add test instance

Latex File Name Bug Patch

README.md

@@ -1,6 +1,8 @@
 > **Note**
 >
-> 2023.7.5: 对Gradio依赖进行了调整。请及时**更新代码**。安装依赖时，请严格选择`requirements.txt`中**指定的版本**：
+> 2023.7.5: Gradio依赖调整。请及时**更新代码**
+>
+> 2023.7.8: pydantic出现兼容问题，已修改 `requirements.txt`。安装依赖时，请严格选择`requirements.txt`中**指定的版本**
 >
 > `pip install -r requirements.txt`
 
@@ -41,6 +43,7 @@ Markdown[中英互译](https://www.bilibili.com/video/BV1yo4y157jV/) | [函数
 chat分析报告生成 | [函数插件] 运行后自动生成总结汇报
 [PDF论文全文翻译功能](https://www.bilibili.com/video/BV1KT411x7Wn) | [函数插件] PDF论文提取题目&摘要+翻译全文（多线程）
 [Arxiv小助手](https://www.bilibili.com/video/BV1LM4y1279X) | [函数插件] 输入arxiv文章url即可一键翻译摘要+下载PDF
+Latex论文一键校对 | [函数插件] 仿Grammarly对Latex文章进行语法、拼写纠错+输出对照PDF
 [谷歌学术统合小助手](https://www.bilibili.com/video/BV19L411U7ia) | [函数插件] 给定任意谷歌学术搜索页面URL，让gpt帮你[写relatedworks](https://www.bilibili.com/video/BV1GP411U7Az/)
 互联网信息聚合+GPT | [函数插件] 一键[让GPT从互联网获取信息](https://www.bilibili.com/video/BV1om4y127ck)回答问题，让信息永不过时
 ⭐Arxiv论文精细翻译 | [函数插件] 一键[以超高质量翻译arxiv论文](https://www.bilibili.com/video/BV1dz4y1v77A/)，目前最好的论文翻译工具
@@ -48,8 +51,9 @@ chat分析报告生成 | [函数插件] 运行后自动生成总结汇报
 多线程函数插件支持 | 支持多线调用chatgpt，一键处理[海量文本](https://www.bilibili.com/video/BV1FT411H7c5/)或程序
 启动暗色[主题](https://github.com/binary-husky/gpt_academic/issues/173) | 在浏览器url后面添加```/?__theme=dark```可以切换dark主题
 [多LLM模型](https://www.bilibili.com/video/BV1wT411p7yf)支持 | 同时被GPT3.5、GPT4、[清华ChatGLM](https://github.com/THUDM/ChatGLM-6B)、[复旦MOSS](https://github.com/OpenLMLab/MOSS)同时伺候的感觉一定会很不错吧？
+ChatGLM2微调模型 | 支持加载ChatGLM2微调模型，提供ChatGLM2微调插件
 更多LLM模型接入，支持[huggingface部署](https://huggingface.co/spaces/qingxu98/gpt-academic) | 加入Newbing接口(新必应)，引入清华[Jittorllms](https://github.com/Jittor/JittorLLMs)支持[LLaMA](https://github.com/facebookresearch/llama)和[盘古α](https://openi.org.cn/pangu/)
-更多新功能展示(图像生成等) …… | 见本文档结尾处 ……
+更多新功能展示 (图像生成等) …… | 见本文档结尾处 ……
 
 </div>
 
@@ -149,9 +153,9 @@ 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
 
-#（最后一步-选择1）在Linux环境下，用`--net=host`更方便快捷
+#（最后一步-Linux操作系统）用`--net=host`更方便快捷
 docker run --rm -it --net=host gpt-academic
-#（最后一步-选择2）在macOS/windows环境下，只能用-p选项将容器上的端口(例如50923)暴露给主机上的端口
+#（最后一步-MacOS/Windows操作系统）只能用-p选项将容器上的端口(例如50923)暴露给主机上的端口
 docker run --rm -it -e WEB_PORT=50923 -p 50923:50923 gpt-academic
 ```
 P.S. 如果需要依赖Latex的插件功能，请见Wiki。另外，您也可以直接使用docker-compose获取Latex功能（修改docker-compose.yml，保留方案4并删除其他方案）。
@@ -282,6 +286,8 @@ Tip：不指定文件直接点击 `载入对话历史存档` 可以查看历史h
 
 ### II：版本:
 - version 3.5(Todo): 使用自然语言调用本项目的所有函数插件（高优先级）
+- version 3.45: 支持自定义ChatGLM2微调模型
+- version 3.44: 正式支持Azure，优化界面易用性
 - version 3.4: +arxiv论文翻译、latex论文批改功能
 - version 3.3: +互联网信息综合功能
 - version 3.2: 函数插件支持更多参数接口 (保存对话功能, 解读任意语言代码+同时询问任意的LLM组合)

config.py

@@ -8,7 +8,7 @@
 """
 
 # [step 1]>> API_KEY = "sk-123456789xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx123456789"。极少数情况下，还需要填写组织（格式如org-123456789abcdefghijklmno的），请向下翻，找 API_ORG 设置项
-API_KEY = "sk-此处填API密钥"    # 可同时填写多个API-KEY，用英文逗号分割，例如API_KEY = "sk-openaikey1,sk-openaikey2,fkxxxx-api2dkey1,fkxxxx-api2dkey2"
+API_KEY = "此处填API密钥"    # 可同时填写多个API-KEY，用英文逗号分割，例如API_KEY = "sk-openaikey1,sk-openaikey2,fkxxxx-api2dkey3,azure-apikey4"
 
 
 # [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
@@ -74,6 +74,10 @@ AVAIL_LLM_MODELS = ["gpt-3.5-turbo-16k", "gpt-3.5-turbo", "azure-gpt-3.5", "api2
 # P.S. 其他可用的模型还包括 ["gpt-3.5-turbo-0613", "gpt-3.5-turbo-16k-0613", "newbing-free", "jittorllms_rwkv", "jittorllms_pangualpha", "jittorllms_llama"]
 
 
+# ChatGLM(2) Finetune Model Path （如果使用ChatGLM2微调模型，需要把"chatglmft"加入AVAIL_LLM_MODELS中）
+ChatGLM_PTUNING_CHECKPOINT = "" # 例如"/home/hmp/ChatGLM2-6B/ptuning/output/6b-pt-128-1e-2/checkpoint-100"
+
+
 # 本地LLM模型如ChatGLM的执行方式 CPU/GPU
 LOCAL_MODEL_DEVICE = "cpu" # 可选 "cuda"
 
@@ -110,9 +114,8 @@ SLACK_CLAUDE_USER_TOKEN = ''
 
 # 如果需要使用AZURE 详情请见额外文档 docs\use_azure.md
 AZURE_ENDPOINT = "https://你亲手写的api名称.openai.azure.com/"
-AZURE_API_KEY = "填入azure openai api的密钥"
-AZURE_API_VERSION = "2023-05-15"        # 一般不修改
-AZURE_ENGINE = "填入你亲手写的部署名"     # 读 docs\use_azure.md
+AZURE_API_KEY = "填入azure openai api的密钥"    # 建议直接在API_KEY处填写，该选项即将被弃用
+AZURE_ENGINE = "填入你亲手写的部署名"            # 读 docs\use_azure.md
 
 
 # 使用Newbing

crazy_functional.py

@@ -352,6 +352,32 @@ def get_crazy_functions():
         })
     except:
         print('Load function plugin failed')
+        
+    try:
+        from crazy_functions.交互功能函数模板 import 交互功能模板函数
+        function_plugins.update({
+            "交互功能模板函数": {
+                "Color": "stop",
+                "AsButton": False,
+                "Function": HotReload(交互功能模板函数)
+            }
+        })
+    except:
+        print('Load function plugin failed')
+
+    # try:
+    #     from crazy_functions.chatglm微调工具 import 微调数据集生成
+    #     function_plugins.update({
+    #         "黑盒模型学习: 微调数据集生成 (先上传数据集)": {
+    #             "Color": "stop",
+    #             "AsButton": False,
+    #             "AdvancedArgs": True,
+    #             "ArgsReminder": "针对数据集输入（如 绿帽子*深蓝色衬衫*黑色运动裤）给出指令，例如您可以将以下命令复制到下方: --llm_to_learn=azure-gpt-3.5 --prompt_prefix='根据下面的服装类型提示，想象一个穿着者，对这个人外貌、身处的环境、内心世界、过去经历进行描写。要求：100字以内，用第二人称。' --system_prompt=''",
+    #             "Function": HotReload(微调数据集生成)
+    #         }
+    #     })
+    # except:
+    #     print('Load function plugin failed')
 
     try:
         from crazy_functions.Latex输出PDF结果 import Latex英文纠错加PDF对比

crazy_functions/chatglm微调工具.py

@@ -0,0 +1,141 @@
+from toolbox import CatchException, update_ui, promote_file_to_downloadzone
+from .crazy_utils import request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency
+import datetime, json
+
+def fetch_items(list_of_items, batch_size):
+    for i in range(0, len(list_of_items), batch_size):
+        yield list_of_items[i:i + batch_size]
+
+def string_to_options(arguments):
+    import argparse
+    import shlex
+
+    # Create an argparse.ArgumentParser instance
+    parser = argparse.ArgumentParser()
+
+    # Add command-line arguments
+    parser.add_argument("--llm_to_learn", type=str, help="LLM model to learn", default="gpt-3.5-turbo")
+    parser.add_argument("--prompt_prefix", type=str, help="Prompt prefix", default='')
+    parser.add_argument("--system_prompt", type=str, help="System prompt", default='')
+    parser.add_argument("--batch", type=int, help="System prompt", default=50)
+    parser.add_argument("--pre_seq_len", type=int, help="pre_seq_len", default=50)
+    parser.add_argument("--learning_rate", type=float, help="learning_rate", default=2e-2)
+    parser.add_argument("--num_gpus", type=int, help="num_gpus", default=1)
+    parser.add_argument("--json_dataset", type=str, help="json_dataset", default="")
+    parser.add_argument("--ptuning_directory", type=str, help="ptuning_directory", default="")
+
+
+
+    # Parse the arguments
+    args = parser.parse_args(shlex.split(arguments))
+
+    return args
+
+@CatchException
+def 微调数据集生成(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+    """
+    txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
+    llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数
+    chatbot         聊天显示框的句柄，用于显示给用户
+    history         聊天历史，前情提要
+    system_prompt   给gpt的静默提醒
+    web_port        当前软件运行的端口号
+    """
+    history = []    # 清空历史，以免输入溢出
+    chatbot.append(("这是什么功能？", "[Local Message] 微调数据集生成"))
+    if ("advanced_arg" in plugin_kwargs) and (plugin_kwargs["advanced_arg"] == ""): plugin_kwargs.pop("advanced_arg")
+    args = plugin_kwargs.get("advanced_arg", None)
+    if args is None: 
+        chatbot.append(("没给定指令", "退出"))
+        yield from update_ui(chatbot=chatbot, history=history); return
+    else:
+        arguments = string_to_options(arguments=args)
+
+    dat = []
+    with open(txt, 'r', encoding='utf8') as f:
+        for line in f.readlines():
+            json_dat = json.loads(line)
+            dat.append(json_dat["content"])
+
+    llm_kwargs['llm_model'] = arguments.llm_to_learn
+    for batch in fetch_items(dat, arguments.batch):
+        res = yield from request_gpt_model_multi_threads_with_very_awesome_ui_and_high_efficiency(
+            inputs_array=[f"{arguments.prompt_prefix}\n\n{b}" for b in (batch)],
+            inputs_show_user_array=[f"Show Nothing" for _ in (batch)],
+            llm_kwargs=llm_kwargs,
+            chatbot=chatbot,
+            history_array=[[] for _ in (batch)],
+            sys_prompt_array=[arguments.system_prompt for _ in (batch)],
+            max_workers=10  # OpenAI所允许的最大并行过载
+        )
+    
+        with open(txt+'.generated.json', 'a+', encoding='utf8') as f:
+            for b, r in zip(batch, res[1::2]):
+                f.write(json.dumps({"content":b, "summary":r}, ensure_ascii=False)+'\n')
+
+    promote_file_to_downloadzone(txt+'.generated.json', rename_file='generated.json', chatbot=chatbot)
+    return
+
+
+
+@CatchException
+def 启动微调(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+    """
+    txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
+    llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数
+    chatbot         聊天显示框的句柄，用于显示给用户
+    history         聊天历史，前情提要
+    system_prompt   给gpt的静默提醒
+    web_port        当前软件运行的端口号
+    """
+    import subprocess
+    history = []    # 清空历史，以免输入溢出
+    chatbot.append(("这是什么功能？", "[Local Message] 微调数据集生成"))
+    if ("advanced_arg" in plugin_kwargs) and (plugin_kwargs["advanced_arg"] == ""): plugin_kwargs.pop("advanced_arg")
+    args = plugin_kwargs.get("advanced_arg", None)
+    if args is None: 
+        chatbot.append(("没给定指令", "退出"))
+        yield from update_ui(chatbot=chatbot, history=history); return
+    else:
+        arguments = string_to_options(arguments=args)
+      
+
+
+    pre_seq_len = arguments.pre_seq_len             # 128
+    learning_rate = arguments.learning_rate                               # 2e-2
+    num_gpus = arguments.num_gpus                   # 1
+    json_dataset = arguments.json_dataset                 # 't_code.json'
+    ptuning_directory = arguments.ptuning_directory       # '/home/hmp/ChatGLM2-6B/ptuning'
+
+    command = f"torchrun --standalone --nnodes=1 --nproc-per-node={num_gpus} main.py \
+        --do_train \
+        --train_file AdvertiseGen/{json_dataset} \
+        --validation_file AdvertiseGen/{json_dataset} \
+        --preprocessing_num_workers 20 \
+        --prompt_column content \
+        --response_column summary \
+        --overwrite_cache \
+        --model_name_or_path THUDM/chatglm2-6b \
+        --output_dir output/clothgen-chatglm2-6b-pt-{pre_seq_len}-{learning_rate} \
+        --overwrite_output_dir \
+        --max_source_length 256 \
+        --max_target_length 256 \
+        --per_device_train_batch_size 1 \
+        --per_device_eval_batch_size 1 \
+        --gradient_accumulation_steps 16 \
+        --predict_with_generate \
+        --max_steps 100 \
+        --logging_steps 10 \
+        --save_steps 20 \
+        --learning_rate {learning_rate} \
+        --pre_seq_len {pre_seq_len} \
+        --quantization_bit 4"
+
+    process = subprocess.Popen(command, shell=True, cwd=ptuning_directory)
+    try:
+        process.communicate(timeout=3600*24)
+    except subprocess.TimeoutExpired:
+        process.kill()
+    return

crazy_functions/crazy_functions_test.py

@@ -211,22 +211,36 @@ def test_Latex():
     # # for cookies, cb, hist, msg in silence_stdout(编译Latex)(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
     #     cli_printer.print(cb)   #  print(cb)
 
+def test_chatglm_finetune():
+    from crazy_functions.chatglm微调工具 import 微调数据集生成, 启动微调
+    txt = 'build/dev.json'
+    plugin_kwargs = {"advanced_arg":"--llm_to_learn=gpt-3.5-turbo --prompt_prefix='根据下面的服装类型提示，想象一个穿着者，对这个人外貌、身处的环境、内心世界、人设进行描写。要求：100字以内，用第二人称。' --system_prompt=''" }
+
+    # for cookies, cb, hist, msg in (微调数据集生成)(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+    #     cli_printer.print(cb)
+
+    plugin_kwargs = {"advanced_arg":
+    "      --pre_seq_len=128 --learning_rate=2e-2 --num_gpus=1 --json_dataset='t_code.json' --ptuning_directory='/home/hmp/ChatGLM2-6B/ptuning'     " }
+
+    for cookies, cb, hist, msg in (启动微调)(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+        cli_printer.print(cb)
 
 
-# test_解析一个Python项目()
-# test_Latex英文润色()
-# test_Markdown中译英()
-# test_批量翻译PDF文档()
-# test_谷歌检索小助手()
-# test_总结word文档()
-# test_下载arxiv论文并翻译摘要()
-# test_解析一个Cpp项目()
-# test_联网回答问题()
-# test_解析ipynb文件()
-# test_数学动画生成manim()
-# test_Langchain知识库()
-# test_Langchain知识库读取()
 if __name__ == "__main__":
-    test_Latex()
+    # test_解析一个Python项目()
+    # test_Latex英文润色()
+    # test_Markdown中译英()
+    # test_批量翻译PDF文档()
+    # test_谷歌检索小助手()
+    # test_总结word文档()
+    # test_下载arxiv论文并翻译摘要()
+    # test_解析一个Cpp项目()
+    # test_联网回答问题()
+    # test_解析ipynb文件()
+    # 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()
     input("程序完成，回车退出。")
     print("退出。")

crazy_functions/latex_utils.py

@@ -189,6 +189,18 @@ def rm_comments(main_file):
     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
@@ -197,15 +209,11 @@ def merge_tex_files_(project_foler, main_file, mode):
     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)
-        if os.path.exists(fp):  
-            # e.g., \input{srcs/07_appendix.tex}
-            with open(fp, 'r', encoding='utf-8', errors='replace') as fx:
-                c = fx.read()
-        else:  
-            # e.g., \input{srcs/07_appendix}
-            assert os.path.exists(fp+'.tex'), f'即找不到{fp}，也找不到{fp}.tex，Tex源文件缺失！'
-            with open(fp+'.tex', 'r', encoding='utf-8', errors='replace') as fx:
-                c = fx.read()
+        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
@@ -324,7 +332,7 @@ def split_subprocess(txt, project_folder, return_dict, opts):
     # 吸收在42行以内的begin-end组合
     text, mask = set_forbidden_text_begin_end(text, mask, r"\\begin\{([a-z\*]*)\}(.*?)\\end\{\1\}", re.DOTALL, limit_n_lines=42)
     # 吸收匿名公式
-    text, mask = set_forbidden_text(text, mask, [ r"\$\$(.*?)\$\$",  r"\\\[.*?\\\]" ], re.DOTALL)
+    text, mask = set_forbidden_text(text, mask, [ r"\$\$([^$]+)\$\$",  r"\\\[.*?\\\]" ], re.DOTALL)
     # 吸收其他杂项
     text, mask = set_forbidden_text(text, mask, [ r"\\section\{(.*?)\}", r"\\section\*\{(.*?)\}", r"\\subsection\{(.*?)\}", r"\\subsubsection\{(.*?)\}" ])
     text, mask = set_forbidden_text(text, mask, [ r"\\bibliography\{(.*?)\}", r"\\bibliographystyle\{(.*?)\}" ])

crazy_functions/交互功能函数模板.py

@@ -0,0 +1,63 @@
+from toolbox import CatchException, update_ui
+from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
+
+
+@CatchException
+def 交互功能模板函数(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, web_port):
+    """
+    txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
+    llm_kwargs      gpt模型参数, 如温度和top_p等, 一般原样传递下去就行
+    plugin_kwargs   插件模型的参数, 如温度和top_p等, 一般原样传递下去就行
+    chatbot         聊天显示框的句柄，用于显示给用户
+    history         聊天历史，前情提要
+    system_prompt   给gpt的静默提醒
+    web_port        当前软件运行的端口号
+    """
+    history = []    # 清空历史，以免输入溢出
+    chatbot.append(("这是什么功能？", "交互功能函数模板。在执行完成之后, 可以将自身的状态存储到cookie中, 等待用户的再次调用。"))
+    yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+
+    state = chatbot._cookies.get('plugin_state_0001', None) # 初始化插件状态
+
+    if state is None:
+        chatbot._cookies['lock_plugin'] = 'crazy_functions.交互功能函数模板->交互功能模板函数'      # 赋予插件锁定 锁定插件回调路径，当下一次用户提交时，会直接转到该函数
+        chatbot._cookies['plugin_state_0001'] = 'wait_user_keyword'                              # 赋予插件状态
+
+        chatbot.append(("第一次调用：", "请输入关键词, 我将为您查找相关壁纸, 建议使用英文单词, 插件锁定中，请直接提交即可。"))
+        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        return
+
+    if state == 'wait_user_keyword':
+        chatbot._cookies['lock_plugin'] = None          # 解除插件锁定，避免遗忘导致死锁
+        chatbot._cookies['plugin_state_0001'] = None    # 解除插件状态，避免遗忘导致死锁
+
+        # 解除插件锁定
+        chatbot.append((f"获取关键词：{txt}", ""))
+        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        page_return = get_image_page_by_keyword(txt)
+        inputs=inputs_show_user=f"Extract all image urls in this html page, pick the first 5 images and show them with markdown format: \n\n {page_return}"
+        gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
+            inputs=inputs, inputs_show_user=inputs_show_user,
+            llm_kwargs=llm_kwargs, chatbot=chatbot, history=[], 
+            sys_prompt="When you want to show an image, use markdown format. e.g. ![image_description](image_url). If there are no image url provided, answer 'no image url provided'"
+        )
+        chatbot[-1] = [chatbot[-1][0], gpt_say]
+        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        return
+
+
+
+# ---------------------------------------------------------------------------------
+
+def get_image_page_by_keyword(keyword):
+    import requests
+    from bs4 import BeautifulSoup
+    response = requests.get(f'https://wallhaven.cc/search?q={keyword}', timeout=2)
+    res = "image urls: \n"
+    for image_element in BeautifulSoup(response.content, 'html.parser').findAll("img"):
+        try:
+            res += image_element["data-src"]
+            res += "\n"
+        except:
+            pass
+    return res

crazy_functions/总结word文档.py

@@ -14,17 +14,19 @@ def 解析docx(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot
             doc = Document(fp)
             file_content = "\n".join([para.text for para in doc.paragraphs])
         else:
-            import win32com.client
-            word = win32com.client.Dispatch("Word.Application")
-            word.visible = False
-            # 打开文件
-            print('fp', os.getcwd())
-            doc = word.Documents.Open(os.getcwd() + '/' + fp)
-            # file_content = doc.Content.Text
-            doc = word.ActiveDocument
-            file_content = doc.Range().Text
-            doc.Close()
-            word.Quit()
+            try:
+                import win32com.client
+                word = win32com.client.Dispatch("Word.Application")
+                word.visible = False
+                # 打开文件
+                doc = word.Documents.Open(os.getcwd() + '/' + fp)
+                # file_content = doc.Content.Text
+                doc = word.ActiveDocument
+                file_content = doc.Range().Text
+                doc.Close()
+                word.Quit()
+            except:
+                raise RuntimeError('请先将.doc文档转换为.docx文档。')
 
         print(file_content)
         # private_upload里面的文件名在解压zip后容易出现乱码（rar和7z格式正常），故可以只分析文章内容，不输入文件名

crazy_functions/批量总结PDF文档.py

@@ -1,121 +1,107 @@
-from toolbox import update_ui
+from toolbox import update_ui, promote_file_to_downloadzone, gen_time_str
 from toolbox import CatchException, report_execption, write_results_to_file
-import re
-import unicodedata
-fast_debug = False
 from .crazy_utils import request_gpt_model_in_new_thread_with_ui_alive
+from .crazy_utils import read_and_clean_pdf_text
+from .crazy_utils import input_clipping
 
-def is_paragraph_break(match):
-    """
-    根据给定的匹配结果来判断换行符是否表示段落分隔。
-    如果换行符前为句子结束标志（句号，感叹号，问号），且下一个字符为大写字母，则换行符更有可能表示段落分隔。
-    也可以根据之前的内容长度来判断段落是否已经足够长。
-    """
-    prev_char, next_char = match.groups()
 
-    # 句子结束标志
-    sentence_endings = ".!?"
-
-    # 设定一个最小段落长度阈值
-    min_paragraph_length = 140
-
-    if prev_char in sentence_endings and next_char.isupper() and len(match.string[:match.start(1)]) > min_paragraph_length:
-        return "\n\n" 
-    else:
-        return " "
-
-def normalize_text(text):
-    """
-    通过把连字（ligatures）等文本特殊符号转换为其基本形式来对文本进行归一化处理。
-    例如，将连字 "fi" 转换为 "f" 和 "i"。
-    """
-    # 对文本进行归一化处理，分解连字
-    normalized_text = unicodedata.normalize("NFKD", text)
-
-    # 替换其他特殊字符
-    cleaned_text = re.sub(r'[^\x00-\x7F]+', '', normalized_text)
-
-    return cleaned_text
-
-def clean_text(raw_text):
-    """
-    对从 PDF 提取出的原始文本进行清洗和格式化处理。
-    1. 对原始文本进行归一化处理。
-    2. 替换跨行的连词
-    3. 根据 heuristic 规则判断换行符是否是段落分隔，并相应地进行替换
-    """
-    # 对文本进行归一化处理
-    normalized_text = normalize_text(raw_text)
-
-    # 替换跨行的连词
-    text = re.sub(r'(\w+-\n\w+)', lambda m: m.group(1).replace('-\n', ''), normalized_text)
-
-    # 根据前后相邻字符的特点，找到原文本中的换行符
-    newlines = re.compile(r'(\S)\n(\S)')
-
-    # 根据 heuristic 规则，用空格或段落分隔符替换原换行符
-    final_text = re.sub(newlines, lambda m: m.group(1) + is_paragraph_break(m) + m.group(2), text)
-
-    return final_text.strip()
 
 def 解析PDF(file_manifest, project_folder, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt):
-    import time, glob, os, fitz
-    print('begin analysis on:', file_manifest)
-    for index, fp in enumerate(file_manifest):
-        with fitz.open(fp) as doc:
-            file_content = ""
-            for page in doc:
-                file_content += page.get_text()
-            file_content = clean_text(file_content)
-            print(file_content)
+    file_write_buffer = []
+    for file_name in file_manifest:
+        print('begin analysis on:', file_name)
+        ############################## <第 0 步，切割PDF> ##################################
+        # 递归地切割PDF文件，每一块（尽量是完整的一个section，比如introduction，experiment等，必要时再进行切割）
+        # 的长度必须小于 2500 个 Token
+        file_content, page_one = read_and_clean_pdf_text(file_name) # （尝试）按照章节切割PDF
+        file_content = file_content.encode('utf-8', 'ignore').decode()   # avoid reading non-utf8 chars
+        page_one = str(page_one).encode('utf-8', 'ignore').decode()  # avoid reading non-utf8 chars
+        
+        TOKEN_LIMIT_PER_FRAGMENT = 2500
 
-        prefix = "接下来请你逐文件分析下面的论文文件，概括其内容" if index==0 else ""
-        i_say = prefix + f'请对下面的文章片段用中文做一个概述，文件名是{os.path.relpath(fp, project_folder)}，文章内容是 ```{file_content}```'
-        i_say_show_user = prefix + f'[{index + 1}/{len(file_manifest)}] 请对下面的文章片段做一个概述: {os.path.abspath(fp)}'
-        chatbot.append((i_say_show_user, "[Local Message] waiting gpt response."))
-        yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
+        from .crazy_utils import breakdown_txt_to_satisfy_token_limit_for_pdf
+        from request_llm.bridge_all import model_info
+        enc = model_info["gpt-3.5-turbo"]['tokenizer']
+        def get_token_num(txt): return len(enc.encode(txt, disallowed_special=()))
+        paper_fragments = breakdown_txt_to_satisfy_token_limit_for_pdf(
+            txt=file_content,  get_token_fn=get_token_num, limit=TOKEN_LIMIT_PER_FRAGMENT)
+        page_one_fragments = breakdown_txt_to_satisfy_token_limit_for_pdf(
+            txt=str(page_one), get_token_fn=get_token_num, limit=TOKEN_LIMIT_PER_FRAGMENT//4)
+        # 为了更好的效果，我们剥离Introduction之后的部分（如果有）
+        paper_meta = page_one_fragments[0].split('introduction')[0].split('Introduction')[0].split('INTRODUCTION')[0]
+        
+        ############################## <第 1 步，从摘要中提取高价值信息，放到history中> ##################################
+        final_results = []
+        final_results.append(paper_meta)
 
-        if not fast_debug: 
-            msg = '正常'
-            # ** gpt request **
-            gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
-                inputs=i_say, 
-                inputs_show_user=i_say_show_user, 
-                llm_kwargs=llm_kwargs,
-                chatbot=chatbot, 
-                history=[],
-                sys_prompt="总结文章。"
-            )  # 带超时倒计时
-                
+        ############################## <第 2 步，迭代地历遍整个文章，提取精炼信息> ##################################
+        i_say_show_user = f'首先你在中文语境下通读整篇论文。'; gpt_say = "[Local Message] 收到。"           # 用户提示
+        chatbot.append([i_say_show_user, gpt_say]); yield from update_ui(chatbot=chatbot, history=[])    # 更新UI
 
-            chatbot[-1] = (i_say_show_user, gpt_say)
-            history.append(i_say_show_user); history.append(gpt_say)
-            yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
-            if not fast_debug: time.sleep(2)
+        iteration_results = []
+        last_iteration_result = paper_meta  # 初始值是摘要
+        MAX_WORD_TOTAL = 4096 * 0.7
+        n_fragment = len(paper_fragments)
+        if n_fragment >= 20: print('文章极长，不能达到预期效果')
+        for i in range(n_fragment):
+            NUM_OF_WORD = MAX_WORD_TOTAL // n_fragment
+            i_say = f"Read this section, recapitulate the content of this section with less than {NUM_OF_WORD} Chinese characters: {paper_fragments[i]}"
+            i_say_show_user = f"[{i+1}/{n_fragment}] Read this section, recapitulate the content of this section with less than {NUM_OF_WORD} Chinese characters: {paper_fragments[i][:200]}"
+            gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(i_say, i_say_show_user,  # i_say=真正给chatgpt的提问， i_say_show_user=给用户看的提问
+                                                                                llm_kwargs, chatbot, 
+                                                                                history=["The main idea of the previous section is?", last_iteration_result], # 迭代上一次的结果
+                                                                                sys_prompt="Extract the main idea of this section with Chinese."  # 提示
+                                                                                ) 
+            iteration_results.append(gpt_say)
+            last_iteration_result = gpt_say
 
-    all_file = ', '.join([os.path.relpath(fp, project_folder) for index, fp in enumerate(file_manifest)])
-    i_say = f'根据以上你自己的分析，对全文进行概括，用学术性语言写一段中文摘要，然后再写一段英文摘要（包括{all_file}）。'
-    chatbot.append((i_say, "[Local Message] waiting gpt response."))
-    yield from update_ui(chatbot=chatbot, history=history) # 刷新界面
-
-    if not fast_debug: 
-        msg = '正常'
-        # ** gpt request **
+        ############################## <第 3 步，整理history，提取总结> ##################################
+        final_results.extend(iteration_results)
+        final_results.append(f'Please conclude this paper discussed above。')
+        # This prompt is from https://github.com/kaixindelele/ChatPaper/blob/main/chat_paper.py
+        NUM_OF_WORD = 1000
+        i_say = """
+1. Mark the title of the paper (with Chinese translation)
+2. list all the authors' names (use English)
+3. mark the first author's affiliation (output Chinese translation only)
+4. mark the keywords of this article (use English)
+5. link to the paper, Github code link (if available, fill in Github:None if not)
+6. summarize according to the following four points.Be sure to use Chinese answers (proper nouns need to be marked in English)
+    - (1):What is the research background of this article?
+    - (2):What are the past methods? What are the problems with them? Is the approach well motivated?
+    - (3):What is the research methodology proposed in this paper?
+    - (4):On what task and what performance is achieved by the methods in this paper? Can the performance support their goals?
+Follow the format of the output that follows:                  
+1. Title: xxx\n\n
+2. Authors: xxx\n\n
+3. Affiliation: xxx\n\n
+4. Keywords: xxx\n\n
+5. Urls: xxx or xxx , xxx \n\n
+6. Summary: \n\n
+    - (1):xxx;\n 
+    - (2):xxx;\n 
+    - (3):xxx;\n
+    - (4):xxx.\n\n
+Be sure to use Chinese answers (proper nouns need to be marked in English), statements as concise and academic as possible,
+do not have too much repetitive information, numerical values using the original numbers.
+        """
+        # This prompt is from https://github.com/kaixindelele/ChatPaper/blob/main/chat_paper.py
+        file_write_buffer.extend(final_results)
+        i_say, final_results = input_clipping(i_say, final_results, max_token_limit=2000)
         gpt_say = yield from request_gpt_model_in_new_thread_with_ui_alive(
-            inputs=i_say, 
-            inputs_show_user=i_say, 
-            llm_kwargs=llm_kwargs,
-            chatbot=chatbot, 
-            history=history,
-            sys_prompt="总结文章。"
-        )  # 带超时倒计时
+            inputs=i_say, inputs_show_user='开始最终总结', 
+            llm_kwargs=llm_kwargs, chatbot=chatbot, history=final_results, 
+            sys_prompt= f"Extract the main idea of this paper with less than {NUM_OF_WORD} Chinese characters"
+        )
+        final_results.append(gpt_say)
+        file_write_buffer.extend([i_say, gpt_say])
+        ############################## <第 4 步，设置一个token上限> ##################################
+        _, final_results = input_clipping("", final_results, max_token_limit=3200)
+        yield from update_ui(chatbot=chatbot, history=final_results) # 注意这里的历史记录被替代了
 
-        chatbot[-1] = (i_say, gpt_say)
-        history.append(i_say); history.append(gpt_say)
-        yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
-        res = write_results_to_file(history)
-        chatbot.append(("完成了吗？", res))
-        yield from update_ui(chatbot=chatbot, history=history, msg=msg) # 刷新界面
+    res = write_results_to_file(file_write_buffer, file_name=gen_time_str())
+    promote_file_to_downloadzone(res.split('\t')[-1], chatbot=chatbot)
+    yield from update_ui(chatbot=chatbot, history=final_results) # 刷新界面
 
 
 @CatchException
@@ -151,10 +137,7 @@ def 批量总结PDF文档(txt, llm_kwargs, plugin_kwargs, chatbot, history, syst
         return
 
     # 搜索需要处理的文件清单
-    file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.pdf', recursive=True)] # + \
-                    # [f for f in glob.glob(f'{project_folder}/**/*.tex', recursive=True)] + \
-                    # [f for f in glob.glob(f'{project_folder}/**/*.cpp', recursive=True)] + \
-                    # [f for f in glob.glob(f'{project_folder}/**/*.c', recursive=True)]
+    file_manifest = [f for f in glob.glob(f'{project_folder}/**/*.pdf', recursive=True)]
     
     # 如果没找到任何文件
     if len(file_manifest) == 0:

crazy_functions/询问多个大语言模型.py

@@ -6,7 +6,7 @@ def 同时问询(txt, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt
     """
     txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
     llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
-    plugin_kwargs   插件模型的参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数，用于灵活调整复杂功能的各种参数
     chatbot         聊天显示框的句柄，用于显示给用户
     history         聊天历史，前情提要
     system_prompt   给gpt的静默提醒
@@ -35,7 +35,7 @@ def 同时问询_指定模型(txt, llm_kwargs, plugin_kwargs, chatbot, history,
     """
     txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
     llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
-    plugin_kwargs   插件模型的参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数，用于灵活调整复杂功能的各种参数
     chatbot         聊天显示框的句柄，用于显示给用户
     history         聊天历史，前情提要
     system_prompt   给gpt的静默提醒

crazy_functions/高级功能函数模板.py

@@ -6,7 +6,7 @@ def 高阶功能模板函数(txt, llm_kwargs, plugin_kwargs, chatbot, history, s
     """
     txt             输入栏用户输入的文本，例如需要翻译的一段话，再例如一个包含了待处理文件的路径
     llm_kwargs      gpt模型参数，如温度和top_p等，一般原样传递下去就行
-    plugin_kwargs   插件模型的参数，如温度和top_p等，一般原样传递下去就行
+    plugin_kwargs   插件模型的参数，用于灵活调整复杂功能的各种参数
     chatbot         聊天显示框的句柄，用于显示给用户
     history         聊天历史，前情提要
     system_prompt   给gpt的静默提醒

docker-compose.yml

@@ -6,7 +6,7 @@
 version: '3'
 services:
   gpt_academic_nolocalllms:
-    image: ghcr.io/binary-husky/gpt_academic_nolocal:master
+    image: ghcr.io/binary-husky/gpt_academic_nolocal:master # (Auto Built by Dockerfile: docs/GithubAction+NoLocal)
     environment:
       # 请查阅 `config.py` 以查看所有的配置信息
       API_KEY:                  '    sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx                                            '
@@ -33,7 +33,7 @@ services:
 version: '3'
 services:
   gpt_academic_with_chatglm:
-    image: ghcr.io/binary-husky/gpt_academic_chatglm_moss:master
+    image: ghcr.io/binary-husky/gpt_academic_chatglm_moss:master  # (Auto Built by Dockerfile: docs/Dockerfile+ChatGLM)
     environment:
       # 请查阅 `config.py` 以查看所有的配置信息
       API_KEY:                  '    sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx,fkxxxxxx-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx  '
@@ -63,7 +63,7 @@ services:
 version: '3'
 services:
   gpt_academic_with_rwkv:
-    image: fuqingxu/gpt_academic:jittorllms    # [option 2] 如果需要运行ChatGLM本地模型
+    image: fuqingxu/gpt_academic:jittorllms
     environment:
       # 请查阅 `config.py` 以查看所有的配置信息
       API_KEY:                  '    sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx,fkxxxxxx-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx  '
@@ -111,7 +111,7 @@ services:
 version: '3'
 services:
   gpt_academic_with_latex:
-    image: ghcr.io/binary-husky/gpt_academic_with_latex:master
+    image: ghcr.io/binary-husky/gpt_academic_with_latex:master  # (Auto Built by Dockerfile: docs/GithubAction+NoLocal+Latex)
     environment:
       # 请查阅 `config.py` 以查看所有的配置信息
       API_KEY:                  '    sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx                              '

docs/use_azure.md

@@ -100,10 +100,12 @@
 # 修改 config.py
 
 ```
-AZURE_ENDPOINT = "填入终结点"
+LLM_MODEL = "azure-gpt-3.5"  # 指定启动时的默认模型，当然事后从下拉菜单选也ok
+
+AZURE_ENDPOINT = "填入终结点" # 见上述图片
 AZURE_API_KEY = "填入azure openai api的密钥"
 AZURE_API_VERSION = "2023-05-15"  # 默认使用 2023-05-15 版本，无需修改
-AZURE_ENGINE = "填入部署名" # 见上图
+AZURE_ENGINE = "填入部署名" # 见上述图片
 
 ```
 

main.py

@@ -4,10 +4,10 @@ def main():
     import gradio as gr
     if gr.__version__ not in ['3.28.3','3.32.2']: assert False, "需要特殊依赖，请务必用 pip install -r requirements.txt 指令安装依赖，详情信息见requirements.txt"
     from request_llm.bridge_all import predict
-    from toolbox import format_io, find_free_port, on_file_uploaded, on_report_generated, get_conf, ArgsGeneralWrapper, DummyWith
+    from toolbox import format_io, find_free_port, on_file_uploaded, on_report_generated, get_conf, ArgsGeneralWrapper, load_chat_cookies, DummyWith
     # 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, LAYOUT, API_KEY, AVAIL_LLM_MODELS, AUTO_CLEAR_TXT = \
-        get_conf('proxies', 'WEB_PORT', 'LLM_MODEL', 'CONCURRENT_COUNT', 'AUTHENTICATION', 'CHATBOT_HEIGHT', 'LAYOUT', 'API_KEY', 'AVAIL_LLM_MODELS', 'AUTO_CLEAR_TXT')
+    proxies, WEB_PORT, LLM_MODEL, CONCURRENT_COUNT, AUTHENTICATION, CHATBOT_HEIGHT, LAYOUT, AVAIL_LLM_MODELS, AUTO_CLEAR_TXT = \
+        get_conf('proxies', 'WEB_PORT', 'LLM_MODEL', 'CONCURRENT_COUNT', 'AUTHENTICATION', 'CHATBOT_HEIGHT', 'LAYOUT', 'AVAIL_LLM_MODELS', 'AUTO_CLEAR_TXT')
 
     # 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
@@ -45,23 +45,23 @@ def main():
     proxy_info = check_proxy(proxies)
 
     gr_L1 = lambda: gr.Row().style()
-    gr_L2 = lambda scale: gr.Column(scale=scale)
+    gr_L2 = lambda scale, elem_id: gr.Column(scale=scale, elem_id=elem_id)
     if LAYOUT == "TOP-DOWN":
         gr_L1 = lambda: DummyWith()
-        gr_L2 = lambda scale: gr.Row()
+        gr_L2 = lambda scale, elem_id: gr.Row()
         CHATBOT_HEIGHT /= 2
 
     cancel_handles = []
     with gr.Blocks(title="ChatGPT 学术优化", theme=set_theme, analytics_enabled=False, css=advanced_css) as demo:
         gr.HTML(title_html)
-        cookies = gr.State({'api_key': API_KEY, 'llm_model': LLM_MODEL})
+        cookies = gr.State(load_chat_cookies())
         with gr_L1():
-            with gr_L2(scale=2):
-                chatbot = gr.Chatbot(label=f"当前模型：{LLM_MODEL}")
-                chatbot.style(height=CHATBOT_HEIGHT)
+            with gr_L2(scale=2, elem_id="gpt-chat"):
+                chatbot = gr.Chatbot(label=f"当前模型：{LLM_MODEL}", elem_id="gpt-chatbot")
+                if LAYOUT == "TOP-DOWN":  chatbot.style(height=CHATBOT_HEIGHT)
                 history = gr.State([])
-            with gr_L2(scale=1):
-                with gr.Accordion("输入区", open=True) as area_input_primary:
+            with gr_L2(scale=1, elem_id="gpt-panel"):
+                with gr.Accordion("输入区", open=True, elem_id="input-panel") as area_input_primary:
                     with gr.Row():
                         txt = gr.Textbox(show_label=False, placeholder="Input question here.").style(container=False)
                     with gr.Row():
@@ -71,14 +71,14 @@ def main():
                         stopBtn = gr.Button("停止", variant="secondary"); stopBtn.style(size="sm")
                         clearBtn = gr.Button("清除", variant="secondary", visible=False); clearBtn.style(size="sm")
                     with gr.Row():
-                        status = gr.Markdown(f"Tip: 按Enter提交, 按Shift+Enter换行。当前模型: {LLM_MODEL} \n {proxy_info}")
-                with gr.Accordion("基础功能区", open=True) as area_basic_fn:
+                        status = gr.Markdown(f"Tip: 按Enter提交, 按Shift+Enter换行。当前模型: {LLM_MODEL} \n {proxy_info}", elem_id="state-panel")
+                with gr.Accordion("基础功能区", open=True, elem_id="basic-panel") as area_basic_fn:
                     with gr.Row():
                         for k in functional:
                             if ("Visible" in functional[k]) and (not functional[k]["Visible"]): continue
                             variant = functional[k]["Color"] if "Color" in functional[k] else "secondary"
                             functional[k]["Button"] = gr.Button(k, variant=variant)
-                with gr.Accordion("函数插件区", open=True) as area_crazy_fn:
+                with gr.Accordion("函数插件区", open=True, elem_id="plugin-panel") as area_crazy_fn:
                     with gr.Row():
                         gr.Markdown("注意：以下“红颜色”标识的函数插件需从输入区读取路径作为参数.")
                     with gr.Row():
@@ -100,7 +100,7 @@ def main():
                     with gr.Row():
                         with gr.Accordion("点击展开“文件上传区”。上传本地文件可供红色函数插件调用。", open=False) as area_file_up:
                             file_upload = gr.Files(label="任何文件, 但推荐上传压缩文件(zip, tar)", file_count="multiple")
-                with gr.Accordion("更换模型 & SysPrompt & 交互界面布局", open=(LAYOUT == "TOP-DOWN")):
+                with gr.Accordion("更换模型 & SysPrompt & 交互界面布局", open=(LAYOUT == "TOP-DOWN"), elem_id="interact-panel"):
                     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",)
@@ -109,7 +109,7 @@ def main():
                     md_dropdown = gr.Dropdown(AVAIL_LLM_MODELS, value=LLM_MODEL, label="更换LLM模型/请求源").style(container=False)
 
                     gr.Markdown(description)
-                with gr.Accordion("备选输入区", open=True, visible=False) as area_input_secondary:
+                with gr.Accordion("备选输入区", open=True, visible=False, elem_id="input-panel2") as area_input_secondary:
                     with gr.Row():
                         txt2 = gr.Textbox(show_label=False, placeholder="Input question here.", label="输入区2").style(container=False)
                     with gr.Row():
@@ -176,16 +176,17 @@ def main():
             return {chatbot: gr.update(label="当前模型："+k)}
         md_dropdown.select(on_md_dropdown_changed, [md_dropdown], [chatbot] )
         # 随变按钮的回调函数注册
-        def route(k, *args, **kwargs):
+        def route(request: gr.Request, k, *args, **kwargs):
             if k in [r"打开插件列表", r"请先从插件列表中选择"]: return
-            yield from ArgsGeneralWrapper(crazy_fns[k]["Function"])(*args, **kwargs)
+            yield from ArgsGeneralWrapper(crazy_fns[k]["Function"])(request, *args, **kwargs)
         click_handle = switchy_bt.click(route,[switchy_bt, *input_combo, gr.State(PORT)], output_combo)
         click_handle.then(on_report_generated, [cookies, file_upload, chatbot], [cookies, file_upload, chatbot])
         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)
         stopBtn2.click(fn=None, inputs=None, outputs=None, cancels=cancel_handles)
-
+        demo.load(lambda: 0, inputs=None, outputs=None, _js='()=>{ChatBotHeight();}')
+        
     # 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

request_llm/bridge_all.py

@@ -16,9 +16,6 @@ from toolbox import get_conf, trimmed_format_exc
 from .bridge_chatgpt import predict_no_ui_long_connection as chatgpt_noui
 from .bridge_chatgpt import predict as chatgpt_ui
 
-from .bridge_azure_test import predict_no_ui_long_connection as azure_noui
-from .bridge_azure_test import predict as azure_ui
-
 from .bridge_chatglm import predict_no_ui_long_connection as chatglm_noui
 from .bridge_chatglm import predict as chatglm_ui
 
@@ -48,10 +45,11 @@ class LazyloadTiktoken(object):
         return encoder.decode(*args, **kwargs)
 
 # Endpoint 重定向
-API_URL_REDIRECT, = get_conf("API_URL_REDIRECT")
+API_URL_REDIRECT, AZURE_ENDPOINT, AZURE_ENGINE = get_conf("API_URL_REDIRECT", "AZURE_ENDPOINT", "AZURE_ENGINE")
 openai_endpoint = "https://api.openai.com/v1/chat/completions"
 api2d_endpoint = "https://openai.api2d.net/v1/chat/completions"
 newbing_endpoint = "wss://sydney.bing.com/sydney/ChatHub"
+azure_endpoint = AZURE_ENDPOINT + f'openai/deployments/{AZURE_ENGINE}/chat/completions?api-version=2023-05-15'
 # 兼容旧版的配置
 try:
     API_URL, = get_conf("API_URL")
@@ -122,9 +120,9 @@ model_info = {
 
     # azure openai
     "azure-gpt-3.5":{
-        "fn_with_ui": azure_ui,
-        "fn_without_ui": azure_noui,
-        "endpoint": get_conf("AZURE_ENDPOINT"),
+        "fn_with_ui": chatgpt_ui,
+        "fn_without_ui": chatgpt_noui,
+        "endpoint": azure_endpoint,
         "max_token": 4096,
         "tokenizer": tokenizer_gpt35,
         "token_cnt": get_token_num_gpt35,
@@ -271,6 +269,24 @@ if "newbing" in AVAIL_LLM_MODELS:   # same with newbing-free
         })
     except:
         print(trimmed_format_exc())
+if "chatglmft" in AVAIL_LLM_MODELS:   # same with newbing-free
+    try:
+        from .bridge_chatglmft import predict_no_ui_long_connection as chatglmft_noui
+        from .bridge_chatglmft import predict as chatglmft_ui
+        # claude
+        model_info.update({
+            "chatglmft": {
+                "fn_with_ui": chatglmft_ui,
+                "fn_without_ui": chatglmft_noui,
+                "endpoint": None,
+                "max_token": 4096,
+                "tokenizer": tokenizer_gpt35,
+                "token_cnt": get_token_num_gpt35,
+            }
+        })
+    except:
+        print(trimmed_format_exc())
+
 
 def LLM_CATCH_EXCEPTION(f):
     """
@@ -374,6 +390,6 @@ def predict(inputs, llm_kwargs, *args, **kwargs):
     additional_fn代表点击的哪个按钮，按钮见functional.py
     """
 
-    method = model_info[llm_kwargs['llm_model']]["fn_with_ui"]
+    method = model_info[llm_kwargs['llm_model']]["fn_with_ui"]  # 如果这里报错，检查config中的AVAIL_LLM_MODELS选项
     yield from method(inputs, llm_kwargs, *args, **kwargs)
 

request_llm/bridge_azure_test.py

@@ -1,237 +0,0 @@
-"""
-    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 logging
-import traceback
-import importlib
-import openai
-import time
-import requests
-import json
-
-# 读取config.py文件中关于AZURE OPENAI API的信息
-from toolbox import get_conf, update_ui, clip_history, trimmed_format_exc
-TIMEOUT_SECONDS, MAX_RETRY, AZURE_ENGINE, AZURE_ENDPOINT, AZURE_API_VERSION, AZURE_API_KEY = \
-    get_conf('TIMEOUT_SECONDS', 'MAX_RETRY',"AZURE_ENGINE","AZURE_ENDPOINT", "AZURE_API_VERSION", "AZURE_API_KEY")
-
-
-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, llm_kwargs, plugin_kwargs, chatbot, history=[], system_prompt='', stream = True, additional_fn=None):
-    """
-    发送至azure openai api，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 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"]
-
-    raw_input = inputs
-    logging.info(f'[raw_input] {raw_input}')
-    chatbot.append((inputs, ""))
-    yield from update_ui(chatbot=chatbot, history=history, msg="等待响应") # 刷新界面
-    
-    payload = generate_azure_payload(inputs, llm_kwargs, history, system_prompt, stream)    
-        
-    history.append(inputs); history.append("")
-
-    retry = 0
-    while True:
-        try:
-            openai.api_type = "azure"            
-            openai.api_version = AZURE_API_VERSION
-            openai.api_base = AZURE_ENDPOINT
-            openai.api_key = AZURE_API_KEY
-            response = openai.ChatCompletion.create(timeout=TIMEOUT_SECONDS, **payload);break
-        except openai.error.AuthenticationError:
-            tb_str = '```\n' + trimmed_format_exc() + '```'
-            chatbot[-1] = [chatbot[-1][0], tb_str]
-            yield from update_ui(chatbot=chatbot, history=history, msg="openai返回错误") # 刷新界面
-            return
-        except:
-            retry += 1
-            traceback.print_exc()
-            if retry > MAX_RETRY: raise TimeoutError
-            if MAX_RETRY!=0: print(f'请求超时，正在重试 ({retry}/{MAX_RETRY}) ……')
-            
-    gpt_replying_buffer = ""    
-    is_head_of_the_stream = True
-    if stream:
-
-        stream_response = response
-
-        while True:
-            try:
-                chunk = next(stream_response)
-                    
-            except StopIteration:                
-                from toolbox import regular_txt_to_markdown; tb_str = '```\n' + trimmed_format_exc() + '```'
-                chatbot[-1] = (chatbot[-1][0], f"[Local Message] 远程返回错误: \n\n{tb_str} \n\n{regular_txt_to_markdown(chunk)}")
-                yield from update_ui(chatbot=chatbot, history=history, msg="远程返回错误:" + chunk) # 刷新界面
-                return            
-            
-            if is_head_of_the_stream and (r'"object":"error"' not in chunk):
-                # 数据流的第一帧不携带content
-                is_head_of_the_stream = False; continue
-            
-            if chunk:
-                #print(chunk)
-                try:                     
-                    if "delta" in chunk["choices"][0]:
-                        if chunk["choices"][0]["finish_reason"] == "stop":
-                            logging.info(f'[response] {gpt_replying_buffer}')
-                            break
-                    status_text = f"finish_reason: {chunk['choices'][0]['finish_reason']}"    
-                    gpt_replying_buffer = gpt_replying_buffer + chunk["choices"][0]["delta"]["content"]                               
-                       
-                    history[-1] = gpt_replying_buffer
-                    chatbot[-1] = (history[-2], history[-1])
-                    yield from update_ui(chatbot=chatbot, history=history, msg=status_text) # 刷新界面
-
-                except Exception as e:
-                    traceback.print_exc()
-                    yield from update_ui(chatbot=chatbot, history=history, msg="Json解析不合常规") # 刷新界面
-                    chunk = get_full_error(chunk, stream_response)
-                    
-                    error_msg = chunk                    
-                    yield from update_ui(chatbot=chatbot, history=history, msg="Json异常" + error_msg) # 刷新界面
-                    return
-
-
-def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="", observe_window=None, console_slience=False):
-    """
-    发送至AZURE OPENAI API，等待回复，一次性完成，不显示中间过程。但内部用stream的方法避免中途网线被掐。
-    inputs：
-        是本次问询的输入
-    sys_prompt:
-        系统静默prompt
-    llm_kwargs：
-        chatGPT的内部调优参数
-    history：
-        是之前的对话列表
-    observe_window = None：
-        用于负责跨越线程传递已经输出的部分，大部分时候仅仅为了fancy的视觉效果，留空即可。observe_window[0]：观测窗。observe_window[1]：看门狗
-    """
-    watch_dog_patience = 5 # 看门狗的耐心, 设置5秒即可
-    payload = generate_azure_payload(inputs, llm_kwargs, history, system_prompt=sys_prompt, stream=True)
-    retry = 0
-    while True:
-        try:
-            openai.api_type = "azure"            
-            openai.api_version = AZURE_API_VERSION
-            openai.api_base = AZURE_ENDPOINT
-            openai.api_key = AZURE_API_KEY
-            response = openai.ChatCompletion.create(timeout=TIMEOUT_SECONDS, **payload);break
-        except:
-            retry += 1
-            traceback.print_exc()
-            if retry > MAX_RETRY: raise TimeoutError
-            if MAX_RETRY!=0: print(f'请求超时，正在重试 ({retry}/{MAX_RETRY}) ……')
-
-    stream_response =  response
-    result = ''
-    while True:
-        try: chunk = next(stream_response)
-        except StopIteration: 
-            break
-        except:
-            chunk = next(stream_response) # 失败了，重试一次？再失败就没办法了。
-        if len(chunk)==0: continue
-
-        json_data = json.loads(str(chunk))['choices'][0]
-        delta = json_data["delta"]
-        if len(delta) == 0:
-            break
-        if "role" in delta:
-            continue
-        if "content" in delta: 
-            result += delta["content"]
-            if not console_slience: print(delta["content"], end='')
-            if observe_window is not None: 
-                # 观测窗，把已经获取的数据显示出去
-                if len(observe_window) >= 1: observe_window[0] += delta["content"]
-                # 看门狗，如果超过期限没有喂狗，则终止
-                if len(observe_window) >= 2000:  
-                    if (time.time()-observe_window[1]) > watch_dog_patience:
-                        raise RuntimeError("用户取消了程序。")
-        else:
-            raise RuntimeError("意外Json结构："+delta)
-    if json_data['finish_reason'] == 'content_filter':
-        raise RuntimeError("由于提问含不合规内容被Azure过滤。")
-    if json_data['finish_reason'] == 'length':
-        raise ConnectionAbortedError("正常结束，但显示Token不足，导致输出不完整，请削减单次输入的文本量。")
-    return result
-
-
-def generate_azure_payload(inputs, llm_kwargs, history, system_prompt, stream):
-    """
-    整合所有信息，选择LLM模型，生成 azure openai api请求，为发送请求做准备
-    """    
-
-    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                
-                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_kwargs['llm_model'],
-        "messages": messages, 
-        "temperature": llm_kwargs['temperature'],  # 1.0,
-        "top_p": llm_kwargs['top_p'],  # 1.0,
-        "n": 1,
-        "stream": stream,
-        "presence_penalty": 0,
-        "frequency_penalty": 0,
-        "engine": AZURE_ENGINE
-    }
-    try:
-        print(f" {llm_kwargs['llm_model']} : {conversation_cnt} : {inputs[:100]} ..........")
-    except:
-        print('输入中可能存在乱码。')
-    return payload
-
-

request_llm/bridge_chatglmft.py

@@ -0,0 +1,210 @@
+
+from transformers import AutoModel, AutoTokenizer
+import time
+import os
+import json
+import threading
+import importlib
+from toolbox import update_ui, get_conf
+from multiprocessing import Process, Pipe
+
+load_message = "ChatGLMFT尚未加载，加载需要一段时间。注意，取决于`config.py`的配置，ChatGLMFT消耗大量的内存（CPU）或显存（GPU），也许会导致低配计算机卡死 ……"
+
+def string_to_options(arguments):
+    import argparse
+    import shlex
+    # Create an argparse.ArgumentParser instance
+    parser = argparse.ArgumentParser()
+    # Add command-line arguments
+    parser.add_argument("--llm_to_learn", type=str, help="LLM model to learn", default="gpt-3.5-turbo")
+    parser.add_argument("--prompt_prefix", type=str, help="Prompt prefix", default='')
+    parser.add_argument("--system_prompt", type=str, help="System prompt", default='')
+    parser.add_argument("--batch", type=int, help="System prompt", default=50)
+    # Parse the arguments
+    args = parser.parse_args(shlex.split(arguments))
+    return args
+
+
+#################################################################################
+class GetGLMFTHandle(Process):
+    def __init__(self):
+        super().__init__(daemon=True)
+        self.parent, self.child = Pipe()
+        self.chatglmft_model = None
+        self.chatglmft_tokenizer = None
+        self.info = ""
+        self.success = True
+        self.check_dependency()
+        self.start()
+        self.threadLock = threading.Lock()
+        
+    def check_dependency(self):
+        try:
+            import sentencepiece
+            self.info = "依赖检测通过"
+            self.success = True
+        except:
+            self.info = "缺少ChatGLMFT的依赖，如果要使用ChatGLMFT，除了基础的pip依赖以外，您还需要运行`pip install -r request_llm/requirements_chatglm.txt`安装ChatGLM的依赖。"
+            self.success = False
+
+    def ready(self):
+        return self.chatglmft_model is not None
+
+    def run(self):
+        # 子进程执行
+        # 第一次运行，加载参数
+        retry = 0
+        while True:
+            try:
+                if self.chatglmft_model is None:
+                    from transformers import AutoConfig
+                    import torch
+                    # conf = 'request_llm/current_ptune_model.json'
+                    # if not os.path.exists(conf): raise RuntimeError('找不到微调模型信息')
+                    # with open(conf, 'r', encoding='utf8') as f:
+                    #     model_args = json.loads(f.read())
+                    ChatGLM_PTUNING_CHECKPOINT, = get_conf('ChatGLM_PTUNING_CHECKPOINT')
+                    assert os.path.exists(ChatGLM_PTUNING_CHECKPOINT), "找不到微调模型检查点"
+                    conf = os.path.join(ChatGLM_PTUNING_CHECKPOINT, "config.json")
+                    with open(conf, 'r', encoding='utf8') as f:
+                        model_args = json.loads(f.read())
+                    if 'model_name_or_path' not in model_args:
+                        model_args['model_name_or_path'] = model_args['_name_or_path']
+                    self.chatglmft_tokenizer = AutoTokenizer.from_pretrained(
+                        model_args['model_name_or_path'], trust_remote_code=True)
+                    config = AutoConfig.from_pretrained(
+                        model_args['model_name_or_path'], trust_remote_code=True)
+
+                    config.pre_seq_len = model_args['pre_seq_len']
+                    config.prefix_projection = model_args['prefix_projection']
+
+                    print(f"Loading prefix_encoder weight from {ChatGLM_PTUNING_CHECKPOINT}")
+                    model = AutoModel.from_pretrained(model_args['model_name_or_path'], config=config, trust_remote_code=True)
+                    prefix_state_dict = torch.load(os.path.join(ChatGLM_PTUNING_CHECKPOINT, "pytorch_model.bin"))
+                    new_prefix_state_dict = {}
+                    for k, v in prefix_state_dict.items():
+                        if k.startswith("transformer.prefix_encoder."):
+                            new_prefix_state_dict[k[len("transformer.prefix_encoder."):]] = v
+                    model.transformer.prefix_encoder.load_state_dict(new_prefix_state_dict)
+
+                    if model_args['quantization_bit'] is not None:
+                        print(f"Quantized to {model_args['quantization_bit']} bit")
+                        model = model.quantize(model_args['quantization_bit'])
+                    model = model.cuda()
+                    if model_args['pre_seq_len'] is not None:
+                        # P-tuning v2
+                        model.transformer.prefix_encoder.float()
+                    self.chatglmft_model = model.eval()
+
+                    break
+                else:
+                    break
+            except Exception as e:
+                retry += 1
+                if retry > 3: 
+                    self.child.send('[Local Message] Call ChatGLMFT fail 不能正常加载ChatGLMFT的参数。')
+                    raise RuntimeError("不能正常加载ChatGLMFT的参数！")
+
+        while True:
+            # 进入任务等待状态
+            kwargs = self.child.recv()
+            # 收到消息，开始请求
+            try:
+                for response, history in self.chatglmft_model.stream_chat(self.chatglmft_tokenizer, **kwargs):
+                    self.child.send(response)
+                    # # 中途接收可能的终止指令（如果有的话）
+                    # if self.child.poll(): 
+                    #     command = self.child.recv()
+                    #     if command == '[Terminate]': break
+            except:
+                from toolbox import trimmed_format_exc
+                self.child.send('[Local Message] Call ChatGLMFT fail.' + '\n```\n' + trimmed_format_exc() + '\n```\n')
+            # 请求处理结束，开始下一个循环
+            self.child.send('[Finish]')
+
+    def stream_chat(self, **kwargs):
+        # 主进程执行
+        self.threadLock.acquire()
+        self.parent.send(kwargs)
+        while True:
+            res = self.parent.recv()
+            if res != '[Finish]':
+                yield res
+            else:
+                break
+        self.threadLock.release()
+    
+global glmft_handle
+glmft_handle = None
+#################################################################################
+def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="", observe_window=[], console_slience=False):
+    """
+        多线程方法
+        函数的说明请见 request_llm/bridge_all.py
+    """
+    global glmft_handle
+    if glmft_handle is None:
+        glmft_handle = GetGLMFTHandle()
+        if len(observe_window) >= 1: observe_window[0] = load_message + "\n\n" + glmft_handle.info
+        if not glmft_handle.success: 
+            error = glmft_handle.info
+            glmft_handle = None
+            raise RuntimeError(error)
+
+    # chatglmft 没有 sys_prompt 接口，因此把prompt加入 history
+    history_feedin = []
+    history_feedin.append(["What can I do?", sys_prompt])
+    for i in range(len(history)//2):
+        history_feedin.append([history[2*i], history[2*i+1]] )
+
+    watch_dog_patience = 5 # 看门狗 (watchdog) 的耐心, 设置5秒即可
+    response = ""
+    for response in glmft_handle.stream_chat(query=inputs, history=history_feedin, max_length=llm_kwargs['max_length'], top_p=llm_kwargs['top_p'], temperature=llm_kwargs['temperature']):
+        if len(observe_window) >= 1:  observe_window[0] = response
+        if len(observe_window) >= 2:  
+            if (time.time()-observe_window[1]) > watch_dog_patience:
+                raise RuntimeError("程序终止。")
+    return response
+
+
+
+def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_prompt='', stream = True, additional_fn=None):
+    """
+        单线程方法
+        函数的说明请见 request_llm/bridge_all.py
+    """
+    chatbot.append((inputs, ""))
+
+    global glmft_handle
+    if glmft_handle is None:
+        glmft_handle = GetGLMFTHandle()
+        chatbot[-1] = (inputs, load_message + "\n\n" + glmft_handle.info)
+        yield from update_ui(chatbot=chatbot, history=[])
+        if not glmft_handle.success: 
+            glmft_handle = None
+            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"]
+
+    # 处理历史信息
+    history_feedin = []
+    history_feedin.append(["What can I do?", system_prompt] )
+    for i in range(len(history)//2):
+        history_feedin.append([history[2*i], history[2*i+1]] )
+
+    # 开始接收chatglmft的回复
+    response = "[Local Message]: 等待ChatGLMFT响应中 ..."
+    for response in glmft_handle.stream_chat(query=inputs, history=history_feedin, max_length=llm_kwargs['max_length'], top_p=llm_kwargs['top_p'], temperature=llm_kwargs['temperature']):
+        chatbot[-1] = (inputs, response)
+        yield from update_ui(chatbot=chatbot, history=history)
+
+    # 总结输出
+    if response == "[Local Message]: 等待ChatGLMFT响应中 ...":
+        response = "[Local Message]: ChatGLMFT响应异常 ..."
+    history.extend([inputs, response])
+    yield from update_ui(chatbot=chatbot, history=history)

request_llm/bridge_chatgpt.py

@@ -22,8 +22,8 @@ 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, update_ui, is_any_api_key, select_api_key, what_keys, clip_history, trimmed_format_exc
-proxies, API_KEY, TIMEOUT_SECONDS, MAX_RETRY, API_ORG = \
-    get_conf('proxies', 'API_KEY', 'TIMEOUT_SECONDS', 'MAX_RETRY', 'API_ORG')
+proxies, TIMEOUT_SECONDS, MAX_RETRY, API_ORG = \
+    get_conf('proxies', 'TIMEOUT_SECONDS', 'MAX_RETRY', 'API_ORG')
 
 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。'
@@ -101,6 +101,8 @@ def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="",
                     if (time.time()-observe_window[1]) > watch_dog_patience:
                         raise RuntimeError("用户取消了程序。")
         else: raise RuntimeError("意外Json结构："+delta)
+    if json_data['finish_reason'] == 'content_filter':
+        raise RuntimeError("由于提问含不合规内容被Azure过滤。")
     if json_data['finish_reason'] == 'length':
         raise ConnectionAbortedError("正常结束，但显示Token不足，导致输出不完整，请削减单次输入的文本量。")
     return result
@@ -247,6 +249,7 @@ def generate_payload(inputs, llm_kwargs, history, system_prompt, stream):
         "Authorization": f"Bearer {api_key}"
     }
     if API_ORG.startswith('org-'): headers.update({"OpenAI-Organization": API_ORG})
+    if llm_kwargs['llm_model'].startswith('azure-'): headers.update({"api-key": api_key})
 
     conversation_cnt = len(history) // 2
 

requirements.txt

@@ -1,4 +1,5 @@
 ./docs/gradio-3.32.2-py3-none-any.whl
+pydantic==1.10.11
 tiktoken>=0.3.3
 requests[socks]
 transformers
@@ -15,4 +16,4 @@ pymupdf
 openai
 numpy
 arxiv
-rich
+rich

theme.py

@@ -1,6 +1,6 @@
 import gradio as gr
 from toolbox import get_conf
-CODE_HIGHLIGHT, ADD_WAIFU = get_conf('CODE_HIGHLIGHT', 'ADD_WAIFU')
+CODE_HIGHLIGHT, ADD_WAIFU, LAYOUT = get_conf('CODE_HIGHLIGHT', 'ADD_WAIFU', 'LAYOUT')
 # gradio可用颜色列表
 # gr.themes.utils.colors.slate (石板色)
 # gr.themes.utils.colors.gray (灰色)
@@ -82,20 +82,76 @@ def adjust_theme():
             button_cancel_text_color_dark="white",
         )
 
+        # Layout = "LEFT-RIGHT"
+        js = """
+            <script>
+            function ChatBotHeight() {
+                function update_height(){
+                    var { panel_height_target, chatbot_height, chatbot } = get_elements();
+                    if (panel_height_target!=chatbot_height)
+                    {
+                        var pixelString = panel_height_target.toString() + 'px';
+                        chatbot.style.maxHeight = pixelString; chatbot.style.height = pixelString; 
+                    }
+                }
+
+                function update_height_slow(){
+                    var { panel_height_target, chatbot_height, chatbot } = get_elements();
+                    if (panel_height_target!=chatbot_height)
+                    {
+                        new_panel_height = (panel_height_target - chatbot_height)*0.5 + chatbot_height;
+                        if (Math.abs(new_panel_height - panel_height_target) < 10){
+                            new_panel_height = panel_height_target;
+                        }
+                        // console.log(chatbot_height, panel_height_target, new_panel_height);
+                        var pixelString = new_panel_height.toString() + 'px';
+                        chatbot.style.maxHeight = pixelString; chatbot.style.height = pixelString; 
+                    }
+                }
+
+                update_height();
+                setInterval(function() {
+                    update_height_slow()
+                }, 50); // 每100毫秒执行一次
+            }
+
+            function get_elements() {
+                var chatbot = document.querySelector('#gpt-chatbot > div.wrap.svelte-18telvq');
+                if (!chatbot) {
+                    chatbot = document.querySelector('#gpt-chatbot');
+                }
+                const panel1 = document.querySelector('#input-panel');
+                const panel2 = document.querySelector('#basic-panel');
+                const panel3 = document.querySelector('#plugin-panel');
+                const panel4 = document.querySelector('#interact-panel');
+                const panel5 = document.querySelector('#input-panel2');
+                const panel_active = document.querySelector('#state-panel');
+                var panel_height_target = (20-panel_active.offsetHeight) + panel1.offsetHeight + panel2.offsetHeight + panel3.offsetHeight + panel4.offsetHeight + panel5.offsetHeight + 21;
+                var panel_height_target = parseInt(panel_height_target);
+                var chatbot_height = chatbot.style.height;
+                var chatbot_height = parseInt(chatbot_height);
+                return { panel_height_target, chatbot_height, chatbot };
+            }
+            </script>
+        """
+
+        if LAYOUT=="TOP-DOWN": 
+            js = ""
+            
         # 添加一个萌萌的看板娘
         if ADD_WAIFU:
-            js = """
+            js += """
                 <script src="file=docs/waifu_plugin/jquery.min.js"></script>
                 <script src="file=docs/waifu_plugin/jquery-ui.min.js"></script>
                 <script src="file=docs/waifu_plugin/autoload.js"></script>
             """
-            gradio_original_template_fn = gr.routes.templates.TemplateResponse
-            def gradio_new_template_fn(*args, **kwargs):
-                res = gradio_original_template_fn(*args, **kwargs)
-                res.body = res.body.replace(b'</html>', f'{js}</html>'.encode("utf8"))
-                res.init_headers()
-                return res
-            gr.routes.templates.TemplateResponse = gradio_new_template_fn   # override gradio template
+        gradio_original_template_fn = gr.routes.templates.TemplateResponse
+        def gradio_new_template_fn(*args, **kwargs):
+            res = gradio_original_template_fn(*args, **kwargs)
+            res.body = res.body.replace(b'</html>', f'{js}</html>'.encode("utf8"))
+            res.init_headers()
+            return res
+        gr.routes.templates.TemplateResponse = gradio_new_template_fn   # override gradio template
     except:
         set_theme = None
         print('gradio版本较旧, 不能自定义字体和颜色')

toolbox.py

@@ -4,6 +4,7 @@ import time
 import inspect
 import re
 import os
+import gradio
 from latex2mathml.converter import convert as tex2mathml
 from functools import wraps, lru_cache
 pj = os.path.join
@@ -40,7 +41,7 @@ def ArgsGeneralWrapper(f):
     """
     装饰器函数，用于重组输入参数，改变输入参数的顺序与结构。
     """
-    def decorated(cookies, max_length, llm_model, txt, txt2, top_p, temperature, chatbot, history, system_prompt, plugin_advanced_arg, *args):
+    def decorated(request: gradio.Request, cookies, max_length, llm_model, txt, txt2, top_p, temperature, chatbot, history, system_prompt, plugin_advanced_arg, *args):
         txt_passon = txt
         if txt == "" and txt2 != "": txt_passon = txt2
         # 引入一个有cookie的chatbot
@@ -54,13 +55,21 @@ def ArgsGeneralWrapper(f):
             'top_p':top_p,
             'max_length': max_length,
             'temperature':temperature,
+            'client_ip': request.client.host,
         }
         plugin_kwargs = {
             "advanced_arg": plugin_advanced_arg,
         }
         chatbot_with_cookie = ChatBotWithCookies(cookies)
         chatbot_with_cookie.write_list(chatbot)
-        yield from f(txt_passon, llm_kwargs, plugin_kwargs, chatbot_with_cookie, history, system_prompt, *args)
+        if cookies.get('lock_plugin', None) is None:
+            # 正常状态
+            yield from f(txt_passon, llm_kwargs, plugin_kwargs, chatbot_with_cookie, history, system_prompt, *args)
+        else:
+            # 处理个别特殊插件的锁定状态
+            module, fn_name = cookies['lock_plugin'].split('->')
+            f_hot_reload = getattr(importlib.import_module(module, fn_name), fn_name)
+            yield from f_hot_reload(txt_passon, llm_kwargs, plugin_kwargs, chatbot_with_cookie, history, system_prompt, *args)
     return decorated
 
 
@@ -68,8 +77,21 @@ def update_ui(chatbot, history, msg='正常', **kwargs):  # 刷新界面
     """
     刷新用户界面
     """
-    assert isinstance(chatbot, ChatBotWithCookies), "在传递chatbot的过程中不要将其丢弃。必要时，可用clear将其清空，然后用for+append循环重新赋值。"
-    yield chatbot.get_cookies(), chatbot, history, msg
+    assert isinstance(chatbot, ChatBotWithCookies), "在传递chatbot的过程中不要将其丢弃。必要时, 可用clear将其清空, 然后用for+append循环重新赋值。"
+    cookies = chatbot.get_cookies()
+
+    # 解决插件锁定时的界面显示问题
+    if cookies.get('lock_plugin', None):
+        label = cookies.get('llm_model', "") + " | " + "正在锁定插件" + cookies.get('lock_plugin', None)
+        chatbot_gr = gradio.update(value=chatbot, label=label)
+        if cookies.get('label', "") != label: cookies['label'] = label   # 记住当前的label
+    elif cookies.get('label', None):
+        chatbot_gr = gradio.update(value=chatbot, label=cookies.get('llm_model', ""))
+        cookies['label'] = None    # 清空label
+    else:
+        chatbot_gr = chatbot
+
+    yield cookies, chatbot_gr, history, msg
 
 def update_ui_lastest_msg(lastmsg, chatbot, history, delay=1):  # 刷新界面
     """
@@ -192,7 +214,7 @@ def write_results_to_file(history, file_name=None):
                 # remove everything that cannot be handled by utf8
                 f.write(content.encode('utf-8', 'ignore').decode())
             f.write('\n\n')
-    res = '以上材料已经被写入' + os.path.abspath(f'./gpt_log/{file_name}')
+    res = '以上材料已经被写入:\t' + os.path.abspath(f'./gpt_log/{file_name}')
     print(res)
     return res
 
@@ -445,8 +467,11 @@ def promote_file_to_downloadzone(file, rename_file=None, chatbot=None):
     import shutil
     if rename_file is None: rename_file = f'{gen_time_str()}-{os.path.basename(file)}'
     new_path = os.path.join(f'./gpt_log/', rename_file)
+    # 如果已经存在，先删除
     if os.path.exists(new_path) and not os.path.samefile(new_path, file): os.remove(new_path)
+    # 把文件复制过去
     if not os.path.exists(new_path): shutil.copyfile(file, new_path)
+    # 将文件添加到chatbot cookie中，避免多用户干扰
     if chatbot:
         if 'file_to_promote' in chatbot._cookies: current = chatbot._cookies['file_to_promote']
         else: current = []
@@ -505,16 +530,24 @@ def on_report_generated(cookies, files, chatbot):
     chatbot.append(['报告如何远程获取？', f'报告已经添加到右侧“文件上传区”（可能处于折叠状态），请查收。{file_links}'])
     return cookies, report_files, chatbot
 
+def load_chat_cookies():
+    API_KEY, LLM_MODEL, AZURE_API_KEY = get_conf('API_KEY', 'LLM_MODEL', 'AZURE_API_KEY')
+    if is_any_api_key(AZURE_API_KEY):
+        if is_any_api_key(API_KEY): API_KEY = API_KEY + ',' + AZURE_API_KEY
+        else: API_KEY = AZURE_API_KEY
+    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}$", key)
+    return bool(API_MATCH_ORIGINAL)
+
+def is_azure_api_key(key):
     API_MATCH_AZURE = re.match(r"[a-zA-Z0-9]{32}$", key)
-    return bool(API_MATCH_ORIGINAL) or bool(API_MATCH_AZURE)
+    return bool(API_MATCH_AZURE)
 
 def is_api2d_key(key):
-    if key.startswith('fk') and len(key) == 41:
-        return True
-    else:
-        return False
+    API_MATCH_API2D = re.match(r"fk[a-zA-Z0-9]{6}-[a-zA-Z0-9]{32}$", key)
+    return bool(API_MATCH_API2D)
 
 def is_any_api_key(key):
     if ',' in key:
@@ -523,10 +556,10 @@ def is_any_api_key(key):
             if is_any_api_key(k): return True
         return False
     else:
-        return is_openai_api_key(key) or is_api2d_key(key)
+        return is_openai_api_key(key) or is_api2d_key(key) or is_azure_api_key(key)
 
 def what_keys(keys):
-    avail_key_list = {'OpenAI Key':0, "API2D Key":0}
+    avail_key_list = {'OpenAI Key':0, "Azure Key":0, "API2D Key":0}
     key_list = keys.split(',')
 
     for k in key_list:
@@ -537,7 +570,11 @@ def what_keys(keys):
         if is_api2d_key(k): 
             avail_key_list['API2D Key'] += 1
 
-    return f"检测到： OpenAI Key {avail_key_list['OpenAI Key']} 个，API2D Key {avail_key_list['API2D Key']} 个"
+    for k in key_list:
+        if is_azure_api_key(k): 
+            avail_key_list['Azure Key'] += 1
+
+    return f"检测到： OpenAI Key {avail_key_list['OpenAI Key']} 个, Azure Key {avail_key_list['Azure Key']} 个, API2D Key {avail_key_list['API2D Key']} 个"
 
 def select_api_key(keys, llm_model):
     import random
@@ -552,8 +589,12 @@ def select_api_key(keys, llm_model):
         for k in key_list:
             if is_api2d_key(k): avail_key_list.append(k)
 
+    if llm_model.startswith('azure-'):
+        for k in key_list:
+            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。")
+        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）")
 
     api_key = random.choice(avail_key_list) # 随机负载均衡
     return api_key

version

@@ -1,5 +1,5 @@
 {
-  "version": 3.43,
+  "version": 3.45,
   "show_feature": true,
-  "new_feature": "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": "支持加载自定义的ChatGLM2微调模型 <-> [改善UI] 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"
 }