何运行: docker run --rm -it --net=host gpt-academic-nolocal
+FROM python:3.11
+
+# 指定路径
+WORKDIR /gpt
+
+# 装载项目文件
+COPY . .
+
+# 安装依赖
+RUN pip3 install -r requirements.txt
+
+# 安装语音插件的额外依赖
+RUN pip3 install pyOpenSSL scipy git+https://github.com/aliyun/alibabacloud-nls-python-sdk.git
+
+# 可选步骤，用于预热模块
+RUN python3  -c 'from check_proxy import warm_up_modules; warm_up_modules()'
+
+# 启动
+CMD ["python3", "-u", "main.py"]
--- a/docs/use_audio.md
+++ b/docs/use_audio.md
@ -28,6 +28,16 @@ ALIYUN_APPKEY = "RoPlZrM88DnAFkZK"   # 此appkey已经失效
 参考 https://help.aliyun.com/document_detail/450255.html
 先有阿里云开发者账号，登录之后，需要开通 智能语音交互 的功能，可以免费获得一个token，然后在 全部项目 中，创建一个项目，可以获得一个appkey.

+- 进阶功能
+进一步填写ALIYUN_ACCESSKEY和ALIYUN_SECRET实现自动获取ALIYUN_TOKEN
+```
+ALIYUN_APPKEY = "RoP1ZrM84DnAFkZK"
+ALIYUN_TOKEN = ""
+ALIYUN_ACCESSKEY = "LTAI5q6BrFUzoRXVGUWnekh1"
+ALIYUN_SECRET = "eHmI20AVWIaQZ0CiTD2bGQVsaP9i68"
+```
+
+
 ## 3.启动

 启动gpt-academic `python main.py`
@ -48,7 +58,7 @@ III `[把特殊软件（如腾讯会议）的外放声音用VoiceMeeter截留]`

 VI 两种音频监听模式切换时，需要刷新页面才有效。

+VII 非localhost运行+非https情况下无法打开录音功能的坑：https://blog.csdn.net/weixin_39461487/article/details/109594434
+
 ## 5.点击函数插件区“实时音频采集” 或者其他音频交互功能

-
-
--- a/request_llm/bridge_chatglm.py
+++ b/request_llm/bridge_chatglm.py
@ -37,15 +37,23 @@ class GetGLMHandle(Process):
        # 子进程执行
        # 第一次运行，加载参数
        retry = 0
+        LOCAL_MODEL_QUANT, device = get_conf('LOCAL_MODEL_QUANT', 'LOCAL_MODEL_DEVICE')
+
+        if LOCAL_MODEL_QUANT == "INT4":         # INT4
+            _model_name_ = "THUDM/chatglm2-6b-int4"
+        elif LOCAL_MODEL_QUANT == "INT8":       # INT8
+            _model_name_ = "THUDM/chatglm2-6b-int8"
+        else:
+            _model_name_ = "THUDM/chatglm2-6b"  # FP16
+
        while True:
            try:
                if self.chatglm_model is None:
-                    self.chatglm_tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)
-                    device, = get_conf('LOCAL_MODEL_DEVICE')
+                    self.chatglm_tokenizer = AutoTokenizer.from_pretrained(_model_name_, trust_remote_code=True)
                    if device=='cpu':
-                        self.chatglm_model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).float()
+                        self.chatglm_model = AutoModel.from_pretrained(_model_name_, trust_remote_code=True).float()
                    else:
-                        self.chatglm_model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).half().cuda()
+                        self.chatglm_model = AutoModel.from_pretrained(_model_name_, trust_remote_code=True).half().cuda()
                    self.chatglm_model = self.chatglm_model.eval()
                    break
                else:
@ -136,11 +144,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
            return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    # 处理历史信息
    history_feedin = []
--- a/request_llm/bridge_chatglmft.py
+++ b/request_llm/bridge_chatglmft.py
@ -185,11 +185,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
            return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    # 处理历史信息
    history_feedin = []
--- a/request_llm/bridge_chatgpt.py
+++ b/request_llm/bridge_chatgpt.py
@ -129,11 +129,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
        return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    raw_input = inputs
    logging.info(f'[raw_input] {raw_input}')
--- a/request_llm/bridge_claude.py
+++ b/request_llm/bridge_claude.py
@ -116,11 +116,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
        return
    
    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    raw_input = inputs
    logging.info(f'[raw_input] {raw_input}')
--- a/request_llm/bridge_internlm.py
+++ b/request_llm/bridge_internlm.py
@ -290,11 +290,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
        return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    # 处理历史信息
    history_feedin = []
--- a/request_llm/bridge_jittorllms_llama.py
+++ b/request_llm/bridge_jittorllms_llama.py
@ -154,11 +154,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
            return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    # 处理历史信息
    history_feedin = []
--- a/request_llm/bridge_jittorllms_pangualpha.py
+++ b/request_llm/bridge_jittorllms_pangualpha.py
@ -154,11 +154,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
            return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    # 处理历史信息
    history_feedin = []
--- a/request_llm/bridge_jittorllms_rwkv.py
+++ b/request_llm/bridge_jittorllms_rwkv.py
@ -154,11 +154,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
            return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    # 处理历史信息
    history_feedin = []
--- a/request_llm/bridge_moss.py
+++ b/request_llm/bridge_moss.py
@ -224,11 +224,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
        yield from update_ui(chatbot=chatbot, history=history)

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    # 处理历史信息
    history_feedin = []
--- a/request_llm/bridge_newbingfree.py
+++ b/request_llm/bridge_newbingfree.py
@ -224,11 +224,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
            return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    history_feedin = []
    for i in range(len(history)//2):
--- a/request_llm/bridge_stackclaude.py
+++ b/request_llm/bridge_stackclaude.py
@ -248,14 +248,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
            return

    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]:
-            inputs = core_functional[additional_fn]["PreProcess"](
-                inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + \
-            inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    history_feedin = []
    for i in range(len(history)//2):
--- a/request_llm/bridge_tgui.py
+++ b/request_llm/bridge_tgui.py
@ -96,11 +96,8 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
        additional_fn代表点击的哪个按钮，按钮见functional.py
    """
    if additional_fn is not None:
-        import core_functional
-        importlib.reload(core_functional)    # 热更新prompt
-        core_functional = core_functional.get_core_functions()
-        if "PreProcess" in core_functional[additional_fn]: inputs = core_functional[additional_fn]["PreProcess"](inputs)  # 获取预处理函数（如果有的话）
-        inputs = core_functional[additional_fn]["Prefix"] + inputs + core_functional[additional_fn]["Suffix"]
+        from core_functional import handle_core_functionality
+        inputs, history = handle_core_functionality(additional_fn, inputs, history)

    raw_input = "What I would like to say is the following: " + inputs
    history.extend([inputs, ""])
--- a/request_llm/edge_gpt_free.py
+++ b/request_llm/edge_gpt_free.py
@ -519,7 +519,11 @@ class _ChatHub:
        resp_txt_no_link = ""
        while not final:
            msg = await self.wss.receive()
-            objects = msg.data.split(DELIMITER)
+            try:
+                objects = msg.data.split(DELIMITER)
+            except :
+                continue
+            
            for obj in objects:
                if obj is None or not obj:
                    continue
--- a/requirements.txt
+++ b/requirements.txt
@ -18,3 +18,4 @@ openai
 numpy
 arxiv
 rich
+pypdf2==2.12.1
--- a/setup.py
+++ b/setup.py
@ -0,0 +1,50 @@
+import setuptools, glob, os, fnmatch
+
+with open("README.md", "r", encoding="utf-8") as fh:
+    long_description = fh.read()
+
+    
+def _process_requirements():
+    packages = open('requirements.txt').read().strip().split('\n')
+    requires = []
+    for pkg in packages:
+        if pkg.startswith('git+ssh'):
+            return_code = os.system('pip install {}'.format(pkg))
+            assert return_code == 0, 'error, status_code is: {}, exit!'.format(return_code)
+        if pkg.startswith('./docs'):
+            continue
+        else:
+            requires.append(pkg)
+    return requires
+
+def package_files(directory):
+    import subprocess
+    list_of_files = subprocess.check_output("git ls-files", shell=True).splitlines()
+    return [str(k) for k in list_of_files]
+
+extra_files = package_files('./')
+
+setuptools.setup(
+    name="void-terminal",
+    version="0.0.0",
+    author="Qingxu",
+    author_email="505030475@qq.com",
+    description="LLM based APIs",
+    long_description=long_description,
+    long_description_content_type="text/markdown",
+    url="https://github.com/binary-husky/gpt-academic",
+    project_urls={
+        "Bug Tracker": "https://github.com/binary-husky/gpt-academic/issues",
+    },
+    classifiers=[
+        "Programming Language :: Python :: 3",
+        "License :: OSI Approved :: MIT License",
+        "Operating System :: OS Independent",
+    ],
+    package_dir={"": "."},
+    package_data={"": extra_files},
+    include_package_data=True,
+    packages=setuptools.find_packages(where="."),
+    python_requires=">=3.9",
+    install_requires=_process_requirements(),
+)
--- a/toolbox.py
+++ b/toolbox.py
@ -538,7 +538,11 @@ def load_chat_cookies():
    return {'api_key': API_KEY, 'llm_model': LLM_MODEL}

 def is_openai_api_key(key):
-    API_MATCH_ORIGINAL = re.match(r"sk-[a-zA-Z0-9]{48}$", key)
+    CUSTOM_API_KEY_PATTERN, = get_conf('CUSTOM_API_KEY_PATTERN')
+    if len(CUSTOM_API_KEY_PATTERN) != 0:
+        API_MATCH_ORIGINAL = re.match(CUSTOM_API_KEY_PATTERN, key)
+    else:
+        API_MATCH_ORIGINAL = re.match(r"sk-[a-zA-Z0-9]{48}$", key)
    return bool(API_MATCH_ORIGINAL)

 def is_azure_api_key(key):
@ -594,7 +598,7 @@ def select_api_key(keys, llm_model):
            if is_azure_api_key(k): avail_key_list.append(k)

    if len(avail_key_list) == 0:
-        raise RuntimeError(f"您提供的api-key不满足要求，不包含任何可用于{llm_model}的api-key。您可能选择了错误的模型或请求源（右下角更换模型菜单中可切换openai,azure和api2d请求源）")
+        raise RuntimeError(f"您提供的api-key不满足要求，不包含任何可用于{llm_model}的api-key。您可能选择了错误的模型或请求源（右下角更换模型菜单中可切换openai,azure,claude,api2d等请求源）。")

    api_key = random.choice(avail_key_list) # 随机负载均衡
    return api_key
@ -670,13 +674,14 @@ def read_single_conf_with_lru_cache(arg):

    # 在读取API_KEY时，检查一下是不是忘了改config
    if arg == 'API_KEY':
-        print亮蓝(f"[API_KEY] 本项目现已支持OpenAI和API2D的api-key。也支持同时填写多个api-key，如API_KEY=\"openai-key1,openai-key2,api2d-key3\"")
+        print亮蓝(f"[API_KEY] 本项目现已支持OpenAI和Azure的api-key。也支持同时填写多个api-key，如API_KEY=\"openai-key1,openai-key2,azure-key3\"")
        print亮蓝(f"[API_KEY] 您既可以在config.py中修改api-key(s)，也可以在问题输入区输入临时的api-key(s)，然后回车键提交后即可生效。")
        if is_any_api_key(r):
            print亮绿(f"[API_KEY] 您的 API_KEY 是: {r[:15]}*** API_KEY 导入成功")
        else:
-            print亮红( "[API_KEY] 正确的 API_KEY 是'sk'开头的51位密钥（OpenAI），或者 'fk'开头的41位密钥，请在config文件中修改API密钥之后再运行。")
+            print亮红( "[API_KEY] 您的 API_KEY 不满足任何一种已知的密钥格式，请在config文件中修改API密钥之后再运行。")
    if arg == 'proxies':
+        if not read_single_conf_with_lru_cache('USE_PROXY'): r = None   # 检查USE_PROXY，防止proxies单独起作用
        if r is None:
            print亮红('[PROXY] 网络代理状态：未配置。无代理状态下很可能无法访问OpenAI家族的模型。建议：检查USE_PROXY选项是否修改。')
        else:
@ -685,6 +690,7 @@ def read_single_conf_with_lru_cache(arg):
    return r


+@lru_cache(maxsize=128)
 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 = []
--- a/6
+++ b/6
@ -1,5 +1,5 @@
 {
-  "version": 3.46,
+  "version": 3.47,
  "show_feature": true,
-  "new_feature": "临时修复theme的文件丢失问题 <-> 新增实时语音对话插件（自动断句，脱手对话） <-> 支持加载自定义的ChatGLM2微调模型 <-> 动态ChatBot窗口高度 <-> 修复Azure接口的BUG <-> 完善多语言模块 <-> 完善本地Latex矫错和翻译功能 <-> 增加gpt-3.5-16k的支持 <-> 新增最强Arxiv论文翻译插件 <-> 修复gradio复制按钮BUG <-> 修复PDF翻译的BUG, 新增HTML中英双栏对照 <-> 添加了OpenAI图片生成插件"
-}
+  "new_feature": "优化一键升级 <-> 提高arxiv翻译速度和成功率 <-> 支持自定义APIKEY格式 <-> 临时修复theme的文件丢失问题 <-> 新增实时语音对话插件（自动断句，脱手对话） <-> 支持加载自定义的ChatGLM2微调模型 <-> 动态ChatBot窗口高度 <-> 修复Azure接口的BUG <-> 完善多语言模块 <-> 完善本地Latex矫错和翻译功能 <-> 增加gpt-3.5-16k的支持"
+}
Author	SHA1	Message	Date
505030475	ec1cfaadba	pip	2023-07-28 12:28:04 +08:00
505030475	2747c23868	Merge branch 'master' of github.com:binary-husky/chatgpt_academic	2023-07-28 10:35:50 +08:00
binary-husky	f446dbb62d	Update README.md	2023-07-28 09:54:03 +08:00
binary-husky	8d37d94e2c	Update README.md	2023-07-28 09:53:17 +08:00
505030475	4216c5196e	verify ignore history practice	2023-07-27 22:30:55 +08:00
binary-husky	2df660a718	Merge pull request #992 from yangchuansheng/master Update README.md	2023-07-26 22:46:43 +08:00
binary-husky	bb496a9c2c	Update README.md	2023-07-26 22:46:21 +08:00
binary-husky	4e0737c0c2	Update README.md	2023-07-26 22:46:02 +08:00
binary-husky	4bb3cba5c8	Update README.md	2023-07-26 18:53:42 +08:00
qingxu fu	08b9b0d140	improve audio assistant documents	2023-07-26 18:51:33 +08:00
qingxu fu	3577a72a3b	add audio assistant docker compose solution	2023-07-26 18:39:32 +08:00
qingxu fu	0328d6f498	add ALIYUN ACCESSKEY SECRET	2023-07-26 18:28:15 +08:00
qingxu fu	d437305a4f	add audio assistant docker	2023-07-26 18:16:59 +08:00
qingxu fu	c4899bcb20	long-term aliyun access	2023-07-26 18:09:28 +08:00
Carson Yang	4295764f8c	Update README.md 添加 Sealos 部署方案	2023-07-25 16:38:37 +08:00
binary-husky	e4e2430255	version 3.47	2023-07-24 19:58:47 +08:00
binary-husky	1732127a28	Merge pull request #979 from fenglui/master 增加chatGLM int4配置支持小显存也可以选择chatGLM	2023-07-24 19:52:27 +08:00
binary-husky	56bb8b6498	improve re efficiency	2023-07-24 18:50:29 +08:00
binary-husky	e93b6fa3a6	Add GLM INT8	2023-07-24 18:19:57 +08:00
binary-husky	dd4ba0ea22	Merge branch 'master' of https://github.com/fenglui/gpt_academic into fenglui-master	2023-07-24 18:06:15 +08:00
binary-husky	c2701c9ce5	Merge pull request #986 from one-pr/git-clone 默认仅 clone 最新的代码，减小 git clone 的大小	2023-07-24 17:48:35 +08:00
woclass	2f019ce359	优化 README.md 中的其他 git clone	2023-07-24 15:14:48 +08:00
woclass	c5b147aeb7	默认仅 clone 最新的代码，减小 git clone 的大小	2023-07-24 15:14:42 +08:00
fenglui	5813d65e52	增加chatGLM int4配置支持小显存也可以选择chatGLM	2023-07-22 08:29:15 +08:00
binary-husky	a393edfaa4	ALLOW CUSTOM API KEY PATTERN	2023-07-21 22:49:07 +08:00
binary-husky	dd7a01cda5	Merge pull request #976 from fenglui/master fix msg.data.split(DELIMITER) exception when msg.data is int	2023-07-21 17:02:29 +08:00
fenglui	00a3b91f95	fix msg.data.split(DELIMITER) exception when msg.data is int	2023-07-21 03:51:33 +08:00
qingxu fu	61ba544282	add latex test samples	2023-07-20 19:49:23 +08:00
qingxu fu	b5b8c123e4	latex plugin stability improvement	2023-07-20 19:39:22 +08:00
qingxu fu	d9ceba959f	expand range after failure	2023-07-20 18:39:02 +08:00
qingxu fu	6b5b040701	remove pdf merge	2023-07-20 18:29:06 +08:00
qingxu fu	4f4c09a5f3	增强Latex修复能力	2023-07-20 18:08:22 +08:00
qingxu fu	067bc97cce	Merge branch 'interface-interlm' of https://github.com/binary-husky/chatgpt_academic into interface-interlm	2023-07-20 12:46:52 +08:00
qingxu fu	7368580cd6	concat pdf after translation	2023-07-20 12:46:48 +08:00