fix(sse): SSE 改为按会话隔离 + 修 trace 租户 bug + 清理冗余

SSE 串台修复： - SseEmitterManager 新增按 sessionId 维度的 connect/sendEvent/disconnect，每个会话一个 SSE 连接，替代原 userId+token 维度（同用户多会话串台） - SseMessageDto 增加 sessionId + eventDto 字段，跨实例按会话路由 - SseTopicListener 优先按 sessionId 路由，回退原 userId/群发逻辑 - SseMessageUtils 增加 sessionId 重载，对话链路全部切换 - ChatServiceFacade / MyMcpClientListener / WorkflowStarter 切到 sessionId - 通知/全局 /sse 端点保留 userId 模式（通知按用户） trace 租户 bug 修复： - trace_run / trace_node 加入 tenant.excludes，监控表跨租户全局可见，绕过异步线程租户上下文不传播导致 node 查不到的问题冗余清理： - 删除零使用的 @TraceNode 注解 + TraceNodeAspect 切面 + 对应单测 - TraceProperties 删除 recordDetail/maxInputLength/maxOutputLength （仅对已删切面生效，编程式埋点用 RagTracePayloadBuilder 不受影响） - TracePayloadUtils 删除 input/output/asString 方法 - TraceConstants 删除无引用的 NODE_METHOD/HTTP/DB/CACHE/TASK/STREAM - TraceContext 删除预留的 copyNodeStack - TraceNodeVo 删除未用的 children 字段 - TraceRecordServiceImpl.getDetail 删除重复 enrich - application.yml 删除 trace.payload 无效配置项 - RagTraceNodeTypes 删除无引用的 NODE_STREAM Co-Authored-By: Claude <noreply@anthropic.com>
2026-09-13 00:14:59 +00:00 · 2026-07-23 15:35:10 +08:00
parent 11bb1dba0f
commit d1a820728f
21 changed files with 259 additions and 405 deletions
--- a/ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyMcpClientListener.java
+++ b/ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/observability/MyMcpClientListener.java
@@ -41,14 +41,14 @@ public class MyMcpClientListener implements McpClientListener {

    private static final ObjectMapper OBJECT_MAPPER = new ObjectMapper();

-    private final Long userId;
+    private final String sessionId;

-    public MyMcpClientListener(Long userId) {
-        this.userId = userId;
+    public MyMcpClientListener(String sessionId) {
+        this.sessionId = sessionId;
    }

    public MyMcpClientListener() {
-        this.userId = null;
+        this.sessionId = null;
    }

    // ==================== 工具执行 ====================
@@ -146,8 +146,8 @@ public class MyMcpClientListener implements McpClientListener {
     * 推送 MCP 事件到前端
     */
    private void pushMcpEvent(String name, String status, String result) {
-        if (userId == null) {
-            log.warn("userId 为空，无法推送 MCP 事件");
+        if (sessionId == null) {
+            log.warn("sessionId 为空，无法推送 MCP 事件");
            return;
        }
        try {
@@ -157,7 +157,7 @@ public class MyMcpClientListener implements McpClientListener {
            content.put("result", result);

            String json = OBJECT_MAPPER.writeValueAsString(content);
-            SseMessageUtils.sendEvent(userId, SseEventDto.builder()
+            SseMessageUtils.sendEvent(sessionId, SseEventDto.builder()
                .event("mcp")
                .content(json)
                .build());
--- a/ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/ChatServiceFacade.java
+++ b/ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/service/chat/impl/ChatServiceFacade.java
@@ -72,7 +72,6 @@ import org.ruoyi.domain.vo.agent.AgentVo;
 import org.ruoyi.domain.vo.knowledge.KnowledgeInfoVo;
 import org.ruoyi.factory.ChatServiceFactory;
 import org.ruoyi.mcp.service.core.LangChain4jMcpToolProviderService;
-import org.ruoyi.mcp.service.core.ToolProviderFactory;
 import org.ruoyi.observability.*;
 import org.ruoyi.service.agent.IAgentService;
 import org.ruoyi.service.chat.AbstractChatService;
@@ -81,7 +80,6 @@ import org.ruoyi.service.chat.impl.memory.PersistentChatMemoryStore;
 import org.ruoyi.service.knowledge.IKnowledgeInfoService;
 import org.ruoyi.service.retrieval.KnowledgeRetrievalService;
 import org.ruoyi.service.knowledge.retriever.CustomVectorRetriever;
-import org.ruoyi.service.vector.VectorStoreService;
 import org.ruoyi.trace.RagTraceNodeTypes;
 import org.ruoyi.trace.RagTracePayloadBuilder;
 import org.springframework.stereotype.Service;
@@ -120,8 +118,6 @@ public class ChatServiceFacade implements IChatService {

    private final IKnowledgeInfoService knowledgeInfoService;

-    private final VectorStoreService vectorStoreService;
-
    private final KnowledgeRetrievalService knowledgeRetrievalService;

    private final SseEmitterManager sseEmitterManager;
@@ -130,7 +126,6 @@ public class ChatServiceFacade implements IChatService {

    private final IWorkFlowStarterService workFlowStarterService;

-    private final ToolProviderFactory toolProviderFactory;

    private final IAgentService agentService;

@@ -156,13 +151,13 @@ public class ChatServiceFacade implements IChatService {
     */
    public SseEmitter sseChat(ChatRequest chatRequest) {

-        // 4. 具体的服务实现
+        // 具体的服务实现
        Long userId = LoginHelper.getUserId();
        String tokenValue = StpUtil.getTokenValue();
-        SseEmitter emitter = sseEmitterManager.connect(userId, tokenValue);
+        // 每个会话一个 SSE 连接，避免同用户多会话串台
+        SseEmitter emitter = sseEmitterManager.connect(String.valueOf(chatRequest.getSessionId()));

-        // 0. 智能体解析：传入 agentId 时按智能体绑定的模型覆盖 model 字段
-        //    （前端默认走智能体；enableThinking 不再作为对话模式开关，Supervisor 多 Agent 编排成为默认智能体路径）
+        // 智能体解析：传入 agentId 时按智能体绑定的模型覆盖 model 字段
        AgentVo agentVo = null;
        if (chatRequest.getAgentId() != null) {
            agentVo = agentService.queryById(chatRequest.getAgentId());
@@ -176,14 +171,14 @@ public class ChatServiceFacade implements IChatService {
            }
        }

-        // 1. 根据模型名称查询完整配置
+        // 根据模型名称查询完整配置
        ChatModelVo chatModelVo = chatModelService.selectModelByName(chatRequest.getModel());
        if (chatModelVo == null) {
            throw new IllegalArgumentException("模型不存在: " + chatRequest.getModel());
        }

-        // 2. 构建上下文消息列表（系统提示词 + 历史消息 + 当前用户消息）
-        //    注意：RAG 检索增强统一在 handleAgentChat 中执行一次，此处不再重复检索
+        // 构建上下文消息列表（系统提示词 + 历史消息 + 当前用户消息）
+        // 注意：RAG 检索增强统一在 handleAgentChat 中执行一次，此处不再重复检索
        List<ChatMessage> contextMessages = buildContextMessages(chatRequest, agentVo);

        chatRequest.setEmitter(emitter);
@@ -242,13 +237,14 @@ public class ChatServiceFacade implements IChatService {
        ChatModel plannerModel = chatService.buildChatModel(chatModelVo);

        Long userId = chatRequest.getUserId();
+        String sessionId = String.valueOf(chatRequest.getSessionId());

        // 工具装配：智能体有关联工具ID时按ID装配，否则回退到原有硬编码 MCP 客户端
        ToolProvider toolProvider;
        if (agentVo != null && agentVo.getMcpToolIds() != null && !agentVo.getMcpToolIds().isEmpty()) {
            toolProvider = langChain4jMcpToolProviderService.getToolProvider(agentVo.getMcpToolIds());
        } else {
-            toolProvider = buildDefaultMcpToolProvider(userId);
+            toolProvider = buildDefaultMcpToolProvider(sessionId);
        }

        // Skills 装配：智能体有勾选技能名时按名过滤磁盘 skills，否则加载全部
@@ -321,16 +317,14 @@ public class ChatServiceFacade implements IChatService {
        promptBuilder.append(augmentedInput);
        String prompt = promptBuilder.toString();

-        String tokenValue = chatRequest.getTokenValue();
-
        // 异步执行 supervisor，避免阻塞 HTTP 请求线程导致 SSE 事件被缓冲
        CompletableFuture.runAsync(() -> {
            TraceStreamSpan llmSpan = null;
            try (TraceScope ignored = openTraceScope(traceRun, userId)) {
                llmSpan = startLlmCallSpan(traceRun, chatRequest);
                String result = supervisor.invoke(prompt);
-                SseMessageUtils.sendContent(userId, result);
-                SseMessageUtils.sendDone(userId);
+                SseMessageUtils.sendContent(sessionId, result);
+                SseMessageUtils.sendDone(sessionId);
                // 保存助手回复到数据库（智能体对话为默认路径后，需在此落库以保留历史）
                if (StringUtils.isNotBlank(result)) {
                    chatMessageService.saveChatMessage(userId, chatRequest.getSessionId(),
@@ -347,12 +341,12 @@ public class ChatServiceFacade implements IChatService {
                }
                finishTraceRun(traceRun, TraceConstants.STATUS_ERROR, e);
                log.error("Supervisor 执行失败", e);
-                SseMessageUtils.sendError(userId, e.getMessage());
+                SseMessageUtils.sendError(sessionId, e.getMessage());
            } finally {
                if (llmSpan != null) {
                    llmSpan.detach();
                }
-                SseMessageUtils.completeConnection(userId, tokenValue);
+                SseMessageUtils.completeConnection(sessionId);
            }
        });
        return chatRequest.getEmitter();
@@ -461,8 +455,10 @@ public class ChatServiceFacade implements IChatService {

    /**
     * 兜底 MCP 工具装配（无智能体时使用，保留原有 3 个硬编码客户端逻辑）
+     *
+     * @param sessionId 会话ID，用于 MCP 工具事件按会话推送 SSE
     */
-    private ToolProvider buildDefaultMcpToolProvider(Long userId) {
+    private ToolProvider buildDefaultMcpToolProvider(String sessionId) {
        String npxCommand = resolveNpxCommand();
        McpTransport playwrightTransport = new StdioMcpTransport.Builder()
            .command(List.of(npxCommand, "-y", "@playwright/mcp@latest"))
@@ -470,7 +466,7 @@ public class ChatServiceFacade implements IChatService {
            .build();
        McpClient playwrightMcpClient = new DefaultMcpClient.Builder()
            .transport(playwrightTransport)
-            .listener(new MyMcpClientListener(userId))
+            .listener(new MyMcpClientListener(sessionId))
            .build();

        String userDir = System.getProperty("user.dir");
@@ -481,7 +477,7 @@ public class ChatServiceFacade implements IChatService {
            .build();
        McpClient filesystemMcpClient = new DefaultMcpClient.Builder()
            .transport(filesystemTransport)
-            .listener(new MyMcpClientListener(userId))
+            .listener(new MyMcpClientListener(sessionId))
            .build();

        return McpToolProvider.builder()
@@ -566,16 +562,15 @@ public class ChatServiceFacade implements IChatService {

        // 4. 获取用户信息
        Long userId = LoginHelper.getUserId();
-        String tokenValue = StpUtil.getTokenValue();

-        // 5. 建立 SSE 连接（用于前端监听）
-        sseEmitterManager.connect(userId, tokenValue);
+        // 5. 建立 SSE 连接（用于前端监听，按会话隔离）
+        sseEmitterManager.connect(String.valueOf(chatRequest.getSessionId()));

        // 保存用户消息
        chatMessageService.saveChatMessage(userId, chatRequest.getSessionId(), chatRequest.getContent(), RoleType.USER.getName(), chatRequest.getModel());

        // 6. 创建组合 handler：同时发送到 SSE 和外部 handler
-        StreamingChatResponseHandler combinedHandler = createCombinedHandler(userId, tokenValue, externalHandler);
+        StreamingChatResponseHandler combinedHandler = createCombinedHandler(String.valueOf(chatRequest.getSessionId()), externalHandler);

        // 7. 发起对话
        StreamingChatModel streamingChatModel = chatService.buildStreamingChatModel(chatModelVo, chatRequest);
@@ -799,12 +794,11 @@ public class ChatServiceFacade implements IChatService {
    /**
     * 创建组合响应处理器 - 同时发送到 SSE 和外部 handler
     *
-     * @param userId          用户ID
-     * @param tokenValue      会话令牌
+     * @param sessionId       会话ID（SSE 按会话隔离推送）
     * @param externalHandler 外部响应处理器（可为 null）
     * @return 组合的流式响应处理器
     */
-    protected StreamingChatResponseHandler createCombinedHandler(Long userId, String tokenValue,
+    protected StreamingChatResponseHandler createCombinedHandler(String sessionId,
                                                                  StreamingChatResponseHandler externalHandler) {
        return new StreamingChatResponseHandler() {

@@ -817,7 +811,7 @@ public class ChatServiceFacade implements IChatService {
                messageBuffer.append(partialResponse);

                // 2. 发送内容事件到 SSE（前端可通过 SSE 监听）
-                SseMessageUtils.sendContent(userId, partialResponse);
+                SseMessageUtils.sendContent(sessionId, partialResponse);

                // 3. 转发给外部 handler（Workflow 等模块可处理）
                if (externalHandler != null) {
@@ -828,7 +822,7 @@ public class ChatServiceFacade implements IChatService {
            @Override
            public void onPartialThinking(PartialThinking partialThinking) {
                // 发送推理内容到 SSE（前端通过 reasoning 事件监听）
-                SseMessageUtils.sendReasoning(userId, partialThinking.text());
+                SseMessageUtils.sendReasoning(sessionId, partialThinking.text());

                // 转发给外部 handler
                if (externalHandler != null) {
@@ -840,10 +834,10 @@ public class ChatServiceFacade implements IChatService {
            public void onCompleteResponse(ChatResponse completeResponse) {
                try {
                    // 1. 发送完成事件
-                    SseMessageUtils.sendDone(userId);
+                    SseMessageUtils.sendDone(sessionId);

                    // 2. 关闭 SSE 连接
-                    SseMessageUtils.completeConnection(userId, tokenValue);
+                    SseMessageUtils.completeConnection(sessionId);

                    // 3. 转发给外部 handler
                    if (externalHandler != null) {
@@ -857,7 +851,7 @@ public class ChatServiceFacade implements IChatService {
            @Override
            public void onError(Throwable error) {
                // 发送错误事件
-                SseMessageUtils.sendError(userId, error.getMessage());
+                SseMessageUtils.sendError(sessionId, error.getMessage());
                log.error("流式响应错误: {}", error.getMessage(), error);

                // 转发给外部 handler
--- a/ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/trace/RagTraceNodeTypes.java
+++ b/ruoyi-modules/ruoyi-chat/src/main/java/org/ruoyi/trace/RagTraceNodeTypes.java
@@ -14,5 +14,4 @@ public final class RagTraceNodeTypes {
    public static final String NODE_RETRIEVAL = "RETRIEVAL";
    public static final String NODE_RERANK = "RERANK";
    public static final String NODE_LLM_CALL = "LLM_CALL";
-    public static final String NODE_STREAM = "STREAM";
 }