Merge branch 'master' into frontier

config.py

@@ -302,7 +302,7 @@ ARXIV_CACHE_DIR = "gpt_log/arxiv_cache"
 
 
 # 除了连接OpenAI之外，还有哪些场合允许使用代理，请尽量不要修改
-WHEN_TO_USE_PROXY = ["Download_LLM", "Download_Gradio_Theme", "Connect_Grobid",
+WHEN_TO_USE_PROXY = ["Connect_OpenAI", "Download_LLM", "Download_Gradio_Theme", "Connect_Grobid",
                      "Warmup_Modules", "Nougat_Download", "AutoGen", "Connect_OpenAI_Embedding"]
 
 

main.py

@@ -1,4 +1,4 @@
-import os, json; os.environ['no_proxy'] = '*' # 避免代理网络产生意外污染
+import os; os.environ['no_proxy'] = '*' # 避免代理网络产生意外污染
 
 help_menu_description = \
 """Github源代码开源和更新[地址🚀](https://github.com/binary-husky/gpt_academic),

request_llms/bridge_all.py

@@ -273,6 +273,7 @@ model_info = {
         "token_cnt": get_token_num_gpt4,
         "openai_disable_system_prompt": True,
         "openai_disable_stream": True,
+        "openai_force_temperature_one": True,
     },
 
     "o1-mini": {
@@ -284,6 +285,7 @@ model_info = {
         "token_cnt": get_token_num_gpt4,
         "openai_disable_system_prompt": True,
         "openai_disable_stream": True,
+        "openai_force_temperature_one": True,
     },
 
     "o1-2024-12-17": {
@@ -295,6 +297,7 @@ model_info = {
         "token_cnt": get_token_num_gpt4,
         "openai_disable_system_prompt": True,
         "openai_disable_stream": True,
+        "openai_force_temperature_one": True,
     },
 
     "o1": {
@@ -306,6 +309,7 @@ model_info = {
         "token_cnt": get_token_num_gpt4,
         "openai_disable_system_prompt": True,
         "openai_disable_stream": True,
+        "openai_force_temperature_one": True,
     },
 
     "gpt-4-turbo": {

request_llms/bridge_chatgpt.py

@@ -23,8 +23,13 @@ from loguru import logger
 from toolbox import get_conf, update_ui, is_any_api_key, select_api_key, what_keys, clip_history
 from toolbox import trimmed_format_exc, is_the_upload_folder, read_one_api_model_name, log_chat
 from toolbox import ChatBotWithCookies, have_any_recent_upload_image_files, encode_image
-proxies, TIMEOUT_SECONDS, MAX_RETRY, API_ORG, AZURE_CFG_ARRAY = \
+proxies, WHEN_TO_USE_PROXY, TIMEOUT_SECONDS, MAX_RETRY, API_ORG, AZURE_CFG_ARRAY = \
-    get_conf('proxies', 'TIMEOUT_SECONDS', 'MAX_RETRY', 'API_ORG', 'AZURE_CFG_ARRAY')
+    get_conf('proxies', 'WHEN_TO_USE_PROXY', 'TIMEOUT_SECONDS', 'MAX_RETRY', 'API_ORG', 'AZURE_CFG_ARRAY')
+
+if "Connect_OpenAI" not in WHEN_TO_USE_PROXY:
+    if proxies is not None:
+        logger.error("虽然您配置了代理设置，但不会在连接OpenAI的过程中起作用，请检查WHEN_TO_USE_PROXY配置。")
+        proxies = None
 
 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。'
@@ -180,14 +185,20 @@ def predict_no_ui_long_connection(inputs:str, llm_kwargs:dict, history:list=[],
                 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。")
             else:
                 raise RuntimeError("OpenAI拒绝了请求：" + error_msg)
-        if ('data: [DONE]' in chunk_decoded): break # api2d 正常完成
+        if ('data: [DONE]' in chunk_decoded): break # api2d & one-api 正常完成
         # 提前读取一些信息 （用于判断异常）
         if has_choices and not choice_valid:
             # 一些垃圾第三方接口的出现这样的错误
             continue
         json_data = chunkjson['choices'][0]
         delta = json_data["delta"]
-        if len(delta) == 0: break
+
+        if len(delta) == 0:
+            is_termination_certain = False
+            if (has_choices) and (chunkjson['choices'][0].get('finish_reason', 'null') == 'stop'): is_termination_certain = True
+            if is_termination_certain: break
+            else: continue # 对于不符合规范的狗屎接口，这里需要继续
+
         if (not has_content) and has_role: continue
         if (not has_content) and (not has_role): continue # raise RuntimeError("发现不标准的第三方接口："+delta)
         if has_content: # has_role = True/False
@@ -285,6 +296,8 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
         history.extend([inputs, ""])
 
     retry = 0
+    previous_ui_reflesh_time = 0
+    ui_reflesh_min_interval = 0.0
     while True:
         try:
             # make a POST request to the API endpoint, stream=True
@@ -297,13 +310,13 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
             yield from update_ui(chatbot=chatbot, history=history, msg="请求超时"+retry_msg) # 刷新界面
             if retry > MAX_RETRY: raise TimeoutError
 
-
     if not stream:
         # 该分支仅适用于不支持stream的o1模型，其他情形一律不适用
         yield from handle_o1_model_special(response, inputs, llm_kwargs, chatbot, history)
         return
 
     if stream:
+        reach_termination = False   # 处理一些 new-api 的奇葩异常
         gpt_replying_buffer = ""
         is_head_of_the_stream = True
         stream_response =  response.iter_lines()
@@ -316,11 +329,14 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
                 error_msg = chunk_decoded
                 # 首先排除一个one-api没有done数据包的第三方Bug情形
                 if len(gpt_replying_buffer.strip()) > 0 and len(error_msg) == 0:
-                    yield from update_ui(chatbot=chatbot, history=history, msg="检测到有缺陷的非OpenAI官方接口，建议选择更稳定的接口。")
+                    yield from update_ui(chatbot=chatbot, history=history, msg="检测到有缺陷的接口，建议选择更稳定的接口。")
+                    if not reach_termination:
+                        reach_termination = True
+                        log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
                     break
                 # 其他情况，直接返回报错
                 chatbot, history = handle_error(inputs, llm_kwargs, chatbot, history, chunk_decoded, error_msg)
-                yield from update_ui(chatbot=chatbot, history=history, msg="非OpenAI官方接口返回了错误:" + chunk.decode()) # 刷新界面
+                yield from update_ui(chatbot=chatbot, history=history, msg="接口返回了错误:" + chunk.decode()) # 刷新界面
                 return
 
             # 提前读取一些信息 （用于判断异常）
@@ -330,6 +346,8 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
                 # 数据流的第一帧不携带content
                 is_head_of_the_stream = False; continue
 
+            if "error" in chunk_decoded: logger.error(f"接口返回了未知错误: {chunk_decoded}")
+
             if chunk:
                 try:
                     if has_choices and not choice_valid:
@@ -338,14 +356,25 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
                     if ('data: [DONE]' not in chunk_decoded) and len(chunk_decoded) > 0 and (chunkjson is None):
                         # 传递进来一些奇怪的东西
                         raise ValueError(f'无法读取以下数据，请检查配置。\n\n{chunk_decoded}')
-                    # 前者是API2D的结束条件，后者是OPENAI的结束条件
+                    # 前者是API2D & One-API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):
+                    one_api_terminate = ('data: [DONE]' in chunk_decoded)
-                        # 判定为数据流的结束，gpt_replying_buffer也写完了
+                    openai_terminate = (has_choices) and (len(chunkjson['choices'][0]["delta"]) == 0)
+                    if one_api_terminate or openai_terminate:
+                        is_termination_certain = False
+                        if one_api_terminate: is_termination_certain = True # 抓取符合规范的结束条件
+                        elif (has_choices) and (chunkjson['choices'][0].get('finish_reason', 'null') == 'stop'): is_termination_certain = True # 抓取符合规范的结束条件
+                        if is_termination_certain:
+                            reach_termination = True
                             log_chat(llm_model=llm_kwargs["llm_model"], input_str=inputs, output_str=gpt_replying_buffer)
-                        break
+                            break # 对于符合规范的接口，这里可以break
-                    # 处理数据流的主体
+                        else:
+                            continue # 对于不符合规范的狗屎接口，这里需要继续
+                    # 到这里，我们已经可以假定必须包含choice了
+                    try:
                         status_text = f"finish_reason: {chunkjson['choices'][0].get('finish_reason', 'null')}"
-                    # 如果这里抛出异常，一般是文本过长，详情见get_full_error的输出
+                    except:
+                        logger.error(f"一些垃圾第三方接口出现这样的错误，兼容一下吧: {chunk_decoded}")
+                    # 处理数据流的主体
                     if has_content:
                         # 正常情况
                         gpt_replying_buffer = gpt_replying_buffer + chunkjson['choices'][0]["delta"]["content"]
@@ -354,21 +383,26 @@ def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot:ChatBotWith
                         continue
                     else:
                         # 至此已经超出了正常接口应该进入的范围，一些垃圾第三方接口会出现这样的错误
-                        if chunkjson['choices'][0]["delta"]["content"] is None: continue # 一些垃圾第三方接口出现这样的错误，兼容一下吧
+                        if chunkjson['choices'][0]["delta"].get("content", None) is None:
+                            logger.error(f"一些垃圾第三方接口出现这样的错误，兼容一下吧: {chunk_decoded}")
+                            continue
                         gpt_replying_buffer = gpt_replying_buffer + chunkjson['choices'][0]["delta"]["content"]
 
                     history[-1] = gpt_replying_buffer
                     chatbot[-1] = (history[-2], history[-1])
+                    if time.time() - previous_ui_reflesh_time > ui_reflesh_min_interval:
                         yield from update_ui(chatbot=chatbot, history=history, msg=status_text) # 刷新界面
+                        previous_ui_reflesh_time = time.time()
                 except Exception as e:
                     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)
                     chunk_decoded = chunk.decode()
                     error_msg = chunk_decoded
                     chatbot, history = handle_error(inputs, llm_kwargs, chatbot, history, chunk_decoded, error_msg)
-                    yield from update_ui(chatbot=chatbot, history=history, msg="Json解析异常" + error_msg) # 刷新界面
                     logger.error(error_msg)
+                    yield from update_ui(chatbot=chatbot, history=history, msg="Json解析异常" + error_msg) # 刷新界面
                     return
+        yield from update_ui(chatbot=chatbot, history=history, msg="完成") # 刷新界面
         return  # return from stream-branch
 
 def handle_o1_model_special(response, inputs, llm_kwargs, chatbot, history):
@@ -536,6 +570,8 @@ def generate_payload(inputs:str, llm_kwargs:dict, history:list, system_prompt:st
         "n": 1,
         "stream": stream,
     }
-
+    openai_force_temperature_one = model_info[llm_kwargs['llm_model']].get('openai_force_temperature_one', False)
+    if openai_force_temperature_one:
+        payload.pop('temperature')
     return headers,payload
 

themes/common.js

@@ -1070,26 +1070,31 @@ function restore_chat_from_local_storage(event) {
 }
 
 
-function clear_conversation(a, b, c) {
-    update_conversation_metadata();
-    let stopButton = document.getElementById("elem_stop");
-    stopButton.click();
-    // console.log("clear_conversation");
-    return reset_conversation(a, b);
-}
-
 
 function reset_conversation(a, b) {
     // console.log("js_code_reset");
     a = btoa(unescape(encodeURIComponent(JSON.stringify(a))));
-    setCookie("js_previous_chat_cookie", a, 1);
+    localStorage.setItem("js_previous_chat_cookie", a);
     b = btoa(unescape(encodeURIComponent(JSON.stringify(b))));
-    setCookie("js_previous_history_cookie", b, 1);
+    localStorage.setItem("js_previous_history_cookie", b);
     // gen_restore_btn();
     return [[], [], "已重置"];
 }
 
 
+// clear -> 将 history 缓存至 history_cache -> 点击复原 -> restore_previous_chat() -> 触发elem_update_history -> 读取 history_cache
+function restore_previous_chat() {
+    // console.log("restore_previous_chat");
+    let chat = localStorage.getItem("js_previous_chat_cookie");
+    chat = JSON.parse(decodeURIComponent(escape(atob(chat))));
+    push_data_to_gradio_component(chat, "gpt-chatbot", "obj");
+    let history = localStorage.getItem("js_previous_history_cookie");
+    history = JSON.parse(decodeURIComponent(escape(atob(history))));
+    push_data_to_gradio_component(history, "history-ng", "obj");
+    // document.querySelector("#elem_update_history").click(); // in order to call set_history_gr_state, and send history state to server
+}
+
+
 // clear -> 将 history 缓存至 history_cache -> 点击复原 -> restore_previous_chat() -> 触发elem_update_history -> 读取 history_cache
 function restore_previous_chat() {
     // console.log("restore_previous_chat");

themes/init.js

@@ -1,5 +1,13 @@
+function remove_legacy_cookie() {
+    setCookie("web_cookie_cache", "", -1);
+    setCookie("js_previous_chat_cookie", "", -1);
+    setCookie("js_previous_history_cookie", "", -1);
+}
+
+
 async function GptAcademicJavaScriptInit(dark, prompt, live2d, layout, tts) {
     // 第一部分，布局初始化
+    remove_legacy_cookie();
     audio_fn_init();
     minor_ui_adjustment();
     ButtonWithDropdown_init();