Merge branch 'main' into tts-response

# Conflicts: # main/xiaozhi-server/core/handle/intentHandler.py
2026-07-22 07:03:53 +08:00 · 2025-05-23 09:15:15 +08:00
parent fdc6dcb26e 0a9e5d2cea
commit 8119897818
23 changed files with 224 additions and 108 deletions
@@ -142,10 +142,11 @@

 #### 🚀 部署方式选择

-| 部署方式 | 特点 | 适用场景 | Docker部署文档 | 源码部署文档 |
-|---------|------|---------|---------|---------|
-| **最简化安装** | 智能对话、IOT功能，数据存储在配置文件 | 低配置环境，无需数据库 | [Docker只运行Server](./docs/Deployment.md#%E6%96%B9%E5%BC%8F%E4%B8%80docker%E5%8F%AA%E8%BF%90%E8%A1%8Cserver) | [本地源码只运行Server](./docs/Deployment.md#%E6%96%B9%E5%BC%8F%E4%BA%8C%E6%9C%AC%E5%9C%B0%E6%BA%90%E7%A0%81%E5%8F%AA%E8%BF%90%E8%A1%8Cserver)|
-| **全模块安装** | 智能对话、IOT、OTA、智控台，数据存储在数据库 | 完整功能体验 |[Docker运行全模块](./docs/Deployment_all.md#%E6%96%B9%E5%BC%8F%E4%B8%80docker%E8%BF%90%E8%A1%8C%E5%85%A8%E6%A8%A1%E5%9D%97) | [本地源码运行全模块](./docs/Deployment_all.md#%E6%96%B9%E5%BC%8F%E4%BA%8C%E6%9C%AC%E5%9C%B0%E6%BA%90%E7%A0%81%E8%BF%90%E8%A1%8C%E5%85%A8%E6%A8%A1%E5%9D%97) |
+| 部署方式 | 特点 | 适用场景 | Docker部署文档 | 源码部署文档 | 视频教程 | 
+|---------|------|---------|---------|---------|---------|
+| **最简化安装** | 智能对话、IOT功能，数据存储在配置文件 | 低配置环境，无需数据库 | [Docker只运行Server](./docs/Deployment.md#%E6%96%B9%E5%BC%8F%E4%B8%80docker%E5%8F%AA%E8%BF%90%E8%A1%8Cserver) | [本地源码只运行Server](./docs/Deployment.md#%E6%96%B9%E5%BC%8F%E4%BA%8C%E6%9C%AC%E5%9C%B0%E6%BA%90%E7%A0%81%E5%8F%AA%E8%BF%90%E8%A1%8Cserver)| - | 
+| **全模块安装** | 智能对话、IOT、OTA、智控台，数据存储在数据库 | 完整功能体验 |[Docker运行全模块](./docs/Deployment_all.md#%E6%96%B9%E5%BC%8F%E4%B8%80docker%E8%BF%90%E8%A1%8C%E5%85%A8%E6%A8%A1%E5%9D%97) | [本地源码运行全模块](./docs/Deployment_all.md#%E6%96%B9%E5%BC%8F%E4%BA%8C%E6%9C%AC%E5%9C%B0%E6%BA%90%E7%A0%81%E8%BF%90%E8%A1%8C%E5%85%A8%E6%A8%A1%E5%9D%97) | [本地源码视频教程](https://www.bilibili.com/video/BV1wBJhz4Ewe) | 
+

 > 💡 提示：以下是按最新代码部署后的测试平台，有需要可烧录测试，并发为6个，每天会清空数据

@@ -1,9 +1,6 @@
 package xiaozhi.modules.sys.controller;

-import java.util.Arrays;
-import java.util.List;
-import java.util.Map;
-import java.util.UUID;
+import java.util.*;
 import java.util.concurrent.TimeUnit;

 import org.apache.commons.lang3.StringUtils;
@@ -38,7 +35,7 @@ import xiaozhi.modules.sys.utils.WebSocketClientManager;
 * 服务端管理控制器
 */
@RestController
-@RequestMapping("admin/server")
+@RequestMapping("/admin/server")
@Tag(name = "服务端管理")
@AllArgsConstructor
 public class ServerSideManageController {
@@ -46,6 +43,7 @@ public class ServerSideManageController {
    private static final ObjectMapper objectMapper;
    static {
        objectMapper = new ObjectMapper();
+        // 忽略json字符串中存在，但pojo中不存在对应字段的情况
        objectMapper.configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false);
    }

@@ -53,8 +51,11 @@ public class ServerSideManageController {
    @GetMapping("/server-list")
    @RequiresPermissions("sys:role:superAdmin")
    public Result<List<String>> getWsServerList() {
-        String cachedWs = sysParamsService.getValue(Constant.SERVER_WEBSOCKET, true);
-        return new Result<List<String>>().ok(Arrays.asList(cachedWs.split(";")));
+        String wsText = sysParamsService.getValue(Constant.SERVER_WEBSOCKET, true);
+        if (StringUtils.isBlank(wsText)) {
+            return new Result<List<String>>().ok(Collections.emptyList());
+        }
+        return new Result<List<String>>().ok(Arrays.asList(wsText.split(";")));
    }

    @Operation(summary = "通知python服务端更新配置")
@@ -10,9 +10,6 @@ import xiaozhi.modules.sys.enums.ServerActionEnum;

 /**
 * 发送python服务端操作DTO
- *
- * @Author CAIXYPROMISE
- * @since 2025/5/15 4:55
 */
@Data
@NoArgsConstructor
@@ -7,9 +7,6 @@ import java.util.Map;

 /**
 * 服务端动作DTO
- *
- * @Author CAIXYPROMISE
- * @since 2025/5/15 5:10
 */
@Data
 public class ServerActionPayloadDTO
@@ -6,9 +6,6 @@ import xiaozhi.common.exception.RenException;

 /**
 * 服务端动作枚举
- *
- * @Author CAIXYPROMISE
- * @since 2025/5/15 4:57
 */
 public enum ServerActionEnum
 {
@@ -7,9 +7,6 @@ import org.apache.commons.lang3.StringUtils;

 /**
 * 服务端调用响应枚举
- *
- * @Author CAIXYPROMISE
- * @since 2025/5/15 5:50
 */
 public enum ServerActionResponseEnum
 {
@@ -0,0 +1,18 @@
+-- 修改自定义TTS接口请求定义
+update `ai_model_provider` set `fields` =
+'[{"key":"url","label":"服务地址","type":"string"},{"key":"method","label":"请求方式","type":"string"},{"key":"params","label":"请求参数","type":"dict","dict_name":"params"},{"key":"headers","label":"请求头","type":"dict","dict_name":"headers"},{"key":"format","label":"音频格式","type":"string"},{"key":"output_dir","label":"输出目录","type":"string"}]'
+where `id` = 'SYSTEM_TTS_custom';
+
+-- 修改自定义TTS配置说明
+UPDATE `ai_model_config` SET
+`doc_link` = NULL,
+`remark` = '自定义TTS配置说明：
+1. 自定义的TTS接口服务，请求参数可自定义，可接入众多TTS服务
+2. 以本地部署的KokoroTTS为例
+3. 如果只有cpu运行：docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:latest
+4. 如果只有gpu运行：docker run --gpus all -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-gpu:latest
+配置说明：
+1. 在params中配置请求参数,使用JSON格式
+   例如KokoroTTS：{ "input": "{prompt_text}", "speed": 1, "voice": "zm_yunxi", "stream": true, "download_format": "mp3", "response_format": "mp3", "return_download_link": true }
+2. 在headers中配置请求头
+3. 设置返回音频格式' WHERE `id` = 'TTS_CustomTTS';
@@ -155,4 +155,11 @@ databaseChangeLog:
      changes:
        - sqlFile:
            encoding: utf8
-            path: classpath:db/changelog/202505201744.sql
+            path: classpath:db/changelog/202505201744.sql
+  - changeSet:
+      id: 202505151451
+      author: hsoftxl
+      changes:
+        - sqlFile:
+            encoding: utf8
+            path: classpath:db/changelog/202505151451.sql
@@ -28,7 +28,7 @@
      </div>
      <div class="settings-btn" @click="handleChatHistory"
        :class="{ 'disabled-btn': device.memModelId === 'Memory_nomem' }">
-        <el-tooltip v-if="device.memModelId === 'Memory_nomem'" content="未开启记忆" placement="top">
+        <el-tooltip v-if="device.memModelId === 'Memory_nomem'" content="请先在“配置角色”界面开启记忆" placement="top">
          <span>聊天记录</span>
        </el-tooltip>
        <span v-else>聊天记录</span>
@@ -35,10 +35,11 @@
          OTA管理
        </div>
        <el-dropdown v-if="isSuperAdmin" trigger="click" class="equipment-management more-dropdown"
-          :class="{ 'active-tab': $route.path === '/dict-management' || $route.path === '/params-management' || $route.path === '/provider-management' }" @visible-change="handleParamDropdownVisibleChange">
+          :class="{ 'active-tab': $route.path === '/dict-management' || $route.path === '/params-management' || $route.path === '/provider-management' || $route.path === '/server-side-management' }"
+          @visible-change="handleParamDropdownVisibleChange">
          <span class="el-dropdown-link">
            <img loading="lazy" alt="" src="@/assets/header/param_management.png"
-              :style="{ filter: $route.path === '/dict-management' || $route.path === '/params-management' || $route.path === '/provider-management' ? 'brightness(0) invert(1)' : 'None' }" />
+              :style="{ filter: $route.path === '/dict-management' || $route.path === '/params-management' || $route.path === '/provider-management' || $route.path === '/server-side-management' ? 'brightness(0) invert(1)' : 'None' }" />
            参数字典
            <i class="el-icon-arrow-down el-icon--right" :class="{ 'rotate-down': paramDropdownVisible }"></i>
          </span>
@@ -328,6 +329,12 @@ export default {
  line-height: 30px;
 }

+.custom-search-input::v-deep .el-input__suffix-inner {
+  display: flex;
+  align-items: center;
+  height: 100%;
+}
+
 .avatar-img {
  width: 21px;
  height: 21px;
@@ -66,9 +66,12 @@
          <div :key="rowIndex" style="display: flex; gap: 20px; margin-bottom: 0;">
            <el-form-item v-for="field in row" :key="field.prop" :label="field.label" :prop="field.prop"
              style="flex: 1;">
-              <el-input v-model="form.configJson[field.prop]" :placeholder="field.placeholder" :type="field.type"
-                class="custom-input-bg" :show-password="field.type === 'password'">
-              </el-input>
+              <template v-if="field.type === 'json-textarea'">
+                <el-input v-model="fieldJsonMap[field.prop]" type="textarea" :rows="3" placeholder="请输入JSON格式变量(示例:{'key':'value'})"
+                  class="custom-input-bg" @change="(val) => handleJsonChange(field.prop, val)"></el-input>
+              </template>
+              <el-input v-else v-model="form.configJson[field.prop]" :placeholder="field.placeholder" :type="field.type"
+                class="custom-input-bg" :show-password="field.type === 'password'"></el-input>
            </el-form-item>
          </div>
        </template>
@@ -112,6 +115,7 @@ export default {
      pendingProviderType: null,
      pendingModelData: null,
      dynamicCallInfoFields: [],
+      fieldJsonMap: {}, // 用于存储JSON字段的字符串形式
      form: {
        id: "",
        modelType: "",
@@ -175,9 +179,7 @@ export default {
        sort: 0,
        configJson: {}
      };
-      this.dynamicCallInfoFields.forEach(field => {
-        this.$set(this.form.configJson, field.prop, '');
-      });
+      this.fieldJsonMap = {};
    },
    resetProviders() {
      this.providers = [];
@@ -203,7 +205,14 @@ export default {
    handleSave() {
      this.saving = true; // 开始保存加载

-      const provideCode = this.form.configJson.type;
+      // 处理所有JSON字段
+      Object.keys(this.fieldJsonMap).forEach(key => {
+        const parsed = this.validateJson(this.fieldJsonMap[key]);
+        if (parsed !== null) {
+          this.form.configJson[key] = parsed;
+        }
+      });
+
      const formData = {
        id: this.modelData.id,
        modelCode: this.form.modelCode,
@@ -213,13 +222,11 @@ export default {
        docLink: this.form.docLink,
        remark: this.form.remark,
        sort: this.form.sort || 0,
-        configJson: {
-          ...this.form.configJson,
-        }
+        configJson: { ...this.form.configJson }
      };

      this.$emit("save", {
-        provideCode,
+        provideCode: this.form.configJson.type,
        formData,
        done: () => {
          this.saving = false; // 保存完成后回调
@@ -240,7 +247,6 @@ export default {
          value: String(item.providerCode)
        }));
        this.providersLoaded = true;
-
        this.allProvidersData = data;

        if (this.pendingProviderType) {
@@ -255,7 +261,7 @@ export default {
          this.dynamicCallInfoFields = JSON.parse(provider.fields || '[]').map(f => ({
            label: f.label,
            prop: f.key,
-            type: f.type === 'password' ? 'password' : 'text',
+            type: f.type === 'dict' ? 'json-textarea' : (f.type === 'password' ? 'password' : 'text'),
            placeholder: `请输入${f.label}`
          }));

@@ -272,6 +278,9 @@ export default {
      this.dynamicCallInfoFields.forEach(field => {
        if (!configJson.hasOwnProperty(field.prop)) {
          configJson[field.prop] = '';
+        } else if (field.type === 'json-textarea') {
+          this.$set(this.fieldJsonMap, field.prop, this.formatJson(configJson[field.prop]));
+          configJson[field.prop] = this.ensureObject(configJson[field.prop]);
        } else if (typeof configJson[field.prop] !== 'string') {
          configJson[field.prop] = String(configJson[field.prop]);
        }
@@ -287,10 +296,43 @@ export default {
        docLink: model.docLink,
        remark: model.remark,
        sort: Number(model.sort) || 0,
-        configJson: {
-          ...configJson
-        }
+        configJson: { ...configJson }
      };
+    },
+    handleJsonChange(field, value) {
+      const parsed = this.validateJson(value);
+      if (parsed !== null) {
+        this.form.configJson[field] = parsed;
+      }
+    },
+    validateJson(value) {
+      try {
+        const parsed = JSON.parse(value);
+        if (typeof parsed === 'object' && parsed !== null && !Array.isArray(parsed)) {
+          return parsed;
+        }
+        this.$message.error({
+          message: '必须输入字典格式（如 {"key":"value"}），保存则使用原数据',
+          showClose: true
+        });
+        return null;
+      } catch (e) {
+        this.$message.error({
+          message: 'JSON格式错误（如 {"key":"value"}），保存则使用原数据',
+          showClose: true
+        });
+        return null;
+      }
+    },
+    formatJson(obj) {
+      try {
+        return JSON.stringify(obj, null, 2);
+      } catch {
+        return '';
+      }
+    },
+    ensureObject(value) {
+      return typeof value === 'object' ? value : {};
    }
  }
 };
@@ -316,7 +358,6 @@ export default {
  justify-content: center;
 }

-
 .custom-close-btn {
  position: absolute;
  top: 20px;
@@ -452,21 +493,14 @@ export default {
  height: 32px;
 }

-
 .custom-form .el-form-item {
  margin-bottom: 20px;
 }

-
-.custom-input-bg .el-input__inner {
-  height: 32px;
-}
-
-
 .custom-form .el-form-item__label {
  color: #3d4566;
  font-weight: normal;
  text-align: right;
  padding-right: 20px;
 }
-</style>
+</style>
@@ -102,7 +102,7 @@ async def main():
        await asyncio.wait(
            [stdin_task, ws_task, ota_task] if ota_task else [stdin_task, ws_task],
            timeout=3.0,
-            return_when=asyncio.ALL_COMPLETED
+            return_when=asyncio.ALL_COMPLETED,
        )
        print("服务器已关闭，程序退出。")

@@ -686,17 +686,24 @@ TTS:
    speed: 1
    output_dir: tmp/
  CustomTTS:
-    # 自定义的TTS接口服务，请求参数可自定义
-    # 要求接口使用GET方式请求，并返回音频文件
+    # 自定义的TTS接口服务，请求参数可自定义，可接入众多TTS服务
+    # 以本地部署的KokoroTTS为例
+    # 如果只有cpu运行：docker run -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-cpu:latest
+    # 如果只有gpu运行：docker run --gpus all -p 8880:8880 ghcr.io/remsky/kokoro-fastapi-gpu:latest
+    # 要求接口使用POST方式请求，并返回音频文件
    type: custom
-    url: "http://127.0.0.1:9880/tts"
+    method: POST
+    url: "http://127.0.0.1:8880/v1/audio/speech"
    params: # 自定义请求参数
-      # text: "{prompt_text}" # {prompt_text}会被替换为实际的提示词内容
-      # speaker: jok老师
-      # speed: 1
-      # foo: bar
-      # testabc: 123456
+      input: "{prompt_text}"
+      response_format: "mp3"
+      download_format: "mp3"
+      voice: "zf_xiaoxiao"
+      lang_code: "z"
+      return_download_link: true
+      speed: 1
+      stream: false
    headers: # 自定义请求头
      # Authorization: Bearer xxxx
-    format: wav # 接口返回的音频格式
+    format: mp3 # 接口返回的音频格式
    output_dir: tmp/
@@ -1,6 +1,12 @@
 from config.logger import setup_logging
 import json
-from plugins_func.register import FunctionRegistry, ActionResponse, Action, ToolType
+from plugins_func.register import (
+    FunctionRegistry,
+    ActionResponse,
+    Action,
+    ToolType,
+    DeviceTypeRegistry,
+)
 from plugins_func.functions.hass_init import append_devices_to_prompt

 TAG = __name__
@@ -10,6 +16,7 @@ class FunctionHandler:
    def __init__(self, conn):
        self.conn = conn
        self.config = conn.config
+        self.device_type_registry = DeviceTypeRegistry()
        self.function_registry = FunctionRegistry()
        self.register_nessary_functions()
        self.register_config_functions()
@@ -54,7 +61,7 @@ class FunctionHandler:
        self.function_registry.register_function("plugin_loader")
        self.function_registry.register_function("get_time")
        self.function_registry.register_function("get_lunar")
-        self.function_registry.register_function("handle_device")
+        self.function_registry.register_function("handle_speaker_volume_or_screen_brightness")

    def register_config_functions(self):
        """注册配置中的函数,可以不同客户端使用不同的配置"""
@@ -40,8 +40,8 @@ async def checkWakeupWords(conn, text):
    if not enable_wakeup_words_response_cache:
        return False
    """检查是否是唤醒词"""
-    _, text = remove_punctuation_and_length(text)
-    if text in conn.config.get("wakeup_words"):
+    _, filtered_text = remove_punctuation_and_length(text)
+    if filtered_text in conn.config.get("wakeup_words"):
        await send_stt_message(conn, text)
        conn.tts_first_text_index = 0
        conn.tts_last_text_index = 0
@@ -12,11 +12,12 @@ TAG = __name__

 async def handle_user_intent(conn, text):
    # 检查是否有明确的退出命令
-    if await check_direct_exit(conn, text):
+    filtered_text = remove_punctuation_and_length(text)[1]
+    if await check_direct_exit(conn, filtered_text):
        return True
    # 4月4日因流式改造暂时关闭唤醒词加速功能
-    # # 检查是否是唤醒词
-    # if await checkWakeupWords(conn, text):
+    # 检查是否是唤醒词
+    # if await checkWakeupWords(conn, filtered_text):
    #     return True

    if conn.intent_type == "function_call":
@@ -1,9 +1,8 @@
 import json
 import asyncio
-from config.logger import setup_logging
 from plugins_func.register import (
-    device_type_registry,
-    register_function,
+    FunctionItem,
+    register_device_function,
    ActionResponse,
    Action,
    ToolType,
@@ -177,7 +176,7 @@ class IotDescriptor:
                self.methods.append(method)


-def register_device_type(descriptor):
+def register_device_type(descriptor, device_type_registry):
    """注册设备类型及其功能"""
    device_name = descriptor["name"]
    type_id = device_type_registry.generate_device_type_id(descriptor)
@@ -213,10 +212,12 @@ def register_device_type(descriptor):
            },
        }
        query_func = create_iot_query_function(device_name, prop_name, prop_info)
-        decorated_func = register_function(func_name, func_desc, ToolType.IOT_CTL)(
-            query_func
+        decorated_func = register_device_function(
+            func_name, func_desc, ToolType.IOT_CTL
+        )(query_func)
+        functions[func_name] = FunctionItem(
+            func_name, func_desc, decorated_func, ToolType.IOT_CTL
        )
-        functions[func_name] = decorated_func

    # 为每个方法创建控制函数
    for method_name, method_info in descriptor["methods"].items():
@@ -267,10 +268,12 @@ def register_device_type(descriptor):
            },
        }
        control_func = create_iot_function(device_name, method_name, method_info)
-        decorated_func = register_function(func_name, func_desc, ToolType.IOT_CTL)(
-            control_func
+        decorated_func = register_device_function(
+            func_name, func_desc, ToolType.IOT_CTL
+        )(control_func)
+        functions[func_name] = FunctionItem(
+            func_name, func_desc, decorated_func, ToolType.IOT_CTL
        )
-        functions[func_name] = decorated_func

    device_type_registry.register_device_type(type_id, functions)
    return type_id
@@ -289,7 +292,6 @@ async def handleIotDescriptors(conn, descriptors):
    functions_changed = False

    for descriptor in descriptors:
-
        # 如果descriptor没有properties和methods，则直接跳过
        if "properties" not in descriptor and "methods" not in descriptor:
            continue
@@ -319,13 +321,16 @@ async def handleIotDescriptors(conn, descriptors):

        if conn.load_function_plugin:
            # 注册或获取设备类型
-            type_id = register_device_type(descriptor)
+            device_type_registry = conn.func_handler.device_type_registry
+            type_id = register_device_type(descriptor, device_type_registry)
            device_functions = device_type_registry.get_device_functions(type_id)

            # 在连接级注册设备函数
            if hasattr(conn, "func_handler"):
-                for func_name in device_functions:
-                    conn.func_handler.function_registry.register_function(func_name)
+                for func_name, func_item in device_functions.items():
+                    conn.func_handler.function_registry.register_function(
+                        func_name, func_item
+                    )
                    conn.logger.bind(tag=TAG).info(
                        f"注册IOT函数到function handler: {func_name}"
                    )
@@ -39,14 +39,14 @@ async def handleAudioMessage(conn, audio):
        if len(conn.asr_audio) < 15:
            conn.asr_server_receive = True
        else:
-            text, _ = await conn.asr.speech_to_text(conn.asr_audio, conn.session_id)
-            conn.logger.bind(tag=TAG).info(f"识别文本: {text}")
-            text_len, _ = remove_punctuation_and_length(text)
+            raw_text, _ = await conn.asr.speech_to_text(conn.asr_audio, conn.session_id)  # 确保ASR模块返回原始文本
+            conn.logger.bind(tag=TAG).info(f"识别文本: {raw_text}")
+            text_len, _ = remove_punctuation_and_length(raw_text)
            if text_len > 0:
                # 使用自定义模块进行上报
-                enqueue_asr_report(conn, text, copy.deepcopy(conn.asr_audio))
+                enqueue_asr_report(conn, raw_text, copy.deepcopy(conn.asr_audio))

-                await startToChat(conn, text)
+                await startToChat(conn, raw_text)
            else:
                conn.asr_server_receive = True
        conn.asr_audio.clear()
@@ -45,17 +45,17 @@ async def handleTextMessage(conn, message):
                conn.client_have_voice = False
                conn.asr_audio.clear()
                if "text" in msg_json:
-                    text = msg_json["text"]
-                    _, text = remove_punctuation_and_length(text)
+                    original_text = msg_json["text"]  # 保留原始文本
+                    filtered_len, filtered_text = remove_punctuation_and_length(original_text)

                    # 识别是否是唤醒词
-                    is_wakeup_words = text in conn.config.get("wakeup_words")
+                    is_wakeup_words = filtered_text in conn.config.get("wakeup_words")
                    # 是否开启唤醒词回复
                    enable_greeting = conn.config.get("enable_greeting", True)

                    if is_wakeup_words and not enable_greeting:
                        # 如果是唤醒词，且关闭了唤醒词回复，就不用回答
-                        await send_stt_message(conn, text)
+                        await send_stt_message(conn, original_text)
                        await send_tts_message(conn, "stop", None)
                    elif is_wakeup_words:
                        # 上报纯文字数据（复用ASR上报功能，但不提供音频数据）
@@ -63,9 +63,9 @@ async def handleTextMessage(conn, message):
                        await startToChat(conn, "嘿，你好呀")
                    else:
                        # 上报纯文字数据（复用ASR上报功能，但不提供音频数据）
-                        enqueue_asr_report(conn, text, [])
+                        enqueue_asr_report(conn, original_text, [])
                        # 否则需要LLM对文字内容进行答复
-                        await startToChat(conn, text)
+                        await startToChat(conn, original_text)
        elif msg_json["type"] == "iot":
            conn.logger.bind(tag=TAG).info(f"收到iot消息：{message}")
            if "descriptors" in msg_json:
@@ -53,6 +53,8 @@ class IntentProvider(IntentProviderBase):

        prompt = (
            "你是一个意图识别助手。请分析用户的最后一句话，判断用户意图并调用相应的函数。\n\n"
+            "- 如果用户使用疑问词（如'怎么'、'为什么'、'如何'）询问退出相关的问题（例如'怎么退出了？'），注意这不是让你退出，请返回 {'function_call': {'name': 'continue_chat'}\n"
+            "- 仅当用户明确使用'退出系统'、'结束对话'、'我不想和你说话了'等指令时，才触发 handle_exit_intent\n\n"
            f"{functions_desc}\n"
            "处理步骤:\n"
            "1. 分析用户输入，确定用户意图\n"
@@ -1,4 +1,5 @@
 import os
+import json
 import uuid
 import requests
 from pydub import AudioSegment
@@ -16,11 +17,21 @@ class TTSProvider(TTSProviderBase):
    def __init__(self, config, delete_audio_file):
        super().__init__(config, delete_audio_file)
        self.url = config.get("url")
+        self.method = config.get("method", "GET")
        self.headers = config.get("headers", {})
-        self.params = config.get("params")
        self.format = config.get("format", "wav")
        self.output_file = config.get("output_dir", "tmp/")

+        self.params = config.get("params")
+
+        if isinstance(self.params, str):
+            try:
+                self.params = json.loads(self.params)
+            except json.JSONDecodeError:
+                raise ValueError("Custom TTS配置参数出错,无法将字符串解析为对象")
+        elif not isinstance(self.params, dict):
+            raise TypeError("Custom TTS配置参数出错, 请参考配置说明")
+
    def generate_filename(self):
        return os.path.join(
            self.output_file,
@@ -35,7 +46,10 @@ class TTSProvider(TTSProviderBase):
                v = v.replace("{prompt_text}", text)
            request_params[k] = v

-        resp = requests.get(self.url, params=request_params, headers=self.headers)
+        if self.method.upper() == "POST":
+            resp = requests.post(self.url, json=request_params, headers=self.headers)
+        else:
+            resp = requests.get(self.url, params=request_params, headers=self.headers)
        if resp.status_code == 200:
            with open(tmp_file, "wb") as file:
                file.write(resp.content)
@@ -54,20 +54,29 @@ def _handle_device_action(conn, func, success_message, error_message, *args, **k
 handle_device_function_desc = {
    "type": "function",
    "function": {
-        "name": "handle_device",
+        "name": "handle_speaker_volume_or_screen_brightness",
        "description": (
-            "用户想要获取或者设置设备的音量/亮度大小，或者用户觉得声音/亮度过高或过低，或者用户想提高或降低音量/亮度。"
-            "比如用户说现在亮度多少，参数为：device_type:Screen,action:get。"
-            "比如用户说设置音量为50，参数为：device_type:Speaker,action:set,value:50。"
-            "比如用户说亮度太高了，参数为：device_type:Screen,action:lower。"
-            "比如用户说调大音量，参数为：device_type:Speaker,action:raise。"
+            "用户想要获取或者设置设备的音量/亮度大小，或者用户觉得声音/亮度过高或过低，或者用户想提高或降低音量/亮度。\n"
+            "**严格限制**：仅当用户明确操作 **Speaker（音量）或Screen（亮度）** 时才能调用此函数！\n"
+            "对于其他设备（如AC、Battery、Switch等），请不要调用此函数，而是继续正常的对话。\n\n"
+            "示例：\n"
+            "- 用户说『现在亮度多少』 → 调用函数：device_type: Screen, action: get\n"
+            "- 用户说『设置音量为50』 → 调用函数：device_type: Speaker, action: set, value: 50\n"
+            "- 用户说『亮度太高了』 → 调用函数：device_type: Screen, action: lower\n"
+            "- 用户说『调大音量』 → 调用函数：device_type: Speaker, action: raise\n\n"
+            "**拒绝调用示例**（应继续对话而非调用本函数）：\n"
+            "- 用户说『空调调低一度』 → 不调用（设备类型为AC）\n"
+            "- 用户说『开关灯』 → 不调用（设备类型为Switch）\n"
+            "- 用户说『电量多少』 → 不调用（设备类型为Battery）\n"
        ),
        "parameters": {
            "type": "object",
            "properties": {
                "device_type": {
                    "type": "string",
-                    "description": "设备类型，可选值：Speaker(音量),Screen(亮度)"
+                    "description": "设备类型，**严格限定为Speaker（音量）或Screen（亮度）**，其他设备类型禁止调用此函数",
+                    "enum": ["Speaker", "Screen"]
+
                },
                "action": {
                    "type": "string",
@@ -84,8 +93,8 @@ handle_device_function_desc = {
 }


-@register_function('handle_device', handle_device_function_desc, ToolType.IOT_CTL)
-def handle_device(conn, device_type: str, action: str, value: int = None):
+@register_function('handle_speaker_volume_or_screen_brightness', handle_device_function_desc, ToolType.IOT_CTL)
+def handle_speaker_volume_or_screen_brightness(conn, device_type: str, action: str, value: int = None):
    if device_type == "Speaker":
        method_name, property_name, device_name = "SetVolume", "volume", "音量"
    elif device_type == "Screen":
@@ -77,7 +77,6 @@ class DeviceTypeRegistry:

 # 初始化函数注册字典
 all_function_registry = {}
-device_type_registry = DeviceTypeRegistry()


 def register_function(name, desc, type=None):
@@ -91,13 +90,29 @@ def register_function(name, desc, type=None):
    return decorator


+def register_device_function(name, desc, type=None):
+    """注册设备级别的函数到函数注册字典的装饰器"""
+
+    def decorator(func):
+        logger.bind(tag=TAG).debug(f"设备函数 '{name}' 已加载")
+        return func
+
+    return decorator
+
+
 class FunctionRegistry:
    def __init__(self):
        self.function_registry = {}
        self.logger = setup_logging()

-    def register_function(self, name):
-        # 查找all_function_registry中是否有对应的函数
+    def register_function(self, name, func_item=None):
+        # 如果提供了func_item，直接注册
+        if func_item:
+            self.function_registry[name] = func_item
+            self.logger.bind(tag=TAG).debug(f"函数 '{name}' 直接注册成功")
+            return func_item
+
+        # 否则从all_function_registry中查找
        func = all_function_registry.get(name)
        if not func:
            self.logger.bind(tag=TAG).error(f"函数 '{name}' 未找到")