Merge branch 'master' into frontier

2025-12-06 14:36:48 +00:00 · 2024-05-18 15:52:08 +08:00
--- a/config.py
+++ b/config.py
@@ -32,7 +32,8 @@ else:

 # [step 3]>> 模型选择是 (注意: LLM_MODEL是默认选中的模型, 它*必须*被包含在AVAIL_LLM_MODELS列表中 )
 LLM_MODEL = "gpt-3.5-turbo-16k" # 可选 ↓↓↓
-AVAIL_LLM_MODELS = ["gpt-4-1106-preview", "gpt-4-turbo-preview", "gpt-4-vision-preview", "gpt-4-turbo", "gpt-4-turbo-2024-04-09",
+AVAIL_LLM_MODELS = ["gpt-4-1106-preview", "gpt-4-turbo-preview", "gpt-4-vision-preview",
+                    "gpt-4o", "gpt-4-turbo", "gpt-4-turbo-2024-04-09",
                    "gpt-3.5-turbo-1106", "gpt-3.5-turbo-16k", "gpt-3.5-turbo", "azure-gpt-3.5",
                    "gpt-4", "gpt-4-32k", "azure-gpt-4", "glm-4", "glm-4v", "glm-3-turbo",
                    "gemini-pro", "chatglm3"
@@ -44,7 +45,7 @@ AVAIL_LLM_MODELS = ["gpt-4-1106-preview", "gpt-4-turbo-preview", "gpt-4-vision-p
 #   "spark", "sparkv2", "sparkv3", "sparkv3.5",
 #   "qwen-turbo", "qwen-plus", "qwen-max", "qwen-local",
 #   "moonshot-v1-128k", "moonshot-v1-32k", "moonshot-v1-8k",
-#   "gpt-3.5-turbo-0613", "gpt-3.5-turbo-16k-0613", "gpt-3.5-turbo-0125"
+#   "gpt-3.5-turbo-0613", "gpt-3.5-turbo-16k-0613", "gpt-3.5-turbo-0125", "gpt-4o-2024-05-13"
 #   "claude-3-haiku-20240307","claude-3-sonnet-20240229","claude-3-opus-20240229", "claude-2.1", "claude-instant-1.2",
 #   "moss", "llama2", "chatglm_onnx", "internlm", "jittorllms_pangualpha", "jittorllms_llama",
 #   "deepseek-chat" ,"deepseek-coder",
@@ -104,6 +105,10 @@ TIMEOUT_SECONDS = 30
 WEB_PORT = -1


+# 是否自动打开浏览器页面
+AUTO_OPEN_BROWSER = True
+
+
 # 如果OpenAI不响应（网络卡顿、代理失败、KEY失效），重试的次数限制
 MAX_RETRY = 2

@@ -197,7 +202,7 @@ ALIYUN_SECRET=""    # （无需填写）


 # GPT-SOVITS 文本转语音服务的运行地址（将语言模型的生成文本朗读出来）
-TTS_TYPE = "DISABLE" # LOCAL / LOCAL_SOVITS_API / DISABLE
+TTS_TYPE = "DISABLE" # EDGE_TTS / LOCAL_SOVITS_API / DISABLE
 GPT_SOVITS_URL = ""
 EDGE_TTS_VOICE = "zh-CN-XiaoxiaoNeural"

--- a/core_functional.py
+++ b/core_functional.py
@@ -33,6 +33,8 @@ def get_core_functions():
            "AutoClearHistory": False,
            # [6] 文本预处理 （可选参数，默认 None，举例：写个函数移除所有的换行符）
            "PreProcess": None,
+            # [7] 模型选择 （可选参数。如不设置，则使用当前全局模型；如设置，则用指定模型覆盖全局模型。）
+            # "ModelOverride": "gpt-3.5-turbo", # 主要用途：强制点击此基础功能按钮时，使用指定的模型。
        },


--- a/docs/translate_english.json
+++ b/docs/translate_english.json
@@ -313,7 +313,7 @@
    "注意": "Attention",
    "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": "The function plugins marked in 'red' below need to read the path from the input area as a parameter",
    "更多函数插件": "More function plugins",
-    "打开插件列表": "Open plugin list",
+    "点击这里搜索插件列表": "Click Here to Search the Plugin List",
    "高级参数输入区": "Advanced parameter input area",
    "这里是特殊函数插件的高级参数输入区": "Here is the advanced parameter input area for special function plugins",
    "请先从插件列表中选择": "Please select from the plugin list first",
--- a/main.py
+++ b/main.py
@@ -122,7 +122,7 @@ def main():
                            predefined_btns.update({k: functional[k]["Button"]})
                with gr.Accordion("函数插件区", open=True, elem_id="plugin-panel") as area_crazy_fn:
                    with gr.Row():
-                        gr.Markdown("插件可读取“输入区”文本/路径作为参数（上传文件自动修正路径）")
+                        gr.Markdown("<small>插件可读取“输入区”文本/路径作为参数（上传文件自动修正路径）</small>")
                    with gr.Row(elem_id="input-plugin-group"):
                        plugin_group_sel = gr.Dropdown(choices=all_plugin_groups, label='', show_label=False, value=DEFAULT_FN_GROUPS,
                                                      multiselect=True, interactive=True, elem_classes='normal_mut_select').style(container=False)
@@ -142,7 +142,7 @@ def main():
                                if not plugin.get("AsButton", True): dropdown_fn_list.append(k)     # 排除已经是按钮的插件
                                elif plugin.get('AdvancedArgs', False): dropdown_fn_list.append(k)  # 对于需要高级参数的插件，亦在下拉菜单中显示
                            with gr.Row():
-                                dropdown = gr.Dropdown(dropdown_fn_list, value=r"打开插件列表", label="", show_label=False).style(container=False)
+                                dropdown = gr.Dropdown(dropdown_fn_list, value=r"点击这里搜索插件列表", label="", show_label=False).style(container=False)
                            with gr.Row():
                                plugin_advanced_arg = gr.Textbox(show_label=True, label="高级参数输入区", visible=False,
                                                                 placeholder="这里是特殊函数插件的高级参数输入区").style(container=False)
@@ -314,7 +314,7 @@ def main():
        )
        # 随变按钮的回调函数注册
        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
+            if k in [r"点击这里搜索插件列表", r"请先从插件列表中选择"]: return
            yield from ArgsGeneralWrapper(plugins[k]["Function"])(request, *args, **kwargs)
        click_handle = switchy_bt.click(route,[switchy_bt, *input_combo], output_combo)
        click_handle.then(on_report_generated, [cookies, file_upload, chatbot], [cookies, file_upload, chatbot]).then(None, [switchy_bt], None, _js=r"(fn)=>on_plugin_exe_complete(fn)")
@@ -364,8 +364,9 @@ def main():
        def warm_up_mods(): time.sleep(6); warm_up_modules()

        threading.Thread(target=auto_updates, name="self-upgrade", daemon=True).start() # 查看自动更新
-        threading.Thread(target=open_browser, name="open-browser", daemon=True).start() # 打开浏览器页面
        threading.Thread(target=warm_up_mods, name="warm-up",      daemon=True).start() # 预热tiktoken模块
+        if get_conf('AUTO_OPEN_BROWSER'):
+            threading.Thread(target=open_browser, name="open-browser", daemon=True).start() # 打开浏览器页面

    # 运行一些异步任务：自动更新、打开浏览器页面、预热tiktoken模块
    run_delayed_tasks()
--- a/request_llms/bridge_all.py
+++ b/request_llms/bridge_all.py
@@ -179,6 +179,24 @@ model_info = {
        "token_cnt": get_token_num_gpt4,
    },

+    "gpt-4o": {
+        "fn_with_ui": chatgpt_ui,
+        "fn_without_ui": chatgpt_noui,
+        "endpoint": openai_endpoint,
+        "max_token": 128000,
+        "tokenizer": tokenizer_gpt4,
+        "token_cnt": get_token_num_gpt4,
+    },
+
+    "gpt-4o-2024-05-13": {
+        "fn_with_ui": chatgpt_ui,
+        "fn_without_ui": chatgpt_noui,
+        "endpoint": openai_endpoint,
+        "max_token": 128000,
+        "tokenizer": tokenizer_gpt4,
+        "token_cnt": get_token_num_gpt4,
+    },
+
    "gpt-4-turbo-preview": {
        "fn_with_ui": chatgpt_ui,
        "fn_without_ui": chatgpt_noui,
@@ -971,6 +989,13 @@ if len(AZURE_CFG_ARRAY) > 0:
            AVAIL_LLM_MODELS += [azure_model_name]


+# -=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=-=-=
+# -=-=-=-=-=-=-=-=-=- ☝️ 以上是模型路由 -=-=-=-=-=-=-=-=-=
+# -=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=-=-=
+
+# -=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=-=-=
+# -=-=-=-=-=-=-= 👇 以下是多模型路由切换函数 -=-=-=-=-=-=-=
+# -=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=--=-=-=-=-=-=-=-=


 def LLM_CATCH_EXCEPTION(f):
@@ -1007,13 +1032,11 @@ def predict_no_ui_long_connection(inputs:str, llm_kwargs:dict, history:list, sys
    model = llm_kwargs['llm_model']
    n_model = 1
    if '&' not in model:
-
-        # 如果只询问1个大语言模型：
+        # 如果只询问“一个”大语言模型（多数情况）：
        method = model_info[model]["fn_without_ui"]
        return method(inputs, llm_kwargs, history, sys_prompt, observe_window, console_slience)
    else:
-
-        # 如果同时询问多个大语言模型，这个稍微啰嗦一点，但思路相同，您不必读这个else分支
+        # 如果同时询问“多个”大语言模型，这个稍微啰嗦一点，但思路相同，您不必读这个else分支
        executor = ThreadPoolExecutor(max_workers=4)
        models = model.split('&')
        n_model = len(models)
@@ -1066,8 +1089,26 @@ def predict_no_ui_long_connection(inputs:str, llm_kwargs:dict, history:list, sys
        res = '<br/><br/>\n\n---\n\n'.join(return_string_collect)
        return res

+# 根据基础功能区 ModelOverride 参数调整模型类型，用于 `predict` 中
+import importlib
+import core_functional
+def execute_model_override(llm_kwargs, additional_fn, method):
+    functional = core_functional.get_core_functions()
+    if (additional_fn in functional) and 'ModelOverride' in functional[additional_fn]:
+        # 热更新Prompt & ModelOverride
+        importlib.reload(core_functional)
+        functional = core_functional.get_core_functions()
+        model_override = functional[additional_fn]['ModelOverride']
+        if model_override not in model_info:
+            raise ValueError(f"模型覆盖参数 '{model_override}' 指向一个暂不支持的模型，请检查配置文件。")
+        method = model_info[model_override]["fn_with_ui"]
+        llm_kwargs['llm_model'] = model_override
+        return llm_kwargs, additional_fn, method
+    # 默认返回原参数
+    return llm_kwargs, additional_fn, method

-def predict(inputs:str, llm_kwargs:dict, *args, **kwargs):
+def predict(inputs:str, llm_kwargs:dict, plugin_kwargs:dict, chatbot,
+            history:list=[], system_prompt:str='', stream:bool=True, additional_fn:str=None):
    """
    发送至LLM，流式获取输出。
    用于基础的对话功能。
@@ -1086,6 +1127,11 @@ def predict(inputs:str, llm_kwargs:dict, *args, **kwargs):
    """

    inputs = apply_gpt_academic_string_mask(inputs, mode="show_llm")
-    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)
+
+    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
+
+    if additional_fn: # 根据基础功能区 ModelOverride 参数调整模型类型
+        llm_kwargs, additional_fn, method = execute_model_override(llm_kwargs, additional_fn, method)
+
+    yield from method(inputs, llm_kwargs, plugin_kwargs, chatbot, history, system_prompt, stream, additional_fn)

--- a/themes/common.js
+++ b/themes/common.js
@@ -331,6 +331,9 @@ function addCopyButton(botElement, index, is_last_in_arr) {
                toast_push('正在合成语音 & 自动朗读已开启 (再次点击此按钮可禁用自动朗读)。', 3000);
                // toast_push('正在合成语音', 3000);
                const readText = botElement.innerText;
+                prev_chatbot_index = index;
+                prev_text = readText;
+                prev_text_already_pushed = readText;
                push_text_to_audio(readText);
                setCookie("js_auto_read_cookie", "True", 365);
            }
@@ -1033,7 +1036,7 @@ async function GptAcademicJavaScriptInit(dark, prompt, live2d, layout, tts) {


 function reset_conversation(a, b) {
-    console.log("js_code_reset");
+    // console.log("js_code_reset");
    a = btoa(unescape(encodeURIComponent(JSON.stringify(a))));
    setCookie("js_previous_chat_cookie", a, 1);
    gen_restore_btn();
@@ -1173,7 +1176,7 @@ async function on_plugin_exe_complete(fn_name) {
 // -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
 //  第 8 部分: TTS语音生成函数
 // -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
-
+audio_debug = false;
 class AudioPlayer {
    constructor() {
        this.audioCtx = new (window.AudioContext || window.webkitAudioContext)();
@@ -1321,14 +1324,14 @@ function trigger(T, fire) {
 }


-prev_text = "";
-prev_text_already_pushed = "";
+prev_text = ""; // previous text, this is used to check chat changes
+prev_text_already_pushed = ""; // previous text already pushed to audio, this is used to check where we should continue to play audio
 prev_chatbot_index = -1;
 const delay_live_text_update = trigger(3000, on_live_stream_terminate);

 function on_live_stream_terminate(latest_text) {
    // remove `prev_text_already_pushed` from `latest_text`
-    console.log("on_live_stream_terminate", latest_text)
+    if (audio_debug) console.log("on_live_stream_terminate", latest_text);
    remaining_text = latest_text.slice(prev_text_already_pushed.length);
    if ((!isEmptyOrWhitespaceOnly(remaining_text)) && remaining_text.length != 0) {
        prev_text_already_pushed = latest_text;
@@ -1393,19 +1396,19 @@ function process_latest_text_output(text, chatbot_index) {
        delay_live_text_update(text); // in case of no \n or 。 in the text, this timer will finally commit
    }
    else if (chatbot_index == prev_chatbot_index && !is_continue) {
-        console.log('---------------------')
-        console.log('text twisting!')
-        console.log('[new message begin]', 'text', text, 'prev_text_already_pushed', prev_text_already_pushed)
-        console.log('---------------------')
+        if (audio_debug) console.log('---------------------');
+        if (audio_debug) console.log('text twisting!');
+        if (audio_debug) console.log('[new message begin]', 'text', text, 'prev_text_already_pushed', prev_text_already_pushed);
+        if (audio_debug) console.log('---------------------');
        prev_text_already_pushed = "";
        delay_live_text_update(text); // in case of no \n or 。 in the text, this timer will finally commit
    }
    else {
        // on_new_message_begin, we have to clear `prev_text_already_pushed`
-        console.log('---------------------')
-        console.log('new message begin!')
-        console.log('[new message begin]', 'text', text, 'prev_text_already_pushed', prev_text_already_pushed)
-        console.log('---------------------')
+        if (audio_debug) console.log('---------------------');
+        if (audio_debug) console.log('new message begin!');
+        if (audio_debug) console.log('[new message begin]', 'text', text, 'prev_text_already_pushed', prev_text_already_pushed);
+        if (audio_debug) console.log('---------------------');
        prev_text_already_pushed = "";
        process_increased_text(text);
        delay_live_text_update(text); // in case of no \n or 。 in the text, this timer will finally commit
@@ -1433,7 +1436,7 @@ async function push_text_to_audio(text) {
            // Call the async postData function and log the response
            post_text(url, payload, send_index);
            send_index = send_index + 1;
-            console.log(send_index, audio_buf_text)
+            if (audio_debug) console.log(send_index, audio_buf_text);
            // sleep 2 seconds
            if (allow_auto_read_tts_flag) {
                await delay(3000);
@@ -1450,10 +1453,10 @@ to_be_processed = [];
 async function UpdatePlayQueue(cnt, audio_buf_wave) {
    if (cnt != recv_index) {
        to_be_processed.push([cnt, audio_buf_wave]);
-        console.log('cache', cnt);
+        if (audio_debug) console.log('cache', cnt);
    }
    else {
-        console.log('processing', cnt);
+        if (audio_debug) console.log('processing', cnt);
        recv_index = recv_index + 1;
        if (audio_buf_wave) {
            audioPlayer.enqueueAudio(audio_buf_wave);
@@ -1463,7 +1466,7 @@ async function UpdatePlayQueue(cnt, audio_buf_wave) {
            find_any = false;
            for (i = to_be_processed.length - 1; i >= 0; i--) {
                if (to_be_processed[i][0] == recv_index) {
-                    console.log('processing cached', recv_index);
+                    if (audio_debug) console.log('processing cached', recv_index);
                    if (to_be_processed[i][1]) {
                        audioPlayer.enqueueAudio(to_be_processed[i][1]);
                    }
--- a/2
+++ b/2
@@ -1,5 +1,5 @@
 {
  "version": 3.76,
  "show_feature": true,
-  "new_feature": "添加TTS语音输出（EdgeTTS和SoVits语音克隆） <-> Doc2x PDF翻译 <-> 添加回溯对话按钮"
+  "new_feature": "上传文件时显示进度条 <-> 添加TTS语音输出（EdgeTTS和SoVits语音克隆） <-> Doc2x PDF翻译 <-> 添加回溯对话按钮"
 }