Compare commits

..
93 changed files with 4405 additions and 8574 deletions
-1
View File
@@ -80,7 +80,6 @@ VAD:
7、[如何开启声纹识别](./voiceprint-integration.md)<br/>
8、[新闻插件源配置指南](./newsnow_plugin_config.md)<br/>
9、[知识库ragflow集成指南](./ragflow-integration.md)<br/>
10、[如何部署上下文源](./context-provider-integration.md)<br/>
### 11、语音克隆、本地语音部署相关教程
1、[如何在智控台克隆音色](./huoshan-streamTTS-voice-cloning.md)<br/>
-224
View File
@@ -1,224 +0,0 @@
# 上下文源使用教程
## 概述
`上下文源`,就是为小智系统提示词的上下文添加【数据源】。
`上下文源` 在小智在唤醒那一刻,获取外部系统的数据,并将其动态注入到大模型的系统提示词(System Prompt)中。
让其做到唤醒时感知世界某个事物的状态。
它和MCP、记忆有本质的区别:`上下文源`是强制让小智感知世界的数据;`记忆(Mem)`是让他知道之前聊了什么内容;`MCP(functionc all)`是当需要调用某项能力/知识的时候使用调用。
通过这个功能,在小智唤醒的一刹那,“感知”到:
- 人体健康传感器状态(体温、血压、血氧状态等)
- 业务系统的实时数据(服务器负载、待办数据、股票信息等)
- 任何可以通过 HTTP API 获取的文本信息
**注意**:该功能只是方便小智在唤醒的时候感知事物的状态,而如果想要小智唤醒后实时获取事物的状态,建议在此功能上再结合MCP工具的调用。
## 工作原理
1. **配置源**:用户配置一个或多个 HTTP API 地址。
2. **触发请求**:当系统构建 Prompt 时,如果发现模板中包含 `{{ dynamic_context }}` 占位符,会请求所有配置的 API。
3. **自动注入**:系统会自动将 API 返回的数据格式化为 Markdown 列表,替换 `{{ dynamic_context }}` 占位符。
## 接口规范
为了让小智正确解析数据,您的 API 需要满足以下规范:
- **请求方式**`GET`
- **请求头**:系统会自动添加 `device-id` 字段到 Request Header。
- **响应格式**:必须返回 JSON 格式,且包含 `code``data` 字段。
### 响应示例
**情况 1:返回键值对**
```json
{
"code": 0,
"msg": "success",
"data": {
"客厅温度": "26℃",
"客厅湿度": "45%",
"大门状态": "已关闭"
}
}
```
*注入效果:*
```markdown
<context>
- **客厅温度:** 26℃
- **客厅湿度:** 45%
- **大门状态:** 已关闭
</context>
```
**情况 2:返回列表**
```json
{
"code": 0,
"data": [
"您有10个待办事项",
"当前汽车的行驶速度是100km每小时"
]
}
```
*注入效果:*
```markdown
<context>
- 您有10个待办事项
- 当前汽车的行驶速度是100km每小时
</context>
```
## 配置指南
### 方式 1:智控台配置(全模块部署)
1. 登录智控台,进入**角色配置**页面。
2. 找到**上下文源**配置项(点击“编辑源”按钮)。
3. 点击**添加**,输入您的 API 地址。
4. 如果 API 需要鉴权,可以在**请求头**部分添加 `Authorization` 或其他 Header。
5. 保存配置。
### 方式 2:配置文件配置(单模块部署)
编辑 `xiaozhi-server/data/.config.yaml` 文件,添加 `context_providers` 配置段:
```yaml
# 上下文源配置
context_providers:
- url: "http://api.example.com/data"
headers:
Authorization: "Bearer your-token"
- url: "http://another-api.com/data"
```
## 启用功能
默认情况下,系统的提示词模板文件(`data/.agent-base-prompt.txt`)中已经预置了 `{{ dynamic_context }}` 占位符,您无需手动添加。
**示例:**
```markdown
<context>
【重要!以下信息已实时提供,无需调用工具查询,请直接使用:】
- **设备ID** {{device_id}}
- **当前时间:** {{current_time}}
...
{{ dynamic_context }}
</context>
```
**注意**:如果您不需要使用此功能,可以选择**不配置任何上下文源**,也可以从提示词模板文件中**删除** `{{ dynamic_context }}` 占位符。
## 附录:Mock 测试服务示例
为了方便您测试和开发,我们提供了一个简单的 Python Mock Server 脚本。您可以运行此脚本在本地模拟 API 接口。
**mock_api_server.py**
```python
import http.server
import socketserver
import json
from urllib.parse import urlparse, parse_qs
# 设置端口号
PORT = 8081
class MockRequestHandler(http.server.SimpleHTTPRequestHandler):
def do_GET(self):
# 解析路径和参数
parsed_path = urlparse(self.path)
path = parsed_path.path
query = parse_qs(parsed_path.query)
response_data = {}
status_code = 200
print(f"收到请求: {path}, 参数: {query}")
# Case 1: 模拟健康数据 (返回字典 Dict)
# 路径参数风格: /health
# device_id 从 Header 获取
if path == "/health":
device_id = self.headers.get("device-id", "unknown_device")
print(f"device_id: {device_id}")
response_data = {
"code": 0,
"msg": "success",
"data": {
"测试设备ID": device_id,
"心率": "80 bpm",
"血压": "120/80 mmHg",
"状态": "良好"
}
}
# Case 2: 模拟新闻列表 (返回列表 List)
# 无参数: /news/list
elif path == "/news/list":
response_data = {
"code": 0,
"msg": "success",
"data": [
"今日头条:Python 3.14 发布",
"科技新闻:AI 助手改变生活",
"本地新闻:明日有大雨,记得带伞"
]
}
# Case 3: 模拟天气简报 (返回字符串 String)
# 无参数: /weather/simple
elif path == "/weather/simple":
response_data = {
"code": 0,
"msg": "success",
"data": "今日晴转多云,气温 20-25 度,空气质量优,适合出行。"
}
# Case 4: 模拟设备详情 (Query参数风格)
# 参数风格: /device/info
# device_id 从 Header 获取
elif path == "/device/info":
device_id = self.headers.get("device-id", "unknown_device")
response_data = {
"code": 0,
"msg": "success",
"data": {
"查询方式": "Header参数",
"设备ID": device_id,
"电量": "85%",
"固件": "v2.0.1"
}
}
# Case 5: 404 Not Found
else:
status_code = 404
response_data = {"error": "接口不存在"}
# 发送响应
self.send_response(status_code)
self.send_header('Content-type', 'application/json; charset=utf-8')
self.end_headers()
self.wfile.write(json.dumps(response_data, ensure_ascii=False).encode('utf-8'))
# 启动服务
# 允许地址重用,防止快速重启报错
socketserver.TCPServer.allow_reuse_address = True
with socketserver.TCPServer(("", PORT), MockRequestHandler) as httpd:
print(f"==================================================")
print(f"Mock API Server 已启动: http://localhost:{PORT}")
print(f"可用接口列表:")
print(f"1. [字典] http://localhost:{PORT}/health")
print(f"2. [列表] http://localhost:{PORT}/news/list")
print(f"3. [文本] http://localhost:{PORT}/weather/simple")
print(f"4. [参数] http://localhost:{PORT}/device/info")
print(f"==================================================")
try:
httpd.serve_forever()
except KeyboardInterrupt:
print("\n服务已停止")
```
-1
View File
@@ -71,7 +71,6 @@ docker logs -f mcp-endpoint-server
请你保留好上面两个`接口地址`,下一步要用到。
# 2、全模块部署时,怎么配置MCP接入点
首先,你要开启MCP接入点功能。在智控台,点击顶部`参数字典`,在下拉菜单中,点击`系统功能配置`页面。在页面上勾选`MCP接入点`,点击`保存配置`。在`角色配置`页面,点击`编辑功能`按钮,即可看到`mcp接入点`功能。
如果你是全模块部署,使用管理员账号,登录智控台,点击顶部`参数字典`,选择`参数管理`功能。
+2 -12
View File
@@ -156,14 +156,6 @@ services:
编辑`ragflow/docker`文件夹下的`.env`文件,找到以下配置,逐个搜索,逐个修改!逐个搜索,逐个修改!
下面对于`.env`文件的修改,60%的人会忽略`MYSQL_USER`配置导致ragflow启动不成功,因此,需要强调三次:
强调第一次:如果你的`.env`文件如果没有`MYSQL_USER`配置,请在配置文件增加这项!
强调第二次:如果你的`.env`文件如果没有`MYSQL_USER`配置,请在配置文件增加这项!
强调第三次:如果你的`.env`文件如果没有`MYSQL_USER`配置,请在配置文件增加这项!
``` env
# 端口设置
SVR_WEB_HTTP_PORT=8008 # HTTP端口
@@ -238,11 +230,9 @@ docker-compose -f docker-compose.yml up -d
在弹框中,点击"Create new Key"按钮,生成一个API Key。复制这个`API Key`,你稍后会用到。
# 第二步 配置到智控台
确保你的智控台版本是`0.8.7`或以上。使用超级管理员账号登录到智控台。
确保你的智控台版本是`0.8.7`或以上。使用超级管理员账号登录到智控台。在顶部导航栏中,点击`模型配置`,在左侧导航栏中,点击`知识库`。
首先,你要先开启知识库功能。在顶部导航栏中,点击`参数字典`,在下拉菜单中,点击`系统功能配置`页面。在页面上勾选`知识库`,点击`保存配置`。即可在导航栏看到`知识库`功能
在顶部导航栏中,点击`模型配置`,在左侧导航栏中,点击`知识库`。在列表中找到`RAG_RAGFlow`,点击`编辑`按钮。
在列表中找到`RAG_RAGFlow`,点击`编辑`按钮
在`服务地址`中,填写`http://你的ragflow服务的局域网IP:8008`,例如我的ragflow服务的局域网IP是`192.168.1.100`,那么我就填写`http://192.168.1.100:8008`。
-2
View File
@@ -164,8 +164,6 @@ http://192.168.1.25:8005/voiceprint/health?key=abcd
# 2、全模块部署时,怎么配置声纹识别
## 第一步 配置接口
首先,你要开启声纹识别功能。在智控台,点击顶部`参数字典`,在下拉菜单中,点击`系统功能配置`页面。在页面上勾选`声纹识别`,点击`保存配置`。即可在新建智能体的卡片上看到`声纹识别`按钮。
如果你是全模块部署,使用管理员账号,登录智控台,点击顶部`参数字典`,选择`参数管理`功能。
然后搜索参数`server.voice_print`,此时,它的值应该是`null`值。
@@ -141,11 +141,6 @@ public interface Constant {
*/
String SERVER_MQTT_SECRET = "server.mqtt_signature_key";
/**
* WebSocket认证开关
*/
String SERVER_AUTH_ENABLED = "server.auth.enabled";
/**
* 无记忆
*/
@@ -304,7 +299,7 @@ public interface Constant {
/**
* 版本号
*/
public static final String VERSION = "0.8.10";
public static final String VERSION = "0.8.8";
/**
* 无效固件URL
@@ -228,16 +228,4 @@ public interface ErrorCode {
// 智能体模板相关错误码(补充)
int AGENT_TEMPLATE_NOT_FOUND = 10183; // 默认智能体未找到
// 知识库适配器相关错误码
int RAG_ADAPTER_TYPE_NOT_SUPPORTED = 10184; // 不支持的适配器类型
int RAG_CONFIG_VALIDATION_FAILED = 10185; // RAG配置验证失败
int RAG_ADAPTER_CREATION_FAILED = 10186; // 适配器创建失败
int RAG_ADAPTER_INIT_FAILED = 10187; // 适配器初始化失败
int RAG_ADAPTER_CONNECTION_FAILED = 10188; // 适配器连接测试失败
int RAG_ADAPTER_OPERATION_FAILED = 10189; // 适配器操作失败
int RAG_ADAPTER_NOT_FOUND = 10190; // 适配器未找到
int RAG_ADAPTER_CACHE_ERROR = 10191; // 适配器缓存错误
int RAG_ADAPTER_TYPE_NOT_FOUND = 10192; // 适配器类型未找到
}
@@ -44,7 +44,6 @@ import xiaozhi.modules.agent.entity.AgentEntity;
import xiaozhi.modules.agent.entity.AgentTemplateEntity;
import xiaozhi.modules.agent.service.AgentChatAudioService;
import xiaozhi.modules.agent.service.AgentChatHistoryService;
import xiaozhi.modules.agent.service.AgentContextProviderService;
import xiaozhi.modules.agent.service.AgentPluginMappingService;
import xiaozhi.modules.agent.service.AgentService;
import xiaozhi.modules.agent.service.AgentTemplateService;
@@ -65,7 +64,6 @@ public class AgentController {
private final AgentChatHistoryService agentChatHistoryService;
private final AgentChatAudioService agentChatAudioService;
private final AgentPluginMappingService agentPluginMappingService;
private final AgentContextProviderService agentContextProviderService;
private final RedisUtils redisUtils;
@GetMapping("/list")
@@ -137,8 +135,6 @@ public class AgentController {
agentChatHistoryService.deleteByAgentId(id, true, true);
// 删除关联的插件
agentPluginMappingService.deleteByAgentId(id);
// 删除关联的上下文源配置
agentContextProviderService.deleteByAgentId(id);
// 再删除智能体
agentService.deleteById(id);
return new Result<>();
@@ -1,9 +0,0 @@
package xiaozhi.modules.agent.dao;
import org.apache.ibatis.annotations.Mapper;
import xiaozhi.common.dao.BaseDao;
import xiaozhi.modules.agent.entity.AgentContextProviderEntity;
@Mapper
public interface AgentContextProviderDao extends BaseDao<AgentContextProviderEntity> {
}
@@ -69,9 +69,6 @@ public class AgentUpdateDTO implements Serializable {
@Schema(description = "排序", example = "1", nullable = true)
private Integer sort;
@Schema(description = "上下文源配置", nullable = true)
private List<ContextProviderDTO> contextProviders;
@Data
@Schema(description = "插件函数信息")
public static class FunctionInfo implements Serializable {
@@ -1,19 +0,0 @@
package xiaozhi.modules.agent.dto;
import java.io.Serializable;
import java.util.Map;
import io.swagger.v3.oas.annotations.media.Schema;
import lombok.Data;
@Data
@Schema(description = "上下文源配置DTO")
public class ContextProviderDTO implements Serializable {
private static final long serialVersionUID = 1L;
@Schema(description = "URL地址")
private String url;
@Schema(description = "请求头")
private Map<String, Object> headers;
}
@@ -1,43 +0,0 @@
package xiaozhi.modules.agent.entity;
import java.util.Date;
import java.util.List;
import com.baomidou.mybatisplus.annotation.IdType;
import com.baomidou.mybatisplus.annotation.TableField;
import com.baomidou.mybatisplus.annotation.TableId;
import com.baomidou.mybatisplus.annotation.TableName;
import com.baomidou.mybatisplus.extension.handlers.JacksonTypeHandler;
import io.swagger.v3.oas.annotations.media.Schema;
import lombok.Data;
import xiaozhi.modules.agent.dto.ContextProviderDTO;
@Data
@TableName(value = "ai_agent_context_provider", autoResultMap = true)
@Schema(description = "智能体上下文源配置")
public class AgentContextProviderEntity {
@TableId(type = IdType.ASSIGN_UUID)
@Schema(description = "主键")
private String id;
@Schema(description = "智能体ID")
private String agentId;
@Schema(description = "上下文源配置")
@TableField(typeHandler = JacksonTypeHandler.class)
private List<ContextProviderDTO> contextProviders;
@Schema(description = "创建者")
private Long creator;
@Schema(description = "创建时间")
private Date createdAt;
@Schema(description = "更新者")
private Long updater;
@Schema(description = "更新时间")
private Date updatedAt;
}
@@ -1,25 +0,0 @@
package xiaozhi.modules.agent.service;
import xiaozhi.common.service.BaseService;
import xiaozhi.modules.agent.entity.AgentContextProviderEntity;
public interface AgentContextProviderService extends BaseService<AgentContextProviderEntity> {
/**
* 根据智能体ID获取上下文源配置
* @param agentId 智能体ID
* @return 上下文源配置实体
*/
AgentContextProviderEntity getByAgentId(String agentId);
/**
* 保存或更新上下文源配置
* @param entity 实体
*/
void saveOrUpdateByAgentId(AgentContextProviderEntity entity);
/**
* 根据智能体ID删除上下文源配置
* @param agentId 智能体ID
*/
void deleteByAgentId(String agentId);
}
@@ -1,35 +0,0 @@
package xiaozhi.modules.agent.service.impl;
import org.springframework.stereotype.Service;
import com.baomidou.mybatisplus.core.conditions.query.QueryWrapper;
import xiaozhi.common.service.impl.BaseServiceImpl;
import xiaozhi.modules.agent.dao.AgentContextProviderDao;
import xiaozhi.modules.agent.entity.AgentContextProviderEntity;
import xiaozhi.modules.agent.service.AgentContextProviderService;
@Service
public class AgentContextProviderServiceImpl extends BaseServiceImpl<AgentContextProviderDao, AgentContextProviderEntity> implements AgentContextProviderService {
@Override
public AgentContextProviderEntity getByAgentId(String agentId) {
return baseDao.selectOne(new QueryWrapper<AgentContextProviderEntity>().eq("agent_id", agentId));
}
@Override
public void saveOrUpdateByAgentId(AgentContextProviderEntity entity) {
AgentContextProviderEntity exist = getByAgentId(entity.getAgentId());
if (exist != null) {
entity.setId(exist.getId());
updateById(entity);
} else {
insert(entity);
}
}
@Override
public void deleteByAgentId(String agentId) {
baseDao.delete(new QueryWrapper<AgentContextProviderEntity>().eq("agent_id", agentId));
}
}
@@ -32,12 +32,10 @@ import xiaozhi.modules.agent.dao.AgentDao;
import xiaozhi.modules.agent.dto.AgentCreateDTO;
import xiaozhi.modules.agent.dto.AgentDTO;
import xiaozhi.modules.agent.dto.AgentUpdateDTO;
import xiaozhi.modules.agent.entity.AgentContextProviderEntity;
import xiaozhi.modules.agent.entity.AgentEntity;
import xiaozhi.modules.agent.entity.AgentPluginMapping;
import xiaozhi.modules.agent.entity.AgentTemplateEntity;
import xiaozhi.modules.agent.service.AgentChatHistoryService;
import xiaozhi.modules.agent.service.AgentContextProviderService;
import xiaozhi.modules.agent.service.AgentPluginMappingService;
import xiaozhi.modules.agent.service.AgentService;
import xiaozhi.modules.agent.service.AgentTemplateService;
@@ -64,7 +62,6 @@ public class AgentServiceImpl extends BaseServiceImpl<AgentDao, AgentEntity> imp
private final AgentChatHistoryService agentChatHistoryService;
private final AgentTemplateService agentTemplateService;
private final ModelProviderService modelProviderService;
private final AgentContextProviderService agentContextProviderService;
@Override
public PageData<AgentEntity> adminAgentList(Map<String, Object> params) {
@@ -88,13 +85,6 @@ public class AgentServiceImpl extends BaseServiceImpl<AgentDao, AgentEntity> imp
agent.setChatHistoryConf(Constant.ChatHistoryConfEnum.RECORD_TEXT_AUDIO.getCode());
}
}
// 查询上下文源配置
AgentContextProviderEntity contextProviderEntity = agentContextProviderService.getByAgentId(id);
if (contextProviderEntity != null) {
agent.setContextProviders(contextProviderEntity.getContextProviders());
}
// 无需额外查询插件列表,已通过SQL查询出来
return agent;
}
@@ -341,14 +331,6 @@ public class AgentServiceImpl extends BaseServiceImpl<AgentDao, AgentEntity> imp
agentChatHistoryService.deleteByAgentId(existingEntity.getId(), true, false);
}
// 更新上下文源配置
if (dto.getContextProviders() != null) {
AgentContextProviderEntity contextEntity = new AgentContextProviderEntity();
contextEntity.setAgentId(agentId);
contextEntity.setContextProviders(dto.getContextProviders());
agentContextProviderService.saveOrUpdateByAgentId(contextEntity);
}
boolean b = validateLLMIntentParams(dto.getLlmModelId(), dto.getIntentModelId());
if (!b) {
throw new RenException(ErrorCode.LLM_INTENT_PARAMS_MISMATCH);
@@ -413,20 +395,7 @@ public class AgentServiceImpl extends BaseServiceImpl<AgentDao, AgentEntity> imp
entity.setIntentModelId(template.getIntentModelId());
entity.setSystemPrompt(template.getSystemPrompt());
entity.setSummaryMemory(template.getSummaryMemory());
// 根据记忆模型类型设置默认的chatHistoryConf值
if (template.getMemModelId() != null) {
if (template.getMemModelId().equals("Memory_nomem")) {
// 无记忆功能的模型,默认不记录聊天记录
entity.setChatHistoryConf(0);
} else {
// 有记忆功能的模型,默认记录文本和语音
entity.setChatHistoryConf(2);
}
} else {
entity.setChatHistoryConf(template.getChatHistoryConf());
}
entity.setChatHistoryConf(template.getChatHistoryConf());
entity.setLangCode(template.getLangCode());
entity.setLanguage(template.getLanguage());
}
@@ -5,7 +5,6 @@ import com.baomidou.mybatisplus.extension.handlers.JacksonTypeHandler;
import io.swagger.v3.oas.annotations.media.Schema;
import lombok.Data;
import lombok.EqualsAndHashCode;
import xiaozhi.modules.agent.dto.ContextProviderDTO;
import xiaozhi.modules.agent.entity.AgentEntity;
import xiaozhi.modules.agent.entity.AgentPluginMapping;
@@ -22,7 +21,4 @@ public class AgentInfoVO extends AgentEntity
@Schema(description = "插件列表Id")
@TableField(typeHandler = JacksonTypeHandler.class)
private List<AgentPluginMapping> functions;
@Schema(description = "上下文源配置")
private List<ContextProviderDTO> contextProviders;
}
@@ -20,12 +20,10 @@ import xiaozhi.common.redis.RedisUtils;
import xiaozhi.common.utils.ConvertUtils;
import xiaozhi.common.utils.JsonUtils;
import xiaozhi.modules.agent.dao.AgentVoicePrintDao;
import xiaozhi.modules.agent.entity.AgentContextProviderEntity;
import xiaozhi.modules.agent.entity.AgentEntity;
import xiaozhi.modules.agent.entity.AgentPluginMapping;
import xiaozhi.modules.agent.entity.AgentTemplateEntity;
import xiaozhi.modules.agent.entity.AgentVoicePrintEntity;
import xiaozhi.modules.agent.service.AgentContextProviderService;
import xiaozhi.modules.agent.service.AgentMcpAccessPointService;
import xiaozhi.modules.agent.service.AgentPluginMappingService;
import xiaozhi.modules.agent.service.AgentService;
@@ -55,7 +53,6 @@ public class ConfigServiceImpl implements ConfigService {
private final TimbreService timbreService;
private final AgentPluginMappingService agentPluginMappingService;
private final AgentMcpAccessPointService agentMcpAccessPointService;
private final AgentContextProviderService agentContextProviderService;
private final VoiceCloneService cloneVoiceService;
private final AgentVoicePrintDao agentVoicePrintDao;
@@ -181,13 +178,6 @@ public class ConfigServiceImpl implements ConfigService {
mcpEndpoint = mcpEndpoint.replace("/mcp/", "/call/");
result.put("mcp_endpoint", mcpEndpoint);
}
// 获取上下文源配置
AgentContextProviderEntity contextProviderEntity = agentContextProviderService.getByAgentId(agent.getId());
if (contextProviderEntity != null && contextProviderEntity.getContextProviders() != null && !contextProviderEntity.getContextProviders().isEmpty()) {
result.put("context_providers", contextProviderEntity.getContextProviders());
}
// 获取声纹信息
buildVoiceprintConfig(agent.getId(), result);
@@ -72,12 +72,10 @@ public class DeviceController {
return new Result<String>().error(ErrorCode.MCA_NOT_NULL);
}
// 生成六位验证码
String code;
String key;
String code = String.valueOf(Math.random()).substring(2, 8);
String key = RedisKeys.getDeviceCaptchaKey(code);
String existsMac = null;
do {
code = String.valueOf(Math.random()).substring(2, 8);
key = RedisKeys.getDeviceCaptchaKey(code);
existsMac = (String) redisUtils.get(key);
} while (StringUtils.isNotBlank(existsMac));
@@ -1,8 +1,6 @@
package xiaozhi.modules.device.service.impl;
import java.nio.charset.StandardCharsets;
import java.security.InvalidKeyException;
import java.security.NoSuchAlgorithmException;
import java.time.Instant;
import java.util.Base64;
import java.util.Date;
@@ -171,22 +169,7 @@ public class DeviceServiceImpl extends BaseServiceImpl<DeviceDao, DeviceEntity>
DeviceReportRespDTO.Websocket websocket = new DeviceReportRespDTO.Websocket();
// 从系统参数获取WebSocket URL,如果未配置则使用默认值
String wsUrl = sysParamsService.getValue(Constant.SERVER_WEBSOCKET, true);
// 检查是否启用认证并生成token
String authEnabled = sysParamsService.getValue(Constant.SERVER_AUTH_ENABLED, true);
if ("true".equalsIgnoreCase(authEnabled)) {
try {
// 生成token
String token = generateWebSocketToken(clientId, macAddress);
websocket.setToken(token);
} catch (Exception e) {
log.error("生成WebSocket token失败: {}", e.getMessage());
websocket.setToken("");
}
} else {
websocket.setToken("");
}
websocket.setToken("");
if (StringUtils.isBlank(wsUrl) || wsUrl.equals("null")) {
log.error("WebSocket地址未配置,请登录智控台,在参数管理找到【server.websocket】配置");
wsUrl = "ws://xiaozhi.server.com:8000/xiaozhi/v1/";
@@ -206,7 +189,7 @@ public class DeviceServiceImpl extends BaseServiceImpl<DeviceDao, DeviceEntity>
// 添加MQTT UDP配置
// 从系统参数获取MQTT Gateway地址,仅在配置有效时使用
String mqttUdpConfig = sysParamsService.getValue(Constant.SERVER_MQTT_GATEWAY, true);
String mqttUdpConfig = sysParamsService.getValue(Constant.SERVER_MQTT_GATEWAY, false);
if (mqttUdpConfig != null && !mqttUdpConfig.equals("null") && !mqttUdpConfig.isEmpty()) {
try {
String groupId = deviceById != null && deviceById.getBoard() != null ? deviceById.getBoard()
@@ -511,40 +494,6 @@ public class DeviceServiceImpl extends BaseServiceImpl<DeviceDao, DeviceEntity>
return Base64.getEncoder().encodeToString(signature);
}
/**
* 生成WebSocket认证token 遵循Python端AuthManager的实现逻辑:token = signature.timestamp
*
* @param clientId 客户端ID
* @param username 用户名 (通常为deviceId/macAddress)
* @return 认证token字符串
*/
private String generateWebSocketToken(String clientId, String username)
throws NoSuchAlgorithmException, InvalidKeyException {
// 从系统参数获取密钥
String secretKey = sysParamsService.getValue(Constant.SERVER_SECRET, false);
if (StringUtils.isBlank(secretKey)) {
throw new IllegalStateException("WebSocket认证密钥未配置(server.secret)");
}
// 获取当前时间戳(秒)
long timestamp = System.currentTimeMillis() / 1000;
// 构建签名内容: clientId|username|timestamp
String content = String.format("%s|%s|%d", clientId, username, timestamp);
// 生成HMAC-SHA256签名
Mac hmac = Mac.getInstance("HmacSHA256");
SecretKeySpec keySpec = new SecretKeySpec(secretKey.getBytes(StandardCharsets.UTF_8), "HmacSHA256");
hmac.init(keySpec);
byte[] signature = hmac.doFinal(content.getBytes(StandardCharsets.UTF_8));
// Base64 URL-safe编码签名(去除填充符=)
String signatureBase64 = Base64.getUrlEncoder().withoutPadding().encodeToString(signature);
// 返回格式: signature.timestamp
return String.format("%s.%d", signatureBase64, timestamp);
}
/**
* 构建MQTT配置信息
*
@@ -555,7 +504,7 @@ public class DeviceServiceImpl extends BaseServiceImpl<DeviceDao, DeviceEntity>
private DeviceReportRespDTO.MQTT buildMqttConfig(String macAddress, String groupId)
throws Exception {
// 从环境变量或系统参数获取签名密钥
String signatureKey = sysParamsService.getValue("server.mqtt_signature_key", true);
String signatureKey = sysParamsService.getValue("server.mqtt_signature_key", false);
if (StringUtils.isBlank(signatureKey)) {
log.warn("缺少MQTT_SIGNATURE_KEY,跳过MQTT配置生成");
return null;
@@ -1,22 +0,0 @@
package xiaozhi.modules.knowledge.config;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import xiaozhi.modules.knowledge.rag.KnowledgeBaseAdapterFactory;
/**
* 知识库配置类
* 配置知识库相关的Bean
*/
@Configuration
public class KnowledgeBaseConfig {
/**
* 提供KnowledgeBaseAdapterFactory的Bean实例
* @return KnowledgeBaseAdapterFactory实例
*/
@Bean
public KnowledgeBaseAdapterFactory knowledgeBaseAdapterFactory() {
return new KnowledgeBaseAdapterFactory();
}
}
@@ -77,7 +77,7 @@ public class KnowledgeFilesDTO implements Serializable {
return STATUS_UNSTART;
}
// RAGFlow根据run字段的值直接映射到对应的状态码
// 根据run字段的值直接映射到对应的状态码
switch (run.toUpperCase()) {
case "RUNNING":
return STATUS_RUNNING;
@@ -1,200 +0,0 @@
package xiaozhi.modules.knowledge.rag;
import java.util.List;
import java.util.Map;
import org.springframework.web.multipart.MultipartFile;
import xiaozhi.common.page.PageData;
import xiaozhi.modules.knowledge.dto.KnowledgeFilesDTO;
/**
* 知识库API适配器抽象基类
* 定义通用的知识库操作接口,支持多种后端API实现
*/
public abstract class KnowledgeBaseAdapter {
/**
* 获取适配器类型标识
*
* @return 适配器类型(如:ragflow, milvus, pinecone等)
*/
public abstract String getAdapterType();
/**
* 初始化适配器配置
*
* @param config 配置参数
*/
public abstract void initialize(Map<String, Object> config);
/**
* 验证配置是否有效
*
* @param config 配置参数
* @return 验证结果
*/
public abstract boolean validateConfig(Map<String, Object> config);
/**
* 分页查询文档列表
*
* @param datasetId 知识库ID
* @param queryParams 查询参数
* @param page 页码
* @param limit 每页数量
* @return 分页数据
*/
public abstract PageData<KnowledgeFilesDTO> getDocumentList(String datasetId,
Map<String, Object> queryParams,
Integer page,
Integer limit);
/**
* 根据文档ID获取文档详情
*
* @param datasetId 知识库ID
* @return 文档详情
*/
public abstract KnowledgeFilesDTO getDocumentById(String datasetId, String documentId);
/**
* 上传文档到知识库
*
* @param datasetId 知识库ID
* @param file 上传的文件
* @param name 文档名称
* @param metaFields 元数据字段
* @param chunkMethod 分块方法
* @param parserConfig 解析器配置
* @return 上传的文档信息
*/
public abstract KnowledgeFilesDTO uploadDocument(String datasetId,
MultipartFile file,
String name,
Map<String, Object> metaFields,
String chunkMethod,
Map<String, Object> parserConfig);
/**
* 根据状态分页查询文档列表
*
* @param datasetId 知识库ID
* @param status 文档解析状态
* @param page 页码
* @param limit 每页数量
* @return 分页数据
*/
public abstract PageData<KnowledgeFilesDTO> getDocumentListByStatus(String datasetId,
Integer status,
Integer page,
Integer limit);
/**
* 删除文档
*
* @param datasetId 知识库ID
* @param documentId 文档ID
*/
public abstract void deleteDocument(String datasetId, String documentId);
/**
* 解析文档(切块)
*
* @param datasetId 知识库ID
* @param documentIds 文档ID列表
* @return 解析结果
*/
public abstract boolean parseDocuments(String datasetId, List<String> documentIds);
/**
* 列出指定文档的切片
*
* @param datasetId 知识库ID
* @param documentId 文档ID
* @param keywords 关键词过滤
* @param page 页码
* @param pageSize 每页数量
* @param chunkId 切片ID
* @return 切片列表信息
*/
public abstract Map<String, Object> listChunks(String datasetId,
String documentId,
String keywords,
Integer page,
Integer pageSize,
String chunkId);
/**
* 召回测试 - 从知识库中检索相关切片
*
* @param question 用户查询
* @param datasetIds 数据集ID列表
* @param documentIds 文档ID列表
* @param retrievalParams 检索参数
* @return 召回测试结果
*/
public abstract Map<String, Object> retrievalTest(String question,
List<String> datasetIds,
List<String> documentIds,
Map<String, Object> retrievalParams);
/**
* 测试连接
*
* @return 连接测试结果
*/
public abstract boolean testConnection();
/**
* 获取适配器状态信息
*
* @return 状态信息
*/
public abstract Map<String, Object> getStatus();
/**
* 获取支持的配置参数
*
* @return 配置参数说明
*/
public abstract Map<String, Object> getSupportedConfig();
/**
* 获取默认配置
*
* @return 默认配置
*/
public abstract Map<String, Object> getDefaultConfig();
/**
* 创建数据集
*
* @param createParams 创建参数
* @return 数据集ID
*/
public abstract String createDataset(Map<String, Object> createParams);
/**
* 更新数据集
*
* @param datasetId 数据集ID
* @param updateParams 更新参数
*/
public abstract void updateDataset(String datasetId, Map<String, Object> updateParams);
/**
* 删除数据集
*
* @param datasetId 数据集ID
*/
public abstract void deleteDataset(String datasetId);
/**
* 获取数据集的文档数量
*
* @param datasetId 数据集ID
* @return 文档数量
*/
public abstract Integer getDocumentCount(String datasetId);
}
@@ -1,197 +0,0 @@
package xiaozhi.modules.knowledge.rag;
import java.util.HashMap;
import java.util.Map;
import java.util.Set;
import java.util.concurrent.ConcurrentHashMap;
import lombok.extern.slf4j.Slf4j;
import xiaozhi.common.exception.ErrorCode;
import xiaozhi.common.exception.RenException;
/**
* 知识库适配器工厂类
* 负责创建和管理不同类型的知识库API适配器
*/
@Slf4j
public class KnowledgeBaseAdapterFactory {
// 注册的适配器类型映射
private static final Map<String, Class<? extends KnowledgeBaseAdapter>> adapterRegistry = new HashMap<>();
// 适配器实例缓存
private static final Map<String, KnowledgeBaseAdapter> adapterCache = new ConcurrentHashMap<>();
static {
// 注册内置适配器类型
registerAdapter("ragflow", xiaozhi.modules.knowledge.rag.impl.RAGFlowAdapter.class);
// 可以在这里注册更多适配器类型
}
/**
* 注册新的适配器类型
*
* @param adapterType 适配器类型标识
* @param adapterClass 适配器类
*/
public static void registerAdapter(String adapterType, Class<? extends KnowledgeBaseAdapter> adapterClass) {
if (adapterRegistry.containsKey(adapterType)) {
log.warn("适配器类型 '{}' 已存在,将被覆盖", adapterType);
}
adapterRegistry.put(adapterType, adapterClass);
log.info("注册适配器类型: {} -> {}", adapterType, adapterClass.getSimpleName());
}
/**
* 获取适配器实例
*
* @param adapterType 适配器类型
* @param config 配置参数
* @return 适配器实例
*/
public static KnowledgeBaseAdapter getAdapter(String adapterType, Map<String, Object> config) {
String cacheKey = buildCacheKey(adapterType, config);
// 检查缓存中是否已存在实例
if (adapterCache.containsKey(cacheKey)) {
log.debug("从缓存获取适配器实例: {}", cacheKey);
return adapterCache.get(cacheKey);
}
// 创建新的适配器实例
KnowledgeBaseAdapter adapter = createAdapter(adapterType, config);
// 缓存适配器实例
adapterCache.put(cacheKey, adapter);
log.info("创建并缓存适配器实例: {}", cacheKey);
return adapter;
}
/**
* 获取适配器实例(无配置)
*
* @param adapterType 适配器类型
* @return 适配器实例
*/
public static KnowledgeBaseAdapter getAdapter(String adapterType) {
return getAdapter(adapterType, null);
}
/**
* 获取所有已注册的适配器类型
*
* @return 适配器类型集合
*/
public static Set<String> getRegisteredAdapterTypes() {
return adapterRegistry.keySet();
}
/**
* 检查适配器类型是否已注册
*
* @param adapterType 适配器类型
* @return 是否已注册
*/
public static boolean isAdapterTypeRegistered(String adapterType) {
return adapterRegistry.containsKey(adapterType);
}
/**
* 清除适配器缓存
*/
public static void clearCache() {
int cacheSize = adapterCache.size();
adapterCache.clear();
log.info("清除适配器缓存,共清除 {} 个实例", cacheSize);
}
/**
* 移除特定适配器类型的缓存
*
* @param adapterType 适配器类型
*/
public static void removeCacheByType(String adapterType) {
int removedCount = 0;
for (String cacheKey : adapterCache.keySet()) {
if (cacheKey.startsWith(adapterType + "@")) {
adapterCache.remove(cacheKey);
removedCount++;
}
}
log.info("移除适配器类型 '{}' 的缓存,共移除 {} 个实例", adapterType, removedCount);
}
/**
* 获取适配器工厂状态信息
*
* @return 状态信息
*/
public static Map<String, Object> getFactoryStatus() {
Map<String, Object> status = new HashMap<>();
status.put("registeredAdapterTypes", adapterRegistry.keySet());
status.put("cachedAdapterCount", adapterCache.size());
status.put("cacheKeys", adapterCache.keySet());
return status;
}
/**
* 创建适配器实例
*
* @param adapterType 适配器类型
* @param config 配置参数
* @return 适配器实例
*/
private static KnowledgeBaseAdapter createAdapter(String adapterType, Map<String, Object> config) {
if (!adapterRegistry.containsKey(adapterType)) {
throw new RenException(ErrorCode.RAG_ADAPTER_TYPE_NOT_SUPPORTED,
"不支持的适配器类型: " + adapterType);
}
try {
Class<? extends KnowledgeBaseAdapter> adapterClass = adapterRegistry.get(adapterType);
KnowledgeBaseAdapter adapter = adapterClass.getDeclaredConstructor().newInstance();
// 初始化适配器
if (config != null) {
adapter.initialize(config);
// 验证配置
if (!adapter.validateConfig(config)) {
throw new RenException(ErrorCode.RAG_CONFIG_VALIDATION_FAILED,
"适配器配置验证失败: " + adapterType);
}
}
log.info("成功创建适配器实例: {}", adapterType);
return adapter;
} catch (Exception e) {
log.error("创建适配器实例失败: {}", adapterType, e);
throw new RenException(ErrorCode.RAG_ADAPTER_CREATION_FAILED,
"创建适配器失败: " + adapterType + ", 错误: " + e.getMessage());
}
}
/**
* 构建缓存键
*
* @param adapterType 适配器类型
* @param config 配置参数
* @return 缓存键
*/
private static String buildCacheKey(String adapterType, Map<String, Object> config) {
if (config == null || config.isEmpty()) {
return adapterType + "@default";
}
// 基于配置参数生成缓存键
StringBuilder keyBuilder = new StringBuilder(adapterType + "@");
// 使用配置的哈希值作为缓存键的一部分
int configHash = config.hashCode();
keyBuilder.append(configHash);
return keyBuilder.toString();
}
}
@@ -35,8 +35,6 @@ import xiaozhi.common.utils.MessageUtils;
import xiaozhi.modules.knowledge.dao.KnowledgeBaseDao;
import xiaozhi.modules.knowledge.dto.KnowledgeBaseDTO;
import xiaozhi.modules.knowledge.entity.KnowledgeBaseEntity;
import xiaozhi.modules.knowledge.rag.KnowledgeBaseAdapter;
import xiaozhi.modules.knowledge.rag.KnowledgeBaseAdapterFactory;
import xiaozhi.modules.knowledge.service.KnowledgeBaseService;
import xiaozhi.modules.model.dao.ModelConfigDao;
import xiaozhi.modules.model.entity.ModelConfigEntity;
@@ -108,7 +106,7 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
if (pageData != null && pageData.getList() != null) {
for (KnowledgeBaseDTO knowledgeBase : pageData.getList()) {
try {
Integer documentCount = getDocumentCountFromRAG(knowledgeBase.getDatasetId(),
Integer documentCount = getDocumentCountFromRAGFlow(knowledgeBase.getDatasetId(),
knowledgeBase.getRagModelId());
knowledgeBase.setDocumentCount(documentCount);
} catch (Exception e) {
@@ -148,10 +146,10 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
checkDuplicateKnowledgeBaseName(knowledgeBaseDTO, null);
String datasetId = null;
// 调用RAG API创建数据集
// 调用RAGFlow API创建数据集
try {
Map<String, Object> ragConfig = getValidatedRAGConfig(knowledgeBaseDTO.getRagModelId());
datasetId = createDatasetInRAG(
datasetId = createDatasetInRAGFlow(
knowledgeBaseDTO.getName(),
knowledgeBaseDTO.getDescription(),
ragConfig);
@@ -164,13 +162,13 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
KnowledgeBaseEntity existingEntity = knowledgeBaseDao.selectOne(
new QueryWrapper<KnowledgeBaseEntity>().eq("dataset_id", datasetId));
if (existingEntity != null) {
// 如果datasetId已存在,删除RAG中的数据集并抛出异常
// 如果datasetId已存在,删除RAGFlow中的数据集并抛出异常
try {
Map<String, Object> ragConfig = getValidatedRAGConfig(knowledgeBaseDTO.getRagModelId());
deleteDatasetInRAG(datasetId, ragConfig);
deleteDatasetInRAGFlow(datasetId, ragConfig);
} catch (Exception deleteException) {
// 提供更详细的错误信息,包括异常类型和消息
String errorMessage = "删除重复datasetId的RAG数据集失败: " + deleteException.getClass().getSimpleName();
String errorMessage = "删除重复datasetId的RAGFlow数据集失败: " + deleteException.getClass().getSimpleName();
if (deleteException.getMessage() != null) {
errorMessage += " - " + deleteException.getMessage();
}
@@ -221,17 +219,17 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
// 先校验RAG配置
Map<String, Object> ragConfig = getValidatedRAGConfig(knowledgeBaseDTO.getRagModelId());
// 调用RAG API更新数据集
updateDatasetInRAG(
// 调用RAGFlow API更新数据集
updateDatasetInRAGFlow(
knowledgeBaseDTO.getDatasetId(),
knowledgeBaseDTO.getName(),
knowledgeBaseDTO.getDescription(),
ragConfig);
log.info("RAG API更新成功,datasetId: {}", knowledgeBaseDTO.getDatasetId());
log.info("RAGFlow API更新成功,datasetId: {}", knowledgeBaseDTO.getDatasetId());
} catch (Exception e) {
// 提供更详细的错误信息,包括异常类型和消息
String errorMessage = "更新RAG数据集失败: " + e.getClass().getSimpleName();
String errorMessage = "更新RAGFlow数据集失败: " + e.getClass().getSimpleName();
if (e.getMessage() != null) {
errorMessage += " - " + e.getMessage();
}
@@ -239,7 +237,7 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
throw e;
}
} else {
log.warn("datasetId或ragModelId为空,跳过RAG更新");
log.warn("datasetId或ragModelId为空,跳过RAGFlow更新");
}
KnowledgeBaseEntity entity = ConvertUtils.sourceToTarget(knowledgeBaseDTO, KnowledgeBaseEntity.class);
@@ -290,19 +288,19 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
log.info("找到记录: ID={}, datasetId={}, ragModelId={}",
entity.getId(), entity.getDatasetId(), entity.getRagModelId());
// 先调用RAG API删除数据集
// 先调用RAGFlow API删除数据集
boolean apiDeleteSuccess = false;
if (StringUtils.isNotBlank(entity.getDatasetId()) && StringUtils.isNotBlank(entity.getRagModelId())) {
try {
log.info("开始调用RAG API删除数据集");
log.info("开始调用RAGFlow API删除数据集");
// 在删除前进行RAG配置校验
Map<String, Object> ragConfig = getValidatedRAGConfig(entity.getRagModelId());
deleteDatasetInRAG(entity.getDatasetId(), ragConfig);
log.info("RAG API删除调用完成");
deleteDatasetInRAGFlow(entity.getDatasetId(), ragConfig);
log.info("RAGFlow API删除调用完成");
apiDeleteSuccess = true;
} catch (Exception e) {
// 提供更详细的错误信息,包括异常类型和消息
String errorMessage = "删除RAG数据集失败: " + e.getClass().getSimpleName();
String errorMessage = "删除RAGFlow数据集失败: " + e.getClass().getSimpleName();
if (e.getMessage() != null) {
errorMessage += " - " + e.getMessage();
}
@@ -310,7 +308,7 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
throw e;
}
} else {
log.warn("datasetId或ragModelId为空,跳过RAG删除");
log.warn("datasetId或ragModelId为空,跳过RAGFlow删除");
apiDeleteSuccess = true; // 没有RAG数据集,视为成功
}
@@ -434,133 +432,289 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
}
/**
* RAG配置中提取适配器类型
*
* @param config RAG配置
* @return 适配器类型
* 调用RAGFlow API创建数据集
*/
private String extractAdapterType(Map<String, Object> config) {
if (config == null) {
throw new RenException(ErrorCode.RAG_CONFIG_NOT_FOUND);
private String createDatasetInRAGFlow(String name, String description, Map<String, Object> ragConfig) {
String datasetId = null;
String baseUrl = (String) ragConfig.get("base_url");
String apiKey = (String) ragConfig.get("api_key");
log.info("开始调用RAGFlow API创建数据集, name: {}", name);
log.debug("RAGFlow配置 - baseUrl: {}, apiKey: {}", baseUrl, StringUtils.isBlank(apiKey) ? "未配置" : "已配置");
// 构建请求URL
String url = baseUrl + "/api/v1/datasets";
log.debug("请求URL: {}", url);
// 构建请求体
Map<String, Object> requestBody = new HashMap<>();
String username = SecurityUser.getUser().getUsername();
requestBody.put("name", username + "_" + name);
if (StringUtils.isNotBlank(description)) {
requestBody.put("description", description);
}
log.debug("请求体: {}", requestBody);
// 从配置中提取适配器类型
String adapterType = (String) config.get("type");
// 设置请求头
HttpHeaders headers = new HttpHeaders();
headers.setContentType(MediaType.APPLICATION_JSON);
headers.set("Authorization", "Bearer " + apiKey);
// 验证适配器类型是否存在且非空
if (StringUtils.isBlank(adapterType)) {
throw new RenException(ErrorCode.RAG_ADAPTER_TYPE_NOT_FOUND);
}
// 验证适配器类型是否已注册
if (!KnowledgeBaseAdapterFactory.isAdapterTypeRegistered(adapterType)) {
throw new RenException(ErrorCode.RAG_ADAPTER_TYPE_NOT_SUPPORTED,
"不支持的适配器类型: " + adapterType);
}
return adapterType;
}
/**
* 使用适配器创建数据集
*/
private String createDatasetInRAG(String name, String description, Map<String, Object> ragConfig) {
log.info("开始使用适配器创建数据集, name: {}", name);
HttpEntity<Map<String, Object>> requestEntity = new HttpEntity<>(requestBody, headers);
// 发送POST请求
log.info("发送POST请求到RAGFlow API...");
ResponseEntity<String> response;
try {
// 从RAG配置中提取适配器类型
String adapterType = extractAdapterType(ragConfig);
// 使用适配器工厂获取适配器实例
KnowledgeBaseAdapter adapter = KnowledgeBaseAdapterFactory.getAdapter(adapterType, ragConfig);
// 构建数据集创建参数
Map<String, Object> createParams = new HashMap<>();
String username = SecurityUser.getUser().getUsername();
createParams.put("name", username + "_" + name);
if (StringUtils.isNotBlank(description)) {
createParams.put("description", description);
}
// 调用适配器的创建数据集方法
String datasetId = adapter.createDataset(createParams);
log.info("数据集创建成功,datasetId: {}", datasetId);
return datasetId;
response = restTemplate.exchange(url, HttpMethod.POST, requestEntity, String.class);
} catch (Exception e) {
// 直接传递底层适配器的详细错误信息
log.error("创建数据集失败", e);
if (e instanceof RenException) {
throw (RenException) e;
}
throw new RenException(ErrorCode.RAG_API_ERROR, e.getMessage());
String errorMessage = url + e.getMessage();
log.error(errorMessage, e);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
log.info("RAGFlow API响应状态码: {}", response.getStatusCode());
log.debug("RAGFlow API响应内容: {}", response.getBody());
if (!response.getStatusCode().is2xxSuccessful()) {
String errorMessage = response.getStatusCode() + ", 响应内容: " + response.getBody();
log.error(errorMessage);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
// 解析响应体,提取datasetId
String responseBody = response.getBody();
if (StringUtils.isNotBlank(responseBody)) {
try {
// 解析RAGFlow API响应,支持多种可能的字段名
ObjectMapper objectMapper = new ObjectMapper();
Map<String, Object> responseMap = objectMapper.readValue(responseBody, Map.class);
log.debug("RAGFlow API响应解析结果: {}", responseMap);
// 首先检查响应码
Integer code = (Integer) responseMap.get("code");
String message = (String) responseMap.get("message");
if (code != null && code == 0) {
// 响应码为0表示成功,从data字段中获取datasetId
Object dataObj = responseMap.get("data");
if (dataObj instanceof Map) {
Map<String, Object> dataMap = (Map<String, Object>) dataObj;
datasetId = (String) dataMap.get("id");
if (StringUtils.isBlank(datasetId)) {
// 如果id字段为空,尝试其他可能的字段名
datasetId = (String) dataMap.get("dataset_id");
datasetId = (String) dataMap.get("datasetId");
}
}
} else {
// 如果响应码不为0,说明API调用失败
String errorMessage = code + (message != null ? message : "null");
log.error(errorMessage);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
log.info("从RAGFlow API响应中解析出datasetId: {}", datasetId);
log.debug("完整响应内容: {}", responseBody);
} catch (IOException e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName();
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
} catch (Exception e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName();
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
// 如果解析失败,但响应体不为空,尝试直接使用响应体作为错误信息
String finalErrorMessage = responseBody;
if (e.getMessage() != null) {
finalErrorMessage += errorMessage;
}
throw new RenException(ErrorCode.RAG_API_ERROR, finalErrorMessage);
}
}
if (StringUtils.isBlank(datasetId)) {
log.error("无法从RAGFlow API响应中获取datasetId,响应内容: {}", responseBody);
throw new RenException(ErrorCode.RAG_DATASET_ID_NOT_NULL);
}
log.info("RAGFlow数据集创建成功,datasetId: {}", datasetId);
return datasetId;
}
/**
* 使用适配器更新数据集
* 调用RAGFlow API更新数据集
*/
private void updateDatasetInRAG(String datasetId, String name, String description,
private void updateDatasetInRAGFlow(String datasetId, String name, String description,
Map<String, Object> ragConfig) {
log.info("开始使用适配器更新数据集,datasetId: {}, name: {}", datasetId, name);
String baseUrl = (String) ragConfig.get("base_url");
String apiKey = (String) ragConfig.get("api_key");
try {
// 从RAG配置中提取适配器类型
String adapterType = extractAdapterType(ragConfig);
log.info("开始调用RAGFlow API更新数据集,datasetId: {}, name: {}", datasetId, name);
log.debug("RAGFlow配置 - baseUrl: {}, apiKey: {}", baseUrl, StringUtils.isBlank(apiKey) ? "未配置" : "已配置");
// 使用适配器工厂获取适配器实例
KnowledgeBaseAdapter adapter = KnowledgeBaseAdapterFactory.getAdapter(adapterType, ragConfig);
// 构建数据集更新参数
Map<String, Object> updateParams = new HashMap<>();
String username = SecurityUser.getUser().getUsername();
updateParams.put("name", username + "_" + name);
if (StringUtils.isNotBlank(description)) {
updateParams.put("description", description);
}
// 调用适配器的更新数据集方法
adapter.updateDataset(datasetId, updateParams);
log.info("数据集更新成功,datasetId: {}", datasetId);
} catch (Exception e) {
// 直接传递底层适配器的详细错误信息
log.error("更新数据集失败", e);
if (e instanceof RenException) {
throw (RenException) e;
}
throw new RenException(ErrorCode.RAG_API_ERROR, e.getMessage());
// 构建请求URL
String url = baseUrl + "/api/v1/datasets/" + datasetId;
log.debug("请求URL: {}", url);
// 构建请求体
Map<String, Object> requestBody = new HashMap<>();
requestBody.put("dataset_id", datasetId);
String username = SecurityUser.getUser().getUsername();
requestBody.put("name", username + "_" + name);
if (StringUtils.isNotBlank(description)) {
requestBody.put("description", description);
}
log.debug("请求体: {}", requestBody);
// 设置请求头
HttpHeaders headers = new HttpHeaders();
headers.setContentType(MediaType.APPLICATION_JSON);
headers.set("Authorization", "Bearer " + apiKey);
HttpEntity<Map<String, Object>> requestEntity = new HttpEntity<>(requestBody, headers);
// 发送PUT请求
log.info("发送PUT请求到RAGFlow API...");
ResponseEntity<String> response;
try {
response = restTemplate.exchange(url, HttpMethod.PUT, requestEntity, String.class);
} catch (Exception e) {
String errorMessage = url + e.getMessage();
log.error(errorMessage, e);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
log.info("RAGFlow API响应状态码: {}", response.getStatusCode());
log.debug("RAGFlow API响应内容: {}", response.getBody());
if (!response.getStatusCode().is2xxSuccessful()) {
String errorMessage = response.getStatusCode() + ", 响应内容: " + response.getBody();
log.error(errorMessage);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
// 解析响应体,验证操作是否真正成功
String responseBody = response.getBody();
if (responseBody != null) {
try {
Map<String, Object> responseMap = objectMapper.readValue(responseBody, Map.class);
Integer code = (Integer) responseMap.get("code");
String message = (String) responseMap.get("message");
if (code != null && code != 0) {
String errorMessage = code + (message != null ? message : "null");
log.error(errorMessage);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
} catch (IOException e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName();
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
} catch (Exception e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName();
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
// 如果解析失败,但响应体不为空,尝试直接使用响应体作为错误信息
String finalErrorMessage = responseBody;
if (e.getMessage() != null) {
finalErrorMessage += errorMessage;
}
throw new RenException(ErrorCode.RAG_API_ERROR, finalErrorMessage);
}
}
log.info("RAGFlow数据集更新成功,datasetId: {}", datasetId);
}
/**
* 使用适配器删除数据集
* 调用RAGFlow API删除数据集
*/
private void deleteDatasetInRAG(String datasetId, Map<String, Object> ragConfig) {
log.info("开始使用适配器删除数据集,datasetId: {}", datasetId);
private void deleteDatasetInRAGFlow(String datasetId, Map<String, Object> ragConfig) {
String baseUrl = (String) ragConfig.get("base_url");
String apiKey = (String) ragConfig.get("api_key");
log.info("开始调用RAGFlow API删除数据集,datasetId: {}", datasetId);
log.debug("RAGFlow配置 - baseUrl: {}, apiKey: {}", baseUrl, StringUtils.isBlank(apiKey) ? "未配置" : "已配置");
// 构建请求URL
String url = baseUrl + "/api/v1/datasets";
log.debug("请求URL: {}", url);
// 构建请求体
Map<String, Object> requestBody = new HashMap<>();
requestBody.put("ids", List.of(datasetId));
log.debug("请求体: {}", requestBody);
// 设置请求头
HttpHeaders headers = new HttpHeaders();
headers.setContentType(MediaType.APPLICATION_JSON);
headers.set("Authorization", "Bearer " + apiKey);
HttpEntity<Map<String, Object>> requestEntity = new HttpEntity<>(requestBody, headers);
// 发送DELETE请求
log.info("发送DELETE请求到RAGFlow API...");
ResponseEntity<String> response;
try {
// 从RAG配置中提取适配器类型
String adapterType = extractAdapterType(ragConfig);
// 使用适配器工厂获取适配器实例
KnowledgeBaseAdapter adapter = KnowledgeBaseAdapterFactory.getAdapter(adapterType, ragConfig);
// 调用适配器的删除数据集方法
adapter.deleteDataset(datasetId);
log.info("数据集删除成功,datasetId: {}", datasetId);
response = restTemplate.exchange(url, HttpMethod.DELETE, requestEntity, String.class);
} catch (Exception e) {
// 直接传递底层适配器的详细错误信息
log.error("删除数据集失败", e);
if (e instanceof RenException) {
throw (RenException) e;
}
throw new RenException(ErrorCode.RAG_API_ERROR, e.getMessage());
String errorMessage = url + e.getMessage();
log.error(errorMessage, e);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
log.info("RAGFlow API响应状态码: {}", response.getStatusCode());
log.debug("RAGFlow API响应内容: {}", response.getBody());
if (!response.getStatusCode().is2xxSuccessful()) {
String errorMessage = response.getStatusCode() + response.getBody();
log.error(errorMessage);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
// 解析响应体,验证操作是否真正成功
String responseBody = response.getBody();
if (responseBody != null) {
try {
Map<String, Object> responseMap = objectMapper.readValue(responseBody, Map.class);
Integer code = (Integer) responseMap.get("code");
String message = (String) responseMap.get("message");
if (code != null && code != 0) {
String errorMessage = code + (message != null ? message : "null");
log.error(errorMessage);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
}
} catch (IOException e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName();
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
throw new RenException(ErrorCode.RAG_API_ERROR, errorMessage);
} catch (Exception e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName();
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
// 如果解析失败,但响应体不为空,尝试直接使用响应体作为错误信息
String finalErrorMessage = responseBody;
if (e.getMessage() != null) {
finalErrorMessage += errorMessage;
}
throw new RenException(ErrorCode.RAG_API_ERROR, finalErrorMessage);
}
}
log.info("RAGFlow数据集删除成功,datasetId: {}", datasetId);
}
/**
@@ -606,9 +760,9 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
}
/**
* 从适配器获取知识库的文档数量
* 从RAGFlow API获取知识库的文档数量
*/
private Integer getDocumentCountFromRAG(String datasetId, String ragModelId) {
private Integer getDocumentCountFromRAGFlow(String datasetId, String ragModelId) {
if (StringUtils.isBlank(datasetId) || StringUtils.isBlank(ragModelId)) {
log.warn("datasetId或ragModelId为空,无法获取文档数量");
return 0;
@@ -616,29 +770,79 @@ public class KnowledgeBaseServiceImpl extends BaseServiceImpl<KnowledgeBaseDao,
log.info("开始获取知识库 {} 的文档数量", datasetId);
// 获取RAG配置
Map<String, Object> ragConfig = getValidatedRAGConfig(ragModelId);
String baseUrl = (String) ragConfig.get("base_url");
String apiKey = (String) ragConfig.get("api_key");
// 构建请求URL - 调用RAGFlow API获取文档列表,但不返回文档详情,只获取总数
String url = baseUrl + "/api/v1/datasets/" + datasetId + "/documents?page=1&size=1";
log.debug("请求URL: {}", url);
// 设置请求头
HttpHeaders headers = new HttpHeaders();
headers.setContentType(MediaType.APPLICATION_JSON);
headers.set("Authorization", "Bearer " + apiKey);
HttpEntity<String> requestEntity = new HttpEntity<>(headers);
// 发送GET请求
log.info("发送GET请求到RAGFlow API获取文档数量...");
ResponseEntity<String> response;
try {
// 获取RAG配置
Map<String, Object> ragConfig = getValidatedRAGConfig(ragModelId);
// 从RAG配置中提取适配器类型
String adapterType = extractAdapterType(ragConfig);
// 使用适配器工厂获取适配器实例
KnowledgeBaseAdapter adapter = KnowledgeBaseAdapterFactory.getAdapter(adapterType, ragConfig);
// 调用适配器的获取文档数量方法
Integer documentCount = adapter.getDocumentCount(datasetId);
log.info("获取知识库 {} 的文档数量成功: {}", datasetId, documentCount);
return documentCount;
response = restTemplate.exchange(url, HttpMethod.GET, requestEntity, String.class);
} catch (Exception e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName() + " - 获取知识库文档数量失败";
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
String errorMessage = url + e.getMessage();
log.error(errorMessage, e);
return 0;
}
log.info("RAGFlow API响应状态码: {}", response.getStatusCode());
if (!response.getStatusCode().is2xxSuccessful()) {
log.error("RAGFlow API调用失败,状态码: {}, 响应内容: {}", response.getStatusCode(), response.getBody());
return 0;
}
String responseBody = response.getBody();
log.debug("RAGFlow API响应内容: {}", responseBody);
// 解析响应
try {
ObjectMapper objectMapper = new ObjectMapper();
Map<String, Object> responseMap = objectMapper.readValue(responseBody, Map.class);
Integer code = (Integer) responseMap.get("code");
if (code != null && code == 0) {
Object dataObj = responseMap.get("data");
if (dataObj instanceof Map) {
Map<String, Object> dataMap = (Map<String, Object>) dataObj;
Object totalObj = dataMap.get("total");
if (totalObj instanceof Integer) {
Integer documentCount = (Integer) totalObj;
log.info("获取知识库 {} 的文档数量成功: {}", datasetId, documentCount);
return documentCount;
} else if (totalObj instanceof Long) {
Long documentCount = (Long) totalObj;
log.info("获取知识库 {} 的文档数量成功: {}", datasetId, documentCount);
return documentCount.intValue();
}
}
} else {
log.error("RAGFlow API调用失败,响应码: {}, 响应内容: {}", code, responseBody);
}
} catch (IOException e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName() + " - 解析RAGFlow API响应时发生IO异常";
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
} catch (Exception e) {
// 构建详细的错误信息,包含异常类型和消息
String baseErrorMessage = e.getClass().getSimpleName() + " - 解析RAGFlow API响应失败";
String errorMessage = baseErrorMessage + (e.getMessage() != null ? ": " + e.getMessage() : "");
log.error(errorMessage, e);
}
return 0;
}
}
@@ -6,7 +6,6 @@ import java.util.HashMap;
import java.util.List;
import java.util.Map;
import org.apache.commons.lang3.StringUtils;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.PostMapping;
import org.springframework.web.bind.annotation.PutMapping;
@@ -24,9 +23,7 @@ import xiaozhi.common.exception.ErrorCode;
import xiaozhi.common.exception.RenException;
import xiaozhi.common.page.TokenDTO;
import xiaozhi.common.user.UserDetail;
import xiaozhi.common.utils.JsonUtils;
import xiaozhi.common.utils.Result;
import xiaozhi.common.utils.Sm2DecryptUtil;
import xiaozhi.common.validator.AssertUtils;
import xiaozhi.common.validator.ValidatorUtils;
import xiaozhi.modules.security.dto.LoginDTO;
@@ -35,6 +32,8 @@ import xiaozhi.modules.security.password.PasswordUtils;
import xiaozhi.modules.security.service.CaptchaService;
import xiaozhi.modules.security.service.SysUserTokenService;
import xiaozhi.modules.security.user.SecurityUser;
import xiaozhi.common.utils.Sm2DecryptUtil;
import org.apache.commons.lang3.StringUtils;
import xiaozhi.modules.sys.dto.PasswordDTO;
import xiaozhi.modules.sys.dto.RetrievePasswordDTO;
import xiaozhi.modules.sys.dto.SysUserDTO;
@@ -90,13 +89,13 @@ public class LoginController {
@Operation(summary = "登录")
public Result<TokenDTO> login(@RequestBody LoginDTO login) {
String password = login.getPassword();
// 使用工具类解密并验证验证码
String actualPassword = Sm2DecryptUtil.decryptAndValidateCaptcha(
password, login.getCaptchaId(), captchaService, sysParamsService);
login.setPassword(actualPassword);
// 按照用户名获取用户
SysUserDTO userDTO = sysUserService.getByUsername(login.getUsername());
// 判断用户是否存在
@@ -109,6 +108,8 @@ public class LoginController {
}
return sysUserTokenService.createToken(userDTO.getId());
}
@PostMapping("/register")
@Operation(summary = "注册")
@@ -116,15 +117,15 @@ public class LoginController {
if (!sysUserService.getAllowUserRegister()) {
throw new RenException(ErrorCode.USER_REGISTER_DISABLED);
}
String password = login.getPassword();
// 使用工具类解密并验证验证码
String actualPassword = Sm2DecryptUtil.decryptAndValidateCaptcha(
password, login.getCaptchaId(), captchaService, sysParamsService);
login.setPassword(actualPassword);
// 是否开启手机注册
Boolean isMobileRegister = sysParamsService
.getValueObject(Constant.SysMSMParam.SERVER_ENABLE_MOBILE_REGISTER.getValue(), Boolean.class);
@@ -203,11 +204,11 @@ public class LoginController {
}
String password = dto.getPassword();
// 使用工具类解密并验证验证码
String actualPassword = Sm2DecryptUtil.decryptAndValidateCaptcha(
password, dto.getCaptchaId(), captchaService, sysParamsService);
dto.setPassword(actualPassword);
sysUserService.changePasswordDirectly(userDTO.getId(), dto.getPassword());
@@ -228,7 +229,7 @@ public class LoginController {
config.put("beianIcpNum", sysParamsService.getValue(Constant.SysBaseParam.BEIAN_ICP_NUM.getValue(), true));
config.put("beianGaNum", sysParamsService.getValue(Constant.SysBaseParam.BEIAN_GA_NUM.getValue(), true));
config.put("name", sysParamsService.getValue(Constant.SysBaseParam.SERVER_NAME.getValue(), true));
// SM2公钥
String publicKey = sysParamsService.getValue(Constant.SM2_PUBLIC_KEY, true);
if (StringUtils.isBlank(publicKey)) {
@@ -236,12 +237,6 @@ public class LoginController {
}
config.put("sm2PublicKey", publicKey);
// 获取system-web.menu参数配置
String menuConfig = sysParamsService.getValue("system-web.menu", true);
if (StringUtils.isNotBlank(menuConfig)) {
config.put("systemWebMenu", JsonUtils.parseObject(menuConfig, Object.class));
}
return new Result<Map<String, Object>>().ok(config);
}
}
@@ -174,7 +174,7 @@ public class SysParamsController {
return;
}
if (StringUtils.isBlank(url) || url.equals("null")) {
return;
throw new RenException(ErrorCode.OTA_URL_EMPTY);
}
// 检查是否包含localhost或127.0.0.1
@@ -211,7 +211,7 @@ public class SysParamsController {
return;
}
if (StringUtils.isBlank(url) || url.equals("null")) {
return;
throw new RenException(ErrorCode.MCP_URL_EMPTY);
}
if (url.contains("localhost") || url.contains("127.0.0.1")) {
throw new RenException(ErrorCode.MCP_URL_LOCALHOST);
@@ -242,7 +242,7 @@ public class SysParamsController {
return;
}
if (StringUtils.isBlank(url) || url.equals("null")) {
return;
throw new RenException(ErrorCode.VOICEPRINT_URL_EMPTY);
}
if (url.contains("localhost") || url.contains("127.0.0.1")) {
throw new RenException(ErrorCode.VOICEPRINT_URL_LOCALHOST);
@@ -1,6 +0,0 @@
-- 添加系统功能菜单配置参数
delete from `sys_params` where param_code = 'system-web.menu';
-- 添加系统功能菜单配置参数
INSERT INTO `sys_params` (id, param_code, param_value, value_type, param_type, remark) VALUES
(600, 'system-web.menu', '{"features":{"voiceprintRecognition":{"name":"feature.voiceprintRecognition.name","enabled":false,"description":"feature.voiceprintRecognition.description"},"voiceClone":{"name":"feature.voiceClone.name","enabled":false,"description":"feature.voiceClone.description"},"knowledgeBase":{"name":"feature.knowledgeBase.name","enabled":false,"description":"feature.knowledgeBase.description"},"mcpAccessPoint":{"name":"feature.mcpAccessPoint.name","enabled":false,"description":"feature.mcpAccessPoint.description"},"vad":{"name":"feature.vad.name","enabled":true,"description":"feature.vad.description"},"asr":{"name":"feature.asr.name","enabled":true,"description":"feature.asr.description"}},"groups":{"featureManagement":["voiceprintRecognition","voiceClone","knowledgeBase","mcpAccessPoint"],"voiceManagement":["vad","asr"]}}', 'json', 1, '系统功能菜单配置');
@@ -1,14 +0,0 @@
-- liquibase formatted sql
-- changeset xiaozhi:202512041515
CREATE TABLE ai_agent_context_provider (
id VARCHAR(32) NOT NULL COMMENT '主键',
agent_id VARCHAR(32) NOT NULL COMMENT '智能体ID',
context_providers JSON COMMENT '上下文源配置',
creator BIGINT COMMENT '创建者',
created_at DATETIME COMMENT '创建时间',
updater BIGINT COMMENT '更新者',
updated_at DATETIME COMMENT '更新时间',
PRIMARY KEY (id),
INDEX idx_agent_id (agent_id)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COMMENT='智能体上下文源配置表';
@@ -1,6 +0,0 @@
-- 删除server模块是否开启token认证参数
delete from `sys_params` where param_code = 'server.auth.enabled';
-- 添加server模块是否开启token认证参数
INSERT INTO `sys_params` (id, param_code, param_value, value_type, param_type, remark) VALUES
(122, 'server.auth.enabled', 'true', 'boolean', 1, 'server模块是否开启token认证');
@@ -423,25 +423,3 @@ databaseChangeLog:
- sqlFile:
encoding: utf8
path: classpath:db/changelog/202511131023.sql
- changeSet:
id: 202512031517
author: rainv123
changes:
- sqlFile:
encoding: utf8
path: classpath:db/changelog/202512031517.sql
- changeSet:
id: 202512041515
author: cgd
changes:
- sqlFile:
encoding: utf8
path: classpath:db/changelog/202512041515.sql
- changeSet:
id: 202512131453
author: hrz
changes:
- sqlFile:
encoding: utf8
path: classpath:db/changelog/202512131453.sql
@@ -189,13 +189,4 @@
10180=\u6587\u4EF6\u5185\u5BB9\u4E0D\u80FD\u4E3A\u7A7A
10181=\u97F3\u8272\u514B\u9686\u540D\u79F0\u4E0D\u80FD\u4E3A\u7A7A
10182=\u97F3\u8272\u514B\u9686\u97F3\u9891\u4E0D\u5B58\u5728
10183=\u9ED8\u8BA4\u667A\u80FD\u4F53\u672A\u627E\u5230
10184=\u4E0D\u652F\u6301\u7684\u9002\u914D\u5668\u7C7B\u578B
10185=RAG\u914D\u7F6E\u9A8C\u8BC1\u5931\u8D25
10186=\u9002\u914D\u5668\u521B\u5EFA\u5931\u8D25
10187=\u9002\u914D\u5668\u521D\u59CB\u5316\u5931\u8D25
10188=\u9002\u914D\u5668\u8FDE\u63A5\u6D4B\u8BD5\u5931\u8D25
10189=\u9002\u914D\u5668\u64CD\u4F5C\u5931\u8D25
10190=\u9002\u914D\u5668\u672A\u627E\u5230
10191=\u9002\u914D\u5668\u7F13\u5B58\u9519\u8BEF
10192=\u9002\u914D\u5668\u7C7B\u578B\u672A\u627E\u5230
10183=\u9ED8\u8BA4\u667A\u80FD\u4F53\u672A\u627E\u5230
@@ -189,13 +189,4 @@
10180=Dateiinhalt darf nicht leer sein
10181=Stimmenklon-Name darf nicht leer sein
10182=Stimmenklon-Audio nicht gefunden
10183=Standard-Agent-Vorlage nicht gefunden
10184=Nicht unterstützter Adaptertyp
10185=RAG-Konfigurationsvalidierung fehlgeschlagen
10186=Adapter-Erstellung fehlgeschlagen
10187=Adapter-Initialisierung fehlgeschlagen
10188=Adapter-Verbindungstest fehlgeschlagen
10189=Adapter-Operation fehlgeschlagen
10190=Adapter nicht gefunden
10191=Adapter-Cache-Fehler
10192=Adaptertyp nicht gefunden
10183=Standard-Agent-Vorlage nicht gefunden
@@ -189,13 +189,4 @@
10180=File content cannot be empty
10181=Voice clone name cannot be empty
10182=Voice clone audio not found
10183=Default agent template not found
10184=Unsupported adapter type
10185=RAG configuration validation failed
10186=Adapter creation failed
10187=Adapter initialization failed
10188=Adapter connection test failed
10189=Adapter operation failed
10190=Adapter not found
10191=Adapter cache error
10192=Adapter type not found
10183=Default agent template not found
@@ -189,13 +189,4 @@
10180=Nội dung tệp không thể để trống
10181=Tên nhân bản giọng nói không thể để trống
10182=Không tìm thấy âm thanh nhân bản giọng nói
10183=Không tìm thấy mẫu agent mặc định
10184=Loại bộ chuyển đổi không được hỗ trợ
10185=Kiểm tra cấu hình RAG thất bại
10186=Tạo bộ chuyển đổi thất bại
10187=Khởi tạo bộ chuyển đổi thất bại
10188=Kiểm tra kết nối bộ chuyển đổi thất bại
10189=Thao tác bộ chuyển đổi thất bại
10190=Không tìm thấy bộ chuyển đổi
10191=Lỗi bộ nhớ đệm bộ chuyển đổi
10192=Không tìm thấy loại bộ chuyển đổi
10183=Không tìm thấy mẫu agent mặc định
@@ -189,13 +189,4 @@
10180=\u6587\u4EF6\u5185\u5BB9\u4E0D\u80FD\u4E3A\u7A7A
10181=\u97F3\u8272\u514B\u9686\u540D\u79F0\u4E0D\u80FD\u4E3A\u7A7A
10182=\u97F3\u8272\u514B\u9686\u97F3\u9891\u4E0D\u5B58\u5728
10183=\u9ED8\u8BA4\u667A\u80FD\u4F53\u672A\u627E\u5230
10184=\u4E0D\u652F\u6301\u7684\u9002\u914D\u5668\u7C7B\u578B
10185=RAG\u914D\u7F6E\u9A8C\u8BC1\u5931\u8D25
10186=\u9002\u914D\u5668\u521B\u5EFA\u5931\u8D25
10187=\u9002\u914D\u5668\u521D\u59CB\u5316\u5931\u8D25
10188=\u9002\u914D\u5668\u8FDE\u63A5\u6D4B\u8BD5\u5931\u8D25
10189=\u9002\u914D\u5668\u64CD\u4F5C\u5931\u8D25
10190=\u9002\u914D\u5668\u672A\u627E\u5230
10191=\u9002\u914D\u5668\u7F13\u5B58\u9519\u8BEF
10192=\u9002\u914D\u5668\u7C7B\u578B\u672A\u627E\u5230
10183=\u9ED8\u8BA4\u667A\u80FD\u4F53\u672A\u627E\u5230
@@ -189,13 +189,4 @@
10180=\u6587\u4ef6\u5185\u5bb9\u4e0d\u80fd\u70ba\u7a7a
10181=\u97f3\u8272\u514b\u9686\u540d\u7a31\u4e0d\u80fd\u70ba\u7a7a
10182=\u97f3\u8272\u514b\u9686\u97f3\u983b\u4e0d\u5b58\u5728
10183=\u9ed8\u8ba4\u667a\u80fd\u4f53\u672a\u627e\u5230
10184=\u4E0D\u652F\u6301\u7684\u9002\u914D\u5668\u985E\u578B
10185=RAG\u914D\u7F6E\u9A57\u8B49\u5931\u6557
10186=\u9002\u914D\u5668\u5275\u5EFA\u5931\u6557
10187=\u9002\u914D\u5668\u521D\u59CB\u5316\u5931\u6557
10188=\u9002\u914D\u5668\u9023\u63A5\u6E2C\u8A66\u5931\u6557
10189=\u9002\u914D\u5668\u64CD\u4F5C\u5931\u6557
10190=\u9002\u914D\u5668\u672A\u627E\u5230
10191=\u9002\u914D\u5668\u7F13\u5B58\u932F\u8AA4
10192=\u9002\u914D\u5668\u985E\u578B\u672A\u627E\u5230
10183=\u9ed8\u8ba4\u667a\u80fd\u4f53\u672a\u627e\u5230
@@ -235,7 +235,7 @@ function showAbout() {
title: t('settings.aboutApp', { appName: import.meta.env.VITE_APP_TITLE }),
content: t('settings.aboutContent', {
appName: import.meta.env.VITE_APP_TITLE,
version: '0.8.10'
version: '0.8.8'
}),
showCancel: false,
confirmText: t('common.confirm'),
@@ -200,13 +200,6 @@ export default {
confirm() {
this.saving = true;
// 校验模型ID不能为纯文字或空格
if (this.formData.id && !this.validateModelId(this.formData.id)) {
this.$message.error(this.$t('modelConfigDialog.invalidModelId'));
this.saving = false;
return;
}
if (!this.formData.supplier) {
this.$message.error(this.$t('addModelDialog.requiredSupplier'));
this.saving = false;
@@ -261,38 +254,6 @@ export default {
this.providerFields = [];
this.currentProvider = null;
},
// 校验模型ID:不能为纯文字或空格
validateModelId(modelId) {
if (!modelId || typeof modelId !== 'string') {
return false;
}
// 去除首尾空格
const trimmedId = modelId.trim();
// 检查是否为空或纯空格
if (trimmedId === '') {
return false;
}
// 检查是否只包含字母(纯文字)
if (/^[a-zA-Z]+$/.test(trimmedId)) {
return false;
}
// 检查是否包含空格
if (/\s/.test(trimmedId)) {
return false;
}
// 允许字母、数字、下划线、连字符
if (!/^[a-zA-Z0-9_-]+$/.test(trimmedId)) {
return false;
}
return true;
}
}
}
</script>
@@ -404,7 +365,7 @@ export default {
.custom-input-bg .el-input__inner,
.custom-input-bg .el-textarea__inner {
background-color: #ffffff;
background-color: #f6f8fc;
}
@@ -1,329 +0,0 @@
<template>
<el-dialog
:visible.sync="dialogVisible"
width="900px"
:title="$t('contextProviderDialog.title')"
:close-on-click-modal="false"
custom-class="context-provider-dialog"
append-to-body
>
<div class="dialog-content">
<el-empty v-if="localProviders.length === 0" :description="$t('contextProviderDialog.noContextApi')">
<el-button type="primary" icon="el-icon-plus" @click="addProvider(0)">{{ $t('contextProviderDialog.add') }}</el-button>
</el-empty>
<div
v-for="(provider, pIndex) in localProviders"
:key="pIndex"
class="provider-item"
>
<el-card class="provider-card" shadow="hover" :body-style="{ padding: '15px 20px' }">
<!-- URL Row -->
<div class="input-row">
<span class="label-text">{{ $t('contextProviderDialog.apiUrl') }}</span>
<el-input
v-model="provider.url"
:placeholder="$t('contextProviderDialog.apiUrlPlaceholder')"
size="small"
class="flex-1"
></el-input>
</div>
<!-- Headers Section -->
<div class="headers-section">
<div class="label-text" style="margin-top: 6px;">{{ $t('contextProviderDialog.requestHeaders') }}</div>
<div class="headers-list">
<div
v-for="(header, hIndex) in provider.headers"
:key="hIndex"
class="header-row"
>
<el-input
v-model="header.key"
:placeholder="$t('contextProviderDialog.headerKeyPlaceholder')"
size="small"
style="width: 180px;"
></el-input>
<span class="separator">:</span>
<el-input
v-model="header.value"
:placeholder="$t('contextProviderDialog.headerValuePlaceholder')"
size="small"
class="flex-1"
></el-input>
<div class="row-controls">
<el-button
type="primary"
icon="el-icon-plus"
circle
size="mini"
plain
@click="addHeader(pIndex, hIndex + 1)"
></el-button>
<el-button
type="danger"
icon="el-icon-minus"
circle
size="mini"
plain
@click="removeHeader(pIndex, hIndex)"
></el-button>
</div>
</div>
<!-- Empty Headers State -->
<div v-if="provider.headers.length === 0" class="header-row empty-header">
<span class="no-header-text">{{ $t('contextProviderDialog.noHeaders') }}</span>
<el-button
type="text"
icon="el-icon-plus"
size="mini"
@click="addHeader(pIndex, 0)"
>{{ $t('contextProviderDialog.addHeader') }}</el-button>
</div>
</div>
</div>
</el-card>
<!-- Provider Block Controls (Right Side) -->
<div class="block-controls">
<el-button
type="primary"
icon="el-icon-plus"
circle
size="medium"
@click="addProvider(pIndex + 1)"
></el-button>
<el-button
type="danger"
icon="el-icon-minus"
circle
size="medium"
@click="removeProvider(pIndex)"
></el-button>
</div>
</div>
</div>
<span slot="footer" class="dialog-footer">
<el-button @click="dialogVisible = false">{{ $t('contextProviderDialog.cancel') }}</el-button>
<el-button type="primary" @click="handleConfirm">{{ $t('contextProviderDialog.confirm') }}</el-button>
</span>
</el-dialog>
</template>
<script>
export default {
name: 'ContextProviderDialog',
props: {
visible: {
type: Boolean,
default: false
},
providers: {
type: Array,
default: () => []
}
},
data() {
return {
localProviders: []
};
},
computed: {
dialogVisible: {
get() {
return this.visible;
},
set(val) {
this.$emit('update:visible', val);
}
}
},
watch: {
visible(val) {
if (val) {
this.initLocalData();
}
}
},
methods: {
initLocalData() {
// 深拷贝并将 headers 对象转换为数组
this.localProviders = this.providers.map(p => {
const headers = p.headers || {};
return {
url: p.url || '',
headers: Object.entries(headers).map(([key, value]) => ({ key, value }))
};
});
// 如果为空,添加一个默认块
if (this.localProviders.length === 0) {
this.localProviders.push({ url: '', headers: [{ key: '', value: '' }] });
}
},
addProvider(index) {
this.localProviders.splice(index, 0, {
url: '',
headers: [{ key: '', value: '' }]
});
},
removeProvider(index) {
this.localProviders.splice(index, 1);
},
addHeader(pIndex, hIndex) {
this.localProviders[pIndex].headers.splice(hIndex, 0, { key: '', value: '' });
},
removeHeader(pIndex, hIndex) {
this.localProviders[pIndex].headers.splice(hIndex, 1);
},
handleConfirm() {
const result = this.localProviders
.filter(p => p.url.trim() !== '')
.map(p => {
const headersObj = {};
p.headers.forEach(h => {
if (h.key.trim()) {
headersObj[h.key.trim()] = h.value;
}
});
return {
url: p.url.trim(),
headers: headersObj
};
});
this.$emit('confirm', result);
this.dialogVisible = false;
}
}
};
</script>
<style scoped>
.dialog-content {
max-height: 60vh;
overflow-y: auto;
padding: 20px 25px;
}
.dialog-content::-webkit-scrollbar {
width: 6px;
}
.dialog-content::-webkit-scrollbar-thumb {
background: #dcdfe6;
border-radius: 3px;
}
.dialog-content::-webkit-scrollbar-track {
background: #f5f7fa;
}
.provider-item {
display: flex;
gap: 15px;
margin-bottom: 20px;
align-items: center;
}
.provider-card {
flex: 1;
border-radius: 12px;
border: 1px solid #e4e7ed;
border-left: 4px solid #409EFF; /* 左侧强调色 */
background-color: #fff;
transition: all 0.3s ease;
box-shadow: 0 2px 12px 0 rgba(0, 0, 0, 0.05);
}
.provider-card:hover {
box-shadow: 0 8px 20px rgba(0, 0, 0, 0.1);
transform: translateY(-2px);
}
.block-controls {
display: flex;
flex-direction: row;
gap: 8px;
}
.input-row {
display: flex;
align-items: center;
gap: 12px;
margin-bottom: 18px;
}
.label-text {
width: 60px;
font-weight: 600;
color: #606266;
text-align: right;
font-size: 13px;
white-space: nowrap;
line-height: 32px; /* 垂直居中对齐 */
}
.flex-1 {
flex: 1;
}
.headers-section {
display: flex;
gap: 12px;
align-items: flex-start;
}
.headers-list {
flex: 1;
display: flex;
flex-direction: column;
gap: 10px;
background: #fcfcfc;
padding: 15px;
border-radius: 8px;
border: 1px dashed #dcdfe6;
transition: all 0.3s;
}
.headers-list:hover {
border-color: #c0c4cc;
background: #fff;
}
.header-row {
display: flex;
align-items: center;
gap: 10px;
}
.separator {
color: #909399;
font-weight: bold;
margin: 0 2px;
}
.row-controls {
display: flex;
gap: 6px;
margin-left: 8px;
flex-shrink: 0;
opacity: 0.6;
transition: opacity 0.2s;
}
.header-row:hover .row-controls {
opacity: 1;
}
.empty-header {
justify-content: center;
padding: 10px;
color: #909399;
font-size: 13px;
}
.no-header-text {
margin-right: 8px;
}
</style>
+2 -10
View File
@@ -23,7 +23,7 @@
<div class="settings-btn" @click="handleConfigure">
{{ $t('home.configureRole') }}
</div>
<div v-if="featureStatus.voiceprintRecognition" class="settings-btn" @click="handleVoicePrint">
<div class="settings-btn" @click="handleVoicePrint">
{{ $t('home.voiceprintRecognition') }}
</div>
<div class="settings-btn" @click="handleDeviceManage">
@@ -49,15 +49,7 @@ import i18n from '@/i18n';
export default {
name: 'DeviceItem',
props: {
device: { type: Object, required: true },
featureStatus: {
type: Object,
default: () => ({
voiceprintRecognition: false,
voiceClone: false,
knowledgeBase: false
})
}
device: { type: Object, required: true }
},
data() {
return { switchValue: false }
@@ -106,7 +106,7 @@
</div>
<!-- MCP区域 -->
<div class="mcp-access-point" v-if="featureStatus.mcpAccessPoint">
<div class="mcp-access-point">
<div class="mcp-container">
<!-- 左侧区域 -->
<div class="mcp-left">
@@ -171,7 +171,6 @@
<script>
import Api from '@/apis/api';
import i18n from '@/i18n';
import featureManager from '@/utils/featureManager';
export default {
i18n,
@@ -206,11 +205,6 @@ export default {
mcpUrl: "",
mcpStatus: "disconnected",
mcpTools: [],
// 功能状态
featureStatus: {
mcpAccessPoint: false
}
}
},
computed: {
@@ -255,9 +249,6 @@ export default {
// 右侧默认指向第一个
this.currentFunction = this.selectedList[0] || null;
// 加载功能状态
this.loadFeatureStatus();
// 加载MCP数据
this.loadMcpAddress();
this.loadMcpTools();
@@ -268,19 +259,6 @@ export default {
}
},
methods: {
/**
* 加载功能状态
*/
async loadFeatureStatus() {
// 确保featureManager已初始化完成
await featureManager.waitForInitialization();
const config = featureManager.getConfig();
this.featureStatus = {
mcpAccessPoint: config.mcpAccessPoint || false
};
},
copyUrl() {
const textarea = document.createElement('textarea');
textarea.value = this.mcpUrl;
@@ -478,7 +456,6 @@ export default {
.function-column {
position: relative;
width: auto;
height:700px;
padding: 10px;
overflow-y: auto;
border-right: 1px solid #EBEEF5;
@@ -486,12 +463,6 @@ export default {
overflow-x: hidden;
}
.mcp-access-point {
position: relative;
z-index: 1;
background: white;
}
.function-column::-webkit-scrollbar {
display: none;
}
+6 -33
View File
@@ -26,7 +26,7 @@
<span class="nav-text">{{ $t("header.smartManagement") }}</span>
</div>
<!-- 普通用户显示音色克隆 -->
<div v-if="!isSuperAdmin && featureStatus.voiceClone" class="equipment-management"
<div v-if="!isSuperAdmin" class="equipment-management"
:class="{ 'active-tab': $route.path === '/voice-clone-management' }" @click="goVoiceCloneManagement">
<img loading="lazy" alt="" src="@/assets/header/voice.png" :style="{
filter:
@@ -38,7 +38,7 @@
</div>
<!-- 超级管理员显示音色克隆下拉菜单 -->
<el-dropdown v-if="isSuperAdmin && featureStatus.voiceClone" trigger="click" class="equipment-management more-dropdown" :class="{
<el-dropdown v-if="isSuperAdmin" trigger="click" class="equipment-management more-dropdown" :class="{
'active-tab':
$route.path === '/voice-clone-management' ||
$route.path === '/voice-resource-management',
@@ -72,7 +72,7 @@
}" />
<span class="nav-text">{{ $t("header.modelConfig") }}</span>
</div>
<div v-if="featureStatus.knowledgeBase" class="equipment-management"
<div class="equipment-management"
:class="{ 'active-tab': $route.path === '/knowledge-base-management' || $route.path === '/knowledge-file-upload' }"
@click="goKnowledgeBaseManagement">
<img loading="lazy" alt="" src="@/assets/header/knowledge_base.png" :style="{
@@ -89,8 +89,7 @@
$route.path === '/server-side-management' ||
$route.path === '/agent-template-management' ||
$route.path === '/ota-management' ||
$route.path === '/user-management' ||
$route.path === '/feature-management',
$route.path === '/user-management',
}" @visible-change="handleParamDropdownVisibleChange">
<span class="el-dropdown-link">
<img loading="lazy" alt="" src="@/assets/header/param_management.png" :style="{
@@ -101,8 +100,7 @@
$route.path === '/server-side-management' ||
$route.path === '/agent-template-management' ||
$route.path === '/ota-management' ||
$route.path === '/user-management' ||
$route.path === '/feature-management'
$route.path === '/user-management'
? 'brightness(0) invert(1)'
: 'None',
}" />
@@ -131,9 +129,6 @@
<el-dropdown-item @click.native="goServerSideManagement">
{{ $t("header.serverSideManagement") }}
</el-dropdown-item>
<el-dropdown-item @click.native="goFeatureManagement">
{{ $t("header.featureManagement") }}
</el-dropdown-item>
</el-dropdown-menu>
</el-dropdown>
</div>
@@ -191,7 +186,6 @@ import userApi from "@/apis/module/user";
import i18n, { changeLanguage } from "@/i18n";
import { mapActions, mapGetters } from "vuex";
import ChangePasswordDialog from "./ChangePasswordDialog.vue"; // 引入修改密码弹窗组件
import featureManager from "@/utils/featureManager"; // 引入功能管理工具类
export default {
name: "HeaderBar",
@@ -223,11 +217,6 @@ export default {
label: "label",
children: "children",
},
// 功能状态
featureStatus: {
voiceClone: false, // 音色克隆功能状态
knowledgeBase: false, // 知识库功能状态
},
};
},
computed: {
@@ -297,14 +286,12 @@ export default {
];
},
},
async mounted() {
mounted() {
this.fetchUserInfo();
this.checkScreenSize();
window.addEventListener("resize", this.checkScreenSize);
// 从localStorage加载搜索历史
this.loadSearchHistory();
// 等待featureManager初始化完成后再加载功能状态
await this.loadFeatureStatus();
},
//移除事件监听器
beforeDestroy() {
@@ -351,20 +338,6 @@ export default {
goAgentTemplateManagement() {
this.$router.push("/agent-template-management");
},
// 跳转到功能管理
goFeatureManagement() {
this.$router.push("/feature-management");
},
// 加载功能状态
async loadFeatureStatus() {
// 等待featureManager初始化完成
await featureManager.waitForInitialization();
const config = featureManager.getConfig();
this.featureStatus.voiceClone = config.voiceClone;
this.featureStatus.knowledgeBase = config.knowledgeBase;
},
// 获取用户信息
fetchUserInfo() {
userApi.getUserInfo(({ data }) => {
File diff suppressed because it is too large Load Diff
+2 -55
View File
@@ -22,7 +22,6 @@ export default {
'header.clearHistory': 'Clear History',
'header.providerManagement': 'Provider Management',
'header.serverSideManagement': 'Server Management',
'header.featureManagement': 'System Feature Management',
'header.changePassword': 'Change Password',
'header.logout': 'Logout',
'header.searchPlaceholder': 'Search by name..',
@@ -230,26 +229,6 @@ export default {
'voicePrintDialog.requiredName': 'Please enter name',
'voicePrintDialog.requiredAudioVector': 'Please select audio vector',
// Context provider dialog related
'contextProviderDialog.title': 'Edit Source',
'contextProviderDialog.noContextApi': 'No Context API',
'contextProviderDialog.add': 'Add',
'contextProviderDialog.apiUrl': 'API URL',
'contextProviderDialog.apiUrlPlaceholder': 'http://api.example.com/data',
'contextProviderDialog.requestHeaders': 'Request Headers',
'contextProviderDialog.headerKeyPlaceholder': 'Key',
'contextProviderDialog.headerValuePlaceholder': 'Value',
'contextProviderDialog.noHeaders': 'No Headers',
'contextProviderDialog.addHeader': 'Add Header',
'contextProviderDialog.cancel': 'Cancel',
'contextProviderDialog.confirm': 'Confirm',
// Role config page - context provider related
'roleConfig.contextProvider': 'Context',
'roleConfig.contextProviderSuccess': 'Successfully added {count} sources.',
'roleConfig.contextProviderDocLink': 'How to deploy context provider',
'roleConfig.editContextProvider': 'Edit Source',
// Voice print page related
'voicePrint.pageTitle': 'Voice Print Recognition',
'voicePrint.name': 'Name',
@@ -713,7 +692,7 @@ export default {
'paramManagement.deleteFailed': 'Deletion failed, please try again',
'paramManagement.operationCancelled': 'Deletion cancelled',
'paramManagement.operationClosed': 'Operation closed',
'paramManagement.updateSuccess': 'Update successful. Some configurations will take effect only after restarting the xiaozhi-server module.',
'paramManagement.updateSuccess': 'Update successful',
'paramManagement.addSuccess': 'Add successful',
'paramManagement.updateFailed': 'Update failed',
'paramManagement.addFailed': 'Add failed',
@@ -852,7 +831,7 @@ export default {
'modelConfig.enableSuccess': 'Enable successful',
'modelConfig.disableSuccess': 'Disable successful',
'modelConfig.operationFailed': 'Operation failed',
'modelConfig.setDefaultSuccess': 'Set default model successful, please restart the xiaozhi-server module manually in time',
'modelConfig.setDefaultSuccess': 'Set default model successful',
'modelConfig.itemsPerPage': '{items} items/page',
'modelConfig.firstPage': 'First Page',
'modelConfig.prevPage': 'Previous Page',
@@ -870,7 +849,6 @@ export default {
'modelConfigDialog.setDefault': 'Set as Default',
'modelConfigDialog.modelId': 'Model ID',
'modelConfigDialog.enterModelId': 'If not filled in, it will be generated automatically',
'modelConfigDialog.invalidModelId': 'Model ID cannot be pure text or spaces, please use letters, numbers, underscores, or hyphens',
'modelConfigDialog.modelName': 'Model Name',
'modelConfigDialog.enterModelName': 'Please enter model name',
'modelConfigDialog.modelCode': 'Model Code',
@@ -1283,35 +1261,4 @@ export default {
'knowledgeFileUpload.content': 'Content:',
'knowledgeFileUpload.testQuestionRequired': 'Please enter test question',
'knowledgeBaseDialog.descriptionRequired': 'Please enter knowledge base description',
// Feature Management page text
'featureManagement.selectAll': 'Select All',
'featureManagement.deselectAll': 'Deselect All',
'featureManagement.save': 'Save Configuration',
'featureManagement.reset': 'Reset',
'featureManagement.group.featureManagement': 'Enable/Disable the feature/section',
'featureManagement.group.voiceManagement': 'Visible to users during agent configuration',
'featureManagement.noFeatures': 'No features available',
'featureManagement.contactAdmin': 'Please contact administrator to configure features',
'featureManagement.saveSuccess': 'Feature configuration saved successfully',
'featureManagement.resetConfirm': 'Are you sure you want to reset all feature configurations?',
'featureManagement.confirm': 'Confirm',
'featureManagement.cancel': 'Cancel',
'featureManagement.resetSuccess': 'Feature configuration reset successfully',
'featureManagement.noChanges': 'No changes to save',
// Feature names and descriptions
'feature.voiceprintRecognition.name': 'Voiceprint Recognition',
'feature.voiceprintRecognition.description': 'Verify user identity through voiceprint recognition technology, providing secure voice interaction experience',
'feature.voiceClone.name': 'Voice Clone',
'feature.voiceClone.description': 'Clone specific voice timbre using AI technology to achieve personalized voice synthesis',
'feature.knowledgeBase.name': 'Knowledge Base',
'feature.knowledgeBase.description': 'Build and manage knowledge base system to provide professional knowledge support for AI assistants',
'feature.mcpAccessPoint.name': 'MCP Access Point',
'feature.mcpAccessPoint.description': 'Provide MCP protocol access points to support integration of external tools and services',
'feature.vad.name': 'Voice Activity Detection',
'feature.vad.description': 'Automatically detect voice activity to optimize voice interaction response efficiency',
'feature.asr.name': 'Speech Recognition',
'feature.asr.description': 'Convert speech to text to enable natural language interaction functionality',
}
File diff suppressed because it is too large Load Diff
+2 -55
View File
@@ -22,7 +22,6 @@ export default {
'header.clearHistory': '清空历史',
'header.providerManagement': '字段管理',
'header.serverSideManagement': '服务端管理',
'header.featureManagement': '系统功能配置',
'header.changePassword': '修改密码',
'header.logout': '退出登录',
'header.searchPlaceholder': '输入名称搜索..',
@@ -230,26 +229,6 @@ export default {
'voicePrintDialog.requiredName': '请输入姓名',
'voicePrintDialog.requiredAudioVector': '请选择音频向量',
// 上下文源对话框相关
'contextProviderDialog.title': '编辑源',
'contextProviderDialog.noContextApi': '暂无上下文API',
'contextProviderDialog.add': '添加',
'contextProviderDialog.apiUrl': '接口地址',
'contextProviderDialog.apiUrlPlaceholder': 'http://api.example.com/data',
'contextProviderDialog.requestHeaders': '请求头',
'contextProviderDialog.headerKeyPlaceholder': 'Key',
'contextProviderDialog.headerValuePlaceholder': 'Value',
'contextProviderDialog.noHeaders': '暂无 Headers',
'contextProviderDialog.addHeader': '添加 Header',
'contextProviderDialog.cancel': '取消',
'contextProviderDialog.confirm': '确定',
// 角色配置页面-上下文源相关
'roleConfig.contextProvider': '上下文源',
'roleConfig.contextProviderSuccess': '已成功添加 {count} 个源。',
'roleConfig.contextProviderDocLink': '如何部署上下文源',
'roleConfig.editContextProvider': '编辑源',
// 声纹页面相关
'voicePrint.pageTitle': '声纹识别',
'voicePrint.name': '姓名',
@@ -713,7 +692,7 @@ export default {
'paramManagement.deleteFailed': '删除失败,请重试',
'paramManagement.operationCancelled': '已取消删除操作',
'paramManagement.operationClosed': '操作已关闭',
'paramManagement.updateSuccess': '修改成功,部分配置需重启xiaozhi-server模块才生效',
'paramManagement.updateSuccess': '修改成功',
'paramManagement.addSuccess': '新增成功',
'paramManagement.updateFailed': '更新失败',
'paramManagement.addFailed': '新增失败',
@@ -852,7 +831,7 @@ export default {
'modelConfig.enableSuccess': '启用成功',
'modelConfig.disableSuccess': '禁用成功',
'modelConfig.operationFailed': '操作失败',
'modelConfig.setDefaultSuccess': '设置默认模型成功,请及时手动重启xiaozhi-server模块',
'modelConfig.setDefaultSuccess': '设置默认模型成功',
'modelConfig.itemsPerPage': '{items}条/页',
'modelConfig.firstPage': '首页',
'modelConfig.prevPage': '上一页',
@@ -870,7 +849,6 @@ export default {
'modelConfigDialog.setDefault': '设为默认',
'modelConfigDialog.modelId': '模型ID',
'modelConfigDialog.enterModelId': '未填写将自动生成模型ID',
'modelConfigDialog.invalidModelId': '模型ID不能为纯文字或空格,请使用字母、数字、下划线或连字符组合',
'modelConfigDialog.modelName': '模型名称',
'modelConfigDialog.enterModelName': '请输入模型名称',
'modelConfigDialog.modelCode': '模型编码',
@@ -1283,35 +1261,4 @@ export default {
'knowledgeFileUpload.content': '内容:',
'knowledgeFileUpload.testQuestionRequired': '请输入测试问题',
'knowledgeBaseDialog.descriptionRequired': '请输入知识库描述',
// 系统功能配置页面文本
'featureManagement.selectAll': '全选',
'featureManagement.deselectAll': '取消全选',
'featureManagement.save': '保存配置',
'featureManagement.reset': '重置',
'featureManagement.group.featureManagement': '是否开启功能/板块',
'featureManagement.group.voiceManagement': '配置智能体时是否对用户可见',
'featureManagement.noFeatures': '暂无功能',
'featureManagement.contactAdmin': '请联系管理员配置功能',
'featureManagement.saveSuccess': '功能配置保存成功',
'featureManagement.resetConfirm': '确定要重置所有功能配置吗?',
'featureManagement.confirm': '确定',
'featureManagement.cancel': '取消',
'featureManagement.resetSuccess': '功能配置重置成功',
'featureManagement.noChanges': '没有需要保存的更改',
// 功能名称和描述
'feature.voiceprintRecognition.name': '声纹识别',
'feature.voiceprintRecognition.description': '通过声纹识别技术验证用户身份,提供安全的语音交互体验',
'feature.voiceClone.name': '音色克隆',
'feature.voiceClone.description': '使用AI技术克隆特定音色,实现个性化语音合成',
'feature.knowledgeBase.name': '知识库',
'feature.knowledgeBase.description': '构建和管理知识库系统,为AI助手提供专业知识支持',
'feature.mcpAccessPoint.name': 'MCP接入点',
'feature.mcpAccessPoint.description': '提供MCP协议接入点,支持外部工具和服务的集成',
'feature.vad.name': '语音活动检测',
'feature.vad.description': '自动检测语音活动,优化语音交互的响应效率',
'feature.asr.name': '语音识别',
'feature.asr.description': '将语音转换为文本,实现自然语言交互功能',
}
+2 -55
View File
@@ -22,7 +22,6 @@ export default {
'header.clearHistory': '清空歷史',
'header.providerManagement': '字段管理',
'header.serverSideManagement': '服務端管理',
'header.featureManagement': '系統功能配置',
'header.changePassword': '修改密碼',
'header.logout': '退出登錄',
'header.searchPlaceholder': '輸入名稱搜索..',
@@ -230,26 +229,6 @@ export default {
'voicePrintDialog.requiredName': '請輸入姓名',
'voicePrintDialog.requiredAudioVector': '請選擇音頻向量',
// 上下文源對話框相關
'contextProviderDialog.title': '編輯源',
'contextProviderDialog.noContextApi': '暫無上下文API',
'contextProviderDialog.add': '添加',
'contextProviderDialog.apiUrl': '接口地址',
'contextProviderDialog.apiUrlPlaceholder': 'http://api.example.com/data',
'contextProviderDialog.requestHeaders': '請求頭',
'contextProviderDialog.headerKeyPlaceholder': 'Key',
'contextProviderDialog.headerValuePlaceholder': 'Value',
'contextProviderDialog.noHeaders': '暫無 Headers',
'contextProviderDialog.addHeader': '添加 Header',
'contextProviderDialog.cancel': '取消',
'contextProviderDialog.confirm': '確定',
// 角色配置頁面-上下文源相關
'roleConfig.contextProvider': '上下文源',
'roleConfig.contextProviderSuccess': '已成功添加 {count} 個源。',
'roleConfig.contextProviderDocLink': '如何部署上下文源',
'roleConfig.editContextProvider': '編輯源',
// 聲紋頁面相關
'voicePrint.pageTitle': '聲紋識別',
'voicePrint.name': '姓名',
@@ -713,7 +692,7 @@ export default {
'paramManagement.deleteFailed': '刪除失敗,請重試',
'paramManagement.operationCancelled': '已取消刪除操作',
'paramManagement.operationClosed': '操作已關閉',
'paramManagement.updateSuccess': '修改成功,部分配置需重啟xiaozhi-server模組才生效',
'paramManagement.updateSuccess': '修改成功',
'paramManagement.addSuccess': '新增成功',
'paramManagement.updateFailed': '更新失敗',
'paramManagement.addFailed': '新增失敗',
@@ -852,7 +831,7 @@ export default {
'modelConfig.enableSuccess': '啟用成功',
'modelConfig.disableSuccess': '禁用成功',
'modelConfig.operationFailed': '操作失敗',
'modelConfig.setDefaultSuccess': '設置默認模型成功,請及時手動重啟xiaozhi-server模組',
'modelConfig.setDefaultSuccess': '設置默認模型成功',
'modelConfig.itemsPerPage': '{items}條/頁',
'modelConfig.firstPage': '首頁',
'modelConfig.prevPage': '上一頁',
@@ -870,7 +849,6 @@ export default {
'modelConfigDialog.setDefault': '設為默認',
'modelConfigDialog.modelId': '模型ID',
'modelConfigDialog.enterModelId': '未填冩將自動生成模型ID',
'modelConfigDialog.invalidModelId': '模型ID不能為純文字或空格,請使用字母、數字、底線或連字符組合',
'modelConfigDialog.modelName': '模型名稱',
'modelConfigDialog.enterModelName': '請輸入模型名稱',
'modelConfigDialog.modelCode': '模型編碼',
@@ -1283,35 +1261,4 @@ export default {
'knowledgeFileUpload.content': '內容:',
'knowledgeFileUpload.testQuestionRequired': '請輸入測試問題',
'knowledgeBaseDialog.descriptionRequired': '請輸入知识库描述',
// 功能管理頁面文本
'featureManagement.selectAll': '全選',
'featureManagement.deselectAll': '取消全選',
'featureManagement.save': '儲存配置',
'featureManagement.reset': '重置',
'featureManagement.group.featureManagement': '是否開啟功能/板块',
'featureManagement.group.voiceManagement': '配置智能体時是否對用戶可見',
'featureManagement.noFeatures': '暫無功能',
'featureManagement.contactAdmin': '請聯繫管理員配置功能',
'featureManagement.saveSuccess': '功能配置儲存成功',
'featureManagement.resetConfirm': '確定要重置所有功能配置嗎?',
'featureManagement.confirm': '確定',
'featureManagement.cancel': '取消',
'featureManagement.resetSuccess': '功能配置重置成功',
'featureManagement.noChanges': '沒有需要儲存的更改',
// 功能名稱和描述
'feature.voiceprintRecognition.name': '聲紋識別',
'feature.voiceprintRecognition.description': '通過聲紋識別技術驗證用戶身份,提供安全的語音交互體驗',
'feature.voiceClone.name': '音色複刻',
'feature.voiceClone.description': '使用AI技術複刻特定音色,實現個性化語音合成',
'feature.knowledgeBase.name': '知識庫',
'feature.knowledgeBase.description': '構建和管理知識庫系統,為AI助手提供專業知識支持',
'feature.mcpAccessPoint.name': 'MCP接入點',
'feature.mcpAccessPoint.description': '提供MCP協議接入點,支持外部工具和服務的整合',
'feature.vad.name': '語音活動檢測',
'feature.vad.description': '自動檢測語音活動,優化語音交互的響應效率',
'feature.asr.name': '語音識別',
'feature.asr.description': '將語音轉換為文本,實現自然語言交互功能',
}
-1
View File
@@ -8,7 +8,6 @@ import store from './store';
import i18n from './i18n';
import './styles/global.scss';
import { register as registerServiceWorker } from './registerServiceWorker';
import featureManager from './utils/featureManager';
// 创建事件总线,用于组件间通信
Vue.prototype.$eventBus = new Vue();
+3 -18
View File
@@ -76,8 +76,7 @@ const routes = [
return import('../views/ModelConfig.vue')
}
},
{
path: '/params-management',
{ path: '/params-management',
name: 'ParamsManagement',
component: function () {
return import('../views/ParamsManagement.vue')
@@ -87,8 +86,7 @@ const routes = [
title: '参数管理'
}
},
{
path: '/knowledge-base-management',
{ path: '/knowledge-base-management',
name: 'KnowledgeBaseManagement',
component: function () {
return import('../views/KnowledgeBaseManagement.vue')
@@ -98,8 +96,7 @@ const routes = [
title: '知识库管理'
}
},
{
path: '/knowledge-file-upload',
{ path: '/knowledge-file-upload',
name: 'KnowledgeFileUpload',
component: function () {
return import('../views/KnowledgeFileUpload.vue')
@@ -184,18 +181,6 @@ const routes = [
return import('../views/TemplateQuickConfig.vue')
}
},
// 功能配置页面路由
{
path: '/feature-management',
name: 'FeatureManagement',
component: function () {
return import('../views/FeatureManagement.vue')
},
meta: {
requiresAuth: true,
title: '功能配置'
}
},
]
const router = new VueRouter({
base: process.env.VUE_APP_PUBLIC_PATH || '/',
@@ -1,340 +0,0 @@
//功能配置工具
import Api from "@/apis/api";
class FeatureManager {
constructor() {
this.defaultFeatures = {
voiceprintRecognition: {
name: 'feature.voiceprintRecognition.name',
enabled: false,
description: 'feature.voiceprintRecognition.description'
},
voiceClone: {
name: 'feature.voiceClone.name',
enabled: false,
description: 'feature.voiceClone.description'
},
knowledgeBase: {
name: 'feature.knowledgeBase.name',
enabled: false,
description: 'feature.knowledgeBase.description'
},
mcpAccessPoint: {
name: 'feature.mcpAccessPoint.name',
enabled: false,
description: 'feature.mcpAccessPoint.description'
},
vad: {
name: 'feature.vad.name',
enabled: false,
description: 'feature.vad.description'
},
asr: {
name: 'feature.asr.name',
enabled: false,
description: 'feature.asr.description'
}
};
this.currentFeatures = { ...this.defaultFeatures }; // 当前内存中的配置
this.initialized = false;
this.initPromise = null;
}
/**
* 等待初始化完成
*/
async waitForInitialization() {
if (!this.initPromise) {
this.initPromise = this.init();
}
await this.initPromise;
return this.initialized;
}
/**
* 初始化功能配置
*/
async init() {
try {
// 从pub-config接口获取配置
const config = await this.getConfigFromPubConfig();
if (config) {
this.currentFeatures = { ...config }; // 保存到内存
this.initialized = true;
return;
}
} catch (error) {
console.warn('从pub-config接口获取配置失败:', error);
}
// pub-config接口失败,使用默认配置
this.currentFeatures = { ...this.defaultFeatures }; // 保存默认配置到内存
this.initialized = true;
}
/**
* 从pub-config接口获取配置
*/
async getConfigFromPubConfig() {
return new Promise((resolve) => {
// 直接调用pub-config接口获取配置
Api.user.getPubConfig((result) => {
// 检查返回结果的结构
if (result && result.status === 200) {
// 检查是否有data字段
if (result.data) {
// 检查是否有code字段,如果有则按照code判断
if (result.data.code !== undefined) {
if (result.data.code === 0 && result.data.data && result.data.data.systemWebMenu) {
try {
let config;
if (typeof result.data.data.systemWebMenu === 'string') {
// 如果是字符串,需要解析JSON
config = JSON.parse(result.data.data.systemWebMenu);
} else {
// 如果已经是对象,直接使用
config = result.data.data.systemWebMenu;
}
// 检查配置中是否包含features对象
if (config && config.features) {
// 确保knowledgeBase功能存在且配置正确
if (!config.features.knowledgeBase) {
console.warn('配置中缺少knowledgeBase功能,合并默认配置');
config.features = { ...this.defaultFeatures, ...config.features };
}
resolve(config.features);
} else {
console.warn('配置中缺少features对象,使用默认配置');
resolve(this.defaultFeatures);
}
} catch (error) {
console.warn('处理systemWebMenu配置失败:', error);
resolve(null);
}
} else {
console.warn('接口返回code不为0或缺少必要数据,使用默认配置');
resolve(null);
}
} else {
// 如果没有code字段,直接检查systemWebMenu
if (result.data && result.data.systemWebMenu) {
try {
let config;
if (typeof result.data.systemWebMenu === 'string') {
// 如果是字符串,需要解析JSON
config = JSON.parse(result.data.systemWebMenu);
} else {
// 如果已经是对象,直接使用
config = result.data.systemWebMenu;
}
// 检查配置中是否包含features对象
if (config && config.features) {
// 确保knowledgeBase功能存在且配置正确
if (!config.features.knowledgeBase) {
console.warn('配置中缺少knowledgeBase功能,合并默认配置');
config.features = { ...this.defaultFeatures, ...config.features };
}
resolve(config.features);
} else {
console.warn('配置中缺少features对象,使用默认配置');
resolve(this.defaultFeatures);
}
} catch (error) {
console.warn('处理systemWebMenu配置失败:', error);
resolve(null);
}
} else {
console.warn('接口返回缺少systemWebMenu数据,使用默认配置');
resolve(null);
}
}
} else {
console.warn('接口返回数据中缺少data字段,使用默认配置');
resolve(null);
}
} else {
console.warn('pub-config接口调用失败,使用默认配置');
resolve(null);
}
});
});
}
/**
* 获取当前配置
*/
getCurrentConfig() {
// 返回内存中的当前配置
return this.currentFeatures;
}
/**
* 保存配置到后端API
*/
async saveConfig(config) {
try {
// 更新内存中的配置
this.currentFeatures = { ...config };
// 异步保存到后端API
this.saveConfigToAPI(config).catch(error => {
console.warn('保存配置到API失败:', error);
});
// 触发配置变更事件
window.dispatchEvent(new CustomEvent('featureConfigChanged', {
detail: config
}));
} catch (error) {
console.error('保存功能配置失败:', error);
}
}
/**
* 保存配置到后端API
*/
async saveConfigToAPI(config) {
return new Promise((resolve) => {
// 直接使用已知的ID(600)更新参数
Api.admin.updateParam(
{
id: 600,
paramCode: 'system-web.menu',
paramValue: JSON.stringify({
features: config,
groups: {
featureManagement: ["voiceprintRecognition", "voiceClone", "knowledgeBase", "mcpAccessPoint"],
voiceManagement: ["vad", "asr"]
}
}),
valueType: 'json',
remark: '系统功能菜单配置'
},
(updateResult) => {
if (updateResult.code === 0) {
resolve();
} else {
// 如果更新失败,可能是参数不存在或其他错误,记录但不阻止保存到localStorage
console.warn('更新参数失败:', updateResult.msg);
resolve(); // 不阻止保存到localStorage
}
},
(error) => {
console.warn('更新参数失败:', error);
resolve(); // 不阻止保存到localStorage
}
);
});
}
/**
* 获取所有功能配置
*/
getAllFeatures() {
return this.getCurrentConfig();
}
/**
* 获取简化的配置对象(用于首页组件)
*/
getConfig() {
const features = this.getAllFeatures();
return {
voiceprintRecognition: features.voiceprintRecognition?.enabled || false,
voiceClone: features.voiceClone?.enabled || false,
knowledgeBase: features.knowledgeBase?.enabled || false,
mcpAccessPoint: features.mcpAccessPoint?.enabled || false,
vad: features.vad?.enabled || false,
asr: features.asr?.enabled || false
};
}
/**
* 获取指定功能的状态
*/
getFeatureStatus(featureKey) {
const features = this.getAllFeatures();
return features[featureKey]?.enabled || false;
}
/**
* 设置功能状态
*/
setFeatureStatus(featureKey, enabled) {
const features = this.getAllFeatures();
if (features[featureKey]) {
features[featureKey].enabled = enabled;
this.saveConfig(features);
return true;
}
return false;
}
/**
* 启用功能
*/
enableFeature(featureKey) {
return this.setFeatureStatus(featureKey, true);
}
/**
* 禁用功能
*/
disableFeature(featureKey) {
return this.setFeatureStatus(featureKey, false);
}
/**
* 切换功能状态
*/
toggleFeature(featureKey) {
const currentStatus = this.getFeatureStatus(featureKey);
return this.setFeatureStatus(featureKey, !currentStatus);
}
/**
* 重置所有功能为默认状态
*/
resetToDefault() {
this.saveConfig(this.defaultFeatures);
}
/**
* 批量更新功能状态
*/
updateFeatures(featureUpdates) {
const features = this.getAllFeatures();
Object.keys(featureUpdates).forEach(featureKey => {
if (features[featureKey]) {
features[featureKey].enabled = featureUpdates[featureKey];
}
});
this.saveConfig(features);
}
/**
* 获取已启用的功能列表
*/
getEnabledFeatures() {
const features = this.getAllFeatures();
return Object.keys(features).filter(key => features[key].enabled);
}
/**
* 检查功能是否启用
*/
isFeatureEnabled(featureKey) {
return this.getFeatureStatus(featureKey);
}
}
// 创建单例实例
const featureManager = new FeatureManager();
export default featureManager;
@@ -1,601 +0,0 @@
<template>
<div class="welcome">
<HeaderBar />
<div class="operation-bar">
<h2 class="page-title">{{ $t('header.featureManagement') }}</h2>
</div>
<div class="main-wrapper">
<div class="content-panel">
<div class="content-area">
<el-card class="feature-card" shadow="never">
<div class="config-header">
<div class="header-icon">
<img loading="lazy" src="@/assets/home/equipment.png" alt="" />
</div>
<div class="header-actions">
<el-button @click="!isSaving && toggleSelectAll()" class="btn-select-all" :disabled="isSaving">
{{ isAllSelected ? $t('featureManagement.deselectAll') : $t('featureManagement.selectAll') }}
</el-button>
<el-button type="primary" class="save-btn" @click="handleSave" :disabled="isSaving">
{{ isSaving ? $t('featureManagement.saving') : $t('featureManagement.save') }}
</el-button>
<el-button class="reset-btn" @click="handleReset" :disabled="isSaving">
{{ $t('featureManagement.reset') }}
</el-button>
</div>
</div>
<div class="divider"></div>
<!-- 功能分组容器 - 左右布局 -->
<div class="feature-groups-container">
<!-- 功能管理分组 -->
<div v-if="featureManagementFeatures.length > 0" class="feature-group">
<h3 class="group-title">{{ $t('featureManagement.group.featureManagement') }}</h3>
<div class="features-grid">
<div
v-for="feature in featureManagementFeatures"
:key="feature.id"
class="feature-card-item"
:class="{ 'feature-enabled': feature.enabled, 'feature-disabled': isSaving }"
@click="!isSaving && toggleFeature(feature)"
>
<div class="feature-header">
<h3 class="feature-name">{{ $t(`feature.${feature.id}.name`) }}</h3>
<el-checkbox
v-model="feature.enabled"
@change="!isSaving && toggleFeature(feature)"
class="feature-checkbox"
:disabled="isSaving"
/>
</div>
<p class="feature-description">{{ $t(`feature.${feature.id}.description`) }}</p>
</div>
</div>
</div>
<!-- 语音管理分组 -->
<div v-if="voiceManagementFeatures.length > 0" class="feature-group">
<h3 class="group-title">{{ $t('featureManagement.group.voiceManagement') }}</h3>
<div class="features-grid">
<div
v-for="feature in voiceManagementFeatures"
:key="feature.id"
class="feature-card-item"
:class="{ 'feature-enabled': feature.enabled, 'feature-disabled': isSaving }"
@click="!isSaving && toggleFeature(feature)"
>
<div class="feature-header">
<h3 class="feature-name">{{ $t(`feature.${feature.id}.name`) }}</h3>
<el-checkbox
v-model="feature.enabled"
@change="!isSaving && toggleFeature(feature)"
class="feature-checkbox"
:disabled="isSaving"
/>
</div>
<p class="feature-description">{{ $t(`feature.${feature.id}.description`) }}</p>
</div>
</div>
</div>
</div>
<div v-if="filteredFeatures.length === 0" class="empty-state">
<el-empty :description="$t('featureManagement.noFeatures')">
<p class="empty-tip">{{ $t('featureManagement.contactAdmin') }}</p>
</el-empty>
</div>
</el-card>
</div>
</div>
</div>
<el-footer>
<VersionFooter />
</el-footer>
</div>
</template>
<script>
import HeaderBar from "@/components/HeaderBar.vue";
import VersionFooter from "@/components/VersionFooter.vue";
import featureManager from "@/utils/featureManager.js";
export default {
name: "FeatureManagement",
components: {
HeaderBar,
VersionFooter
},
data() {
return {
pendingChanges: false,
featureManagementFeatures: [],
voiceManagementFeatures: [],
isSaving: false // 添加保存状态锁定
}
},
computed: {
// 所有功能列表
filteredFeatures() {
return [...this.featureManagementFeatures, ...this.voiceManagementFeatures]
},
// 判断是否所有功能都已选中
isAllSelected() {
const allFeatures = [...this.featureManagementFeatures, ...this.voiceManagementFeatures]
return allFeatures.length > 0 && allFeatures.every(feature => feature.enabled)
}
},
async created() {
// 等待功能配置管理器初始化完成
try {
console.log('等待功能配置管理器初始化...')
await featureManager.waitForInitialization()
console.log('功能配置管理器初始化完成,开始加载功能配置')
await this.loadFeatures()
this.setupConfigChangeListener()
} catch (error) {
console.error('功能配置管理器初始化等待失败:', error)
await this.loadFeatures()
this.setupConfigChangeListener()
}
},
beforeDestroy() {
this.removeConfigChangeListener()
},
methods: {
// 根据ID列表获取功能
async getFeaturesByIds(featureIds) {
try {
const featureConfig = await featureManager.getAllFeatures()
console.log('获取到的功能配置:', JSON.stringify(featureConfig, null, 2))
console.log('请求的功能ID列表:', featureIds)
const result = featureIds.map(id => {
const feature = featureConfig[id]
console.log(`功能 ${id} 的配置:`, feature)
console.log(`功能 ${id} 的启用状态:`, feature?.enabled)
return {
id: id,
name: this.$t(`feature.${id}.name`),
description: this.$t(`feature.${id}.description`),
enabled: feature?.enabled || false
}
})
console.log('最终返回的功能列表:', JSON.stringify(result, null, 2))
return result
} catch (error) {
console.error('获取功能配置失败:', error)
// 如果获取失败,返回默认配置
return featureIds.map(id => ({
id: id,
name: this.$t(`feature.${id}.name`),
description: this.$t(`feature.${id}.description`),
enabled: false
}))
}
},
// 加载功能配置
async loadFeatures() {
// 保存当前用户的选择状态
const currentFeatureStates = {}
const allCurrentFeatures = [...this.featureManagementFeatures, ...this.voiceManagementFeatures]
allCurrentFeatures.forEach(feature => {
currentFeatureStates[feature.id] = feature.enabled
})
// 重新加载配置
this.featureManagementFeatures = await this.getFeaturesByIds(['voiceprintRecognition', 'voiceClone', 'knowledgeBase', 'mcpAccessPoint'])
this.voiceManagementFeatures = await this.getFeaturesByIds(['vad', 'asr'])
// 恢复用户的选择状态(如果存在)
const allFeatures = [...this.featureManagementFeatures, ...this.voiceManagementFeatures]
allFeatures.forEach(feature => {
if (currentFeatureStates.hasOwnProperty(feature.id)) {
feature.enabled = currentFeatureStates[feature.id]
}
})
},
// 切换功能状态
async toggleFeature(feature) {
// 如果正在保存,阻止操作
if (this.isSaving) {
return
}
feature.enabled = !feature.enabled
this.pendingChanges = true
// 不再立即更新到配置管理器,只在保存时统一更新
},
// 保存配置
async handleSave() {
if (!this.pendingChanges) {
this.$message.info({
message: this.$t('featureManagement.noChanges'),
showClose: true
})
return
}
// 设置保存状态,锁定界面
this.isSaving = true
try {
// 获取当前所有功能的状态并保存
const featureUpdates = {}
const allFeatures = [...this.featureManagementFeatures, ...this.voiceManagementFeatures]
allFeatures.forEach(feature => {
featureUpdates[feature.id] = feature.enabled
})
await featureManager.updateFeatures(featureUpdates)
this.pendingChanges = false
this.$message.success({
message: this.$t('featureManagement.saveSuccess'),
showClose: true
})
setTimeout(() => {
this.loadFeatures()
this.$router.go(0)
}, 1000)
} catch (error) {
console.error('保存配置失败:', error)
this.$message.error({
message: this.$t('featureManagement.saveError'),
showClose: true
})
} finally {
// 无论成功与否,都解除保存状态锁定
this.isSaving = false
}
},
// 设置配置变化监听器
setupConfigChangeListener() {
this.configChangeHandler = () => {
console.log('检测到配置变化,重新加载功能列表')
this.loadFeatures()
}
window.addEventListener('featureConfigReloaded', this.configChangeHandler)
},
// 移除配置变化监听器
removeConfigChangeListener() {
if (this.configChangeHandler) {
window.removeEventListener('featureConfigReloaded', this.configChangeHandler)
}
},
// 重置配置
async handleReset() {
try {
await this.$confirm(
this.$t('featureManagement.resetConfirm'),
this.$t('featureManagement.reset'),
{
confirmButtonText: this.$t('featureManagement.confirm'),
cancelButtonText: this.$t('featureManagement.cancel'),
type: 'warning'
}
)
featureManager.resetToDefault()
this.loadFeatures()
this.pendingChanges = false
this.$message.success({
message: this.$t('featureManagement.resetSuccess'),
showClose: true
})
setTimeout(() => {
this.loadFeatures()
this.$router.go(0)
}, 1000)
} catch (error) {
// 用户取消操作
}
},
// 搜索功能(预留接口)
handleSearch() {
// 搜索功能待实现
},
// 全选/取消全选
toggleSelectAll() {
// 如果正在保存,阻止操作
if (this.isSaving) {
return
}
const allFeatures = [...this.featureManagementFeatures, ...this.voiceManagementFeatures]
const newStatus = !this.isAllSelected
allFeatures.forEach(feature => {
feature.enabled = newStatus
})
this.pendingChanges = true
}
}
}
</script>
<style scoped>
.welcome {
min-width: 900px;
min-height: 506px;
height: 100vh;
display: flex;
position: relative;
flex-direction: column;
background-size: cover;
background: linear-gradient(to bottom right, #dce8ff, #e4eeff, #e6cbfd) center;
-webkit-background-size: cover;
-o-background-size: cover;
overflow: hidden;
}
.operation-bar {
display: flex;
justify-content: space-between;
align-items: center;
padding: 16px 24px;
}
.page-title {
font-size: 24px;
margin: 0;
}
.config-header {
display: flex;
align-items: center;
justify-content: space-between;
padding: 0 0 16px 0;
}
.header-icon {
width: 40px;
height: 40px;
background: #5778ff;
border-radius: 50%;
display: flex;
align-items: center;
justify-content: center;
margin-right: 12px;
}
.header-icon img {
width: 20px;
height: 20px;
}
.header-actions {
display: flex;
align-items: center;
gap: 8px;
margin-left: auto;
}
.divider {
height: 1px;
background: #e0e0e0;
margin-bottom: 20px;
}
.btn-select-all {
background: #e6ebff;
color: #5778ff;
border: 1px solid #adbdff;
border-radius: 18px;
padding: 8px 16px;
height: 32px;
font-size: 14px;
}
.btn-select-all:hover {
background: #d0d8ff;
}
.save-btn {
background: #5778ff;
color: white;
border: none;
border-radius: 18px;
padding: 8px 16px;
height: 32px;
font-size: 14px;
}
.save-btn:hover {
background: #4a6ae8;
}
.reset-btn {
background: #e6ebff;
color: #5778ff;
border: 1px solid #adbdff;
border-radius: 18px;
padding: 8px 16px;
height: 32px;
}
.reset-btn:hover {
background: #d0d8ff;
}
.main-wrapper {
margin: 0 22px 5px 22px;
border-radius: 15px;
min-height: calc(100vh - 24vh);
height: auto;
max-height: 80vh;
box-shadow: 0 2px 12px rgba(0, 0, 0, 0.1);
position: relative;
background: rgba(237, 242, 255, 0.5);
display: flex;
flex-direction: column;
}
.content-panel {
flex: 1;
display: flex;
overflow: hidden;
height: 100%;
border-radius: 15px;
background: transparent;
border: 1px solid #fff;
}
.content-area {
flex: 1;
height: 100%;
min-width: 600px;
overflow: auto;
background-color: white;
display: flex;
flex-direction: column;
}
.feature-card {
background: white;
flex: 1;
display: flex;
flex-direction: column;
border: none;
box-shadow: none;
overflow: hidden;
}
.feature-card ::v-deep .el-card__body {
padding: 24px;
display: flex;
flex-direction: column;
flex: 1;
overflow: hidden;
}
.features-grid {
display: grid;
grid-template-columns: repeat(auto-fill, minmax(300px, 1fr));
gap: 16px;
}
.feature-card-item {
display: flex;
flex-direction: column;
padding: 20px;
border-radius: 12px;
border: 2px solid #e0e0e0;
background-color: white;
cursor: pointer;
transition: all 0.3s ease;
user-select: none;
position: relative;
}
.feature-card-item:hover {
border-color: #869bf0;
box-shadow: 0 4px 12px rgba(0, 0, 0, 0.15);
transform: translateY(-2px);
}
.feature-card-item.feature-enabled {
border-color:#5778ff;
box-shadow: 0 4px 16px rgba(95, 112, 243, 0.2);
transform: translateY(-2px);
}
.feature-header {
display: flex;
justify-content: space-between;
align-items: center;
margin-bottom: 12px;
}
.feature-checkbox ::v-deep .el-checkbox__input {
transform: scale(1.2);
}
.feature-checkbox ::v-deep .el-checkbox__input.is-checked .el-checkbox__inner {
background-color: #5778ff;
border-color: #5778ff;
}
.feature-checkbox ::v-deep .el-checkbox__input.is-checked + .el-checkbox__label {
color: #5778ff;
}
.feature-name {
font-size: 18px;
font-weight: 600;
color: #333;
margin: 0;
transition: color 0.3s ease;
}
.feature-description {
font-size: 14px;
line-height: 1.6;
color: #666;
margin: 0 0 12px 0;
transition: color 0.3s ease;
text-align: left;
}
/* 功能分组容器 - 左右布局 */
.feature-groups-container {
display: flex;
gap: 32px;
align-items: flex-start;
position: relative;
}
/* 分组之间的分隔线 */
.feature-groups-container::before {
content: '';
position: absolute;
left: 50%;
top: 0;
bottom: 0;
width: 1px;
height: 550px;
background: #e0e0e0;
opacity: 0.5;
transform: translateX(-50%);
}
/* 分组样式 */
.feature-group {
flex: 1;
min-width: 0;
margin-bottom: 32px;
}
.group-title {
font-size: 18px;
font-weight: 600;
color: #303133;
margin-bottom: 12px;
padding-left: 12px;
border-left: 4px solid #5f70f3;
text-align: left;
}
.features-grid {
display: grid;
grid-template-columns: repeat(auto-fill, minmax(300px, 1fr));
gap: 16px;
}
</style>
@@ -214,11 +214,10 @@
<el-dialog :title="$t('knowledgeFileUpload.retrievalTest')" :visible.sync="retrievalTestDialogVisible"
width="1200px" class="retrieval-test-dialog">
<div class="retrieval-test-form">
<el-form :model="retrievalTestForm" label-width="80px" @submit.native.prevent="runRetrievalTest">
<el-form :model="retrievalTestForm" label-width="80px">
<el-form-item :label="$t('knowledgeFileUpload.testQuestion')" required>
<el-input v-model="retrievalTestForm.question"
:placeholder="$t('knowledgeFileUpload.testQuestionPlaceholder')" style="width: 100%; max-height: 80px;"
@keyup.enter.native="runRetrievalTest">
:placeholder="$t('knowledgeFileUpload.testQuestionPlaceholder')" style="width: 100%; max-height: 80px;">
</el-input>
</el-form-item>
</el-form>
+2 -2
View File
@@ -38,11 +38,11 @@
position: absolute;
top: 50%;
transform: translateY(-50%);
right: 18%;
right: 15%;
background-color: #fff;
border-radius: 20px;
padding: 35px 0;
width: 450px;
width: 550px;
box-sizing: border-box;
}
+4 -24
View File
@@ -39,9 +39,8 @@
</template>
<template v-else>
<DeviceItem v-for="(item, index) in devices" :key="index" :device="item" :feature-status="featureStatus"
@configure="goToRoleConfig" @deviceManage="handleDeviceManage" @delete="handleDeleteAgent"
@chat-history="handleShowChatHistory" />
<DeviceItem v-for="(item, index) in devices" :key="index" :device="item" @configure="goToRoleConfig"
@deviceManage="handleDeviceManage" @delete="handleDeleteAgent" @chat-history="handleShowChatHistory" />
</template>
</div>
</div>
@@ -62,7 +61,6 @@ import ChatHistoryDialog from '@/components/ChatHistoryDialog.vue';
import DeviceItem from '@/components/DeviceItem.vue';
import HeaderBar from '@/components/HeaderBar.vue';
import VersionFooter from '@/components/VersionFooter.vue';
import featureManager from '@/utils/featureManager';
export default {
name: 'HomePage',
@@ -78,33 +76,15 @@ export default {
skeletonCount: localStorage.getItem('skeletonCount') || 8,
showChatHistory: false,
currentAgentId: '',
currentAgentName: '',
// 功能状态
featureStatus: {
voiceprintRecognition: false,
voiceClone: false,
knowledgeBase: false
}
currentAgentName: ''
}
},
async mounted() {
mounted() {
this.fetchAgentList();
await this.loadFeatureStatus();
},
methods: {
// 加载功能状态
async loadFeatureStatus() {
await featureManager.waitForInitialization();
const config = featureManager.getConfig();
this.featureStatus = {
voiceprintRecognition: config.voiceprintRecognition,
voiceClone: config.voiceClone,
knowledgeBase: config.knowledgeBase
};
},
showAddDialog() {
this.addDeviceDialogVisible = true
},
-8
View File
@@ -156,7 +156,6 @@ import VersionFooter from "@/components/VersionFooter.vue";
import i18n, { changeLanguage } from "@/i18n";
import { getUUID, goToPage, showDanger, showSuccess, sm2Encrypt, validateMobile } from "@/utils";
import { mapState } from "vuex";
import featureManager from "@/utils/featureManager";
export default {
name: "login",
@@ -215,13 +214,6 @@ export default {
this.$store.dispatch("fetchPubConfig").then(() => {
// 根据配置决定默认登录方式
this.isMobileLogin = this.enableMobileRegister;
// pub-config接口调用完成后,重新初始化featureManager以确保使用最新的配置
featureManager.waitForInitialization().then(() => {
console.log('featureManager重新初始化完成,使用pub-config配置');
}).catch(error => {
console.warn('featureManager重新初始化失败:', error);
});
});
},
methods: {
+16 -112
View File
@@ -55,24 +55,10 @@
</div>
</div>
</el-form-item>
<el-form-item :label="$t('roleConfig.contextProvider') + ''" class="context-provider-item">
<div style="display: flex; align-items: center; justify-content: space-between;">
<span style="color: #606266; font-size: 13px;">
{{ $t('roleConfig.contextProviderSuccess', { count: currentContextProviders.length }) }}<a href="https://github.com/xinnan-tech/xiaozhi-esp32-server/blob/main/docs/context-provider-integration.md" target="_blank" class="doc-link">{{ $t('roleConfig.contextProviderDocLink') }}</a>
</span>
<el-button
class="edit-function-btn"
size="small"
@click="openContextProviderDialog"
>
{{ $t('roleConfig.editContextProvider') }}
</el-button>
</div>
</el-form-item>
<el-form-item :label="$t('roleConfig.roleIntroduction') + ''">
<el-input
type="textarea"
rows="8"
rows="9"
resize="none"
:placeholder="$t('roleConfig.pleaseEnterContent')"
v-model="form.systemPrompt"
@@ -85,7 +71,7 @@
<el-form-item :label="$t('roleConfig.memoryHis') + ''">
<el-input
type="textarea"
rows="4"
rows="6"
resize="none"
v-model="form.summaryMemory"
maxlength="2000"
@@ -121,11 +107,7 @@
</div>
<div class="form-column">
<div class="model-row">
<el-form-item
v-if="featureStatus.vad"
:label="$t('roleConfig.vad')"
class="model-item"
>
<el-form-item :label="$t('roleConfig.vad')" class="model-item">
<div class="model-select-wrapper">
<el-select
v-model="form.model.vadModelId"
@@ -143,11 +125,7 @@
</el-select>
</div>
</el-form-item>
<el-form-item
v-if="featureStatus.asr"
:label="$t('roleConfig.asr')"
class="model-item"
>
<el-form-item :label="$t('roleConfig.asr')" class="model-item">
<div class="model-select-wrapper">
<el-select
v-model="form.model.asrModelId"
@@ -289,11 +267,6 @@
@update-functions="handleUpdateFunctions"
@dialog-closed="handleDialogClosed"
/>
<context-provider-dialog
:visible.sync="showContextProviderDialog"
:providers="currentContextProviders"
@confirm="handleUpdateContext"
/>
</div>
</template>
@@ -302,17 +275,14 @@ import Api from "@/apis/api";
import { getServiceUrl } from "@/apis/api";
import RequestService from "@/apis/httpRequest";
import FunctionDialog from "@/components/FunctionDialog.vue";
import ContextProviderDialog from "@/components/ContextProviderDialog.vue";
import HeaderBar from "@/components/HeaderBar.vue";
import i18n from "@/i18n";
import featureManager from "@/utils/featureManager";
export default {
name: "RoleConfigPage",
components: { HeaderBar, FunctionDialog, ContextProviderDialog },
components: { HeaderBar, FunctionDialog },
data() {
return {
showContextProviderDialog: false,
form: {
agentCode: "",
agentName: "",
@@ -350,18 +320,12 @@ export default {
voiceDetails: {}, // 保存完整的音色信息
showFunctionDialog: false,
currentFunctions: [],
currentContextProviders: [],
allFunctions: [],
originalFunctions: [],
playingVoice: false,
isPaused: false,
currentAudio: null,
currentPlayingVoiceId: null,
// 功能状态
featureStatus: {
vad: false, // 语言检测活动功能状态
asr: false, // 语音识别功能状态
},
};
},
methods: {
@@ -392,7 +356,6 @@ export default {
paramInfo: item.params,
};
}),
contextProviders: this.currentContextProviders,
};
Api.agent.updateAgentConfig(this.$route.query.agentId, configData, ({ data }) => {
if (data.code === 0) {
@@ -509,9 +472,6 @@ export default {
};
// 后端只给了最小映射:[{ id, agentId, pluginId }, ...]
const savedMappings = data.data.functions || [];
// 加载上下文配置
this.currentContextProviders = data.data.contextProviders || [];
// 先保证 allFunctions 已经加载(如果没有,则先 fetchAllFunctions
const ensureFuncs = this.allFunctions.length
@@ -644,14 +604,15 @@ export default {
if (type === "Intent" && value !== "Intent_nointent") {
this.fetchAllFunctions();
}
if (type === "Memory") {
if (value === "Memory_nomem") {
// 无记忆功能的模型,默认不记录聊天记录
this.form.chatHistoryConf = 0;
} else {
// 有记忆功能的模型,默认记录文本和语音
this.form.chatHistoryConf = 2;
}
if (type === "Memory" && value === "Memory_nomem") {
this.form.chatHistoryConf = 0;
}
if (
type === "Memory" &&
value !== "Memory_nomem" &&
(this.form.chatHistoryConf === 0 || this.form.chatHistoryConf === null)
) {
this.form.chatHistoryConf = 2;
}
if (type === "LLM") {
// 当LLM类型改变时,更新意图识别选项的可见性
@@ -686,12 +647,6 @@ export default {
this.showFunctionDialog = true;
}
},
openContextProviderDialog() {
this.showContextProviderDialog = true;
},
handleUpdateContext(providers) {
this.currentContextProviders = providers;
},
handleUpdateFunctions(selected) {
this.currentFunctions = selected;
},
@@ -1026,19 +981,6 @@ export default {
this.form.chatHistoryConf = 0;
}
},
// 加载功能状态
async loadFeatureStatus() {
try {
// 确保featureManager已初始化完成
await featureManager.waitForInitialization();
const config = featureManager.getConfig();
this.featureStatus.voiceprintRecognition = config.voiceprintRecognition || false;
this.featureStatus.vad = config.vad || false;
this.featureStatus.asr = config.asr || false;
} catch (error) {
console.error("加载功能状态失败:", error);
}
},
},
watch: {
"form.model.ttsModelId": {
@@ -1061,7 +1003,7 @@ export default {
immediate: true,
},
},
async mounted() {
mounted() {
const agentId = this.$route.query.agentId;
if (agentId) {
this.fetchAgentConfig(agentId);
@@ -1069,8 +1011,6 @@ export default {
}
this.fetchModelOptions();
this.fetchTemplates();
// 加载功能状态,确保featureManager已初始化
await this.loadFeatureStatus();
},
};
</script>
@@ -1230,8 +1170,7 @@ export default {
.template-item {
height: 4vh;
min-width: 60px;
padding: 0 12px;
width: 76px;
border-radius: 8px;
background: #e6ebff;
line-height: 4vh;
@@ -1241,7 +1180,6 @@ export default {
color: #5778ff;
cursor: pointer;
transition: background-color 0.3s ease;
white-space: nowrap;
}
.template-item:hover {
@@ -1359,26 +1297,6 @@ export default {
justify-content: flex-end;
}
.chat-history-options ::v-deep .el-radio-button {
border-color: #5778ff;
}
.chat-history-options ::v-deep .el-radio-button .el-radio-button__inner {
color: #5778ff;
border-color: #5778ff;
background-color: transparent;
}
.chat-history-options ::v-deep .el-radio-button.is-active .el-radio-button__inner {
background-color: #5778ff;
border-color: #5778ff;
color: white;
}
.chat-history-options ::v-deep .el-radio-button .el-radio-button__inner:hover {
color: #5778ff;
}
.header-actions {
display: flex;
align-items: center;
@@ -1426,18 +1344,4 @@ export default {
height: 32px;
margin-left: 8px;
}
.context-provider-item ::v-deep .el-form-item__label {
line-height: 42px !important;
}
.doc-link {
color: #5778ff;
text-decoration: none;
margin-left: 4px;
&:hover {
text-decoration: underline;
}
}
</style>
@@ -56,7 +56,6 @@
3. **洞察需求:** 结合上下文**深入理解用户真实意图**后再决定调用,避免无意义调用。
4. **独立任务:** 除`<context>`已涵盖信息外,用户每个要求(即使相似)都视为**独立任务**,需调用工具获取最新数据,**不可偷懒复用历史结果**。
5. **不确定时:** **切勿猜测或编造答案**。若不确定相关操作,可引导用户澄清或告知能力限制。
6. **多工具调用:** 当用户要求执行多个任务时,你会调用多个工具(数量不定)。**重要:在获取到所有工具结果后,你必须依次总结每个工具的查询结果**,不要遗漏任何一个。例如用户问"设备当前状态,某某地方的天气和社会新闻",你要先说设备状态,再说天气情况,最后说新闻内容。
- **重要例外(无需调用):**
- `查询"现在的时间"、"今天的日期/星期几"、"今天农历"、"{{local_address}}的天气/未来天气"` -> **直接使用`<context>`信息回复**。
- **需要调用的情况(示例):**
@@ -74,7 +73,6 @@
- **今天农历:** {{lunar_date}}
- **用户所在城市:** {{local_address}}
- **当地未来7天天气:** {{weather_info}}
{{ dynamic_context }}
</context>
<memory>
-8
View File
@@ -9,7 +9,6 @@ from core.utils.util import get_local_ip, validate_mcp_endpoint
from core.http_server import SimpleHttpServer
from core.websocket_server import WebSocketServer
from core.utils.util import check_ffmpeg_installed
from core.utils.gc_manager import get_gc_manager
TAG = __name__
logger = setup_logging()
@@ -64,10 +63,6 @@ async def main():
# 添加 stdin 监控任务
stdin_task = asyncio.create_task(monitor_stdin())
# 启动全局GC管理器(5分钟清理一次)
gc_manager = get_gc_manager(interval_seconds=300)
await gc_manager.start()
# 启动 WebSocket 服务器
ws_server = WebSocketServer(config)
ws_task = asyncio.create_task(ws_server.start())
@@ -127,9 +122,6 @@ async def main():
except asyncio.CancelledError:
print("任务被取消,清理资源中...")
finally:
# 停止全局GC管理器
await gc_manager.stop()
# 取消所有任务(关键修复点)
stdin_task.cancel()
ws_task.cancel()
+2 -9
View File
@@ -113,15 +113,6 @@ wakeup_words:
# MCP接入点地址,地址格式为:ws://你的mcp接入点ip或者域名:端口号/mcp/?token=你的token
# 详细教程 https://github.com/xinnan-tech/xiaozhi-esp32-server/blob/main/docs/mcp-endpoint-integration.md
mcp_endpoint: 你的接入点 websocket地址
# 上下文源配置
# 用于在系统提示词中注入动态数据,如健康数据、股票信息等
# 可以添加多个上下文源
context_providers:
- url: ""
headers:
Authorization: ""
# 插件的基础配置
plugins:
# 获取天气插件的配置,这里填写你的api_key
@@ -471,6 +462,7 @@ ASR:
domain: slm # 识别领域,iat:日常用语,medical:医疗,finance:金融等
language: zh_cn # 语言,zh_cn:中文,en_us:英文
accent: mandarin # 方言,mandarin:普通话
dwa: wpgs # 动态修正,wpgs:实时返回中间结果
# 调整音频处理参数以提高长语音识别质量
output_dir: tmp/
@@ -495,6 +487,7 @@ LLM:
temperature: 0.7 # 温度值
max_tokens: 500 # 最大生成token数
top_p: 1
top_k: 50
frequency_penalty: 0 # 频率惩罚
AliAppLLM:
# 定义LLM API类型
+6 -17
View File
@@ -32,16 +32,7 @@ def load_config():
custom_config = read_config(custom_config_path)
if custom_config.get("manager-api", {}).get("url"):
import asyncio
try:
loop = asyncio.get_running_loop()
# 如果已经在事件循环中,使用异步版本
config = asyncio.run_coroutine_threadsafe(
get_config_from_api_async(custom_config), loop
).result()
except RuntimeError:
# 如果不在事件循环中(启动时),创建新的事件循环
config = asyncio.run(get_config_from_api_async(custom_config))
config = get_config_from_api(custom_config)
else:
# 合并配置
config = merge_configs(default_config, custom_config)
@@ -53,13 +44,13 @@ def load_config():
return config
async def get_config_from_api_async(config):
"""从Java API获取配置(异步版本)"""
def get_config_from_api(config):
"""从Java API获取配置"""
# 初始化API客户端
init_service(config)
# 获取服务器配置
config_data = await get_server_config()
config_data = get_server_config()
if config_data is None:
raise Exception("Failed to fetch server config from API")
@@ -68,7 +59,6 @@ async def get_config_from_api_async(config):
"url": config["manager-api"].get("url", ""),
"secret": config["manager-api"].get("secret", ""),
}
auth_enabled = config_data.get("server", {}).get("auth", {}).get("enabled", False)
# server的配置以本地为准
if config.get("server"):
config_data["server"] = {
@@ -78,16 +68,15 @@ async def get_config_from_api_async(config):
"vision_explain": config["server"].get("vision_explain", ""),
"auth_key": config["server"].get("auth_key", ""),
}
config_data["server"]["auth"] = {"enabled": auth_enabled}
# 如果服务器没有prompt_template,则从本地配置读取
if not config_data.get("prompt_template"):
config_data["prompt_template"] = config.get("prompt_template")
return config_data
async def get_private_config_from_api(config, device_id, client_id):
def get_private_config_from_api(config, device_id, client_id):
"""从Java API获取私有配置"""
return await get_agent_models(device_id, client_id, config["selected_module"])
return get_agent_models(device_id, client_id, config["selected_module"])
def ensure_directories(config):
+1 -1
View File
@@ -5,7 +5,7 @@ from config.config_loader import load_config
from config.settings import check_config_file
from datetime import datetime
SERVER_VERSION = "0.8.10"
SERVER_VERSION = "0.8.8"
_logger_initialized = False
+36 -58
View File
@@ -1,4 +1,5 @@
import os
import time
import base64
from typing import Optional, Dict
@@ -19,7 +20,7 @@ class DeviceBindException(Exception):
class ManageApiClient:
_instance = None
_async_clients = {} # 为每个事件循环存储独立的客户端
_client = None
_secret = None
def __new__(cls, config):
@@ -31,7 +32,7 @@ class ManageApiClient:
@classmethod
def _init_client(cls, config):
"""初始化配置(延迟创建客户端)"""
"""初始化持久化连接池"""
cls.config = config.get("manager-api")
if not cls.config:
@@ -46,40 +47,23 @@ class ManageApiClient:
cls._secret = cls.config.get("secret")
cls.max_retries = cls.config.get("max_retries", 6) # 最大重试次数
cls.retry_delay = cls.config.get("retry_delay", 10) # 初始重试延迟(秒)
# 不在这里创建 AsyncClient,延迟到实际使用时创建
cls._async_clients = {}
# NOTE(goody): 2025/4/16 http相关资源统一管理,后续可以增加线程池或者超时
# 后续也可以统一配置apiToken之类的走通用的Auth
cls._client = httpx.Client(
base_url=cls.config.get("url"),
headers={
"User-Agent": f"PythonClient/2.0 (PID:{os.getpid()})",
"Accept": "application/json",
"Authorization": "Bearer " + cls._secret,
},
timeout=cls.config.get("timeout", 30), # 默认超时时间30秒
)
@classmethod
async def _ensure_async_client(cls):
"""确保异步客户端已创建(为每个事件循环创建独立的客户端)"""
import asyncio
try:
loop = asyncio.get_running_loop()
loop_id = id(loop)
# 为每个事件循环创建独立的客户端
if loop_id not in cls._async_clients:
cls._async_clients[loop_id] = httpx.AsyncClient(
base_url=cls.config.get("url"),
headers={
"User-Agent": f"PythonClient/2.0 (PID:{os.getpid()})",
"Accept": "application/json",
"Authorization": "Bearer " + cls._secret,
},
timeout=cls.config.get("timeout", 30),
)
return cls._async_clients[loop_id]
except RuntimeError:
# 如果没有运行中的事件循环,创建一个临时的
raise Exception("必须在异步上下文中调用")
@classmethod
async def _async_request(cls, method: str, endpoint: str, **kwargs) -> Dict:
"""发送单次异步HTTP请求并处理响应"""
# 确保客户端已创建
client = await cls._ensure_async_client()
def _request(cls, method: str, endpoint: str, **kwargs) -> Dict:
"""发送单次HTTP请求并处理响应"""
endpoint = endpoint.lstrip("/")
response = await client.request(method, endpoint, **kwargs)
response = cls._client.request(method, endpoint, **kwargs)
response.raise_for_status()
result = response.json()
@@ -112,23 +96,22 @@ class ManageApiClient:
return False
@classmethod
async def _execute_async_request(cls, method: str, endpoint: str, **kwargs) -> Dict:
"""带重试机制的异步请求执行器"""
import asyncio
def _execute_request(cls, method: str, endpoint: str, **kwargs) -> Dict:
"""带重试机制的请求执行器"""
retry_count = 0
while retry_count <= cls.max_retries:
try:
# 执行异步请求
return await cls._async_request(method, endpoint, **kwargs)
# 执行请求
return cls._request(method, endpoint, **kwargs)
except Exception as e:
# 判断是否应该重试
if retry_count < cls.max_retries and cls._should_retry(e):
retry_count += 1
print(
f"{method} {endpoint} 异步请求失败,将在 {cls.retry_delay:.1f} 秒后进行第 {retry_count} 次重试"
f"{method} {endpoint} 请求失败,将在 {cls.retry_delay:.1f} 秒后进行第 {retry_count} 次重试"
)
await asyncio.sleep(cls.retry_delay)
time.sleep(cls.retry_delay)
continue
else:
# 不重试,直接抛出异常
@@ -136,27 +119,22 @@ class ManageApiClient:
@classmethod
def safe_close(cls):
"""安全关闭所有异步连接池"""
import asyncio
for client in list(cls._async_clients.values()):
try:
asyncio.run(client.aclose())
except Exception:
pass
cls._async_clients.clear()
cls._instance = None
"""安全关闭连接池"""
if cls._client:
cls._client.close()
cls._instance = None
async def get_server_config() -> Optional[Dict]:
def get_server_config() -> Optional[Dict]:
"""获取服务器基础配置"""
return await ManageApiClient._instance._execute_async_request("POST", "/config/server-base")
return ManageApiClient._instance._execute_request("POST", "/config/server-base")
async def get_agent_models(
def get_agent_models(
mac_address: str, client_id: str, selected_module: Dict
) -> Optional[Dict]:
"""获取代理模型配置"""
return await ManageApiClient._instance._execute_async_request(
return ManageApiClient._instance._execute_request(
"POST",
"/config/agent-models",
json={
@@ -167,9 +145,9 @@ async def get_agent_models(
)
async def save_mem_local_short(mac_address: str, short_momery: str) -> Optional[Dict]:
def save_mem_local_short(mac_address: str, short_momery: str) -> Optional[Dict]:
try:
return await ManageApiClient._instance._execute_async_request(
return ManageApiClient._instance._execute_request(
"PUT",
f"/agent/saveMemory/" + mac_address,
json={
@@ -181,14 +159,14 @@ async def save_mem_local_short(mac_address: str, short_momery: str) -> Optional[
return None
async def report(
def report(
mac_address: str, session_id: str, chat_type: int, content: str, audio, report_time
) -> Optional[Dict]:
"""异步聊天记录上报"""
"""带熔断的业务方法示例"""
if not content or not ManageApiClient._instance:
return None
try:
return await ManageApiClient._instance._execute_async_request(
return ManageApiClient._instance._execute_request(
"POST",
f"/agent/chat-history/report",
json={
@@ -96,7 +96,7 @@ class VisionHandler:
current_config = copy.deepcopy(self.config)
read_config_from_api = current_config.get("read_config_from_api", False)
if read_config_from_api:
current_config = await get_private_config_from_api(
current_config = get_private_config_from_api(
current_config,
device_id,
client_id,
+133 -269
View File
@@ -68,12 +68,8 @@ class ConnectionHandler:
self.logger = setup_logging()
self.server = server # 保存server实例的引用
self.need_bind = False # 是否需要绑定设备
self.bind_completed_event = asyncio.Event()
self.bind_code = None # 绑定设备的验证码
self.last_bind_prompt_time = 0 # 上次播放绑定提示的时间戳(秒)
self.bind_prompt_interval = 60 # 绑定提示播放间隔(秒)
self.need_bind = False
self.bind_code = None
self.read_config_from_api = self.config.get("read_config_from_api", False)
self.websocket = None
@@ -92,7 +88,7 @@ class ConnectionHandler:
self.client_listen_mode = "auto"
# 线程任务相关
self.loop = None # 在 handle_connection 中获取运行中的事件循环
self.loop = asyncio.get_event_loop()
self.stop_event = threading.Event()
self.executor = ThreadPoolExecutor(max_workers=5)
@@ -120,7 +116,6 @@ class ConnectionHandler:
self.client_audio_buffer = bytearray()
self.client_have_voice = False
self.client_voice_window = deque(maxlen=5)
self.first_activity_time = 0.0 # 记录首次活动的时间(毫秒)
self.last_activity_time = 0.0 # 统一的活动时间戳(毫秒)
self.client_voice_stop = False
self.last_is_voice = False
@@ -163,13 +158,10 @@ class ConnectionHandler:
self.conn_from_mqtt_gateway = False
# 初始化提示词管理器
self.prompt_manager = PromptManager(self.config, self.logger)
self.prompt_manager = PromptManager(config, self.logger)
async def handle_connection(self, ws):
try:
# 获取运行中的事件循环(必须在异步上下文中)
self.loop = asyncio.get_running_loop()
# 获取并验证headers
self.headers = dict(ws.request.headers)
real_ip = self.headers.get("x-real-ip") or self.headers.get(
@@ -195,7 +187,6 @@ class ConnectionHandler:
self.logger.bind(tag=TAG).info("连接来自:MQTT网关")
# 初始化活动时间戳
self.first_activity_time = time.time() * 1000
self.last_activity_time = time.time() * 1000
# 启动超时检查任务
@@ -204,8 +195,10 @@ class ConnectionHandler:
self.welcome_msg = self.config["xiaozhi"]
self.welcome_msg["session_id"] = self.session_id
# 在后台初始化配置和组件(完全不阻塞主循环)
asyncio.create_task(self._background_initialize())
# 获取差异化配置
self._initialize_private_config()
# 异步初始化
self.executor.submit(self._initialize_components)
try:
async for message in self.websocket:
@@ -267,37 +260,8 @@ class ConnectionHandler:
f"保存记忆后关闭连接失败: {close_error}"
)
async def _discard_message_with_bind_prompt(self):
"""丢弃消息并检查是否需要播放绑定提示"""
current_time = time.time()
# 检查是否需要播放绑定提示
if current_time - self.last_bind_prompt_time >= self.bind_prompt_interval:
self.last_bind_prompt_time = current_time
# 复用现有的绑定提示逻辑
from core.handle.receiveAudioHandle import check_bind_device
asyncio.create_task(check_bind_device(self))
async def _route_message(self, message):
"""消息路由"""
# 检查是否已经获取到真实的绑定状态
if not self.bind_completed_event.is_set():
# 还没有获取到真实状态,等待直到获取到真实状态或超时
try:
await asyncio.wait_for(self.bind_completed_event.wait(), timeout=1)
except asyncio.TimeoutError:
# 超时仍未获取到真实状态,丢弃消息
await self._discard_message_with_bind_prompt()
return
# 已经获取到真实状态,检查是否需要绑定
if self.need_bind:
# 需要绑定,丢弃消息
await self._discard_message_with_bind_prompt()
return
# 不需要绑定,继续处理消息
if isinstance(message, str):
await handleTextMessage(self, message)
elif isinstance(message, bytes):
@@ -427,14 +391,6 @@ class ConnectionHandler:
def _initialize_components(self):
try:
if self.tts is None:
self.tts = self._initialize_tts()
# 打开语音合成通道
asyncio.run_coroutine_threadsafe(
self.tts.open_audio_channels(self), self.loop
)
if self.need_bind:
return
self.selected_module_str = build_module_string(
self.config.get("selected_module", {})
)
@@ -458,10 +414,17 @@ class ConnectionHandler:
# 初始化声纹识别
self._initialize_voiceprint()
# 打开语音识别通道
asyncio.run_coroutine_threadsafe(
self.asr.open_audio_channels(self), self.loop
)
if self.tts is None:
self.tts = self._initialize_tts()
# 打开语音合成通道
asyncio.run_coroutine_threadsafe(
self.tts.open_audio_channels(self), self.loop
)
"""加载记忆"""
self._initialize_memory()
@@ -476,7 +439,6 @@ class ConnectionHandler:
self.logger.bind(tag=TAG).error(f"实例化组件失败: {e}")
def _init_prompt_enhancement(self):
# 更新上下文信息
self.prompt_manager.update_context_info(self, self.client_ip)
enhanced_prompt = self.prompt_manager.build_enhanced_prompt(
@@ -512,11 +474,7 @@ class ConnectionHandler:
def _initialize_asr(self):
"""初始化ASR"""
if (
self._asr is not None
and hasattr(self._asr, "interface_type")
and self._asr.interface_type == InterfaceType.LOCAL
):
if self._asr.interface_type == InterfaceType.LOCAL:
# 如果公共ASR是本地服务,则直接返回
# 因为本地一个实例ASR,可以被多个连接共享
asr = self._asr
@@ -543,48 +501,32 @@ class ConnectionHandler:
except Exception as e:
self.logger.bind(tag=TAG).warning(f"声纹识别初始化失败: {str(e)}")
async def _background_initialize(self):
"""在后台初始化配置和组件(完全不阻塞主循环)"""
try:
# 异步获取差异化配置
await self._initialize_private_config_async()
# 在线程池中初始化组件
self.executor.submit(self._initialize_components)
except Exception as e:
self.logger.bind(tag=TAG).error(f"后台初始化失败: {e}")
async def _initialize_private_config_async(self):
"""从接口异步获取差异化配置(异步版本,不阻塞主循环)"""
def _initialize_private_config(self):
"""如果是从配置文件获取,则进行二次实例化"""
if not self.read_config_from_api:
self.need_bind = False
self.bind_completed_event.set()
return
"""从接口获取差异化的配置进行二次实例化,非全量重新实例化"""
try:
begin_time = time.time()
private_config = await get_private_config_from_api(
private_config = get_private_config_from_api(
self.config,
self.headers.get("device-id"),
self.headers.get("client-id", self.headers.get("device-id")),
)
private_config["delete_audio"] = bool(self.config.get("delete_audio", True))
self.logger.bind(tag=TAG).info(
f"{time.time() - begin_time} 秒,异步获取差异化配置成功: {json.dumps(filter_sensitive_info(private_config), ensure_ascii=False)}"
f"{time.time() - begin_time} 秒,获取差异化配置成功: {json.dumps(filter_sensitive_info(private_config), ensure_ascii=False)}"
)
self.need_bind = False
self.bind_completed_event.set()
except DeviceNotFoundException as e:
self.need_bind = True
self.bind_completed_event.set() # 状态已确定,设置事件
private_config = {}
except DeviceBindException as e:
self.need_bind = True
self.bind_code = e.bind_code
self.bind_completed_event.set() # 状态已确定,设置事件
private_config = {}
except Exception as e:
self.need_bind = True
self.bind_completed_event.set() # 状态已确定,设置事件
self.logger.bind(tag=TAG).error(f"异步获取差异化配置失败: {e}")
self.logger.bind(tag=TAG).error(f"获取差异化配置失败: {e}")
private_config = {}
init_llm, init_tts, init_memory, init_intent = (
@@ -657,14 +599,8 @@ class ConnectionHandler:
self.chat_history_conf = int(private_config["chat_history_conf"])
if private_config.get("mcp_endpoint", None) is not None:
self.config["mcp_endpoint"] = private_config["mcp_endpoint"]
if private_config.get("context_providers", None) is not None:
self.config["context_providers"] = private_config["context_providers"]
# 使用 run_in_executor 在线程池中执行 initialize_modules,避免阻塞主循环
try:
modules = await self.loop.run_in_executor(
None, # 使用默认线程池
initialize_modules,
modules = initialize_modules(
self.logger,
private_config,
init_vad,
@@ -787,47 +723,25 @@ class ConnectionHandler:
self.dialogue.update_system_message(self.prompt)
def chat(self, query, depth=0):
if query is not None:
self.logger.bind(tag=TAG).info(f"大模型收到用户消息: {query}")
self.logger.bind(tag=TAG).info(f"大模型收到用户消息: {query}")
self.llm_finish_task = False
# 为最顶层时新建会话ID和发送FIRST请求
# depth!=0目前只能表示是非用户发起的请求,与对话逻辑无关,这里所有tts对话都应该发送FIRST和LAST,没有例外
if depth == 0:
self.llm_finish_task = False
self.sentence_id = str(uuid.uuid4().hex)
self.dialogue.put(Message(role="user", content=query))
self.tts.tts_text_queue.put(
TTSMessageDTO(
sentence_id=self.sentence_id,
sentence_type=SentenceType.FIRST,
content_type=ContentType.ACTION,
)
)
# 设置最大递归深度,避免无限循环,可根据实际需求调整
MAX_DEPTH = 5
force_final_answer = False # 标记是否强制最终回答
if depth >= MAX_DEPTH:
self.logger.bind(tag=TAG).debug(
f"已达到最大工具调用深度 {MAX_DEPTH},将强制基于现有信息回答"
)
force_final_answer = True
# 添加系统指令,要求 LLM 基于现有信息回答
self.dialogue.put(
Message(
role="user",
content="[系统提示] 已达到最大工具调用次数限制,请你基于目前已经获取的所有信息,直接给出最终答案。不要再尝试调用任何工具。",
)
# 发送开始标记
self.sentence_id = str(uuid.uuid4().hex)
self.tts.tts_text_queue.put(
TTSMessageDTO(
sentence_id=self.sentence_id,
sentence_type=SentenceType.FIRST,
content_type=ContentType.ACTION,
)
)
# Define intent functions
functions = None
# 达到最大深度时,禁用工具调用,强制 LLM 直接回答
if (
self.intent_type == "function_call"
and hasattr(self, "func_handler")
and not force_final_answer
):
if self.intent_type == "function_call" and hasattr(self, "func_handler"):
functions = self.func_handler.get_functions()
response_message = []
@@ -862,8 +776,9 @@ class ConnectionHandler:
# 处理流式响应
tool_call_flag = False
# 支持多个并行工具调用 - 使用列表存储
tool_calls_list = [] # 格式: [{"id": "", "name": "", "arguments": ""}]
function_name = None
function_id = None
function_arguments = ""
content_arguments = ""
self.client_abort = False
emotion_flag = True
@@ -884,7 +799,12 @@ class ConnectionHandler:
if tools_call is not None and len(tools_call) > 0:
tool_call_flag = True
self._merge_tool_calls(tool_calls_list, tools_call)
if tools_call[0].id is not None and tools_call[0].id != "":
function_id = tools_call[0].id
if tools_call[0].function.name is not None and tools_call[0].function.name != "":
function_name = tools_call[0].function.name
if tools_call[0].function.arguments is not None:
function_arguments += tools_call[0].function.arguments
else:
content = response
@@ -907,25 +827,27 @@ class ConnectionHandler:
content_detail=content,
)
)
# 发送结束标记
self.tts.tts_text_queue.put(
TTSMessageDTO(
sentence_id=self.sentence_id,
sentence_type=SentenceType.LAST,
content_type=ContentType.ACTION,
)
)
# 处理function call
if tool_call_flag:
bHasError = False
# 处理基于文本的工具调用格式
if len(tool_calls_list) == 0 and content_arguments:
if function_id is None:
a = extract_json_from_string(content_arguments)
if a is not None:
try:
content_arguments_json = json.loads(a)
tool_calls_list.append(
{
"id": str(uuid.uuid4().hex),
"name": content_arguments_json["name"],
"arguments": json.dumps(
content_arguments_json["arguments"],
ensure_ascii=False,
),
}
function_name = content_arguments_json["name"]
function_arguments = json.dumps(
content_arguments_json["arguments"], ensure_ascii=False
)
function_id = str(uuid.uuid4().hex)
except Exception as e:
bHasError = True
response_message.append(a)
@@ -936,120 +858,95 @@ class ConnectionHandler:
self.logger.bind(tag=TAG).error(
f"function call error: {content_arguments}"
)
if not bHasError and len(tool_calls_list) > 0:
if not bHasError:
# 如需要大模型先处理一轮,添加相关处理后的日志情况
if len(response_message) > 0:
text_buff = "".join(response_message)
self.tts_MessageText = text_buff
self.dialogue.put(Message(role="assistant", content=text_buff))
response_message.clear()
self.logger.bind(tag=TAG).debug(
f"检测到 {len(tool_calls_list)} 个工具调用"
f"function_name={function_name}, function_id={function_id}, function_arguments={function_arguments}"
)
function_call_data = {
"name": function_name,
"id": function_id,
"arguments": function_arguments,
}
# 收集所有工具调用的 Future
futures_with_data = []
for tool_call_data in tool_calls_list:
self.logger.bind(tag=TAG).debug(
f"function_name={tool_call_data['name']}, function_id={tool_call_data['id']}, function_arguments={tool_call_data['arguments']}"
)
future = asyncio.run_coroutine_threadsafe(
self.func_handler.handle_llm_function_call(
self, tool_call_data
),
self.loop,
)
futures_with_data.append((future, tool_call_data))
# 等待协程结束(实际等待时长为最慢的那个)
tool_results = []
for future, tool_call_data in futures_with_data:
result = future.result()
tool_results.append((result, tool_call_data))
# 统一处理所有工具调用结果
if tool_results:
self._handle_function_result(tool_results, depth=depth)
# 使用统一工具处理器处理所有工具调用
result = asyncio.run_coroutine_threadsafe(
self.func_handler.handle_llm_function_call(
self, function_call_data
),
self.loop,
).result()
self._handle_function_result(result, function_call_data, depth=depth)
# 存储对话内容
if len(response_message) > 0:
text_buff = "".join(response_message)
self.tts_MessageText = text_buff
self.dialogue.put(Message(role="assistant", content=text_buff))
if depth == 0:
self.tts.tts_text_queue.put(
TTSMessageDTO(
sentence_id=self.sentence_id,
sentence_type=SentenceType.LAST,
content_type=ContentType.ACTION,
)
)
self.llm_finish_task = True
# 使用lambda延迟计算,只有在DEBUG级别时才执行get_llm_dialogue()
self.logger.bind(tag=TAG).debug(
lambda: json.dumps(
self.dialogue.get_llm_dialogue(), indent=4, ensure_ascii=False
)
self.llm_finish_task = True
# 使用lambda延迟计算,只有在DEBUG级别时才执行get_llm_dialogue()
self.logger.bind(tag=TAG).debug(
lambda: json.dumps(
self.dialogue.get_llm_dialogue(), indent=4, ensure_ascii=False
)
)
return True
def _handle_function_result(self, tool_results, depth):
need_llm_tools = []
for result, tool_call_data in tool_results:
if result.action in [
Action.RESPONSE,
Action.NOTFOUND,
Action.ERROR,
]: # 直接回复前端
text = result.response if result.response else result.result
self.tts.tts_one_sentence(self, ContentType.TEXT, content_detail=text)
self.dialogue.put(Message(role="assistant", content=text))
elif result.action == Action.REQLLM:
# 收集需要 LLM 处理的工具
need_llm_tools.append((result, tool_call_data))
else:
pass
if need_llm_tools:
all_tool_calls = [
{
"id": tool_call_data["id"],
"function": {
"arguments": (
"{}"
if tool_call_data["arguments"] == ""
else tool_call_data["arguments"]
),
"name": tool_call_data["name"],
},
"type": "function",
"index": idx,
}
for idx, (_, tool_call_data) in enumerate(need_llm_tools)
]
self.dialogue.put(Message(role="assistant", tool_calls=all_tool_calls))
for result, tool_call_data in need_llm_tools:
text = result.result
if text is not None and len(text) > 0:
self.dialogue.put(
Message(
role="tool",
tool_call_id=(
str(uuid.uuid4())
if tool_call_data["id"] is None
else tool_call_data["id"]
),
content=text,
)
def _handle_function_result(self, result, function_call_data, depth):
if result.action == Action.RESPONSE: # 直接回复前端
text = result.response
self.tts.tts_one_sentence(self, ContentType.TEXT, content_detail=text)
self.dialogue.put(Message(role="assistant", content=text))
elif result.action == Action.REQLLM: # 调用函数后再请求llm生成回复
text = result.result
if text is not None and len(text) > 0:
function_id = function_call_data["id"]
function_name = function_call_data["name"]
function_arguments = function_call_data["arguments"]
self.dialogue.put(
Message(
role="assistant",
tool_calls=[
{
"id": function_id,
"function": {
"arguments": (
"{}"
if function_arguments == ""
else function_arguments
),
"name": function_name,
},
"type": "function",
"index": 0,
}
],
)
)
self.chat(None, depth=depth + 1)
self.dialogue.put(
Message(
role="tool",
tool_call_id=(
str(uuid.uuid4()) if function_id is None else function_id
),
content=text,
)
)
self.chat(text, depth=depth + 1)
elif result.action == Action.NOTFOUND or result.action == Action.ERROR:
text = result.response if result.response else result.result
self.tts.tts_one_sentence(self, ContentType.TEXT, content_detail=text)
self.dialogue.put(Message(role="assistant", content=text))
else:
pass
def _report_worker(self):
"""聊天记录上报工作线程"""
@@ -1077,8 +974,8 @@ class ConnectionHandler:
def _process_report(self, type, text, audio_data, report_time):
"""处理上报任务"""
try:
# 执行异步上报(在事件循环中运行
asyncio.run(report(self, type, text, audio_data, report_time))
# 执行上报(传入二进制数据
report(self, type, text, audio_data, report_time)
except Exception as e:
self.logger.bind(tag=TAG).error(f"上报处理异常: {e}")
finally:
@@ -1169,6 +1066,7 @@ class ConnectionHandler:
f"关闭线程池时出错: {executor_error}"
)
self.executor = None
self.logger.bind(tag=TAG).info("连接资源已释放")
except Exception as e:
self.logger.bind(tag=TAG).error(f"关闭连接时出错: {e}")
@@ -1198,11 +1096,6 @@ class ConnectionHandler:
except queue.Empty:
break
# 重置音频流控器(取消后台任务并清空队列)
if hasattr(self, "audio_rate_controller") and self.audio_rate_controller:
self.audio_rate_controller.reset()
self.logger.bind(tag=TAG).debug("已重置音频流控器")
self.logger.bind(tag=TAG).debug(
f"清理结束: TTS队列大小={self.tts.tts_text_queue.qsize()}, 音频队列大小={self.tts.tts_audio_queue.qsize()}"
)
@@ -1228,14 +1121,13 @@ class ConnectionHandler:
"""检查连接超时"""
try:
while not self.stop_event.is_set():
last_activity_time = self.last_activity_time
if self.need_bind:
last_activity_time = self.first_activity_time
# 检查是否超时(只有在时间戳已初始化的情况下)
if last_activity_time > 0.0:
if self.last_activity_time > 0.0:
current_time = time.time() * 1000
if current_time - last_activity_time > self.timeout_seconds * 1000:
if (
current_time - self.last_activity_time
> self.timeout_seconds * 1000
):
if not self.stop_event.is_set():
self.logger.bind(tag=TAG).info("连接超时,准备关闭")
# 设置停止事件,防止重复处理
@@ -1254,31 +1146,3 @@ class ConnectionHandler:
self.logger.bind(tag=TAG).error(f"超时检查任务出错: {e}")
finally:
self.logger.bind(tag=TAG).info("超时检查任务已退出")
def _merge_tool_calls(self, tool_calls_list, tools_call):
"""合并工具调用列表
Args:
tool_calls_list: 已收集的工具调用列表
tools_call: 新的工具调用
"""
for tool_call in tools_call:
tool_index = getattr(tool_call, "index", None)
if tool_index is None:
if tool_call.function.name:
# 有 function_name,说明是新的工具调用
tool_index = len(tool_calls_list)
else:
tool_index = len(tool_calls_list) - 1 if tool_calls_list else 0
# 确保列表有足够的位置
if tool_index >= len(tool_calls_list):
tool_calls_list.append({"id": "", "name": "", "arguments": ""})
# 更新工具调用信息
if tool_call.id:
tool_calls_list[tool_index]["id"] = tool_call.id
if tool_call.function.name:
tool_calls_list[tool_index]["name"] = tool_call.function.name
if tool_call.function.arguments:
tool_calls_list[tool_index]["arguments"] += tool_call.function.arguments
@@ -101,7 +101,7 @@ async def checkWakeupWords(conn, text):
}
# 获取音频数据
opus_packets = await audio_to_data(response.get("file_path"), use_cache=False)
opus_packets = audio_to_data(response.get("file_path"))
# 播放唤醒词回复
conn.client_abort = False
@@ -123,7 +123,7 @@ async def max_out_size(conn):
text = "不好意思,我现在有点事情要忙,明天这个时候我们再聊,约好了哦!明天不见不散,拜拜!"
await send_stt_message(conn, text)
file_path = "config/assets/max_output_size.wav"
opus_packets = await audio_to_data(file_path)
opus_packets = audio_to_data(file_path)
conn.tts.tts_audio_queue.put((SentenceType.LAST, opus_packets, text))
conn.close_after_chat = True
@@ -142,7 +142,7 @@ async def check_bind_device(conn):
# 播放提示音
music_path = "config/assets/bind_code.wav"
opus_packets = await audio_to_data(music_path)
opus_packets = audio_to_data(music_path)
conn.tts.tts_audio_queue.put((SentenceType.FIRST, opus_packets, text))
# 逐个播放数字
@@ -150,7 +150,7 @@ async def check_bind_device(conn):
try:
digit = conn.bind_code[i]
num_path = f"config/assets/bind_code/{digit}.wav"
num_packets = await audio_to_data(num_path)
num_packets = audio_to_data(num_path)
conn.tts.tts_audio_queue.put((SentenceType.MIDDLE, num_packets, None))
except Exception as e:
conn.logger.bind(tag=TAG).error(f"播放数字音频失败: {e}")
@@ -162,5 +162,5 @@ async def check_bind_device(conn):
text = f"没有找到该设备的版本信息,请正确配置 OTA地址,然后重新编译固件。"
await send_stt_message(conn, text)
music_path = "config/assets/bind_not_found.wav"
opus_packets = await audio_to_data(music_path)
opus_packets = audio_to_data(music_path)
conn.tts.tts_audio_queue.put((SentenceType.LAST, opus_packets, text))
+34 -41
View File
@@ -10,6 +10,7 @@ TTS上报功能已集成到ConnectionHandler类中。
"""
import time
import opuslib_next
from config.manage_api_client import report as manage_report
@@ -17,7 +18,7 @@ from config.manage_api_client import report as manage_report
TAG = __name__
async def report(conn, type, text, opus_data, report_time):
def report(conn, type, text, opus_data, report_time):
"""执行聊天记录上报操作
Args:
@@ -32,8 +33,8 @@ async def report(conn, type, text, opus_data, report_time):
audio_data = opus_to_wav(conn, opus_data)
else:
audio_data = None
# 执行异步上报
await manage_report(
# 执行上报
manage_report(
mac_address=conn.device_id,
session_id=conn.session_id,
chat_type=type,
@@ -55,49 +56,41 @@ def opus_to_wav(conn, opus_data):
Returns:
bytes: WAV格式的音频数据
"""
decoder = None
try:
decoder = opuslib_next.Decoder(16000, 1) # 16kHz, 单声道
pcm_data = []
decoder = opuslib_next.Decoder(16000, 1) # 16kHz, 单声道
pcm_data = []
for opus_packet in opus_data:
try:
pcm_frame = decoder.decode(opus_packet, 960) # 960 samples = 60ms
pcm_data.append(pcm_frame)
except opuslib_next.OpusError as e:
conn.logger.bind(tag=TAG).error(f"Opus解码错误: {e}", exc_info=True)
for opus_packet in opus_data:
try:
pcm_frame = decoder.decode(opus_packet, 960) # 960 samples = 60ms
pcm_data.append(pcm_frame)
except opuslib_next.OpusError as e:
conn.logger.bind(tag=TAG).error(f"Opus解码错误: {e}", exc_info=True)
if not pcm_data:
raise ValueError("没有有效的PCM数据")
if not pcm_data:
raise ValueError("没有有效的PCM数据")
# 创建WAV文件头
pcm_data_bytes = b"".join(pcm_data)
num_samples = len(pcm_data_bytes) // 2 # 16-bit samples
# 创建WAV文件头
pcm_data_bytes = b"".join(pcm_data)
num_samples = len(pcm_data_bytes) // 2 # 16-bit samples
# WAV文件头
wav_header = bytearray()
wav_header.extend(b"RIFF") # ChunkID
wav_header.extend((36 + len(pcm_data_bytes)).to_bytes(4, "little")) # ChunkSize
wav_header.extend(b"WAVE") # Format
wav_header.extend(b"fmt ") # Subchunk1ID
wav_header.extend((16).to_bytes(4, "little")) # Subchunk1Size
wav_header.extend((1).to_bytes(2, "little")) # AudioFormat (PCM)
wav_header.extend((1).to_bytes(2, "little")) # NumChannels
wav_header.extend((16000).to_bytes(4, "little")) # SampleRate
wav_header.extend((32000).to_bytes(4, "little")) # ByteRate
wav_header.extend((2).to_bytes(2, "little")) # BlockAlign
wav_header.extend((16).to_bytes(2, "little")) # BitsPerSample
wav_header.extend(b"data") # Subchunk2ID
wav_header.extend(len(pcm_data_bytes).to_bytes(4, "little")) # Subchunk2Size
# WAV文件头
wav_header = bytearray()
wav_header.extend(b"RIFF") # ChunkID
wav_header.extend((36 + len(pcm_data_bytes)).to_bytes(4, "little")) # ChunkSize
wav_header.extend(b"WAVE") # Format
wav_header.extend(b"fmt ") # Subchunk1ID
wav_header.extend((16).to_bytes(4, "little")) # Subchunk1Size
wav_header.extend((1).to_bytes(2, "little")) # AudioFormat (PCM)
wav_header.extend((1).to_bytes(2, "little")) # NumChannels
wav_header.extend((16000).to_bytes(4, "little")) # SampleRate
wav_header.extend((32000).to_bytes(4, "little")) # ByteRate
wav_header.extend((2).to_bytes(2, "little")) # BlockAlign
wav_header.extend((16).to_bytes(2, "little")) # BitsPerSample
wav_header.extend(b"data") # Subchunk2ID
wav_header.extend(len(pcm_data_bytes).to_bytes(4, "little")) # Subchunk2Size
# 返回完整的WAV数据
return bytes(wav_header) + pcm_data_bytes
finally:
if decoder is not None:
try:
del decoder
except Exception as e:
conn.logger.bind(tag=TAG).debug(f"释放decoder资源时出错: {e}")
# 返回完整的WAV数据
return bytes(wav_header) + pcm_data_bytes
def enqueue_tts_report(conn, text, opus_data):
+124 -151
View File
@@ -4,7 +4,6 @@ import asyncio
from core.utils import textUtils
from core.utils.util import audio_to_data
from core.providers.tts.dto.dto import SentenceType
from core.utils.audioRateController import AudioRateController
TAG = __name__
@@ -16,19 +15,7 @@ async def sendAudioMessage(conn, sentenceType, audios, text):
await send_tts_message(conn, "start", None)
if sentenceType == SentenceType.FIRST:
# 同一句子的后续消息加入流控队列,其他情况立即发送
if (
hasattr(conn, "audio_rate_controller")
and conn.audio_rate_controller
and getattr(conn, "audio_flow_control", {}).get("sentence_id")
== conn.sentence_id
):
conn.audio_rate_controller.add_message(
lambda: send_tts_message(conn, "sentence_start", text)
)
else:
# 新句子或流控器未初始化,立即发送
await send_tts_message(conn, "sentence_start", text)
await send_tts_message(conn, "sentence_start", text)
await sendAudio(conn, audios)
# 发送句子开始消息
@@ -36,27 +23,36 @@ async def sendAudioMessage(conn, sentenceType, audios, text):
conn.logger.bind(tag=TAG).info(f"发送音频消息: {sentenceType}, {text}")
# 发送结束消息(如果是最后一个文本)
if sentenceType == SentenceType.LAST:
if conn.llm_finish_task and sentenceType == SentenceType.LAST:
await send_tts_message(conn, "stop", None)
conn.client_is_speaking = False
if conn.close_after_chat:
await conn.close()
async def _wait_for_audio_completion(conn):
def calculate_timestamp_and_sequence(conn, start_time, packet_index, frame_duration=60):
"""
等待音频队列清空
计算音频数据包的时间戳和序列号
Args:
conn: 连接对象
start_time: 起始时间性能计数器值
packet_index: 数据包索引
frame_duration: 帧时长毫秒匹配 Opus 编码
Returns:
tuple: (timestamp, sequence)
"""
if hasattr(conn, "audio_rate_controller") and conn.audio_rate_controller:
rate_controller = conn.audio_rate_controller
conn.logger.bind(tag=TAG).debug(
f"等待音频发送完成,队列中还有 {len(rate_controller.queue)} 个包"
)
await rate_controller.queue_empty_event.wait()
conn.logger.bind(tag=TAG).debug("音频发送完成")
# 计算时间戳(使用播放位置计算)
timestamp = int((start_time + packet_index * frame_duration / 1000) * 1000) % (
2**32
)
# 计算序列号
if hasattr(conn, "audio_flow_control"):
sequence = conn.audio_flow_control["sequence"]
else:
sequence = packet_index # 如果没有流控状态,直接使用索引
return timestamp, sequence
async def _send_to_mqtt_gateway(conn, opus_packet, timestamp, sequence):
@@ -81,153 +77,131 @@ async def _send_to_mqtt_gateway(conn, opus_packet, timestamp, sequence):
await conn.websocket.send(complete_packet)
# 播放音频
async def sendAudio(conn, audios, frame_duration=60):
"""
发送音频包使用 AudioRateController 进行精确的流量控制
发送单个opus包支持流控
Args:
conn: 连接对象
audios: 单个opus包(bytes) opus包列表
frame_duration: 帧时长毫秒默认60ms
opus_packet: 单个opus数据
pre_buffer: 快速发送音频
frame_duration: 帧时长毫秒匹配 Opus 编码
"""
if audios is None or len(audios) == 0:
return
# 获取发送延迟配置
send_delay = conn.config.get("tts_audio_send_delay", -1) / 1000.0
is_single_packet = isinstance(audios, bytes)
# 初始化或获取 RateController
rate_controller, flow_control = _get_or_create_rate_controller(
conn, frame_duration, is_single_packet
)
if isinstance(audios, bytes):
# 重置流控状态,第一次读取和会话发生转变时
if not hasattr(conn, "audio_flow_control") or conn.audio_flow_control.get("sentence_id") != conn.sentence_id:
conn.audio_flow_control = {
"last_send_time": 0,
"packet_count": 0,
"start_time": time.perf_counter(),
"sequence": 0, # 添加序列号
"sentence_id": conn.sentence_id,
}
# 统一转换为列表处理
audio_list = [audios] if is_single_packet else audios
# 发送音频包
await _send_audio_with_rate_control(
conn, audio_list, rate_controller, flow_control, send_delay
)
def _get_or_create_rate_controller(conn, frame_duration, is_single_packet):
"""
获取或创建 RateController flow_control
Args:
conn: 连接对象
frame_duration: 帧时长
is_single_packet: 是否单包模式True: TTS流式单包, False: 批量包
Returns:
(rate_controller, flow_control)
"""
# 判断是否需要重置:单包模式且 sentence_id 变化,或者控制器不存在
need_reset = (
is_single_packet
and getattr(conn, "audio_flow_control", {}).get("sentence_id")
!= conn.sentence_id
) or not hasattr(conn, "audio_rate_controller")
if need_reset:
# 创建或获取 rate_controller
if not hasattr(conn, "audio_rate_controller"):
conn.audio_rate_controller = AudioRateController(frame_duration)
else:
conn.audio_rate_controller.reset()
# 初始化 flow_control
conn.audio_flow_control = {
"packet_count": 0,
"sequence": 0,
"sentence_id": conn.sentence_id,
}
# 启动后台发送循环
_start_background_sender(
conn, conn.audio_rate_controller, conn.audio_flow_control
)
return conn.audio_rate_controller, conn.audio_flow_control
def _start_background_sender(conn, rate_controller, flow_control):
"""
启动后台发送循环任务
Args:
conn: 连接对象
rate_controller: 速率控制器
flow_control: 流控状态
"""
async def send_callback(packet):
# 检查是否应该中止
if conn.client_abort:
raise asyncio.CancelledError("客户端已中止")
conn.last_activity_time = time.time() * 1000
await _do_send_audio(conn, packet, flow_control)
conn.client_is_speaking = True
# 使用 start_sending 启动后台循环
rate_controller.start_sending(send_callback)
async def _send_audio_with_rate_control(
conn, audio_list, rate_controller, flow_control, send_delay
):
"""
使用 rate_controller 发送音频包
Args:
conn: 连接对象
audio_list: 音频包列表
rate_controller: 速率控制器
flow_control: 流控状态
send_delay: 固定延迟-1表示使用动态流控
"""
pre_buffer_count = 5
for packet in audio_list:
if conn.client_abort:
return
conn.last_activity_time = time.time() * 1000
# 预缓冲:前5个包直接发送
# 预缓冲:前5个包直接发送,不做延迟
pre_buffer_count = 5
flow_control = conn.audio_flow_control
current_time = time.perf_counter()
if flow_control["packet_count"] < pre_buffer_count:
await _do_send_audio(conn, packet, flow_control)
conn.client_is_speaking = True
# 预缓冲阶段,直接发送不延迟
pass
elif send_delay > 0:
# 固定延迟模式
# 使用固定延迟
await asyncio.sleep(send_delay)
await _do_send_audio(conn, packet, flow_control)
conn.client_is_speaking = True
else:
# 动态流控模式:仅添加到队列,由后台循环负责发送
rate_controller.add_audio(packet)
effective_packet = flow_control["packet_count"] - pre_buffer_count
expected_time = flow_control["start_time"] + (
effective_packet * frame_duration / 1000
)
delay = expected_time - current_time
if delay > 0:
await asyncio.sleep(delay)
else:
# 纠正误差
flow_control["start_time"] += abs(delay)
if conn.conn_from_mqtt_gateway:
# 计算时间戳和序列号
timestamp, sequence = calculate_timestamp_and_sequence(
conn,
flow_control["start_time"],
flow_control["packet_count"],
frame_duration,
)
# 调用通用函数发送带头部的数据包
await _send_to_mqtt_gateway(conn, audios, timestamp, sequence)
else:
# 直接发送opus数据包,不添加头部
await conn.websocket.send(audios)
async def _do_send_audio(conn, opus_packet, flow_control):
"""
执行实际的音频发送
"""
packet_index = flow_control.get("packet_count", 0)
sequence = flow_control.get("sequence", 0)
if conn.conn_from_mqtt_gateway:
# 计算时间戳(基于播放位置)
start_time = time.time()
timestamp = int(start_time * 1000) % (2**32)
await _send_to_mqtt_gateway(conn, opus_packet, timestamp, sequence)
# 更新流控状态
flow_control["packet_count"] += 1
flow_control["sequence"] += 1
flow_control["last_send_time"] = time.perf_counter()
else:
# 直接发送opus数据包
await conn.websocket.send(opus_packet)
# 文件型音频走普通播放
start_time = time.perf_counter()
play_position = 0
# 更新流控状态
flow_control["packet_count"] = packet_index + 1
flow_control["sequence"] = sequence + 1
# 执行预缓冲
pre_buffer_frames = min(5, len(audios))
for i in range(pre_buffer_frames):
if conn.conn_from_mqtt_gateway:
# 计算时间戳和序列号
timestamp, sequence = calculate_timestamp_and_sequence(
conn, start_time, i, frame_duration
)
# 调用通用函数发送带头部的数据包
await _send_to_mqtt_gateway(conn, audios[i], timestamp, sequence)
else:
# 直接发送预缓冲包,不添加头部
await conn.websocket.send(audios[i])
remaining_audios = audios[pre_buffer_frames:]
# 播放剩余音频帧
for i, opus_packet in enumerate(remaining_audios):
if conn.client_abort:
break
# 重置没有声音的状态
conn.last_activity_time = time.time() * 1000
if send_delay > 0:
# 固定延迟模式
await asyncio.sleep(send_delay)
else:
# 计算预期发送时间
expected_time = start_time + (play_position / 1000)
current_time = time.perf_counter()
delay = expected_time - current_time
if delay > 0:
await asyncio.sleep(delay)
if conn.conn_from_mqtt_gateway:
# 计算时间戳和序列号(使用当前的数据包索引确保连续性)
packet_index = pre_buffer_frames + i
timestamp, sequence = calculate_timestamp_and_sequence(
conn, start_time, packet_index, frame_duration
)
# 调用通用函数发送带头部的数据包
await _send_to_mqtt_gateway(conn, opus_packet, timestamp, sequence)
else:
# 直接发送opus数据包,不添加头部
await conn.websocket.send(opus_packet)
play_position += frame_duration
async def send_tts_message(conn, state, text=None):
@@ -246,10 +220,8 @@ async def send_tts_message(conn, state, text=None):
stop_tts_notify_voice = conn.config.get(
"stop_tts_notify_voice", "config/assets/tts_notify.mp3"
)
audios = await audio_to_data(stop_tts_notify_voice, is_opus=True)
audios = audio_to_data(stop_tts_notify_voice, is_opus=True)
await sendAudio(conn, audios)
# 等待所有音频包发送完成
await _wait_for_audio_completion(conn)
# 清除服务端讲话状态
conn.clearSpeakStatus()
@@ -283,4 +255,5 @@ async def send_stt_message(conn, text):
await conn.websocket.send(
json.dumps({"type": "stt", "text": stt_text, "session_id": conn.session_id})
)
conn.client_is_speaking = True
await send_tts_message(conn, "start")
@@ -1,14 +1,12 @@
import time
import asyncio
from typing import Dict, Any
from core.handle.receiveAudioHandle import startToChat
from core.handle.receiveAudioHandle import handleAudioMessage, startToChat
from core.handle.reportHandle import enqueue_asr_report
from core.handle.sendAudioHandle import send_stt_message, send_tts_message
from core.handle.textMessageHandler import TextMessageHandler
from core.handle.textMessageType import TextMessageType
from core.utils.util import remove_punctuation_and_length
from core.providers.asr.dto.dto import InterfaceType
TAG = __name__
@@ -31,18 +29,8 @@ class ListenTextMessageHandler(TextMessageHandler):
elif msg_json["state"] == "stop":
conn.client_have_voice = True
conn.client_voice_stop = True
if conn.asr.interface_type == InterfaceType.STREAM:
# 流式模式下,发送结束请求
asyncio.create_task(conn.asr._send_stop_request())
else:
# 非流式模式:直接触发ASR识别
if len(conn.asr_audio) > 0:
asr_audio_task = conn.asr_audio.copy()
conn.asr_audio.clear()
conn.reset_vad_states()
if len(asr_audio_task) > 0:
await conn.asr.handle_voice_stop(conn, asr_audio_task)
if len(conn.asr_audio) > 0:
await handleAudioMessage(conn, b"")
elif msg_json["state"] == "detect":
conn.client_have_voice = False
conn.asr_audio.clear()
@@ -8,6 +8,8 @@ import asyncio
import requests
import websockets
import opuslib_next
import random
from typing import Optional, Tuple, List
from urllib import parse
from datetime import datetime
from config.logger import setup_logging
@@ -137,13 +139,13 @@ class ASRProvider(ASRProviderBase):
conn.asr_audio.append(audio)
conn.asr_audio = conn.asr_audio[-10:]
# 只在有声音且没有连接时建立连接(排除正在停止的情况)
if audio_have_voice and not self.is_processing and not self.asr_ws:
# 只在有声音且没有连接时建立连接
if audio_have_voice and not self.is_processing:
try:
await self._start_recognition(conn)
except Exception as e:
logger.bind(tag=TAG).error(f"开始识别失败: {str(e)}")
await self._cleanup()
await self._cleanup(conn)
return
if self.asr_ws and self.is_processing and self.server_ready:
@@ -183,8 +185,10 @@ class ASRProvider(ASRProviderBase):
"header": {
"namespace": "SpeechTranscriber",
"name": "StartTranscription",
"status": 20000000,
"message_id": uuid.uuid4().hex,
"task_id": self.task_id,
"status_text": "Gateway:SUCCESS:Success.",
"appkey": self.appkey
},
"payload": {
@@ -203,21 +207,18 @@ class ASRProvider(ASRProviderBase):
async def _forward_results(self, conn):
"""转发识别结果"""
try:
while not conn.stop_event.is_set():
while self.asr_ws and not conn.stop_event.is_set():
try:
response = await asyncio.wait_for(self.asr_ws.recv(), timeout=1.0)
result = json.loads(response)
header = result.get("header", {})
payload = result.get("payload", {})
message_name = header.get("name", "")
status = header.get("status", 0)
if status != 20000000:
if status == 40010004:
logger.bind(tag=TAG).warning(f"请在服务端响应完成后再关闭链接,状态码: {status}")
break
if status in [40000004, 40010003]: # 连接超时或客户端断开
if status in [40000004, 40010004]: # 连接超时或客户端断开
logger.bind(tag=TAG).warning(f"连接问题,状态码: {status}")
break
elif status in [40270002, 40270003]: # 音频问题
@@ -226,12 +227,12 @@ class ASRProvider(ASRProviderBase):
else:
logger.bind(tag=TAG).error(f"识别错误,状态码: {status}, 消息: {header.get('status_text', '')}")
continue
# 收到TranscriptionStarted表示服务器准备好接收音频数据
if message_name == "TranscriptionStarted":
self.server_ready = True
logger.bind(tag=TAG).debug("服务器已准备,开始发送缓存音频...")
# 发送缓存音频
if conn.asr_audio:
for cached_audio in conn.asr_audio[-10:]:
@@ -242,89 +243,89 @@ class ASRProvider(ASRProviderBase):
logger.bind(tag=TAG).warning(f"发送缓存音频失败: {e}")
break
continue
elif message_name == "SentenceEnd":
# 句子结束(每个句子都会触发)
if message_name == "TranscriptionResultChanged":
# 中间结果
text = payload.get("result", "")
if text:
logger.bind(tag=TAG).info(f"识别到文本: {text}")
# 手动模式下累积识别结果
if conn.client_listen_mode == "manual":
if self.text:
self.text += text
else:
self.text = text
# 手动模式下,只有在收到stop信号后才触发处理(仅处理一次)
if conn.client_voice_stop:
audio_data = getattr(conn, 'asr_audio_for_voiceprint', [])
if len(audio_data) > 0:
logger.bind(tag=TAG).debug("收到最终识别结果,触发处理")
await self.handle_voice_stop(conn, audio_data)
# 清理音频缓存
conn.asr_audio.clear()
conn.reset_vad_states()
break
else:
# 自动模式下直接覆盖
self.text = text
conn.reset_vad_states()
audio_data = getattr(conn, 'asr_audio_for_voiceprint', [])
await self.handle_voice_stop(conn, audio_data)
break
self.text = text
elif message_name == "SentenceEnd":
# 最终结果
text = payload.get("result", "")
if text:
self.text = text
conn.reset_vad_states()
# 传递缓存的音频数据
audio_data = getattr(conn, 'asr_audio_for_voiceprint', [])
await self.handle_voice_stop(conn, audio_data)
# 清空缓存
conn.asr_audio_for_voiceprint = []
break
elif message_name == "TranscriptionCompleted":
# 识别完成
self.is_processing = False
break
except asyncio.TimeoutError:
logger.bind(tag=TAG).error("接收结果超时")
break
except websockets.ConnectionClosed:
logger.bind(tag=TAG).info("ASR服务连接已关闭")
self.is_processing = False
continue
except websockets.exceptions.ConnectionClosed:
break
except Exception as e:
logger.bind(tag=TAG).error(f"处理结果失败: {str(e)}")
break
except Exception as e:
logger.bind(tag=TAG).error(f"结果转发失败: {str(e)}")
finally:
# 清理连接的音频缓存
await self._cleanup()
if conn:
if hasattr(conn, 'asr_audio_for_voiceprint'):
conn.asr_audio_for_voiceprint = []
if hasattr(conn, 'asr_audio'):
conn.asr_audio = []
await self._cleanup(conn)
async def _send_stop_request(self):
"""发送停止识别请求(不关闭连接)"""
if self.asr_ws:
async def _cleanup(self, conn):
"""清理资源"""
logger.bind(tag=TAG).debug(f"开始ASR会话清理 | 当前状态: processing={self.is_processing}, server_ready={self.server_ready}")
# 清理连接的音频缓存
if conn and hasattr(conn, 'asr_audio_for_voiceprint'):
conn.asr_audio_for_voiceprint = []
# 判断是否需要发送终止请求
should_stop = self.is_processing or self.server_ready
# 发送停止识别请求
if self.asr_ws and should_stop:
try:
# 先停止音频发送
self.is_processing = False
stop_msg = {
"header": {
"namespace": "SpeechTranscriber",
"name": "StopTranscription",
"status": 20000000,
"message_id": uuid.uuid4().hex,
"task_id": self.task_id,
"status_text": "Client:Stop",
"appkey": self.appkey
}
}
logger.bind(tag=TAG).debug("停止识别请求已发送")
logger.bind(tag=TAG).debug("正在发送ASR终止请求")
await self.asr_ws.send(json.dumps(stop_msg, ensure_ascii=False))
await asyncio.sleep(0.1)
logger.bind(tag=TAG).debug("ASR终止请求已发送")
except Exception as e:
logger.bind(tag=TAG).error(f"发送停止识别请求失败: {e}")
async def _cleanup(self):
"""清理资源(关闭连接)"""
logger.bind(tag=TAG).debug(f"开始ASR会话清理 | 当前状态: processing={self.is_processing}, server_ready={self.server_ready}")
# 状态重置
logger.bind(tag=TAG).error(f"ASR终止请求发送失败: {e}")
# 状态重置(在终止请求发送后)
self.is_processing = False
self.server_ready = False
logger.bind(tag=TAG).debug("ASR状态已重置")
# 清理任务
if self.forward_task and not self.forward_task.done():
self.forward_task.cancel()
try:
await asyncio.wait_for(self.forward_task, timeout=1.0)
except Exception as e:
logger.bind(tag=TAG).debug(f"forward_task取消异常: {e}")
finally:
self.forward_task = None
# 关闭连接
if self.asr_ws:
try:
@@ -335,10 +336,7 @@ class ASRProvider(ASRProviderBase):
logger.bind(tag=TAG).error(f"关闭WebSocket连接失败: {e}")
finally:
self.asr_ws = None
# 清理任务引用
self.forward_task = None
logger.bind(tag=TAG).debug("ASR会话清理完成")
async def speech_to_text(self, opus_data, session_id, audio_format):
@@ -349,11 +347,4 @@ class ASRProvider(ASRProviderBase):
async def close(self):
"""关闭资源"""
await self._cleanup(None)
if hasattr(self, 'decoder') and self.decoder is not None:
try:
del self.decoder
self.decoder = None
logger.bind(tag=TAG).debug("Aliyun decoder resources released")
except Exception as e:
logger.bind(tag=TAG).debug(f"释放Aliyun decoder资源时出错: {e}")
await self._cleanup()
+83 -57
View File
@@ -9,6 +9,7 @@ import asyncio
import traceback
import threading
import opuslib_next
import concurrent.futures
from abc import ABC, abstractmethod
from config.logger import setup_logging
from typing import Optional, Tuple, List
@@ -52,89 +53,121 @@ class ASRProviderBase(ABC):
# 接收音频
async def receive_audio(self, conn, audio, audio_have_voice):
if conn.client_listen_mode == "manual":
# 手动模式:缓存音频用于ASR识别
conn.asr_audio.append(audio)
else:
# 自动/实时模式:使用VAD检测
if conn.client_listen_mode == "auto" or conn.client_listen_mode == "realtime":
have_voice = audio_have_voice
else:
have_voice = conn.client_have_voice
conn.asr_audio.append(audio)
if not have_voice and not conn.client_have_voice:
conn.asr_audio = conn.asr_audio[-10:]
return
conn.asr_audio.append(audio)
if not have_voice and not conn.client_have_voice:
conn.asr_audio = conn.asr_audio[-10:]
return
if conn.client_voice_stop:
asr_audio_task = conn.asr_audio.copy()
conn.asr_audio.clear()
conn.reset_vad_states()
# 自动模式下通过VAD检测到语音停止时触发识别
if conn.client_voice_stop:
asr_audio_task = conn.asr_audio.copy()
conn.asr_audio.clear()
conn.reset_vad_states()
if len(asr_audio_task) > 15:
await self.handle_voice_stop(conn, asr_audio_task)
if len(asr_audio_task) > 15 or conn.client_listen_mode == "manual":
await self.handle_voice_stop(conn, asr_audio_task)
# 处理语音停止
async def handle_voice_stop(self, conn, asr_audio_task: List[bytes]):
"""并行处理ASR和声纹识别"""
try:
total_start_time = time.monotonic()
# 准备音频数据
if conn.audio_format == "pcm":
pcm_data = asr_audio_task
else:
pcm_data = self.decode_opus(asr_audio_task)
combined_pcm_data = b"".join(pcm_data)
# 预先准备WAV数据
wav_data = None
if conn.voiceprint_provider and combined_pcm_data:
wav_data = self._pcm_to_wav(combined_pcm_data)
# 定义ASR任务
asr_task = self.speech_to_text(asr_audio_task, conn.session_id, conn.audio_format)
if conn.voiceprint_provider and wav_data:
voiceprint_task = conn.voiceprint_provider.identify_speaker(wav_data, conn.session_id)
# 并发等待两个结果
asr_result, voiceprint_result = await asyncio.gather(
asr_task, voiceprint_task, return_exceptions=True
)
else:
asr_result = await asr_task
voiceprint_result = None
# 记录识别结果 - 检查是否为异常
if isinstance(asr_result, Exception):
logger.bind(tag=TAG).error(f"ASR识别失败: {asr_result}")
raw_text = ""
else:
raw_text, _ = asr_result
if isinstance(voiceprint_result, Exception):
logger.bind(tag=TAG).error(f"声纹识别失败: {voiceprint_result}")
speaker_name = ""
else:
speaker_name = voiceprint_result
def run_asr():
start_time = time.monotonic()
try:
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
try:
result = loop.run_until_complete(
self.speech_to_text(asr_audio_task, conn.session_id, conn.audio_format)
)
end_time = time.monotonic()
logger.bind(tag=TAG).debug(f"ASR耗时: {end_time - start_time:.3f}s")
return result
finally:
loop.close()
except Exception as e:
end_time = time.monotonic()
logger.bind(tag=TAG).error(f"ASR失败: {e}")
return ("", None)
# 定义声纹识别任务
def run_voiceprint():
if not wav_data:
return None
try:
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
try:
# 使用连接的声纹识别提供者
result = loop.run_until_complete(
conn.voiceprint_provider.identify_speaker(wav_data, conn.session_id)
)
return result
finally:
loop.close()
except Exception as e:
logger.bind(tag=TAG).error(f"声纹识别失败: {e}")
return None
# 使用线程池执行器并行运行
with concurrent.futures.ThreadPoolExecutor(max_workers=2) as thread_executor:
asr_future = thread_executor.submit(run_asr)
if conn.voiceprint_provider and wav_data:
voiceprint_future = thread_executor.submit(run_voiceprint)
# 等待两个线程都完成
asr_result = asr_future.result(timeout=15)
voiceprint_result = voiceprint_future.result(timeout=15)
results = {"asr": asr_result, "voiceprint": voiceprint_result}
else:
asr_result = asr_future.result(timeout=15)
results = {"asr": asr_result, "voiceprint": None}
# 处理结果
raw_text, _ = results.get("asr", ("", None))
speaker_name = results.get("voiceprint", None)
# 记录识别结果
if raw_text:
logger.bind(tag=TAG).info(f"识别文本: {raw_text}")
if speaker_name:
logger.bind(tag=TAG).info(f"识别说话人: {speaker_name}")
# 性能监控
total_time = time.monotonic() - total_start_time
logger.bind(tag=TAG).debug(f"总处理耗时: {total_time:.3f}s")
# 检查文本长度
text_len, _ = remove_punctuation_and_length(raw_text)
self.stop_ws_connection()
if text_len > 0:
# 构建包含说话人信息的JSON字符串
enhanced_text = self._build_enhanced_text(raw_text, speaker_name)
# 使用自定义模块进行上报
await startToChat(conn, enhanced_text)
enqueue_asr_report(conn, enhanced_text, asr_audio_task)
@@ -208,7 +241,6 @@ class ASRProviderBase(ABC):
@staticmethod
def decode_opus(opus_data: List[bytes]) -> List[bytes]:
"""将Opus音频数据解码为PCM数据"""
decoder = None
try:
decoder = opuslib_next.Decoder(16000, 1)
pcm_data = []
@@ -233,9 +265,3 @@ class ASRProviderBase(ABC):
except Exception as e:
logger.bind(tag=TAG).error(f"音频解码过程发生错误: {e}")
return []
finally:
if decoder is not None:
try:
del decoder
except Exception as e:
logger.bind(tag=TAG).debug(f"释放decoder资源时出错: {e}")
@@ -18,6 +18,8 @@ class ASRProvider(ASRProviderBase):
self.interface_type = InterfaceType.STREAM
self.config = config
self.text = ""
self.max_retries = 3
self.retry_delay = 2
self.decoder = opuslib_next.Decoder(16000, 1)
self.asr_ws = None
self.forward_task = None
@@ -54,13 +56,14 @@ class ASRProvider(ASRProviderBase):
async def receive_audio(self, conn, audio, audio_have_voice):
conn.asr_audio.append(audio)
conn.asr_audio = conn.asr_audio[-10:]
# 存储音频数据
if not hasattr(conn, 'asr_audio_for_voiceprint'):
conn.asr_audio_for_voiceprint = []
conn.asr_audio_for_voiceprint.append(audio)
# 当没有音频数据时处理完整语音片段
if conn.client_listen_mode != "manual" and not audio and len(conn.asr_audio_for_voiceprint) > 0:
if not audio and len(conn.asr_audio_for_voiceprint) > 0:
await self.handle_voice_stop(conn, conn.asr_audio_for_voiceprint)
conn.asr_audio_for_voiceprint = []
@@ -176,7 +179,6 @@ class ASRProvider(ASRProviderBase):
payload.get("audio_info", {}).get("duration", 0) > 2000
and not utterances
and not payload["result"].get("text")
and conn.client_listen_mode != "manual"
):
logger.bind(tag=TAG).error(f"识别文本:空")
self.text = ""
@@ -185,44 +187,15 @@ class ASRProvider(ASRProviderBase):
await self.handle_voice_stop(conn, audio_data)
break
# 专门处理没有文本的识别结果(手动模式下可能已经识别完成但是没松按键)
elif not payload["result"].get("text") and not utterances:
if conn.client_listen_mode == "manual" and conn.client_voice_stop and len(audio_data) > 0:
logger.bind(tag=TAG).debug("消息结束收到停止信号,触发处理")
await self.handle_voice_stop(conn, audio_data)
# 清理音频缓存
conn.asr_audio.clear()
conn.reset_vad_states()
break
for utterance in utterances:
if utterance.get("definite", False):
current_text = utterance["text"]
self.text = utterance["text"]
logger.bind(tag=TAG).info(
f"识别到文本: {current_text}"
f"识别到文本: {self.text}"
)
# 手动模式下累积识别结果
if conn.client_listen_mode == "manual":
if self.text:
self.text += current_text
else:
self.text = current_text
# 在接收消息中途时收到停止信号
if conn.client_voice_stop and len(audio_data) > 0:
logger.bind(tag=TAG).debug("消息中途收到停止信号,触发处理")
await self.handle_voice_stop(conn, audio_data)
# 清理音频缓存
conn.asr_audio.clear()
conn.reset_vad_states()
break
else:
# 自动模式下直接覆盖
self.text = current_text
conn.reset_vad_states()
if len(audio_data) > 15: # 确保有足够音频数据
await self.handle_voice_stop(conn, audio_data)
conn.reset_vad_states()
if len(audio_data) > 15: # 确保有足够音频数据
await self.handle_voice_stop(conn, audio_data)
break
elif "error" in payload:
error_msg = payload.get("error", "未知错误")
@@ -254,6 +227,8 @@ class ASRProvider(ASRProviderBase):
conn.asr_audio_for_voiceprint = []
if hasattr(conn, 'asr_audio'):
conn.asr_audio = []
if hasattr(conn, 'has_valid_voice'):
conn.has_valid_voice = False
def stop_ws_connection(self):
if self.asr_ws:
@@ -261,20 +236,6 @@ class ASRProvider(ASRProviderBase):
self.asr_ws = None
self.is_processing = False
async def _send_stop_request(self):
"""发送最后一个音频帧以通知服务器结束"""
if self.asr_ws:
try:
# 发送结束标记的音频帧(gzip压缩的空数据)
empty_payload = gzip.compress(b"")
last_audio_request = bytearray(self.generate_last_audio_default_header())
last_audio_request.extend(len(empty_payload).to_bytes(4, "big"))
last_audio_request.extend(empty_payload)
await self.asr_ws.send(last_audio_request)
logger.bind(tag=TAG).debug("已发送结束音频帧")
except Exception as e:
logger.bind(tag=TAG).debug(f"发送结束音频帧时出错: {e}")
def construct_request(self, reqid):
req = {
"app": {
@@ -409,16 +370,6 @@ class ASRProvider(ASRProviderBase):
pass
self.forward_task = None
self.is_processing = False
# 显式释放decoder资源
if hasattr(self, 'decoder') and self.decoder is not None:
try:
del self.decoder
self.decoder = None
logger.bind(tag=TAG).debug("Doubao decoder resources released")
except Exception as e:
logger.bind(tag=TAG).debug(f"释放Doubao decoder资源时出错: {e}")
# 清理所有连接的音频缓冲区
if hasattr(self, '_connections'):
for conn in self._connections.values():
@@ -426,3 +377,5 @@ class ASRProvider(ASRProviderBase):
conn.asr_audio_for_voiceprint = []
if hasattr(conn, 'asr_audio'):
conn.asr_audio = []
if hasattr(conn, 'has_valid_voice'):
conn.has_valid_voice = False
@@ -1,16 +1,14 @@
import os
import io
import sys
import time
import shutil
import os
import sys
import io
import psutil
import asyncio
from config.logger import setup_logging
from typing import Optional, Tuple, List
from core.providers.asr.base import ASRProviderBase
from funasr import AutoModel
from funasr.utils.postprocess_utils import rich_transcription_postprocess
from core.providers.asr.base import ASRProviderBase
import shutil
from core.providers.asr.dto.dto import InterfaceType
TAG = __name__
@@ -92,17 +90,16 @@ class ASRProvider(ASRProviderBase):
else:
file_path = self.save_audio_to_file(pcm_data, session_id)
# 语音识别 - 使用线程池避免阻塞事件循环
# 语音识别
start_time = time.time()
result = await asyncio.to_thread(
self.model.generate,
result = self.model.generate(
input=combined_pcm_data,
cache={},
language="auto",
use_itn=True,
batch_size_s=60,
)
text = await asyncio.to_thread(rich_transcription_postprocess, result[0]["text"])
text = rich_transcription_postprocess(result[0]["text"])
logger.bind(tag=TAG).debug(
f"语音识别耗时: {time.time() - start_time:.3f}s | 结果: {text}"
)
@@ -1,5 +1,8 @@
import os
import json
import asyncio
import tempfile
import difflib
from typing import Optional, Tuple, List
import dashscope
from config.logger import setup_logging
@@ -13,8 +16,7 @@ logger = setup_logging()
class ASRProvider(ASRProviderBase):
def __init__(self, config: dict, delete_audio_file: bool):
super().__init__()
# 音频文件上传类型,流式文本识别输出
self.interface_type = InterfaceType.NON_STREAM
self.interface_type = InterfaceType.STREAM
"""Qwen3-ASR-Flash ASR初始化"""
# 配置参数
@@ -128,11 +130,27 @@ class ASRProvider(ASRProviderBase):
# 处理流式响应
full_text = ""
last_text = "" # 用于存储上一个文本片段
for chunk in response:
try:
text = chunk["output"]["choices"][0]["message"].content[0]["text"]
# 更新为最新的完整文本
full_text = text.strip()
# 标准化文本片段(去除首尾空格)
normalized_text = text.strip()
# 只有当新文本片段与上一个不同时才处理
if normalized_text != last_text:
# 提取新增的文本部分
# 通过比较当前文本和上一个文本,找到新增的部分
if normalized_text.startswith(last_text):
# 如果当前文本以最后一个文本开头,则新增部分是两者的差集
new_part = normalized_text[len(last_text):]
else:
# 如果不以最后一个文本开头,说明识别结果发生了较大变化,直接使用当前文本
new_part = normalized_text
# 将新增部分添加到完整文本中
full_text += new_part
last_text = normalized_text
# 这里可以实时处理文本片段,例如通过回调函数
except:
pass
@@ -5,7 +5,6 @@ import hashlib
import asyncio
import websockets
import opuslib_next
import gc
from time import mktime
from datetime import datetime
from urllib.parse import urlencode
@@ -35,6 +34,9 @@ class ASRProvider(ASRProviderBase):
self.forward_task = None
self.is_processing = False
self.server_ready = False
self.last_frame_sent = False # 标记是否已发送最终帧
self.best_text = "" # 保存最佳识别结果
self.has_final_result = False # 标记是否收到最终识别结果
# 讯飞配置
self.app_id = config.get("app_id")
@@ -49,6 +51,7 @@ class ASRProvider(ASRProviderBase):
"domain": config.get("domain", "slm"),
"language": config.get("language", "zh_cn"),
"accent": config.get("accent", "mandarin"),
"dwa": config.get("dwa", "wpgs"),
"result": {"encoding": "utf8", "compress": "raw", "format": "plain"},
}
@@ -112,7 +115,7 @@ class ASRProvider(ASRProviderBase):
await self._start_recognition(conn)
except Exception as e:
logger.bind(tag=TAG).error(f"建立ASR连接失败: {str(e)}")
await self._cleanup()
await self._cleanup(conn)
return
# 发送当前音频数据
@@ -122,7 +125,7 @@ class ASRProvider(ASRProviderBase):
await self._send_audio_frame(pcm_frame, STATUS_CONTINUE_FRAME)
except Exception as e:
logger.bind(tag=TAG).warning(f"发送音频数据时发生错误: {e}")
await self._cleanup()
await self._cleanup(conn)
async def _start_recognition(self, conn):
"""开始识别会话"""
@@ -132,10 +135,6 @@ class ASRProvider(ASRProviderBase):
ws_url = self.create_url()
logger.bind(tag=TAG).info(f"正在连接ASR服务: {ws_url[:50]}...")
# 如果为手动模式,设置超时时长为一分钟
if conn.client_listen_mode == "manual":
self.iat_params["eos"] = 60000
self.asr_ws = await websockets.connect(
ws_url,
max_size=1000000000,
@@ -146,6 +145,8 @@ class ASRProvider(ASRProviderBase):
logger.bind(tag=TAG).info("ASR WebSocket连接已建立")
self.server_ready = False
self.last_frame_sent = False
self.best_text = ""
self.forward_task = asyncio.create_task(self._forward_results(conn))
# 发送首帧音频
@@ -194,12 +195,23 @@ class ASRProvider(ASRProviderBase):
await self.asr_ws.send(json.dumps(frame_data, ensure_ascii=False))
# 标记是否发送了最终帧
if status == STATUS_LAST_FRAME:
self.last_frame_sent = True
logger.bind(tag=TAG).info("标记最终帧已发送")
async def _forward_results(self, conn):
"""转发识别结果"""
try:
while not conn.stop_event.is_set():
while self.asr_ws and not conn.stop_event.is_set():
# 获取当前连接的音频数据
audio_data = getattr(conn, "asr_audio_for_voiceprint", [])
try:
response = await asyncio.wait_for(self.asr_ws.recv(), timeout=60)
# 如果已发送最终帧,增加超时时间等待完整结果
timeout = 3.0 if self.last_frame_sent else 30.0
response = await asyncio.wait_for(
self.asr_ws.recv(), timeout=timeout
)
result = json.loads(response)
logger.bind(tag=TAG).debug(f"收到ASR结果: {result}")
@@ -223,27 +235,144 @@ class ASRProvider(ASRProviderBase):
# 解码base64文本
decoded_text = base64.b64decode(text_data).decode("utf-8")
text_json = json.loads(decoded_text)
# 提取文本内容
text_ws = text_json.get("ws", [])
result_text = ""
for i in text_ws:
for j in i.get("cw", []):
w = j.get("w", "")
self.text += w
result_text += w
# 更新识别文本 - 实时更新策略
# 只检查是否为空字符串,不再过滤任何标点符号
# 这样可以确保所有识别到的内容,包括标点符号都能被实时更新
if result_text and result_text.strip():
# 实时更新:正常情况下都更新,提高响应速度
should_update = True
# 保存最佳文本
# 1. 如果是识别完成状态或最终帧后收到的结果,优先保存
# 2. 否则保存最长的有意义文本
# 取消对标点符号的过滤,只检查是否为空
# 这样可以保留所有识别到的内容,包括各种标点符号
is_valid_text = len(result_text.strip()) > 0
if (
self.last_frame_sent or status == 2
) and is_valid_text:
self.best_text = result_text
self.has_final_result = True # 标记已收到最终结果
logger.bind(tag=TAG).debug(
f"保存最终识别结果: {self.best_text}"
)
elif (
len(result_text) > len(self.best_text)
and is_valid_text
and not self.has_final_result
):
self.best_text = result_text
logger.bind(tag=TAG).debug(
f"保存中间最佳文本: {self.best_text}"
)
# 如果已发送最终帧,只过滤空文本
if self.last_frame_sent:
# 只拒绝完全空的结果
if not result_text.strip():
should_update = False
logger.bind(tag=TAG).warning(
f"最终帧后拒绝空文本"
)
if should_update:
# 处理流式识别结果,避免简单替换导致内容丢失
# 1. 如果是中间状态(非最终帧后),可能需要替换为更完整的识别
# 2. 如果是最终帧后收到的结果,可能是对前面文本的补充
if self.last_frame_sent:
# 最终帧后收到的结果可能是标点符号等补充内容
# 检查是否需要合并文本而不是替换
# 如果当前文本是纯标点而前面已有内容,应该追加而不是替换
if len(
self.text
) > 0 and result_text.strip() in [
"",
".",
"?",
"",
"!",
"",
",",
"",
";",
"",
]:
# 对于标点符号,追加到现有文本后
self.text = (
self.text.rstrip().rstrip("。.")
+ result_text
)
else:
# 其他情况保持替换逻辑
self.text = result_text
else:
# 中间状态替换为新的识别结果
self.text = result_text
logger.bind(tag=TAG).info(
f"实时更新识别文本: {self.text} (最终帧已发送: {self.last_frame_sent})"
)
# 识别完成,但如果还没发送最终帧,继续等待
if status == 2:
if conn.client_listen_mode == "manual":
audio_data = getattr(conn, 'asr_audio_for_voiceprint', [])
if len(audio_data) > 0:
logger.bind(tag=TAG).debug("收到最终识别结果,触发处理")
await self.handle_voice_stop(conn, audio_data)
# 清理音频缓存
conn.asr_audio.clear()
logger.bind(tag=TAG).info(
f"识别完成状态已到达,当前识别文本: {self.text}"
)
# 如果还没发送最终帧,继续等待
if not self.last_frame_sent:
logger.bind(tag=TAG).info(
"识别完成但最终帧未发送,继续等待..."
)
continue
# 已发送最终帧且收到完成状态,使用最佳策略选择最终结果
# 优先使用识别完成状态下的最新结果,而不是仅仅基于长度
if self.best_text:
# 如果当前文本是在最终帧发送后或识别完成状态下收到的,优先使用
if (
self.last_frame_sent or status == 2
) and self.text.strip():
logger.bind(tag=TAG).info(
f"使用完成状态下的最新识别结果: {self.text}"
)
elif len(self.best_text) > len(self.text):
logger.bind(tag=TAG).info(
f"使用更长的最佳文本作为最终结果: {self.text} -> {self.best_text}"
)
self.text = self.best_text
logger.bind(tag=TAG).info(f"获取到最终完整文本: {self.text}")
conn.reset_vad_states()
if len(audio_data) > 15: # 确保有足够音频数据
# 准备处理结果
pass
break
except asyncio.TimeoutError:
logger.bind(tag=TAG).error("接收结果超时")
break
if self.last_frame_sent:
# 超时时也使用最佳文本
if self.best_text and len(self.best_text) > len(self.text):
logger.bind(tag=TAG).info(
f"超时,使用最佳文本: {self.text} -> {self.best_text}"
)
self.text = self.best_text
logger.bind(tag=TAG).info(
f"最终帧后超时,使用结果: {self.text}"
)
break
# 如果还没发送最终帧,继续等待
continue
except websockets.ConnectionClosed:
logger.bind(tag=TAG).info("ASR服务连接已关闭")
self.is_processing = False
@@ -260,15 +389,17 @@ class ASRProvider(ASRProviderBase):
if hasattr(e, "__cause__") and e.__cause__:
logger.bind(tag=TAG).error(f"错误原因: {str(e.__cause__)}")
finally:
# 清理连接资源
await self._cleanup()
# 清理连接的音频缓存
if self.asr_ws:
await self.asr_ws.close()
self.asr_ws = None
self.is_processing = False
if conn:
if hasattr(conn, "asr_audio_for_voiceprint"):
conn.asr_audio_for_voiceprint = []
if hasattr(conn, "asr_audio"):
conn.asr_audio = []
if hasattr(conn, "has_valid_voice"):
conn.has_valid_voice = False
async def handle_voice_stop(self, conn, asr_audio_task: List[bytes]):
"""处理语音停止,发送最后一帧并处理识别结果"""
@@ -276,13 +407,22 @@ class ASRProvider(ASRProviderBase):
# 先发送最后一帧表示音频结束
if self.asr_ws and self.is_processing:
try:
await self._send_audio_frame(b"", STATUS_LAST_FRAME)
logger.bind(tag=TAG).debug(f"已发送停止请求")
# 取最后一个有效的音频帧作为最后一帧数据
last_frame = b""
if asr_audio_task:
last_audio = asr_audio_task[-1]
last_frame = self.decoder.decode(last_audio, 960)
await self._send_audio_frame(last_frame, STATUS_LAST_FRAME)
logger.bind(tag=TAG).info("已发送最后一帧")
# 发送最终帧后,给_forward_results适当时间处理最终结果
await asyncio.sleep(0.25)
except Exception as e:
logger.bind(tag=TAG).error(f"发送停止请求失败: {e}")
logger.bind(tag=TAG).info(f"准备处理最终识别结果: {self.text}")
except Exception as e:
logger.bind(tag=TAG).error(f"发送最后一帧失败: {e}")
# 调用父类的handle_voice_stop方法处理识别结果
await super().handle_voice_stop(conn, asr_audio_task)
except Exception as e:
logger.bind(tag=TAG).error(f"处理语音停止失败: {e}")
@@ -296,27 +436,40 @@ class ASRProvider(ASRProviderBase):
self.asr_ws = None
self.is_processing = False
async def _send_stop_request(self):
"""发送停止识别请求(不关闭连接)"""
if self.asr_ws:
try:
# 先停止音频发送
self.is_processing = False
await self._send_audio_frame(b"", STATUS_LAST_FRAME)
logger.bind(tag=TAG).debug("已发送停止请求")
except Exception as e:
logger.bind(tag=TAG).error(f"发送停止请求失败: {e}")
async def _cleanup(self):
"""清理资源(关闭连接)"""
logger.bind(tag=TAG).debug(
async def _cleanup(self, conn):
"""清理资源"""
logger.bind(tag=TAG).info(
f"开始ASR会话清理 | 当前状态: processing={self.is_processing}, server_ready={self.server_ready}"
)
# 发送最后一帧
if self.asr_ws and self.is_processing:
try:
await self._send_audio_frame(b"", STATUS_LAST_FRAME)
await asyncio.sleep(0.1)
logger.bind(tag=TAG).info("已发送最后一帧")
except Exception as e:
logger.bind(tag=TAG).error(f"发送最后一帧失败: {e}")
# 状态重置
self.is_processing = False
self.server_ready = False
logger.bind(tag=TAG).debug("ASR状态已重置")
self.last_frame_sent = False
self.best_text = ""
self.has_final_result = False
logger.bind(tag=TAG).info("ASR状态已重置")
# 清理任务
if self.forward_task and not self.forward_task.done():
self.forward_task.cancel()
try:
await asyncio.wait_for(self.forward_task, timeout=1.0)
except asyncio.CancelledError:
pass
except Exception as e:
logger.bind(tag=TAG).debug(f"forward_task取消异常: {e}")
finally:
self.forward_task = None
# 关闭连接
if self.asr_ws:
@@ -329,10 +482,16 @@ class ASRProvider(ASRProviderBase):
finally:
self.asr_ws = None
# 清理任务引用
self.forward_task = None
# 清理连接的音频缓存
if conn:
if hasattr(conn, "asr_audio_for_voiceprint"):
conn.asr_audio_for_voiceprint = []
if hasattr(conn, "asr_audio"):
conn.asr_audio = []
if hasattr(conn, "has_valid_voice"):
conn.has_valid_voice = False
logger.bind(tag=TAG).debug("ASR会话清理完成")
logger.bind(tag=TAG).info("ASR会话清理完成")
async def speech_to_text(self, opus_data, session_id, audio_format):
"""获取识别结果"""
@@ -353,16 +512,6 @@ class ASRProvider(ASRProviderBase):
pass
self.forward_task = None
self.is_processing = False
# 显式释放decoder资源
if hasattr(self, 'decoder') and self.decoder is not None:
try:
del self.decoder
self.decoder = None
logger.bind(tag=TAG).debug("Xunfei decoder resources released")
except Exception as e:
logger.bind(tag=TAG).debug(f"释放Xunfei decoder资源时出错: {e}")
# 清理所有连接的音频缓冲区
if hasattr(self, "_connections"):
for conn in self._connections.values():
@@ -370,3 +519,5 @@ class ASRProvider(ASRProviderBase):
conn.asr_audio_for_voiceprint = []
if hasattr(conn, "asr_audio"):
conn.asr_audio = []
if hasattr(conn, "has_valid_voice"):
conn.has_valid_voice = False
@@ -5,7 +5,6 @@ import os
import yaml
from config.config_loader import get_project_dir
from config.manage_api_client import save_mem_local_short
import asyncio
from core.utils.util import check_model_key
@@ -194,13 +193,7 @@ class MemoryProvider(MemoryProviderBase):
max_tokens=2000,
temperature=0.2,
)
# 使用异步版本,需要在事件循环中运行
try:
loop = asyncio.get_running_loop()
loop.create_task(save_mem_local_short(self.role_id, result))
except RuntimeError:
# 如果没有运行中的事件循环,创建一个新的
asyncio.run(save_mem_local_short(self.role_id, result))
save_mem_local_short(self.role_id, result)
logger.bind(tag=TAG).info(f"Save memory successful - Role: {self.role_id}")
return self.short_memory
@@ -10,13 +10,10 @@ import concurrent.futures
from contextlib import AsyncExitStack
from typing import Optional, List, Dict, Any
from mcp import ClientSession, StdioServerParameters, Implementation
from mcp.client.session import SamplingFnT, ElicitationFnT, ListRootsFnT, LoggingFnT, MessageHandlerFnT
from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client
from mcp.client.sse import sse_client
from mcp.client.streamable_http import streamablehttp_client
from mcp.shared.session import ProgressFnT
from config.logger import setup_logging
from core.utils.util import sanitize_tool_name
@@ -44,25 +41,13 @@ class ServerMCPClient:
self.tools_dict: Dict[str, Any] = {}
self.name_mapping: Dict[str, str] = {}
async def initialize(self, read_timeout_seconds: timedelta | None = None,
sampling_callback: SamplingFnT | None = None,
elicitation_callback: ElicitationFnT | None = None,
list_roots_callback: ListRootsFnT | None = None,
logging_callback: LoggingFnT | None = None,
message_handler: MessageHandlerFnT | None = None,
client_info: Implementation | None = None):
async def initialize(self):
"""初始化MCP客户端连接"""
if self._worker_task:
return
self._worker_task = asyncio.create_task(
self._worker(read_timeout_seconds=read_timeout_seconds,
sampling_callback=sampling_callback,
elicitation_callback=elicitation_callback,
list_roots_callback=list_roots_callback,
logging_callback=logging_callback,
message_handler=message_handler,
client_info=client_info), name="ServerMCPClientWorker"
self._worker(), name="ServerMCPClientWorker"
)
await self._ready_evt.wait()
@@ -112,15 +97,12 @@ class ServerMCPClient:
for name, tool in self.tools_dict.items()
]
async def call_tool(self, name: str, arguments: dict, read_timeout_seconds: timedelta | None = None, progress_callback: ProgressFnT | None = None, *, meta: dict[str, Any] | None = None) -> Any:
async def call_tool(self, name: str, args: dict) -> Any:
"""调用指定工具
Args:
name: 工具名称
arguments: 工具参数
read_timeout_seconds:
progress_callback: 进度回调函数
meta:
args: 工具参数
Returns:
Any: 工具执行结果
@@ -133,7 +115,7 @@ class ServerMCPClient:
real_name = self.name_mapping.get(name, name)
loop = self._worker_task.get_loop()
coro = self.session.call_tool(real_name, arguments=arguments, read_timeout_seconds=read_timeout_seconds, progress_callback=progress_callback, meta=meta)
coro = self.session.call_tool(real_name, args)
if loop is asyncio.get_running_loop():
return await coro
@@ -162,13 +144,7 @@ class ServerMCPClient:
# 所有检查都通过,连接正常
return True
async def _worker(self, read_timeout_seconds: timedelta | None = None,
sampling_callback: SamplingFnT | None = None,
elicitation_callback: ElicitationFnT | None = None,
list_roots_callback: ListRootsFnT | None = None,
logging_callback: LoggingFnT | None = None,
message_handler: MessageHandlerFnT | None = None,
client_info: Implementation | None = None):
async def _worker(self):
"""MCP客户端工作协程"""
async with AsyncExitStack() as stack:
try:
@@ -232,13 +208,7 @@ class ServerMCPClient:
ClientSession(
read_stream=read_stream,
write_stream=write_stream,
read_timeout_seconds=read_timeout_seconds,
sampling_callback=sampling_callback,
elicitation_callback=elicitation_callback,
list_roots_callback=list_roots_callback,
logging_callback=logging_callback,
message_handler=message_handler,
client_info=client_info
read_timeout_seconds=timedelta(seconds=15),
)
)
await self.session.initialize()
@@ -3,14 +3,7 @@
import asyncio
import os
import json
from datetime import timedelta
from typing import Dict, Any, List
from mcp import Implementation
from mcp.client.session import SamplingFnT, ElicitationFnT, ListRootsFnT, LoggingFnT, MessageHandlerFnT
from mcp.shared.session import ProgressFnT
from mcp.types import LoggingMessageNotificationParams
from config.config_loader import get_project_dir
from config.logger import setup_logging
from .mcp_client import ServerMCPClient
@@ -63,7 +56,7 @@ class ServerMCPManager:
# 初始化服务端MCP客户端
logger.bind(tag=TAG).info(f"初始化服务端MCP客户端: {name}")
client = ServerMCPClient(srv_config)
await client.initialize(logging_callback=self.logging_callback)
await client.initialize()
self.clients[name] = client
client_tools = client.get_available_tools()
self.tools.extend(client_tools)
@@ -116,7 +109,7 @@ class ServerMCPManager:
# 带重试机制的工具调用
for attempt in range(max_retries):
try:
return await target_client.call_tool(tool_name, arguments, progress_callback=self.progress_callback)
return await target_client.call_tool(tool_name, arguments)
except Exception as e:
# 最后一次尝试失败时直接抛出异常
if attempt == max_retries - 1:
@@ -138,7 +131,7 @@ class ServerMCPManager:
config = self.load_config()
if client_name in config:
client = ServerMCPClient(config[client_name])
await client.initialize(logging_callback=self.logging_callback)
await client.initialize()
self.clients[client_name] = client
target_client = client
logger.bind(tag=TAG).info(
@@ -166,11 +159,3 @@ class ServerMCPManager:
except (asyncio.TimeoutError, Exception) as e:
logger.bind(tag=TAG).error(f"关闭服务端MCP客户端 {name} 时出错: {e}")
self.clients.clear()
# 可选回调方法
async def logging_callback(self, params: LoggingMessageNotificationParams):
logger.bind(tag=TAG).info(f"[Server Log - {params.level.upper()}] {params.data}")
async def progress_callback(self, progress: float, total: float | None, message: str | None) -> None:
logger.bind(tag=TAG).info(f"[Progress {progress}/{total}]: {message}")
@@ -36,18 +36,7 @@ class VADProvider(VADProviderBase):
# 至少要多少帧才算有语音
self.frame_window_threshold = 3
def __del__(self):
if hasattr(self, 'decoder') and self.decoder is not None:
try:
del self.decoder
except Exception:
pass
def is_vad(self, conn, opus_packet):
# 手动模式:直接返回True,不进行实时VAD检测,所有音频都缓存
if conn.client_listen_mode == "manual":
return True
try:
pcm_frame = self.decoder.decode(opus_packet, 960)
conn.client_audio_buffer.extend(pcm_frame) # 将新数据加入缓冲区
@@ -1,150 +0,0 @@
import time
import asyncio
from config.logger import setup_logging
TAG = __name__
logger = setup_logging()
class AudioRateController:
"""
音频速率控制器 - 按照60ms帧时长精确控制音频发送
解决高并发下的时间累积误差问题
"""
def __init__(self, frame_duration=60):
"""
Args:
frame_duration: 单个音频帧时长毫秒默认60ms
"""
self.frame_duration = frame_duration
self.queue = []
self.play_position = 0 # 虚拟播放位置(毫秒)
self.start_timestamp = None # 开始时间戳(只读,不修改)
self.pending_send_task = None
self.logger = logger
self.queue_empty_event = asyncio.Event() # 队列清空事件
self.queue_empty_event.set() # 初始为空状态
def reset(self):
"""重置控制器状态"""
if self.pending_send_task and not self.pending_send_task.done():
self.pending_send_task.cancel()
# 取消任务后,任务会在下次事件循环时清理,无需阻塞等待
self.queue.clear()
self.play_position = 0
self.start_timestamp = time.time()
self.queue_empty_event.set() # 队列已清空
def add_audio(self, opus_packet):
"""添加音频包到队列"""
self.queue.append(("audio", opus_packet))
self.queue_empty_event.clear() # 队列非空,清除事件
def add_message(self, message_callback):
"""
添加消息到队列立即发送不占用播放时间
Args:
message_callback: 消息发送回调函数 async def()
"""
self.queue.append(("message", message_callback))
self.queue_empty_event.clear() # 队列非空,清除事件
def _get_elapsed_ms(self):
"""获取已经过的时间(毫秒)"""
if self.start_timestamp is None:
return 0
return (time.time() - self.start_timestamp) * 1000
async def check_queue(self, send_audio_callback):
"""
检查队列并按时发送音频/消息
Args:
send_audio_callback: 发送音频的回调函数 async def(opus_packet)
"""
if self.start_timestamp is None:
self.start_timestamp = time.time()
while self.queue:
item = self.queue[0]
item_type = item[0]
if item_type == "message":
# 消息类型:立即发送,不占用播放时间
_, message_callback = item
self.queue.pop(0)
try:
await message_callback()
except Exception as e:
self.logger.bind(tag=TAG).error(f"发送消息失败: {e}")
raise
elif item_type == "audio":
_, opus_packet = item
# 循环等待直到时间到达
while True:
# 计算时间差
elapsed_ms = self._get_elapsed_ms()
output_ms = self.play_position
if elapsed_ms < output_ms:
# 还不到发送时间,计算等待时长
wait_ms = output_ms - elapsed_ms
# 等待后继续检查(允许被中断)
try:
await asyncio.sleep(wait_ms / 1000)
except asyncio.CancelledError:
self.logger.bind(tag=TAG).debug("音频发送任务被取消")
raise
# 等待结束后重新检查时间(循环回到 while True)
else:
# 时间已到,跳出等待循环
break
# 时间已到,从队列移除并发送
self.queue.pop(0)
self.play_position += self.frame_duration
try:
await send_audio_callback(opus_packet)
except Exception as e:
self.logger.bind(tag=TAG).error(f"发送音频失败: {e}")
raise
self.queue_empty_event.set()
def start_sending(self, send_audio_callback):
"""
启动异步发送任务
Args:
send_audio_callback: 发送音频的回调函数
Returns:
asyncio.Task: 发送任务
"""
async def _send_loop():
try:
while True:
await self.check_queue(send_audio_callback)
# 如果队列空了,短暂等待后再检查(避免 busy loop)
await asyncio.sleep(0.01)
except asyncio.CancelledError:
self.logger.bind(tag=TAG).debug("音频发送循环已停止")
except Exception as e:
self.logger.bind(tag=TAG).error(f"音频发送循环异常: {e}")
self.pending_send_task = asyncio.create_task(_send_loop())
return self.pending_send_task
def stop_sending(self):
"""停止发送任务"""
if self.pending_send_task and not self.pending_send_task.done():
self.pending_send_task.cancel()
self.logger.bind(tag=TAG).debug("已取消音频发送任务")
-4
View File
@@ -19,7 +19,6 @@ class CacheType(Enum):
CONFIG = "config"
DEVICE_PROMPT = "device_prompt"
VOICEPRINT_HEALTH = "voiceprint_health" # 声纹识别健康检查
AUDIO_DATA = "audio_data" # 音频数据缓存
@dataclass
@@ -59,8 +58,5 @@ class CacheConfig:
CacheType.VOICEPRINT_HEALTH: cls(
strategy=CacheStrategy.TTL, ttl=600, max_size=100 # 10分钟过期
),
CacheType.AUDIO_DATA: cls(
strategy=CacheStrategy.TTL, ttl=600, max_size=100 # 10分钟过期
),
}
return configs.get(cache_type, cls())
@@ -1,64 +0,0 @@
import httpx
from typing import Dict, Any, List
from config.logger import setup_logging
TAG = __name__
class ContextDataProvider:
"""数据上下文填充,负责从配置的API获取数据"""
def __init__(self, config: Dict[str, Any], logger=None):
self.config = config
self.logger = logger or setup_logging()
self.context_data = ""
def fetch_all(self, device_id: str) -> str:
"""获取所有配置的上下文数据"""
context_providers = self.config.get("context_providers", [])
if not context_providers:
return ""
formatted_lines = []
for provider in context_providers:
url = provider.get("url")
headers = provider.get("headers", {})
if not url:
continue
try:
headers = headers.copy() if isinstance(headers, dict) else {}
# 将 device_id 添加到请求头
headers["device-id"] = device_id
# 发送请求
response = httpx.get(url, headers=headers, timeout=3)
if response.status_code == 200:
result = response.json()
if isinstance(result, dict):
if result.get("code") == 0:
data = result.get("data")
# 格式化数据
if isinstance(data, dict):
for k, v in data.items():
formatted_lines.append(f"- **{k}** {v}")
elif isinstance(data, list):
for item in data:
formatted_lines.append(f"- {item}")
else:
formatted_lines.append(f"- {data}")
else:
self.logger.bind(tag=TAG).warning(f"API {url} 返回错误码: {result.get('msg')}")
else:
self.logger.bind(tag=TAG).warning(f"API {url} 返回的不是JSON字典")
else:
self.logger.bind(tag=TAG).warning(f"API {url} 请求失败: {response.status_code}")
except Exception as e:
self.logger.bind(tag=TAG).error(f"获取上下文数据 {url} 失败: {e}")
# 将所有格式化后的行拼接成一个字符串
self.context_data = "\n".join(formatted_lines)
if self.context_data:
self.logger.bind(tag=TAG).debug(f"已注入动态上下文数据:\n{self.context_data}")
return self.context_data
@@ -1,122 +0,0 @@
"""
全局GC管理模块
定期执行垃圾回收避免频繁触发GC导致的GIL锁问题
"""
import gc
import asyncio
import threading
from config.logger import setup_logging
TAG = __name__
logger = setup_logging()
class GlobalGCManager:
"""全局垃圾回收管理器"""
def __init__(self, interval_seconds=300):
"""
初始化GC管理器
Args:
interval_seconds: GC执行间隔默认300秒5分钟
"""
self.interval_seconds = interval_seconds
self._task = None
self._stop_event = asyncio.Event()
self._lock = threading.Lock()
async def start(self):
"""启动定时GC任务"""
if self._task is not None:
logger.bind(tag=TAG).warning("GC管理器已经在运行")
return
logger.bind(tag=TAG).info(f"启动全局GC管理器,间隔{self.interval_seconds}")
self._stop_event.clear()
self._task = asyncio.create_task(self._gc_loop())
async def stop(self):
"""停止定时GC任务"""
if self._task is None:
return
logger.bind(tag=TAG).info("停止全局GC管理器")
self._stop_event.set()
if self._task and not self._task.done():
self._task.cancel()
try:
await self._task
except asyncio.CancelledError:
pass
self._task = None
async def _gc_loop(self):
"""GC循环任务"""
try:
while not self._stop_event.is_set():
# 等待指定间隔
try:
await asyncio.wait_for(
self._stop_event.wait(), timeout=self.interval_seconds
)
# 如果stop_event被设置,退出循环
break
except asyncio.TimeoutError:
# 超时表示到了执行GC的时间
pass
# 执行GC
await self._run_gc()
except asyncio.CancelledError:
logger.bind(tag=TAG).info("GC循环任务被取消")
raise
except Exception as e:
logger.bind(tag=TAG).error(f"GC循环任务异常: {e}")
finally:
logger.bind(tag=TAG).info("GC循环任务已退出")
async def _run_gc(self):
"""执行垃圾回收"""
try:
# 在线程池中执行GC,避免阻塞事件循环
loop = asyncio.get_running_loop()
def do_gc():
with self._lock:
before = len(gc.get_objects())
collected = gc.collect()
after = len(gc.get_objects())
return before, collected, after
before, collected, after = await loop.run_in_executor(None, do_gc)
logger.bind(tag=TAG).debug(
f"全局GC执行完成 - 回收对象: {collected}, "
f"对象数量: {before} -> {after}"
)
except Exception as e:
logger.bind(tag=TAG).error(f"执行GC时出错: {e}")
# 全局单例
_gc_manager_instance = None
def get_gc_manager(interval_seconds=300):
"""
获取全局GC管理器实例单例模式
Args:
interval_seconds: GC执行间隔默认300秒5分钟
Returns:
GlobalGCManager实例
"""
global _gc_manager_instance
if _gc_manager_instance is None:
_gc_manager_instance = GlobalGCManager(interval_seconds)
return _gc_manager_instance
@@ -102,9 +102,6 @@ class OpusEncoderUtils:
def _encode(self, frame: np.ndarray) -> Optional[bytes]:
"""编码一帧音频数据"""
try:
# 编码器已释放,跳过编码
if not hasattr(self, 'encoder') or self.encoder is None:
return None
# 将numpy数组转换为bytes
frame_bytes = frame.tobytes()
# opuslib要求输入字节数必须是channels*2的倍数
@@ -131,9 +128,5 @@ class OpusEncoderUtils:
def close(self):
"""关闭编码器并释放资源"""
if hasattr(self, 'encoder') and self.encoder:
try:
del self.encoder
self.encoder = None
except Exception as e:
logging.error(f"Error releasing Opus encoder: {e}")
# opuslib没有明确的关闭方法,Python的垃圾回收会处理
pass
@@ -4,6 +4,7 @@
"""
import os
import cnlunar
from typing import Dict, Any
from config.logger import setup_logging
from jinja2 import Template
@@ -59,11 +60,6 @@ class PromptManager:
self.cache_manager = cache_manager
self.CacheType = CacheType
# 初始化上下文源
from core.utils.context_provider import ContextDataProvider
self.context_provider = ContextDataProvider(config, self.logger)
self.context_data = {}
self._load_base_template()
@@ -188,14 +184,6 @@ class PromptManager:
local_address = self._get_location_info(client_ip)
# 获取天气信息(使用全局缓存)
self._get_weather_info(conn, local_address)
# 获取配置的上下文数据
if hasattr(conn, "device_id") and conn.device_id:
if self.base_prompt_template and "dynamic_context" in self.base_prompt_template:
self.context_data = self.context_provider.fetch_all(conn.device_id)
else:
self.context_data = ""
self.logger.bind(tag=TAG).debug(f"上下文信息更新完成")
except Exception as e:
@@ -242,7 +230,6 @@ class PromptManager:
emojiList=EMOJI_List,
device_id=device_id,
client_ip=client_ip,
dynamic_context=self.context_data,
*args,
**kwargs,
)
+51 -85
View File
@@ -4,7 +4,6 @@ import json
import copy
import wave
import socket
import asyncio
import requests
import subprocess
import numpy as np
@@ -269,82 +268,56 @@ def audio_to_data_stream(
pcm_to_data_stream(raw_data, is_opus, callback)
async def audio_to_data(
audio_file_path: str, is_opus: bool = True, use_cache: bool = True
) -> list[bytes]:
def audio_to_data(audio_file_path: str, is_opus: bool = True) -> list[bytes]:
"""
将音频文件转换为Opus/PCM编码的帧列表
Args:
audio_file_path: 音频文件路径
is_opus: 是否进行Opus编码
use_cache: 是否使用缓存
"""
from core.utils.cache.manager import cache_manager
from core.utils.cache.config import CacheType
# 获取文件后缀名
file_type = os.path.splitext(audio_file_path)[1]
if file_type:
file_type = file_type.lstrip(".")
# 读取音频文件,-nostdin 参数:不要从标准输入读取数据,否则FFmpeg会阻塞
audio = AudioSegment.from_file(
audio_file_path, format=file_type, parameters=["-nostdin"]
)
# 生成缓存键,包含文件路径和编码类型
cache_key = f"{audio_file_path}:{is_opus}"
# 转换为单声道/16kHz采样率/16位小端编码(确保与编码器匹配)
audio = audio.set_channels(1).set_frame_rate(16000).set_sample_width(2)
# 尝试从缓存获取结果
if use_cache:
cached_result = cache_manager.get(CacheType.AUDIO_DATA, cache_key)
if cached_result is not None:
return cached_result
# 获取原始PCM数据(16位小端)
raw_data = audio.raw_data
def _sync_audio_to_data():
# 获取文件后缀名
file_type = os.path.splitext(audio_file_path)[1]
if file_type:
file_type = file_type.lstrip(".")
# 读取音频文件,-nostdin 参数:不要从标准输入读取数据,否则FFmpeg会阻塞
audio = AudioSegment.from_file(
audio_file_path, format=file_type, parameters=["-nostdin"]
)
# 初始化Opus编码器
encoder = opuslib_next.Encoder(16000, 1, opuslib_next.APPLICATION_AUDIO)
# 转换为单声道/16kHz采样率/16位小端编码(确保与编码器匹配)
audio = audio.set_channels(1).set_frame_rate(16000).set_sample_width(2)
# 编码参数
frame_duration = 60 # 60ms per frame
frame_size = int(16000 * frame_duration / 1000) # 960 samples/frame
# 获取原始PCM数据(16位小端)
raw_data = audio.raw_data
datas = []
# 按帧处理所有音频数据(包括最后一帧可能补零)
for i in range(0, len(raw_data), frame_size * 2): # 16bit=2bytes/sample
# 获取当前帧的二进制数据
chunk = raw_data[i : i + frame_size * 2]
# 初始化Opus编码器
encoder = opuslib_next.Encoder(16000, 1, opuslib_next.APPLICATION_AUDIO)
# 如果最后一帧不足,补零
if len(chunk) < frame_size * 2:
chunk += b"\x00" * (frame_size * 2 - len(chunk))
# 编码参数
frame_duration = 60 # 60ms per frame
frame_size = int(16000 * frame_duration / 1000) # 960 samples/frame
if is_opus:
# 转换为numpy数组处理
np_frame = np.frombuffer(chunk, dtype=np.int16)
# 编码Opus数据
frame_data = encoder.encode(np_frame.tobytes(), frame_size)
else:
frame_data = chunk if isinstance(chunk, bytes) else bytes(chunk)
datas = []
# 按帧处理所有音频数据(包括最后一帧可能补零)
for i in range(0, len(raw_data), frame_size * 2): # 16bit=2bytes/sample
# 获取当前帧的二进制数据
chunk = raw_data[i : i + frame_size * 2]
datas.append(frame_data)
# 如果最后一帧不足,补零
if len(chunk) < frame_size * 2:
chunk += b"\x00" * (frame_size * 2 - len(chunk))
if is_opus:
# 转换为numpy数组处理
np_frame = np.frombuffer(chunk, dtype=np.int16)
# 编码Opus数据
frame_data = encoder.encode(np_frame.tobytes(), frame_size)
else:
frame_data = chunk if isinstance(chunk, bytes) else bytes(chunk)
datas.append(frame_data)
return datas
loop = asyncio.get_running_loop()
# 在单独的线程中执行同步的音频处理操作
result = await loop.run_in_executor(None, _sync_audio_to_data)
# 将结果存入缓存,使用配置中定义的TTL(10分钟)
if use_cache:
cache_manager.set(CacheType.AUDIO_DATA, cache_key, result)
return result
return datas
def audio_bytes_to_data_stream(
@@ -399,33 +372,26 @@ def opus_datas_to_wav_bytes(opus_datas, sample_rate=16000, channels=1):
将opus帧列表解码为wav字节流
"""
decoder = opuslib_next.Decoder(sample_rate, channels)
try:
pcm_datas = []
pcm_datas = []
frame_duration = 60 # ms
frame_size = int(sample_rate * frame_duration / 1000) # 960
frame_duration = 60 # ms
frame_size = int(sample_rate * frame_duration / 1000) # 960
for opus_frame in opus_datas:
# 解码为PCM(返回bytes,2字节/采样点)
pcm = decoder.decode(opus_frame, frame_size)
pcm_datas.append(pcm)
for opus_frame in opus_datas:
# 解码为PCM(返回bytes,2字节/采样点)
pcm = decoder.decode(opus_frame, frame_size)
pcm_datas.append(pcm)
pcm_bytes = b"".join(pcm_datas)
pcm_bytes = b"".join(pcm_datas)
# 写入wav字节流
wav_buffer = BytesIO()
with wave.open(wav_buffer, "wb") as wf:
wf.setnchannels(channels)
wf.setsampwidth(2) # 16bit
wf.setframerate(sample_rate)
wf.writeframes(pcm_bytes)
return wav_buffer.getvalue()
finally:
if decoder is not None:
try:
del decoder
except Exception:
pass
# 写入wav字节流
wav_buffer = BytesIO()
with wave.open(wav_buffer, "wb") as wf:
wf.setnchannels(channels)
wf.setsampwidth(2) # 16bit
wf.setframerate(sample_rate)
wf.writeframes(pcm_bytes)
return wav_buffer.getvalue()
def check_vad_update(before_config, new_config):
+9 -31
View File
@@ -1,37 +1,10 @@
import asyncio
import logging
import json
import websockets
from config.logger import setup_logging
class SuppressInvalidHandshakeFilter(logging.Filter):
"""过滤掉无效握手错误日志(如HTTPS访问WS端口)"""
def filter(self, record):
msg = record.getMessage()
suppress_keywords = [
"opening handshake failed",
"did not receive a valid HTTP request",
"connection closed while reading HTTP request",
"line without CRLF",
]
return not any(keyword in msg for keyword in suppress_keywords)
def _setup_websockets_logger():
"""配置 websockets 相关的所有 logger,过滤无效握手错误"""
filter_instance = SuppressInvalidHandshakeFilter()
for logger_name in ["websockets", "websockets.server", "websockets.client"]:
logger = logging.getLogger(logger_name)
logger.addFilter(filter_instance)
_setup_websockets_logger()
from core.connection import ConnectionHandler
from config.config_loader import get_config_from_api_async
from config.config_loader import get_config_from_api
from core.auth import AuthManager, AuthenticationError
from core.utils.modules_initialize import initialize_modules
from core.utils.util import check_vad_update, check_asr_update
@@ -60,6 +33,8 @@ class WebSocketServer:
self._intent = modules["intent"] if "intent" in modules else None
self._memory = modules["memory"] if "memory" in modules else None
self.active_connections = set()
auth_config = self.config["server"].get("auth", {})
self.auth_enable = auth_config.get("enabled", False)
# 设备白名单
@@ -123,11 +98,14 @@ class WebSocketServer:
self._intent,
self, # 传入server实例
)
self.active_connections.add(handler)
try:
await handler.handle_connection(websocket)
except Exception as e:
self.logger.bind(tag=TAG).error(f"处理连接时出错: {e}")
finally:
# 确保从活动连接集合中移除
self.active_connections.discard(handler)
# 强制关闭连接(如果还没有关闭的话)
try:
# 安全地检查WebSocket状态并关闭
@@ -160,8 +138,8 @@ class WebSocketServer:
"""
try:
async with self.config_lock:
# 重新获取配置(使用异步版本)
new_config = await get_config_from_api_async(self.config)
# 重新获取配置
new_config = get_config_from_api(self.config)
if new_config is None:
self.logger.bind(tag=TAG).error("获取新配置失败")
return False
@@ -50,9 +50,7 @@ class BaseASRTester:
raise NotImplementedError
def _calculate_result(self, service_name, latencies, test_count):
"""计算测试结果(修复:正确处理None值,剔除失败测试)"""
# 剔除None值(失败的测试)和无效延迟,只统计有效延迟
valid_latencies = [l for l in latencies if l is not None and l > 0]
valid_latencies = [l for l in latencies if l > 0]
if valid_latencies:
avg_latency = sum(valid_latencies) / len(valid_latencies)
status = f"成功({len(valid_latencies)}/{test_count}次有效)"
@@ -66,45 +64,16 @@ class DoubaoStreamASRTester(BaseASRTester):
def __init__(self):
super().__init__("DoubaoStreamASR")
def _generate_header(
self,
version=0x01,
message_type=0x01,
message_type_specific_flags=0x00,
serial_method=0x01,
compression_type=0x01,
reserved_data=0x00,
extension_header: bytes = b"",
):
"""生成协议头(修复:使用正确的Header格式)"""
def _generate_header(self):
header = bytearray()
header_size = int(len(extension_header) / 4) + 1
header.append((version << 4) | header_size)
header.append((message_type << 4) | message_type_specific_flags)
header.append((serial_method << 4) | compression_type)
header.append(reserved_data)
header.extend(extension_header)
header.append((0x01 << 4) | 0x01)
header.append((0x01 << 4) | 0x00)
header.append((0x01 << 4) | 0x01)
header.append(0x00)
return header
def _generate_audio_default_header(self):
"""生成音频数据Header"""
return self._generate_header(
version=0x01,
message_type=0x02,
message_type_specific_flags=0x00, # 普通音频帧
serial_method=0x01,
compression_type=0x01,
)
def _generate_last_audio_header(self):
"""生成最后一帧音频的Header(标记音频结束)"""
return self._generate_header(
version=0x01,
message_type=0x02,
message_type_specific_flags=0x02, # 0x02表示这是最后一帧
serial_method=0x01,
compression_type=0x01,
)
return self._generate_header()
def _parse_response(self, res: bytes) -> dict:
try:
@@ -141,7 +110,6 @@ class DoubaoStreamASRTester(BaseASRTester):
ws_url = "wss://openspeech.bytedance.com/api/v3/sauc/bigmodel"
appid = self.asr_config["appid"]
access_token = self.asr_config["access_token"]
cluster = self.asr_config.get("cluster", "volcengine_input_common")
uid = self.asr_config.get("uid", "streaming_asr_service")
start_time = time.time()
@@ -162,7 +130,7 @@ class DoubaoStreamASRTester(BaseASRTester):
close_timeout=10
) as ws:
request_params = {
"app": {"appid": appid, "cluster": cluster, "token": access_token},
"app": {"appid": appid, "token": access_token},
"user": {"uid": uid},
"request": {
"reqid": str(uuid.uuid4()),
@@ -198,9 +166,8 @@ class DoubaoStreamASRTester(BaseASRTester):
if audio_data.startswith(b'RIFF'):
audio_data = audio_data[44:]
# 发送音频数据(使用最后一帧标记,告诉服务端音频已结束)
payload = gzip.compress(audio_data)
audio_request = bytearray(self._generate_last_audio_header()) # 修复:使用最后一帧Header
audio_request = bytearray(self._generate_audio_default_header())
audio_request.extend(len(payload).to_bytes(4, "big"))
audio_request.extend(payload)
await ws.send(audio_request)
@@ -208,12 +175,11 @@ class DoubaoStreamASRTester(BaseASRTester):
first_chunk = await ws.recv()
latency = time.time() - start_time
latencies.append(latency)
print(f"[豆包ASR] 第{i+1}次 首词延迟: {latency:.3f}s")
await ws.close()
except Exception as e:
print(f"[豆包ASR] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("豆包流式ASR", latencies, test_count)
@@ -223,12 +189,11 @@ class QwenASRFlashTester(BaseASRTester):
super().__init__("Qwen3ASRFlash")
async def _test_single(self, audio_file_info):
start_time = time.time()
temp_file_path = None
try:
audio_data = audio_file_info['data']
# 优化:将临时文件准备工作移到计时前,减少磁盘IO对性能测试的影响
with tempfile.NamedTemporaryFile(suffix='.wav', delete=False) as f:
temp_file_path = f.name
@@ -256,9 +221,6 @@ class QwenASRFlashTester(BaseASRTester):
dashscope.api_key = api_key
# 统一计时起点:在API调用前开始计时(但文件准备已完成)
start_time = time.time()
response = dashscope.MultiModalConversation.call(
model="qwen3-asr-flash",
messages=messages,
@@ -295,10 +257,10 @@ class QwenASRFlashTester(BaseASRTester):
# print(f"\n[通义ASR] 开始第 {i+1} 次测试...")
latency = await self._test_single(self.test_audio_files[0])
latencies.append(latency)
print(f"[通义ASR] 第{i+1}次 首词延迟: {latency:.3f}s")
# print(f"[通义ASR] 第{i+1}次成功 延迟: {latency:.3f}s")
except Exception as e:
# print(f"[通义ASR] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("通义千问ASR", latencies, test_count)
@@ -306,115 +268,134 @@ class QwenASRFlashTester(BaseASRTester):
class XunfeiStreamASRTester(BaseASRTester):
def __init__(self):
super().__init__("XunfeiStreamASR")
def _create_url(self):
url = "wss://iat-api.xfyun.cn/v2/iat"
"""生成讯飞ASR认证URL"""
url = 'ws://iat.cn-huabei-1.xf-yun.com/v1'
# 生成RFC1123格式的时间戳
now = datetime.now()
date = format_date_time(mktime(now.timetuple()))
signature_origin = f"host: iat-api.xfyun.cn\ndate: {date}\nGET /v2/iat HTTP/1.1"
signature_sha = hmac.new(
self.asr_config["api_secret"].encode('utf-8'),
signature_origin.encode('utf-8'),
hashlib.sha256
).digest()
signature_sha = base64.b64encode(signature_sha).decode()
# 拼接字符串
signature_origin = "host: " + "iat.cn-huabei-1.xf-yun.com" + "\n"
signature_origin += "date: " + date + "\n"
signature_origin += "GET " + "/v1 " + "HTTP/1.1"
authorization_origin = f'api_key="{self.asr_config["api_key"]}", algorithm="hmac-sha256", headers="host date request-line", signature="{signature_sha}"'
authorization = base64.b64encode(authorization_origin.encode()).decode()
# 进行hmac-sha256进行加密
signature_sha = hmac.new(self.asr_config["api_secret"].encode('utf-8'), signature_origin.encode('utf-8'),
digestmod=hashlib.sha256).digest()
signature_sha = base64.b64encode(signature_sha).decode(encoding='utf-8')
v = {"authorization": authorization, "date": date, "host": "iat-api.xfyun.cn"}
return url + "?" + parse.urlencode(v)
authorization_origin = "api_key=\"%s\", algorithm=\"%s\", headers=\"%s\", signature=\"%s\"" % (
self.asr_config["api_key"], "hmac-sha256", "host date request-line", signature_sha)
authorization = base64.b64encode(authorization_origin.encode('utf-8')).decode(encoding='utf-8')
async def test(self, test_count: int = 5):
# 将请求的鉴权参数组合为字典
v = {
"authorization": authorization,
"date": date,
"host": "iat.cn-huabei-1.xf-yun.com"
}
# 拼接鉴权参数,生成url
url = url + '?' + parse.urlencode(v)
return url
async def test(self, test_count=5):
if not self.test_audio_files:
return {"name": "讯飞流式ASR", "latency": 0, "status": "失败: 未找到测试音频"}
if not self.asr_config:
return {"name": "讯飞流式ASR", "latency": 0, "status": "失败: 未配置"}
required = ["app_id", "api_key", "api_secret"]
for k in required:
if k not in self.asr_config:
return {"name": "讯飞流式ASR", "latency": 0, "status": f"失败: 缺少配置 {k}"}
# 检查必要的配置参数
required_keys = ["app_id", "api_key", "api_secret"]
for key in required_keys:
if key not in self.asr_config:
return {"name": "讯飞流式ASR", "latency": 0, "status": f"失败: 缺少配置项 {key}"}
latencies = []
frame_size = 1280
audio_raw = self.test_audio_files[0]['data']
if audio_raw.startswith(b'RIFF'):
audio_raw = audio_raw[44:]
STATUS_FIRST_FRAME = 0
for i in range(test_count):
try:
start_time = time.time()
# 生成认证URL
ws_url = self._create_url()
# 获取音频数据
audio_data = self.test_audio_files[0]['data']
if audio_data.startswith(b'RIFF'):
audio_data = audio_data[44:] # 跳过WAV文件头
# 识别参数
iat_params = {
"domain": self.asr_config.get("domain", "slm"),
"language": self.asr_config.get("language", "zh_cn"),
"accent": self.asr_config.get("accent", "mandarin"),
"dwa": self.asr_config.get("dwa", "wpgs"),
"result": {
"encoding": "utf8",
"compress": "raw",
"format": "plain"
}
}
# 准备首帧数据
first_frame_data = {
"header": {
"status": STATUS_FIRST_FRAME,
"app_id": self.asr_config["app_id"]
},
"parameter": {
"iat": iat_params
},
"payload": {
"audio": {
"audio": base64.b64encode(audio_data[:960]).decode('utf-8'),
"sample_rate": 16000,
"encoding": "raw"
}
}
}
# 启动连接并测量时间
start_time = time.time()
async with websockets.connect(
ws_url,
additional_headers={"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64)"},
max_size=1 << 30,
max_size=1000000000,
ping_interval=None,
ping_timeout=None,
close_timeout=30,
) as ws:
# 第一帧:移除 punc 字段,避免未知参数错误
await ws.send(json.dumps({
"common": {"app_id": self.asr_config["app_id"]},
"business": {
"domain": "iat",
"language": "zh_cn",
"accent": "mandarin",
"dwa": "wpgs",
"vad_eos": 5000
# 已移除 "punc": True
},
"data": {
"status": 0,
"format": "audio/L16;rate=16000",
"encoding": "raw",
"audio": base64.b64encode(audio_raw[:frame_size]).decode()
}
}, ensure_ascii=False))
# 后续所有帧
pos = frame_size
while pos < len(audio_raw):
chunk = audio_raw[pos:pos + frame_size]
status = 2 if (pos + frame_size >= len(audio_raw)) else 1
await ws.send(json.dumps({
"data": {
"status": status,
"format": "audio/L16;rate=16000",
"encoding": "raw",
"audio": base64.b64encode(chunk).decode()
}
}, ensure_ascii=False))
if status == 2:
# 发送首帧数据
await ws.send(json.dumps(first_frame_data, ensure_ascii=False))
print(f"[讯飞ASR] 第{i+1}次测试:已发送首帧,等待响应...")
# 直接等待第一个响应并计算延迟
# 参考豆包和通义千问的实现方式,简化逻辑
response_received = False
while not response_received:
try:
# 设置较大的超时时间
response = await asyncio.wait_for(ws.recv(), timeout=30.0)
# 收到响应立即计算延迟,不管内容是什么
# 这样可以准确测量首包到达时间
latency = time.time() - start_time
latencies.append(latency)
response_received = True
print(f"[讯飞ASR] 第{i+1}次测试:收到首包响应,延迟: {latency:.3f}s")
break
pos += frame_size
# 接收首词
first_token = True
async for message in ws:
data = json.loads(message)
if data.get("code") != 0:
raise Exception(f"讯飞错误: {data.get('message')}")
ws_result = data.get("data", {}).get("result", {}).get("ws")
if ws_result:
text = "".join(cw.get("w", "") for seg in ws_result for cw in seg.get("cw", []))
if text.strip() and first_token:
latency = time.time() - start_time
latencies.append(latency)
print(f"[讯飞ASR] 第{i+1}次 首词延迟: {latency:.3f}s")
first_token = False
break
except asyncio.TimeoutError:
print(f"[讯飞ASR] 第{i+1}次测试:响应超时")
raise Exception("获取响应超时")
except Exception as e:
print(f"[讯飞ASR] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("讯飞流式ASR", latencies, test_count)
class ASRPerformanceSuite:
def __init__(self):
self.testers = []
@@ -457,9 +438,8 @@ class ASRPerformanceSuite:
print(tabulate(table_data, headers=["ASR服务", "首词延迟", "状态"], tablefmt="grid"))
print("\n测试说明:")
print("- 计时起点: 建立连接前(包含握手、发送音频、接收首个识别结果全流程)")
print("- 通义千问优化: 临时文件准备在计时前完成,减少磁盘IO对测试的影响")
print("- 错误处理: 失败的测试不计入平均值,只统计成功测试的延迟")
print("- 测量从发送请求到接收第一个有效识别文本的时间")
print("- 超时控制: DashScope 默认超时,豆包 WebSocket 超时10秒")
print("- 排序规则: 成功的按延迟升序,失败的排在后面")
async def run(self, test_count=5):
@@ -35,12 +35,11 @@ class StreamTTSPerformanceTester:
host = tts_config["host"]
ws_url = f"wss://{host}/ws/v1"
# 统一计时起点:在建立连接前开始计时
start_time = time.time()
async with websockets.connect(ws_url, extra_headers={"X-NLS-Token": token}) as ws:
task_id = str(uuid.uuid4())
message_id = str(uuid.uuid4())
start_request = {
"header": {
"message_id": message_id,
@@ -56,15 +55,14 @@ class StreamTTSPerformanceTester:
"volume": 50,
"speech_rate": 0,
"pitch_rate": 0,
"enable_subtitle": True,
}
}
await ws.send(json.dumps(start_request))
start_response = json.loads(await ws.recv())
if start_response["header"]["name"] != "SynthesisStarted":
raise Exception("启动合成失败")
run_request = {
"header": {
"message_id": str(uuid.uuid4()),
@@ -76,142 +74,23 @@ class StreamTTSPerformanceTester:
"payload": {"text": text}
}
await ws.send(json.dumps(run_request))
while True:
response = await ws.recv()
if isinstance(response, bytes):
latency = time.time() - start_time
latencies.append(latency)
print(f"[阿里云TTS] 第{i+1}次 首词延迟: {latency:.3f}s")
break
elif isinstance(response, str):
data = json.loads(response)
if data["header"]["name"] == "TaskFailed":
raise Exception(f"合成失败: {data['payload']['error_info']}")
except Exception as e:
print(f"[阿里云TTS] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("阿里云TTS", latencies, test_count)
async def test_alibl_tts(self, text=None, test_count=5):
"""测试阿里云百炼CosyVoice流式TTS首词延迟"""
text = text or self.test_texts[0]
latencies = []
for i in range(test_count):
try:
tts_config = self.config["TTS"]["AliBLTTS"]
api_key = tts_config["api_key"]
model = tts_config.get("model", "cosyvoice-v2")
voice = tts_config.get("voice", "longxiaochun_v2")
format_type = tts_config.get("format", "pcm")
sample_rate = int(tts_config.get("sample_rate", "24000"))
ws_url = "wss://dashscope.aliyuncs.com/api-ws/v1/inference/"
headers = {
"Authorization": f"Bearer {api_key}",
"X-DashScope-DataInspection": "enable",
}
start_time = time.time()
async with websockets.connect(
ws_url,
additional_headers=headers,
ping_interval=30,
ping_timeout=10,
close_timeout=10,
max_size=10 * 1024 * 1024,
) as ws:
session_id = uuid.uuid4().hex
# 1. 发送 run-task(启动任务)
run_task_message = {
"header": {
"action": "run-task",
"task_id": session_id,
"streaming": "duplex",
},
"payload": {
"task_group": "audio",
"task": "tts",
"function": "SpeechSynthesizer",
"model": model,
"parameters": {
"text_type": "PlainText",
"voice": voice,
"format": format_type,
"sample_rate": sample_rate,
"volume": 50,
"rate": 1.0,
"pitch": 1.0,
},
"input": {}
},
}
await ws.send(json.dumps(run_task_message))
# 2. 等待 task-started 事件(关键!必须等这个再发文本)
task_started = False
while not task_started:
msg = await ws.recv()
if isinstance(msg, str):
data = json.loads(msg)
header = data.get("header", {})
event = header.get("event")
if event == "task-started":
task_started = True
print(f"[阿里云百炼TTS] 第{i+1}次 任务启动成功")
elif event == "task-failed":
raise Exception(f"启动失败: {header.get('error_message', '未知错误')}")
# 3. 发送 continue-task(发送文本!这是正确动作)
continue_task_message = {
"header": {
"action": "continue-task", # 改回 continue-task
"task_id": session_id,
"streaming": "duplex",
},
"payload": {"input": {"text": text}},
}
await ws.send(json.dumps(continue_task_message))
# 4. 发送 finish-task(结束任务)
finish_task_message = {
"header": {
"action": "finish-task",
"task_id": session_id,
"streaming": "duplex",
},
"payload": {"input": {}}
}
await ws.send(json.dumps(finish_task_message))
# 5. 等待第一个音频数据块
while True:
msg = await asyncio.wait_for(ws.recv(), timeout=15.0)
if isinstance(msg, (bytes, bytearray)) and len(msg) > 0:
latency = time.time() - start_time
print(f"[阿里云百炼TTS] 第{i+1}次 首词延迟: {latency:.3f}s")
latencies.append(latency)
break
elif isinstance(msg, str):
data = json.loads(msg)
event = data.get("header", {}).get("event")
if event == "task-failed":
raise Exception(f"合成失败: {data}")
elif event == "task-finished":
if not latencies or latencies[-1] is None:
raise Exception("任务结束但未收到音频")
except Exception as e:
print(f"[阿里云百炼TTS] 第{i+1}次失败: {str(e)}")
latencies.append(None)
return self._calculate_result("阿里云百炼TTS", latencies, test_count)
async def test_doubao_tts(self, text=None, test_count=5):
"""测试火山引擎流式TTS首词延迟(测试多次取平均)"""
text = text or self.test_texts[0]
@@ -235,12 +114,13 @@ class StreamTTSPerformanceTester:
}
async with websockets.connect(ws_url, additional_headers=ws_header, max_size=1000000000) as ws:
session_id = uuid.uuid4().hex
# 发送会话启动请求
header = bytes([
(0b0001 << 4) | 0b0001,
0b0001 << 4 | 0b1011,
0b0001 << 4 | 0b0000,
(0b0001 << 4) | 0b0001,
0b0001 << 4 | 0b100,
0b0001 << 4 | 0b0000,
0
])
optional = bytearray()
optional.extend((1).to_bytes(4, "big", signed=True))
@@ -249,13 +129,13 @@ class StreamTTSPerformanceTester:
optional.extend(session_id_bytes)
payload = json.dumps({"speaker": speaker}).encode()
await ws.send(header + optional + len(payload).to_bytes(4, "big", signed=True) + payload)
# 发送文本
header = bytes([
(0b0001 << 4) | 0b0001,
0b0001 << 4 | 0b1011,
0b0001 << 4 | 0b0000,
0
(0b0001 << 4) | 0b0001,
0b0001 << 4 | 0b100,
0b0001 << 4 | 0b0000,
0
])
optional = bytearray()
optional.extend((200).to_bytes(4, "big", signed=True))
@@ -264,15 +144,13 @@ class StreamTTSPerformanceTester:
optional.extend(session_id_bytes)
payload = json.dumps({"text": text, "speaker": speaker}).encode()
await ws.send(header + optional + len(payload).to_bytes(4, "big", signed=True) + payload)
first_chunk = await ws.recv()
latency = time.time() - start_time
latencies.append(latency)
print(f"[火山引擎TTS] 第{i+1}次 首词延迟: {latency:.3f}s")
except Exception as e:
print(f"[火山引擎TTS] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("火山引擎TTS", latencies, test_count)
@@ -313,24 +191,22 @@ class StreamTTSPerformanceTester:
first_chunk = await ws.recv()
latency = time.time() - start_time
latencies.append(latency)
print(f"[PaddleSpeechTTS] 第{i+1}次 首词延迟: {latency:.3f}s")
# 发送结束请求
end_request = {
"task": "tts",
"signal": "end"
}
await ws.send(json.dumps(end_request))
# 确保连接正常关闭
try:
await ws.recv()
except websockets.exceptions.ConnectionClosedOK:
pass
except Exception as e:
print(f"[PaddleSpeechTTS] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("PaddleSpeechTTS", latencies, test_count)
@@ -344,32 +220,29 @@ class StreamTTSPerformanceTester:
tts_config = self.config["TTS"]["IndexStreamTTS"]
api_url = tts_config.get("api_url")
voice = tts_config.get("voice")
# 统一计时起点:在建立连接前开始计时
start_time = time.time()
async with aiohttp.ClientSession() as session:
payload = {"text": text, "character": voice}
async with session.post(api_url, json=payload, timeout=10) as resp:
if resp.status != 200:
raise Exception(f"请求失败: {resp.status}, {await resp.text()}")
async for chunk in resp.content.iter_any():
data = chunk[0] if isinstance(chunk, (list, tuple)) else chunk
if not data:
continue
latency = time.time() - start_time
latencies.append(latency)
print(f"[IndexStreamTTS] 第{i+1}次 首词延迟: {latency:.3f}s")
resp.close()
break
else:
latencies.append(None)
latencies.append(0)
except Exception as e:
print(f"[IndexStreamTTS] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("IndexStreamTTS", latencies, test_count)
@@ -384,8 +257,7 @@ class StreamTTSPerformanceTester:
api_url = tts_config["api_url"]
access_token = tts_config["access_token"]
voice = tts_config["voice"]
# 统一计时起点:在建立连接前开始计时
start_time = time.time()
async with aiohttp.ClientSession() as session:
params = {
@@ -401,23 +273,21 @@ class StreamTTSPerformanceTester:
"Authorization": f"Bearer {access_token}",
"Content-Type": "application/json",
}
async with session.get(api_url, params=params, headers=headers, timeout=10) as resp:
if resp.status != 200:
raise Exception(f"请求失败: {resp.status}, {await resp.text()}")
# 接收第一个数据块
async for _ in resp.content.iter_any():
latency = time.time() - start_time
latencies.append(latency)
print(f"[LinkeraiTTS] 第{i+1}次 首词延迟: {latency:.3f}s")
break
else:
latencies.append(None)
latencies.append(0)
except Exception as e:
print(f"[LinkeraiTTS] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("LinkeraiTTS", latencies, test_count)
@@ -435,9 +305,10 @@ class StreamTTSPerformanceTester:
api_secret = tts_config["api_secret"]
api_url = tts_config.get("api_url", "wss://cbm01.cn-huabei-1.xf-yun.com/v1/private/mcd9m97e6")
voice = tts_config.get("voice", "x5_lingxiaoxuan_flow")
# 生成认证URL
auth_url = self._create_xunfei_auth_url(api_key, api_secret, api_url)
start_time = time.time()
async with websockets.connect(
auth_url,
ping_interval=30,
@@ -447,7 +318,10 @@ class StreamTTSPerformanceTester:
) as ws:
# 构造请求
request = self._build_xunfei_request(app_id, text, voice)
# 发送请求后立即计时,确保准确测量从发送文本到接收首块的时间
await ws.send(json.dumps(request))
start_time = time.time()
# 等待第一个音频数据块
first_audio_received = False
while not first_audio_received:
@@ -455,14 +329,14 @@ class StreamTTSPerformanceTester:
data = json.loads(msg)
header = data.get("header", {})
code = header.get("code")
if code != 0:
message = header.get("message", "未知错误")
raise Exception(f"合成失败: {code} - {message}")
payload = data.get("payload", {})
audio_payload = payload.get("audio", {})
if audio_payload:
status = audio_payload.get("status", 0)
audio_data = audio_payload.get("audio", "")
@@ -470,12 +344,10 @@ class StreamTTSPerformanceTester:
# 收到第一个音频数据块
latency = time.time() - start_time
latencies.append(latency)
print(f"[讯飞TTS] 第{i+1}次 首词延迟: {latency:.3f}s")
first_audio_received = True
break
except Exception as e:
print(f"[讯飞TTS] 第{i+1}次测试失败: {str(e)}")
latencies.append(None)
latencies.append(0)
return self._calculate_result("讯飞TTS", latencies, test_count)
@@ -559,9 +431,8 @@ class StreamTTSPerformanceTester:
def _calculate_result(self, service_name, latencies, test_count):
"""计算测试结果(正确处理None值,剔除失败测试)"""
# 剔除失败的测试(None值和<=0延迟),只统计有效延迟
valid_latencies = [l for l in latencies if l is not None and l > 0]
"""计算测试结果"""
valid_latencies = [l for l in latencies if l > 0]
if valid_latencies:
avg_latency = sum(valid_latencies) / len(valid_latencies)
status = f"成功({len(valid_latencies)}/{test_count}次有效)"
@@ -595,10 +466,9 @@ class StreamTTSPerformanceTester:
]
print(tabulate(table_data, headers=["TTS服务", "首词延迟(秒)", "状态"], tablefmt="grid"))
print("\n测试说明:测量从建立连接到接收第一个音频数据块的时间(包含握手、鉴权、发送文本),取多次测试平均值")
print("- 计时起点: 建立WebSocket/HTTP连接前(统一包含网络建连、握手、发送文本全流程)")
print("\n测试说明:测量从发送请求到接收第一个音频数据块的时间,取多次测试平均值")
print("- 超时控制: 单个请求最大等待时间为10秒")
print("- 错误处理: 失败的测试不计入平均值,只统计成功测试的延迟")
print("- 错误处理: 无法连接和超时的列为网络错误")
print("- 排序规则: 按平均耗时从快到慢排序")
@@ -624,12 +494,7 @@ class StreamTTSPerformanceTester:
# 测试阿里云TTS
result = await self.test_aliyun_tts(test_text, test_count)
self.results.append(result)
# 测试阿里云百炼TTS
if self.config.get("TTS", {}).get("AliBLTTS"):
result = await self.test_alibl_tts(test_text, test_count)
self.results.append(result)
# 测试火山引擎TTS
result = await self.test_doubao_tts(test_text, test_count)
self.results.append(result)
+5 -6
View File
@@ -1,4 +1,3 @@
# -*- coding:utf-8 -*-
#--------- 本项目推荐环境是python3.10,以下暂时不推荐升级的依赖
torch==2.2.2
torchaudio==2.2.2
@@ -11,9 +10,9 @@ silero_vad==6.1.0
opuslib_next==1.1.5
pydub==0.25.1
funasr==1.2.7
openai==2.8.1
openai==2.7.1
google-generativeai==0.8.5
edge_tts==7.2.6
edge_tts==7.2.3
httpx==0.28.1
aiohttp==3.13.2
aiohttp_cors==0.8.1
@@ -25,11 +24,11 @@ cozepy==0.20.0
mem0ai==1.0.0
bs4==0.0.2
modelscope==1.23.2
sherpa_onnx==1.12.17
sherpa_onnx==1.12.15
mcp==1.20.0
cnlunar==0.2.0
PySocks==1.7.1
dashscope==1.25.2
dashscope==1.24.6
baidu-aip==4.16.13
chardet==5.2.0
aioconsole==0.8.2
@@ -39,4 +38,4 @@ PyJWT==2.10.1
psutil==7.0.0
portalocker==3.2.0
Jinja2==3.1.6
vosk==0.3.45
vosk==0.3.44
@@ -240,24 +240,6 @@ export class AudioRecorder {
if (this.isRecording) return false;
try {
// 检查是否有WebSocketHandler实例
const { getWebSocketHandler } = await import('../network/websocket.js');
const wsHandler = getWebSocketHandler();
// 如果机器正在说话,发送打断消息
if (wsHandler && wsHandler.isRemoteSpeaking && wsHandler.currentSessionId) {
const abortMessage = {
session_id: wsHandler.currentSessionId,
type: 'abort',
reason: 'wake_word_detected'
};
if (this.websocket && this.websocket.readyState === WebSocket.OPEN) {
this.websocket.send(JSON.stringify(abortMessage));
log('发送打断消息', 'info');
}
}
if (!this.initEncoder()) {
log('无法启动录音: Opus编码器初始化失败', 'error');
return false;