log user name during chat

* 增加对于讯飞星火大模型Spark4.0的支持

* Create github action sync.yml

* 增加对于智谱glm-4-plus的支持

* feat: change arxiv io param

* catch comment source code exception

* upgrade auto comment

* add security patch

---------

Co-authored-by: GH Action - Upstream Sync <action@github.com>
Co-authored-by: binary-husky <qingxu.fu@outlook.com>

.gitignore

@@ -161,3 +161,5 @@ temp.*
 objdump*
 *.min.*.js
 TODO
+experimental_mods
+search_results

crazy_functions/latex_fns/latex_pickle_io.py

@@ -6,12 +6,16 @@ class SafeUnpickler(pickle.Unpickler):
     def get_safe_classes(self):
         from crazy_functions.latex_fns.latex_actions import LatexPaperFileGroup, LatexPaperSplit
         from crazy_functions.latex_fns.latex_toolbox import LinkedListNode
+        from numpy.core.multiarray import scalar
+        from numpy import dtype
         # 定义允许的安全类
         safe_classes = {
             # 在这里添加其他安全的类
             'LatexPaperFileGroup': LatexPaperFileGroup,
             'LatexPaperSplit': LatexPaperSplit,
             'LinkedListNode': LinkedListNode,
+            'scalar': scalar,
+            'dtype': dtype,
         }
         return safe_classes
 
@@ -22,8 +26,6 @@ class SafeUnpickler(pickle.Unpickler):
         for class_name in self.safe_classes.keys():
             if (class_name in f'{module}.{name}'):
                 match_class_name = class_name
-        if module == 'numpy' or module.startswith('numpy.'):
-            return super().find_class(module, name)
         if match_class_name is not None:
             return self.safe_classes[match_class_name]
         # 如果尝试加载未授权的类，则抛出异常

request_llms/bridge_all.py

@@ -385,6 +385,14 @@ model_info = {
         "tokenizer": tokenizer_gpt35,
         "token_cnt": get_token_num_gpt35,
     },
+    "glm-4-plus":{
+        "fn_with_ui": zhipu_ui,
+        "fn_without_ui": zhipu_noui,
+        "endpoint": None,
+        "max_token": 10124 * 8,
+        "tokenizer": tokenizer_gpt35,
+        "token_cnt": get_token_num_gpt35,
+    },
 
     # api_2d (此后不需要在此处添加api2d的接口了，因为下面的代码会自动添加)
     "api2d-gpt-4": {

request_llms/bridge_chatgpt.py

@@ -341,7 +341,7 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
                     # 前者是API2D的结束条件，后者是OPENAI的结束条件
                     if ('data: [DONE]' in chunk_decoded) or (len(chunkjson['choices'][0]["delta"]) == 0):
                         # 判定为数据流的结束，gpt_replying_buffer也写完了
-                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
                         break
                     # 处理数据流的主体
                     status_text = f"finish_reason: {chunkjson['choices'][0].get('finish_reason', 'null')}"
@@ -375,7 +375,7 @@ def handle_o1_model_special(response, inputs, llm_kwargs, chatbot, history):
     try:
         chunkjson = json.loads(response.content.decode())
         gpt_replying_buffer = chunkjson['choices'][0]["message"]["content"]
-        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
         history[-1] = gpt_replying_buffer
         chatbot[-1] = (history[-2], history[-1])
         yield from update_ui(chatbot=chatbot, history=history) # 刷新界面

request_llms/bridge_chatgpt_vision.py

@@ -184,7 +184,7 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
                         # 判定为数据流的结束，gpt_replying_buffer也写完了
                         lastmsg = chatbot[-1][-1] + f"\n\n\n\n「{llm_kwargs['llm_model']}调用结束，该模型不具备上下文对话能力，如需追问，请及时切换模型。」"
                         yield from update_ui_lastest_msg(lastmsg, chatbot, history, delay=1)
-                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
                         break
                     # 处理数据流的主体
                     status_text = f"finish_reason: {chunkjson['choices'][0].get('finish_reason', 'null')}"

request_llms/bridge_claude.py

@@ -216,7 +216,7 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
                 if need_to_pass:
                     pass
                 elif is_last_chunk:
-                    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+                    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
                     # logger.info(f'[response] {gpt_replying_buffer}')
                     break
                 else:

request_llms/bridge_cohere.py

@@ -223,7 +223,7 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
                         chatbot[-1] = (history[-2], history[-1])
                         yield from update_ui(chatbot=chatbot, history=history, msg="正常") # 刷新界面
                     if chunkjson['event_type'] == 'stream-end':
-                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
                         history[-1] = gpt_replying_buffer
                         chatbot[-1] = (history[-2], history[-1])
                         yield from update_ui(chatbot=chatbot, history=history, msg="正常") # 刷新界面

request_llms/bridge_google_gemini.py

@@ -109,7 +109,7 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
             gpt_replying_buffer += paraphrase['text']    # 使用 json 解析库进行处理
             chatbot[-1] = (inputs, gpt_replying_buffer)
             history[-1] = gpt_replying_buffer
-            log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+            log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
             yield from update_ui(chatbot=chatbot, history=history)
         if error_match:
             history = history[-2]  # 错误的不纳入对话

request_llms/bridge_moonshot.py

@@ -166,7 +166,7 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
             history = history[:-2]
             yield from update_ui(chatbot=chatbot, history=history)  # 刷新界面
             break
-    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_bro_result)
+    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_bro_result, user_name=chatbot.get_user())
 
 def predict_no_ui_long_connection(inputs, llm_kwargs, history=[], sys_prompt="", observe_window=None,
                                   console_slience=False):

request_llms/bridge_openrouter.py

@@ -337,7 +337,7 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
                     # 前者是API2D的结束条件，后者是OPENAI的结束条件
                     if ('data: [DONE]' in chunk_decoded) or (len(chunkjson['choices'][0]["delta"]) == 0):
                         # 判定为数据流的结束，gpt_replying_buffer也写完了
-                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
                         break
                     # 处理数据流的主体
                     status_text = f"finish_reason: {chunkjson['choices'][0].get('finish_reason', 'null')}"
@@ -371,7 +371,7 @@ def handle_o1_model_special(response, inputs, llm_kwargs, chatbot, history):
     try:
         chunkjson = json.loads(response.content.decode())
         gpt_replying_buffer = chunkjson['choices'][0]["message"]["content"]
-        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
+        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer, user_name=chatbot.get_user())
         history[-1] = gpt_replying_buffer
         chatbot[-1] = (history[-2], history[-1])
         yield from update_ui(chatbot=chatbot, history=history) # 刷新界面

request_llms/bridge_qwen.py

@@ -59,7 +59,7 @@ def predict(inputs, llm_kwargs, plugin_kwargs, chatbot, history=[], system_promp
         chatbot[-1] = (inputs, response)
         yield from update_ui(chatbot=chatbot, history=history)
 
-    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=response)
+    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=response, user_name=chatbot.get_user())
     # 总结输出
     if response == f"[Local Message] 等待{model_name}响应中 ...":
         response = f"[Local Message] {model_name}响应异常 ..."

request_llms/bridge_taichu.py

@@ -68,5 +68,5 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
         chatbot[-1] = [inputs, response]
         yield from update_ui(chatbot=chatbot, history=history)
     history.extend([inputs, response])
-    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=response)
+    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=response, user_name=chatbot.get_user())
     yield from update_ui(chatbot=chatbot, history=history)

request_llms/bridge_zhipu.py

@@ -97,5 +97,5 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
         chatbot[-1] = [inputs, response]
         yield from update_ui(chatbot=chatbot, history=history)
     history.extend([inputs, response])
-    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=response)
+    log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=response, user_name=chatbot.get_user())
     yield from update_ui(chatbot=chatbot, history=history)

shared_utils/fastapi_server.py

@@ -138,7 +138,9 @@ def start_app(app_block, CONCURRENT_COUNT, AUTHENTICATION, PORT, SSL_KEYFILE, SS
     app_block.is_sagemaker = False
 
     gradio_app = App.create_app(app_block)
-
+    for route in list(gradio_app.router.routes):
+        if route.path == "/proxy={url_path:path}":
+            gradio_app.router.routes.remove(route)
     # --- --- replace gradio endpoint to forbid access to sensitive files --- ---
     if len(AUTHENTICATION) > 0:
         dependencies = []
@@ -154,9 +156,13 @@ def start_app(app_block, CONCURRENT_COUNT, AUTHENTICATION, PORT, SSL_KEYFILE, SS
         @gradio_app.head("/file={path_or_url:path}", dependencies=dependencies)
         @gradio_app.get("/file={path_or_url:path}", dependencies=dependencies)
         async def file(path_or_url: str, request: fastapi.Request):
-            if len(AUTHENTICATION) > 0:
             if not _authorize_user(path_or_url, request, gradio_app):
                 return "越权访问!"
+            stripped = path_or_url.lstrip().lower()
+            if stripped.startswith("https://") or stripped.startswith("http://"):
+                return "账户密码授权模式下, 禁止链接!"
+            if '../' in stripped:
+                return "非法路径!"
             return await endpoint(path_or_url, request)
 
         from fastapi import Request, status
@@ -167,6 +173,26 @@ def start_app(app_block, CONCURRENT_COUNT, AUTHENTICATION, PORT, SSL_KEYFILE, SS
             response.delete_cookie('access-token')
             response.delete_cookie('access-token-unsecure')
             return response
+    else:
+        dependencies = []
+        endpoint = None
+        for route in list(gradio_app.router.routes):
+            if route.path == "/file/{path:path}":
+                gradio_app.router.routes.remove(route)
+            if route.path == "/file={path_or_url:path}":
+                dependencies = route.dependencies
+                endpoint = route.endpoint
+                gradio_app.router.routes.remove(route)
+        @gradio_app.get("/file/{path:path}", dependencies=dependencies)
+        @gradio_app.head("/file={path_or_url:path}", dependencies=dependencies)
+        @gradio_app.get("/file={path_or_url:path}", dependencies=dependencies)
+        async def file(path_or_url: str, request: fastapi.Request):
+            stripped = path_or_url.lstrip().lower()
+            if stripped.startswith("https://") or stripped.startswith("http://"):
+                return "账户密码授权模式下, 禁止链接!"
+            if '../' in stripped:
+                return "非法路径!"
+            return await endpoint(path_or_url, request)
 
     # --- --- enable TTS (text-to-speech) functionality --- ---
     TTS_TYPE = get_conf("TTS_TYPE")

shared_utils/handle_upload.py

@@ -104,6 +104,7 @@ def extract_archive(file_path, dest_dir):
             logger.info("Successfully extracted zip archive to {}".format(dest_dir))
 
     elif file_extension in [".tar", ".gz", ".bz2"]:
+        try:
             with tarfile.open(file_path, "r:*") as tarobj:
                 # 清理提取路径，移除任何不安全的元素
                 for member in tarobj.getmembers():
@@ -115,6 +116,15 @@ def extract_archive(file_path, dest_dir):
 
                 tarobj.extractall(path=dest_dir)
                 logger.info("Successfully extracted tar archive to {}".format(dest_dir))
+        except tarfile.ReadError as e:
+            if file_extension == ".gz":
+                # 一些特别奇葩的项目，是一个gz文件，里面不是tar，只有一个tex文件
+                import gzip
+                with gzip.open(file_path, 'rb') as f_in:
+                    with open(os.path.join(dest_dir, 'main.tex'), 'wb') as f_out:
+                        f_out.write(f_in.read())
+            else:
+                raise e
 
     # 第三方库，需要预先pip install rarfile
     # 此外，Windows上还需要安装winrar软件，配置其Path环境变量，如"C:\Program Files\WinRAR"才可以

shared_utils/key_pattern_manager.py

@@ -14,6 +14,7 @@ openai_regex = re.compile(
     r"sk-[a-zA-Z0-9_-]{92}$|" +
     r"sk-proj-[a-zA-Z0-9_-]{48}$|"+
     r"sk-proj-[a-zA-Z0-9_-]{124}$|"+
+    r"sk-proj-[a-zA-Z0-9_-]{156}$|"+ #新版apikey位数不匹配故修改此正则表达式
     r"sess-[a-zA-Z0-9]{40}$"
 )
 def is_openai_api_key(key):

toolbox.py

@@ -1029,7 +1029,7 @@ def check_repeat_upload(new_pdf_path, pdf_hash):
     # 如果所有页的内容都相同，返回 True
     return False, None
 
-def log_chat(llm_model: str, input_str: str, output_str: str):
+def log_chat(llm_model: str, input_str: str, output_str: str, user_name: str=default_user_name):
     try:
         if output_str and input_str and llm_model:
             uid = str(uuid.uuid4().hex)
@@ -1038,8 +1038,8 @@ def log_chat(llm_model: str, input_str: str, output_str: str):
             logger.bind(chat_msg=True).info(dedent(
             """
             ╭──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
-            [UID]
+            [UID/USER]
-            {uid}
+            {uid}/{user_name}
             [Model]
             {llm_model}
             [Query]
@@ -1047,6 +1047,6 @@ def log_chat(llm_model: str, input_str: str, output_str: str):
             [Response]
             {output_str}
             ╰──────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯
-            """).format(uid=uid, llm_model=llm_model, input_str=input_str, output_str=output_str))
+            """).format(uid=uid, user_name=user_name, llm_model=llm_model, input_str=input_str, output_str=output_str))
     except:
         logger.error(trimmed_format_exc())