智能体更新

oa-back/ruoyi-admin/src/main/java/com/ruoyi/web/controller/oa/CmcContractController.java

         if (cmcContract.getSignDate() == null) {
             if (cmcContract.getParams().size() == 0) {
                 for (int i = 2019; i <= Calendar.getInstance().get(Calendar.YEAR); i++) {
-                    cmcContract.setSignDate(new SimpleDateFormat("yyyy").parse(String.valueOf(i)));
+                    cmcContract.setContractCode("HT" + String.valueOf(i));
                     List<CmcContract> yearContractList = cmcContractService.selectCmcContractList(cmcContract);
+                    cmcContract.setContractCode("-" + (i - 2000));
+                    yearContractList.addAll(cmcContractService.selectCmcContractList(cmcContract));
                     int yearCountI = yearContractList.size();
                     int paidYearCountI = 0;
                     BigDecimal yearAmountI = new BigDecimal(0);
                 params.put("beginTime", beginTimeString + " 00:00:00");
                 params.put("endTime", endTimeString + " 23:59:59");
                 cmcContract.setParams(params);
-                List<CmcContract> monthContractList = cmcContractService.selectCmcContractListByRange(cmcContract);
+                List<CmcContract> monthContractList = cmcContractService.selectCmcContractListByRange(cmcContract); 
                 int monthCount = monthContractList.size();
                 int paidMonthCount = 0;
                 BigDecimal monthAmount = new BigDecimal(0);

oa-back/ruoyi-common/pom.xml

             <artifactId>pinyin4j</artifactId>
             <version>2.5.0</version>
         </dependency>
+
+        <dependency>
+            <groupId>io.milvus</groupId>
+            <artifactId>milvus-sdk-java</artifactId>
+            <version>2.6.2</version>
+        </dependency>
     </dependencies>
 
 </project>

oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/service/MilvusConnectionPool.java → oa-back/ruoyi-common/src/main/java/com/ruoyi/common/utils/milvus/MilvusConnectionPool.java

-package com.ruoyi.llm.service;
+package com.ruoyi.common.utils.milvus;
 
 import io.milvus.v2.client.ConnectConfig;
 import io.milvus.v2.client.MilvusClientV2;
         }
 
         for (int i = 0; i < POOL_SIZE; i++) {
-            MilvusClientV2 client = new MilvusClientV2(
-                    ConnectConfig.builder()
-                            .uri(milvusServiceUrl)
-                            .build());
-            clientPool.add(client);
-            availableClients.offer(client);
+            // MilvusClientV2 client = new MilvusClientV2(
+            //         ConnectConfig.builder()
+            //                 .uri(milvusServiceUrl)
+            //                 .build());
+            // clientPool.add(client);
+            // availableClients.offer(client);
         }
         initialized = true;
         System.out.println("Milvus连接池初始化完成，连接数: " + POOL_SIZE);

oa-back/ruoyi-flowable/src/main/java/com/ruoyi/flowable/agent/MyServiceTask.java

+package com.ruoyi.flowable.agent;
+
+import org.flowable.engine.delegate.DelegateExecution;
+import org.flowable.engine.delegate.JavaDelegate;
+import org.springframework.stereotype.Component;
+
+@Component("MyServiceTask")
+public class MyServiceTask implements JavaDelegate {
+    @Override
+    public void execute(DelegateExecution execution) {
+        System.out.println("========MyServiceTask==========");
+    }
+}

oa-back/ruoyi-flowable/src/main/java/com/ruoyi/flowable/service/impl/FlowTaskServiceImpl.java

     @Resource
     private ICmcTrainApprovalService cmcTrainApprovalService;
     @Resource
+    private ICmcTitleEvalService cmcTitleEvalService;
+    @Resource
+    private ICmcTransferService cmcTransferService;
+    @Resource
+    private ICmcRecruitService cmcRecruitService;
+    @Resource
     private ICmcBudgetService cmcBudgetService;
     @Resource
     private ICmcCheckService cmcCheckService;
         if (flowTaskDto.getProcDefName().contains("承接")) {
             CmcContract cmcContract = cmcContractService.selectCmcContractByContractId(formId);
             if (cmcContract != null)
-                flowTaskDto.setTitle(cmcContractService.selectCmcContractByContractId(formId).getContractName());
+                flowTaskDto.setTitle(cmcContract.getContractName());
         }
         if (flowTaskDto.getProcDefName().contains("分包")) {
             CmcSubContract cmcSubContract = cmcSubContractService.selectCmcSubContractBySubContractId(formId);
             if (cmcSubContract != null)
-                flowTaskDto.setTitle(cmcSubContractService.selectCmcSubContractBySubContractId(formId).getSubContractName());
+                flowTaskDto.setTitle(cmcSubContract.getSubContractName());
         }
         if (flowTaskDto.getProcDefName().equals("采购审批")) {
             CmcProcureApproval cmcProcureApproval = cmcProcureApprovalService.selectCmcProcureApprovalByProcureApplyId(formId);
             if (cmcTrainApproval != null)
                 flowTaskDto.setTitle(cmcTrainApproval.getContent());
         }
+        if (flowTaskDto.getProcDefName().equals("职称评审")) {
+            CmcTitleEval cmcTitleEval = cmcTitleEvalService.selectCmcTitleEvalByTitleEvalId(formId);
+            if (cmcTitleEval != null)
+                flowTaskDto.setTitle(cmcTitleEval.getAnnual() + cmcTitleEval.getInstitude() + "-" + cmcTitleEval.getTitleProfession() + cmcTitleEval.getLevel());
+        }
+        if (flowTaskDto.getProcDefName().equals("调岗审批")) {
+            CmcTransfer cmcTransfer = cmcTransferService.selectCmcTransferByTransferId(formId);
+            if (cmcTransfer != null)
+                flowTaskDto.setTitle(cmcTransfer.getBeforeDept().getDeptName() + "-" + cmcTransfer.getAfterDept().getDeptName());
+        }
+        if (flowTaskDto.getProcDefName().equals("招聘审批")) {
+            CmcRecruit cmcRecruit = cmcRecruitService.selectCmcRecruitByRecruitId(formId);
+            if (cmcRecruit != null)
+                flowTaskDto.setTitle(cmcRecruit.getPostName() + "-" + cmcRecruit.getPostNumber() + "名" );
+        }
     }
 }

oa-back/ruoyi-llm/pom.xml

             <artifactId>ruoyi-framework</artifactId>
         </dependency>
 
-        <!-- 向量数据库-->
-        <dependency>
-            <groupId>io.milvus</groupId>
-            <artifactId>milvus-sdk-java</artifactId>
-            <version>2.6.2</version>
-        </dependency>
-
-        <!-- LangChain4j -->
-        <dependency>
-            <groupId>dev.langchain4j</groupId>
-            <artifactId>langchain4j</artifactId>
-            <version>0.35.0</version>
-        </dependency>
-
-        <!-- LangChain4j Milvus 集成 -->
-        <dependency>
-            <groupId>dev.langchain4j</groupId>
-            <artifactId>langchain4j-milvus</artifactId>
-            <version>0.35.0</version>
-        </dependency>
-
-        <!-- LangChain4j embedding 集成 -->
-        <dependency>
-            <groupId>dev.langchain4j</groupId>
-            <artifactId>langchain4j-embeddings-bge-small-zh-v15</artifactId>
-            <version>0.35.0</version>
-        </dependency>
-
-        <!-- LangChain4j rag 集成 -->
-        <dependency>
-            <groupId>dev.langchain4j</groupId>
-            <artifactId>langchain4j-easy-rag</artifactId>
-            <version>0.35.0</version>
-        </dependency>
-
-        <!-- LangChain4j pdfParser 集成 -->
-        <dependency>
-            <groupId>dev.langchain4j</groupId>
-            <artifactId>langchain4j-document-parser-apache-pdfbox</artifactId>
-            <version>0.35.0</version>
-        </dependency>
-
-        <!-- Solon模型上下文协议 -->
-        <dependency>
-            <groupId>org.noear</groupId>
-            <artifactId>solon-ai-mcp</artifactId>
-            <version>3.9.5</version>
-        </dependency>
-
         <!-- Spring Boot Web -->
         <dependency>
             <groupId>org.springframework.boot</groupId>

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/CmcAgentController.java

     @GetMapping("/list")
     public TableDataInfo list(CmcAgent cmcAgent)
     {
-        startPage();
         List<CmcAgent> list = cmcAgentService.selectCmcAgentList(cmcAgent);
         return getDataTable(list);
     }

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/CmcChatController.java

     @GetMapping("/list")
     public TableDataInfo list(CmcChat cmcChat)
     {
-        startPage();
         List<CmcChat> list = cmcChatService.selectCmcChatList(cmcChat);
         return getDataTable(list);
     }

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/CmcDocumentController.java

     @GetMapping("/list")
     public TableDataInfo list(CmcDocument cmcDocument)
     {
-        startPage();
         List<CmcDocument> list = cmcDocumentService.selectCmcDocumentList(cmcDocument);
         return getDataTable(list);
     }

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/CmcTopicController.java

     @GetMapping("/list")
     public TableDataInfo list(CmcTopic cmcTopic)
     {
-        startPage();
         List<CmcTopic> list = cmcTopicService.selectCmcTopicList(cmcTopic);
         return getDataTable(list);
     }

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/KnowLedgeController.java

 package com.ruoyi.web.llm.controller;
 
 import com.alibaba.fastjson2.JSONArray;
-import com.ruoyi.web.llm.service.ILangChainMilvusService;
-import com.ruoyi.web.llm.service.IMilvusService;
+import com.ruoyi.llm.service.ILangChainMilvusService;
+import com.ruoyi.llm.service.IMilvusService;
 import com.ruoyi.common.core.controller.BaseController;
 import com.ruoyi.common.core.domain.AjaxResult;
 import org.springframework.beans.factory.annotation.Autowired;

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/McpController.java

-package com.ruoyi.web.llm.controller;
-
-import com.alibaba.fastjson2.JSONObject;
-import com.ruoyi.common.core.controller.BaseController;
-import com.ruoyi.llm.domain.CmcChat;
-import com.ruoyi.llm.service.ICmcAgentService;
-import com.ruoyi.llm.service.ICmcChatService;
-import com.ruoyi.llm.service.ICmcTopicService;
-import org.noear.solon.ai.chat.ChatModel;
-import org.noear.solon.ai.chat.ChatResponse;
-import org.noear.solon.ai.chat.ChatSession;
-import org.noear.solon.ai.chat.message.AssistantMessage;
-import org.noear.solon.ai.chat.message.ChatMessage;
-import org.noear.solon.ai.chat.session.InMemoryChatSession;
-import org.noear.solon.ai.mcp.McpChannel;
-import org.noear.solon.ai.mcp.client.McpClientProvider;
-import org.springframework.beans.factory.annotation.Autowired;
-import org.springframework.beans.factory.annotation.Value;
-import org.springframework.web.bind.annotation.GetMapping;
-import org.springframework.web.bind.annotation.RequestMapping;
-import org.springframework.web.bind.annotation.RestController;
-
-import java.io.IOException;
-import java.util.*;
-
-
-/**
- * mcp模型上下文协议Controller
- * 
- * @author cmc
- * @date 2025-04-08
- */
-@RestController
-@RequestMapping("/llm/mcp")
-public class McpController extends BaseController
-{
-    @Autowired
-    private ICmcChatService cmcChatService;
-
-    @Autowired
-    private ICmcTopicService cmcTopicService;
-
-    @Autowired
-    private ICmcAgentService cmcAgentService;
-
-    @Value("${cmc.llmService.url}")
-    private String llmServiceUrl;
-
-    @Value("${cmc.milvusService.url}")
-    private String milvusServiceUrl;
-
-    /**
-     * 自动调用mcp工具问答
-     * @return
-     */
-    @GetMapping("/answer")
-    public AssistantMessage answer(String topicId, String question) throws IOException {
-        McpClientProvider clientProvider = McpClientProvider.builder()
-                .channel(McpChannel.STREAMABLE_STATELESS )
-                .url("http://localhost:8087/mcp/sse")
-                .build();
-        ChatModel chatModel = ChatModel.of(llmServiceUrl)
-                .model("Qwen")
-                .defaultToolAdd(clientProvider)
-                .build();
-
-        List<ChatMessage> messages = new ArrayList<>();
-        CmcChat cmcChat = new CmcChat();
-        cmcChat.setTopicId(topicId);
-        List<CmcChat> cmcChatList = cmcChatService.selectCmcChatList(cmcChat);
-        for (CmcChat chat : cmcChatList) {
-            messages.add(ChatMessage.ofUser(chat.getInput()));
-            messages.add(ChatMessage.ofAssistant(chat.getOutput()));
-        }
-        messages.add(ChatMessage.ofUser(question));
-        ChatSession chatSession =  InMemoryChatSession.builder().messages(messages).build();
-        ChatResponse response = chatModel.prompt(chatSession).call();
-        String resultContent = response.lastChoice().getMessage().getResultContent();
-        AssistantMessage assistantMessage;
-        if (resultContent.startsWith("<tool_call>")) {
-            String content = resultContent.replace("<tool_call>\n", "").replace("\n</tool_call>", "");
-            JSONObject jsonObject = JSONObject.parseObject(content);
-            String name = jsonObject.getString("name");
-            JSONObject arguments = jsonObject.getJSONObject("arguments");
-            if (arguments.getString("templatePath").contains("招标") || arguments.getString("templatePath").contains("询价")) {
-                arguments.put("collectionName", "technical");
-                String agentName = cmcAgentService.selectCmcAgentByAgentId(cmcTopicService.selectCmcTopicByTopicId(topicId).getAgentId()).getAgentName();
-                arguments.put("agentName", agentName);
-                arguments.put("title", question);
-            }
-            resultContent = clientProvider.callTool(name, arguments).getContent();
-            assistantMessage = ChatMessage.ofAssistant(resultContent);
-        }
-        else
-            throw new IOException("模型上下文工具未准备就绪，请重试");
-        return assistantMessage;
-    }
-
-
-}

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/RagController.java

 package com.ruoyi.web.llm.controller;
 
 import com.alibaba.fastjson2.JSONObject;
-import com.ruoyi.web.llm.service.ILangChainMilvusService;
+import com.ruoyi.llm.service.ILangChainMilvusService;
 import com.ruoyi.common.core.controller.BaseController;
-import org.noear.solon.core.util.MimeType;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.web.bind.annotation.GetMapping;
 import org.springframework.web.bind.annotation.RequestMapping;
 import org.springframework.web.bind.annotation.RestController;
-import reactor.core.publisher.Flux;
+import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
 
-import java.io.IOException;
 import java.util.List;
 
 /**
     /**
      * 调用LLM+RAG（知识库）生成回答
      */
-    @GetMapping(value = "/answer", produces = MimeType.TEXT_EVENT_STREAM_UTF8_VALUE)
-    public Flux<String> answerWithCollection(String collectionName, String topicId, String question) throws IOException {
+    @GetMapping("/answer")
+    public SseEmitter answerWithCollection(String collectionName, String topicId, String question) {
         List<JSONObject> contexts = langChainMilvusService.retrieveFromMilvus(collectionName, question, 10);
         return langChainMilvusService.generateAnswerWithCollection(topicId, question, contexts);
     }

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/controller/SessionController.java

 package com.ruoyi.web.llm.controller;
 
 import com.ruoyi.common.core.controller.BaseController;
-import com.ruoyi.web.llm.service.ISessionService;
-import org.noear.solon.core.util.MimeType;
+import com.ruoyi.llm.service.ISessionService;
 import org.springframework.beans.factory.annotation.Autowired;
 import org.springframework.web.bind.annotation.GetMapping;
 import org.springframework.web.bind.annotation.RequestMapping;
 import org.springframework.web.bind.annotation.RestController;
-import reactor.core.publisher.Flux;
-
+import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
 /**
  * session对话Controller
  *
     /**
      * 生成回答
      */
-    @GetMapping(value = "/answer", produces = MimeType.TEXT_EVENT_STREAM_UTF8_VALUE)
-    public Flux<String> answer(String topicId, String question) {
+    @GetMapping("/answer")
+    public SseEmitter answer(String topicId, String question) {
         return sessionService.answer(topicId, question);
     }
 
     /**
      * 调用LLM+RAG（外部文件）生成回答
      */
-    @GetMapping(value = "/answerWithDocument", produces = MimeType.TEXT_EVENT_STREAM_UTF8_VALUE)
-    public Flux<String> answerWithDocument(String topicId, String chatId, String question) throws Exception
+    @GetMapping("/answerWithDocument")
+    public SseEmitter answerWithDocument(String topicId, String chatId, String question) throws Exception
     {
         return sessionService.answerWithDocument(topicId, chatId, question);
     }

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/ISessionService.java

-package com.ruoyi.web.llm.service;
-
-import reactor.core.publisher.Flux;
-
-public interface ISessionService {
-
-    /**
-     * 生成回答
-     */
-    Flux<String> answer(String topicId, String question);
-
-    /**
-     * 调用LLM+RAG（外部文件）生成回答
-     */
-    Flux<String> answerWithDocument(String topicId, String chatId, String question) throws Exception;
-
-}

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/impl/LangChainMilvusServiceImpl.java

-package com.ruoyi.web.llm.service.impl;
-
-import com.alibaba.fastjson2.JSONObject;
-import com.google.gson.JsonObject;
-import com.google.gson.JsonParser;
-import com.ruoyi.common.config.RuoYiConfig;
-import com.ruoyi.llm.domain.CmcChat;
-import com.ruoyi.llm.domain.CmcDocument;
-import com.ruoyi.llm.service.ICmcChatService;
-import com.ruoyi.llm.service.ICmcDocumentService;
-import com.ruoyi.web.llm.service.ILangChainMilvusService;
-import dev.langchain4j.data.document.Document;
-import dev.langchain4j.data.document.parser.TextDocumentParser;
-import dev.langchain4j.data.document.parser.apache.pdfbox.ApachePdfBoxDocumentParser;
-import dev.langchain4j.data.document.splitter.DocumentByParagraphSplitter;
-import dev.langchain4j.data.embedding.Embedding;
-import dev.langchain4j.data.segment.TextSegment;
-import dev.langchain4j.model.embedding.EmbeddingModel;
-import dev.langchain4j.model.embedding.onnx.bgesmallzhv15.BgeSmallZhV15EmbeddingModel;
-import dev.langchain4j.store.embedding.EmbeddingMatch;
-import dev.langchain4j.store.embedding.EmbeddingSearchRequest;
-import dev.langchain4j.store.embedding.inmemory.InMemoryEmbeddingStore;
-import io.milvus.v2.client.ConnectConfig;
-import io.milvus.v2.client.MilvusClientV2;
-import io.milvus.v2.common.IndexParam;
-import io.milvus.v2.service.collection.request.LoadCollectionReq;
-import io.milvus.v2.service.collection.request.ReleaseCollectionReq;
-import io.milvus.v2.service.vector.request.DeleteReq;
-import io.milvus.v2.service.vector.request.InsertReq;
-import io.milvus.v2.service.vector.request.SearchReq;
-import io.milvus.v2.service.vector.request.data.BaseVector;
-import io.milvus.v2.service.vector.request.data.FloatVec;
-import io.milvus.v2.service.vector.response.InsertResp;
-import io.milvus.v2.service.vector.response.SearchResp;
-import org.apache.poi.hwpf.HWPFDocument;
-import org.apache.poi.hwpf.usermodel.Paragraph;
-import org.apache.poi.hwpf.usermodel.Range;
-import org.apache.poi.xwpf.usermodel.XWPFDocument;
-import org.apache.poi.xwpf.usermodel.XWPFParagraph;
-import org.apache.poi.xwpf.usermodel.XWPFStyle;
-import org.noear.solon.ai.chat.ChatModel;
-import org.noear.solon.ai.chat.ChatResponse;
-import org.noear.solon.ai.chat.ChatSession;
-import org.noear.solon.ai.chat.message.ChatMessage;
-import org.noear.solon.ai.chat.prompt.Prompt;
-import org.noear.solon.ai.chat.session.InMemoryChatSession;
-import org.openxmlformats.schemas.wordprocessingml.x2006.main.CTPPrGeneral;
-import org.reactivestreams.Publisher;
-import org.springframework.beans.factory.annotation.Autowired;
-import org.springframework.beans.factory.annotation.Value;
-import org.springframework.stereotype.Service;
-import org.springframework.web.multipart.MultipartFile;
-import reactor.core.publisher.Flux;
-
-import javax.annotation.PostConstruct;
-import javax.annotation.PreDestroy;
-import java.io.*;
-import java.util.*;
-
-@Service
-public class LangChainMilvusServiceImpl implements ILangChainMilvusService
-{
-    @Autowired
-    private ICmcChatService cmcChatService;
-
-    @Autowired
-    private ICmcDocumentService cmcDocumentService;
-
-    private static final EmbeddingModel embeddingModel = new BgeSmallZhV15EmbeddingModel();
-
-    private String processValue = "";
-
-    @Value("${cmc.llmService.url}")
-    private String llmServiceUrl;
-
-    @Value("${cmc.milvusService.url}")
-    private String milvusServiceUrl;
-
-    private MilvusClientV2 milvusClient;
-
-    // 复用 ChatModel 实例
-    private ChatModel chatModel;
-
-    @PostConstruct
-    public void initMilvusClient() {
-        if (milvusServiceUrl == null || milvusServiceUrl.isEmpty()) {
-            throw new IllegalArgumentException("milvusServiceUrl 配置不能为空");
-        }
-        milvusClient = new MilvusClientV2(
-                ConnectConfig.builder()
-                        .uri(milvusServiceUrl)
-                        .build());
-    }
-
-    @PostConstruct
-    public void initChatModel() {
-        // 初始化 ChatModel
-        chatModel = ChatModel.of(llmServiceUrl)
-                .model("Qwen")
-                .build();
-    }
-
-    @PreDestroy
-    public void destroyMilvusClient() {
-        if (milvusClient != null) {
-            try {
-                // 关闭 Milvus 客户端，释放 gRPC 通道
-                milvusClient.close();
-                System.out.println("Milvus client closed successfully");
-            } catch (Exception e) {
-                System.err.println("Error closing Milvus client: " + e.getMessage());
-                e.printStackTrace();
-            }
-        }
-    }
-
-    /**
-     * 上传多文件
-     *
-     * @return 结果
-     */
-    public String getProcess() {
-        return processValue;
-    }
-
-    /**
-     * 导入知识库文件
-     * @return
-     */
-    @Override
-    public int insertLangchainEmbeddingDocument(MultipartFile[] fileList, String collectionName)
-    {
-        processValue = "";
-        int successfullyInsertedFiles = 0;
-
-        for (int i = 0; i < fileList.length; i++) {
-            MultipartFile file = fileList[i];
-            try {
-                // 构建上传目录路径
-                String uploadDir = RuoYiConfig.getProfile() + "/upload/rag/knowledge/" + collectionName;
-                File profilePath = new File(uploadDir);
-
-                // 确保目录存在，创建目录并设置权限
-                if (!profilePath.exists())
-                    profilePath.mkdirs();
-                File transferFile = new File(profilePath, file.getOriginalFilename());
-                if (!transferFile.exists())
-                    file.transferTo(transferFile);
-                List<TextSegment> segments = splitDocument(transferFile);
-
-                // 准备导入数据
-                List<JsonObject> data = new ArrayList<>();
-                // 提取文本和生成嵌入
-                for (TextSegment segment : segments) {
-                    String text = segment.text();
-                    if (text.trim().isEmpty())
-                        continue;
-
-                    JSONObject fastjsonObj = new JSONObject();
-                    fastjsonObj.put("file_name", file.getOriginalFilename());
-                    String[] fileName = file.getOriginalFilename().split("\\.");
-                    fastjsonObj.put("file_type", fileName[fileName.length - 1]);
-                    fastjsonObj.put("title", text.split("\n")[0].split(" ")[1]);
-                    fastjsonObj.put("content", text);
-                    fastjsonObj.put("embedding", embeddingModel.embed(text).content().vectorAsList());
-                    String jsonString = fastjsonObj.toJSONString();
-                    JsonObject jsonObject = JsonParser.parseString(jsonString).getAsJsonObject();
-                    data.add(jsonObject);
-                }
-
-                // 添加一条包含所有二、三级标题的记录
-                String titlesContent = extractAllTitles(transferFile);
-                if (!titlesContent.trim().isEmpty()) {
-                    JSONObject titlesJsonObj = new JSONObject();
-                    titlesJsonObj.put("file_name", file.getOriginalFilename());
-                    titlesJsonObj.put("file_type", "标题");
-                    titlesJsonObj.put("title", "");
-                    titlesJsonObj.put("content", titlesContent);
-                    titlesJsonObj.put("embedding", embeddingModel.embed(titlesContent).content().vectorAsList());
-                    String titlesJsonString = titlesJsonObj.toJSONString();
-                    JsonObject titlesJsonObject = JsonParser.parseString(titlesJsonString).getAsJsonObject();
-                    data.add(titlesJsonObject);
-                }
-
-                // 加载集合
-                LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                        .collectionName(collectionName)
-                        .build();
-                milvusClient.loadCollection(loadCollectionReq);
-
-                try {
-                    // 先删除相同文件名的记录
-                    DeleteReq deleteReq = DeleteReq.builder()
-                            .collectionName(collectionName)
-                            .filter(String.format("file_name == \"%s\"", file.getOriginalFilename()))
-                            .build();
-                    if (milvusClient.delete(deleteReq).getDeleteCnt() > 0)
-                        System.out.println("已删除同名文件记录: " + file.getOriginalFilename());
-
-                    // 构建导入请求
-                    InsertReq insertReq = InsertReq.builder()
-                            .collectionName(collectionName)
-                            .data(data)
-                            .build();
-
-                    // 执行导入并检查结果
-                    InsertResp insertResp = milvusClient.insert(insertReq);
-                    if (insertResp != null && insertResp.getInsertCnt() > 0) {
-                        successfullyInsertedFiles++;
-                        processValue = "上传中: " + String.format("%.2f", (double) (i + 1) / fileList.length * 100) + "%";
-                        System.out.println("成功导入" + (i + 1) + "/" + fileList.length + "文件: " + file.getOriginalFilename() + ", 导入记录数: " + insertResp.getInsertCnt());
-                    } else {
-                        System.err.println("文件导入失败: " + file.getOriginalFilename() + ", 没有记录被导入");
-                    }
-                } finally {
-                    // 释放集合
-                    ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
-                            .collectionName(collectionName)
-                            .build();
-                    milvusClient.releaseCollection(releaseCollectionReq);
-                }
-            } catch (Exception e) {
-                System.err.println("处理文件时出错: " + file.getOriginalFilename() + ", 错误: " + e.getMessage());
-                e.printStackTrace();
-                // 继续处理下一个文件
-            }
-        }
-
-        processValue = "上传完成";
-        return successfullyInsertedFiles;
-    }
-
-    /**
-     * 从Milvus检索相关文档
-     * @return
-     */
-    @Override
-    public List<JSONObject> retrieveFromMilvus(String collectionName, String query, int topK) {
-        List<JSONObject> resultList = new ArrayList<>();
-        List<List<SearchResp.SearchResult>> searchResultList = retrieve(collectionName, query, topK);
-        searchResultList.forEach(searchResult -> {
-            JSONObject result = new JSONObject();
-            result.put("file_name", searchResult.get(0).getEntity().get("file_name"));
-            result.put("content", searchResult.get(0).getEntity().get("content"));
-            resultList.add(result);
-        });
-        return resultList;
-    }
-
-    /**
-     * 从Milvus检索相关文档及相关度
-     * @return
-     */
-    @Override
-    public List<JSONObject> similarityFromMilvus(String collectionName, String query, int topK) {
-        List<JSONObject> resultList = new ArrayList<>();
-        List<List<SearchResp.SearchResult>> searchResultList = retrieve(collectionName, query, topK);
-        searchResultList.forEach(searchResult -> {
-            JSONObject result = new JSONObject();
-            result.put("score", searchResult.get(0).getScore());
-            result.put("file_name", searchResult.get(0).getEntity().get("file_name"));
-            result.put("content", searchResult.get(0).getEntity().get("content"));
-            resultList.add(result);
-        });
-        resultList.removeIf(jsonObject -> jsonObject.getDouble("score") < 0.7);
-        return resultList;
-    }
-
-    /**
-     * 调用LLM生成回答
-     * @return
-     */
-    @Override
-    public Flux<String> generateAnswer(String topicId, String prompt) {
-        List<ChatMessage> messages = new ArrayList<>();
-        if (topicId != null) {
-            CmcChat cmcChat = new CmcChat();
-            cmcChat.setTopicId(topicId);
-            List<CmcChat> cmcChatList = cmcChatService.selectCmcChatList(cmcChat);
-            for (CmcChat chat : cmcChatList) {
-                messages.add(ChatMessage.ofUser(chat.getInput()));
-                messages.add(ChatMessage.ofAssistant(chat.getOutput()));
-            }
-        }
-        messages.add(ChatMessage.ofUser(prompt));
-        ChatSession chatSession =  InMemoryChatSession.builder().messages(messages).build();
-        Prompt prompt1 = Prompt.of(prompt).attrPut("session", chatSession);
-        Publisher<ChatResponse> publisher = chatModel.prompt(prompt1).stream();
-        return Flux.from(publisher)
-                .map(response -> response.lastChoice().getMessage().getContent())
-                .map(this::toSseDataFrame)
-                .concatWith(Flux.just("data: [DONE]\n\n"));
-    }
-
-    private String toSseDataFrame(String data) {
-        if (data == null) {
-            return "data: \n\n";
-        }
-        // SSE 要求每一行都以 data: 开头
-        String normalized = data.replace("\r\n", "\n");
-        StringBuilder sb = new StringBuilder(normalized.length() + 16);
-        sb.append("data: ");
-        int start = 0;
-        while (true) {
-            int idx = normalized.indexOf('\n', start);
-            if (idx < 0) {
-                sb.append(normalized.substring(start));
-                break;
-            }
-            sb.append(normalized, start, idx);
-            sb.append("\n");
-            sb.append("data: ");
-            start = idx + 1;
-        }
-        sb.append("\n\n");
-        return sb.toString();
-    }
-
-    /**
-     * 调用LLM+RAG生成回答
-     * @return
-     */
-    @Override
-    public Flux<String> generateAnswerWithCollection(String topicId, String question, List<JSONObject> contexts) {
-        StringBuilder sb = new StringBuilder();
-        sb.append("问题: ").append(question).append("\n\n");
-        sb.append("根据以下上下文回答问题：\n\n");
-        for (JSONObject context : contexts) {
-            sb.append("文件").append(": ")
-                    .append(context.getString("file_name")).append("\n\n")
-                    .append("上下文").append(": ")
-                    .append(context.getString("content")).append("\n\n");
-        }
-        return generateAnswer(topicId, sb.toString());
-    }
-
-    /**
-     * 调用LLM生成回答
-     */
-    @Override
-    public Flux<String> generateAnswerWithDocument(String topicId, String chatId, String question) throws Exception {
-        CmcDocument cmcDocument = new CmcDocument();
-        cmcDocument.setChatId(chatId);
-        List<CmcDocument> documentList = cmcDocumentService.selectCmcDocumentList(cmcDocument);
-        StringBuilder sb = new StringBuilder("问题: " + question + "\n\n").append("根据以下上下文回答问题：\n\n");
-        for (CmcDocument document : documentList) {
-            File profilePath = new File(RuoYiConfig.getProfile() + document.getPath());
-            InMemoryEmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
-            List<TextSegment> segments = splitDocument(profilePath);
-            List<Embedding> embeddings = embeddingModel.embedAll(segments).content();
-            embeddingStore.addAll(embeddings, segments);
-            Embedding queryEmbedding = embeddingModel.embed(question).content();
-            EmbeddingSearchRequest embeddingSearchRequest = EmbeddingSearchRequest.builder()
-                    .queryEmbedding(queryEmbedding)
-                    .minScore(0.7)
-                    .build();
-            List<EmbeddingMatch<TextSegment>> results = embeddingStore.search(embeddingSearchRequest).matches();
-            results.sort(Comparator.comparingDouble(EmbeddingMatch<TextSegment>::score).reversed());
-            for (EmbeddingMatch<TextSegment> embeddingMatch : results) {
-                String contexts = embeddingMatch.embedded().toString();
-                sb.append("文件").append(": ")
-                        .append(document.getPath()).append("\n\n")
-                        .append("上下文").append(": ")
-                        .append(contexts).append("\n\n");
-            }
-        }
-        return generateAnswer(topicId, sb.toString());
-    }
-
-    /**
-     * 调用LLM生成回答
-     */
-    @Override
-    public Flux<String> generateAnswerWithDocumentAndCollection(String topicId, String question, List<JSONObject> contexts) throws Exception {
-        StringBuilder sb = new StringBuilder("招标文件内容：\n\n");
-        CmcChat cmcChat = new CmcChat();
-        cmcChat.setTopicId(topicId);
-        for (CmcChat chat : cmcChatService.selectCmcChatList(cmcChat)) {
-            CmcDocument cmcDocument = new CmcDocument();
-            cmcDocument.setChatId(chat.getChatId());
-            List<CmcDocument> documentList = cmcDocumentService.selectCmcDocumentList(cmcDocument);
-            //技术标书撰写智能体一个话题只会上传一个招标文件
-            if (documentList.size() == 1) {
-                for (CmcDocument document : documentList) {
-                    File profilePath = new File(RuoYiConfig.getProfile() + document.getPath());
-                    InMemoryEmbeddingStore<TextSegment> embeddingStore = new InMemoryEmbeddingStore<>();
-                    List<TextSegment> segments = splitDocument(profilePath);
-                    List<Embedding> embeddings = embeddingModel.embedAll(segments).content();
-                    embeddingStore.addAll(embeddings, segments);
-                    Embedding queryEmbedding = embeddingModel.embed(question).content();
-                    EmbeddingSearchRequest embeddingSearchRequest = EmbeddingSearchRequest.builder()
-                            .queryEmbedding(queryEmbedding)
-                            .minScore(0.7)
-                            .build();
-                    List<EmbeddingMatch<TextSegment>> results = embeddingStore.search(embeddingSearchRequest).matches();
-                    results.sort(Comparator.comparingDouble(EmbeddingMatch<TextSegment>::score).reversed());
-                    for (EmbeddingMatch<TextSegment> embeddingMatch : results) {
-                        String requests = embeddingMatch.embedded().toString();
-                        sb.append(requests).append("\n\n");
-                    }
-                }
-            }
-        }
-        sb.append("针对本项目招标文件内容，补全以下章节部分：\n\n").append(question);
-//        for (JSONObject context : contexts) {
-//            sb.append("文件").append(": ")
-//                    .append(context.getString("file_name")).append("\n\n")
-//                    .append("段落格式").append(": ")
-//                    .append(context.getString("content")).append("\n\n");
-//        }
-        return generateAnswer(topicId, sb.toString());
-    }
-
-    /**
-     * 获取二级标题下三级标题列表
-     */
-    @Override
-    public List<String> extractSubTitles(String filename, String question) throws IOException {
-        List<String> subTitles = new ArrayList<>();
-        boolean inTargetSection = false;
-
-        InputStream fileInputStream = new FileInputStream(filename);
-        try (XWPFDocument document = new XWPFDocument(fileInputStream)) {
-            for (XWPFParagraph paragraph : document.getParagraphs()) {
-                String text = paragraph.getText().trim();
-                int level = getDocxOutlineLevel(paragraph, document);
-                if (level == 3 && text.contains(question)) {
-                    inTargetSection = true;
-                    continue;
-                }
-
-                if (inTargetSection) {
-                    if (level == 4) {
-                        subTitles.add(text);
-                    }
-                    else if (level == 3) {
-                        break;
-                    }
-                }
-            }
-        }
-        if (subTitles.size() == 0)
-            subTitles.add(question);
-        return subTitles;
-    }
-
-    /**
-     * 检索知识库
-     * @return
-     */
-    private List<List<SearchResp.SearchResult>> retrieve(String collectionName, String query, int topK) {
-        List<BaseVector> queryVector = Collections.singletonList(new FloatVec(embeddingModel.embed(query).content().vector()));
-
-        //  加载集合
-        LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.loadCollection(loadCollectionReq);
-
-        // 构建SearchParam
-        Map<String, Object> searchParams = new HashMap<>();
-        searchParams.put("nprobe", 8);
-        SearchReq searchReq = SearchReq.builder()
-                .collectionName(collectionName)
-                .data(queryVector)
-                .topK(topK)
-                .outputFields(Arrays.asList("file_name", "file_type", "content"))
-                .annsField("embedding")
-                .metricType(IndexParam.MetricType.COSINE)
-                .searchParams(searchParams)
-                .build();
-
-        SearchResp searchResp = milvusClient.search(searchReq);
-        List<List<SearchResp.SearchResult>> searchResultList = searchResp.getSearchResults();
-
-        // 释放集合
-        ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.releaseCollection(releaseCollectionReq);
-
-        return searchResultList;
-    }
-
-    /**
-     * 提取文档中的所有二、三级标题
-     */
-    private String extractAllTitles(File transferFile) throws Exception {
-        StringBuilder titlesBuilder = new StringBuilder();
-        String filename = transferFile.getName().toLowerCase();
-
-        if (filename.endsWith(".docx")) {
-            try (XWPFDocument document = new XWPFDocument(new FileInputStream(transferFile))) {
-                for (XWPFParagraph paragraph : document.getParagraphs()) {
-                    String text = paragraph.getText().trim();
-                    if (text.isEmpty()) continue;
-                    int outlineLevel = getDocxOutlineLevel(paragraph, document);
-                    if (outlineLevel == 2 || outlineLevel == 3) {
-                        titlesBuilder.append(text).append("\n");
-                    }
-                }
-            }
-        } else if (filename.endsWith(".doc")) {
-            try (HWPFDocument document = new HWPFDocument(new FileInputStream(transferFile))) {
-                Range range = document.getRange();
-                for (int i = 0; i < range.numParagraphs(); i++) {
-                    Paragraph paragraph = range.getParagraph(i);
-                    String text = paragraph.text().trim();
-                    if (text.isEmpty()) continue;
-                    int outlineLevel = getDocOutlineLevel(paragraph);
-                    if (outlineLevel == 2 || outlineLevel == 3) {
-                        titlesBuilder.append(text).append("\n");
-                    }
-                }
-            }
-        }
-        return titlesBuilder.toString();
-    }
-
-    /**
-     * 获取DOCX段落的大纲级别
-     */
-    private int getDocxOutlineLevel(XWPFParagraph paragraph, XWPFDocument document) {
-        int level = 0;
-        String styleId = paragraph.getStyleID();
-        if (styleId != null) {
-            XWPFStyle style = document.getStyles().getStyle(styleId);
-            if (style != null) {
-                CTPPrGeneral stylePPr = style.getCTStyle().getPPr();
-                if (stylePPr != null && stylePPr.getOutlineLvl() != null) {
-                    level = stylePPr.getOutlineLvl().getVal().intValue() + 1;
-                }
-            }
-        }
-        return level;
-    }
-
-    /**
-     * 获取DOC段落的大纲级别
-     */
-    private int getDocOutlineLevel(Paragraph paragraph) {
-        int level = 0;
-        short styleIndex = paragraph.getStyleIndex();
-        if (styleIndex >= 1 && styleIndex <= 9) {
-            level = styleIndex;
-        }
-        return level;
-    }
-
-    /**
-     * 检查DOCX段落是否是三级标题
-     */
-    private boolean isDocxLevel3Title(String paraText, XWPFDocument document) {
-        for (XWPFParagraph p : document.getParagraphs()) {
-            if (p.getText().trim().equals(paraText.trim())) {
-                return getDocxOutlineLevel(p, document) == 3;
-            }
-        }
-        return false;
-    }
-
-    /**
-     * 按三级标题分割DOCX内容
-     */
-    private void splitDocxByLevel3(String content, XWPFDocument document, List<TextSegment> segments) {
-        StringBuilder currentLevel3Content = new StringBuilder();
-        String[] paragraphs = content.split("\n");
-        for (String para : paragraphs) {
-            if (para.trim().isEmpty()) continue;
-
-            if (isDocxLevel3Title(para.trim(), document)) {
-                // 保存当前三级标题内容
-                if (currentLevel3Content.length() != 0) {
-                    TextSegment segment = TextSegment.from(currentLevel3Content.toString());
-                    segments.add(segment);
-                    currentLevel3Content = new StringBuilder();
-                }
-                // 开始新的三级标题内容
-                currentLevel3Content.append(para).append("\n");
-            } else {
-                // 普通内容，添加到当前三级标题
-                if (currentLevel3Content.length() != 0) {
-                    currentLevel3Content.append(para).append("\n");
-                }
-            }
-        }
-        // 保存最后一个三级标题内容
-        if (currentLevel3Content.length() != 0) {
-            TextSegment segment = TextSegment.from(currentLevel3Content.toString());
-            segments.add(segment);
-        }
-    }
-
-    /**
-     * 按三级标题分割DOC内容
-     */
-    private void splitDocByLevel3(String content, HWPFDocument document, List<TextSegment> segments) {
-        StringBuilder currentLevel3Content = new StringBuilder();
-        Range level2Range = document.getRange();
-        boolean foundCurrentLevel2 = false;
-
-        for (int j = 0; j < level2Range.numParagraphs(); j++) {
-            Paragraph p = level2Range.getParagraph(j);
-            String paraText = p.text().trim();
-            if (paraText.isEmpty()) continue;
-
-            // 找到当前二级标题的开始
-            int pLevel = getDocOutlineLevel(p);
-            if (pLevel == 2 && paraText.equals(content.split("\n")[0].trim())) {
-                foundCurrentLevel2 = true;
-            }
-
-            if (foundCurrentLevel2) {
-                // 重新获取当前段落的大纲级别
-                pLevel = getDocOutlineLevel(p);
-
-                if (pLevel == 3) {
-                    // 三级标题
-                    // 保存当前三级标题内容
-                    if (currentLevel3Content.length() != 0) {
-                        TextSegment segment = TextSegment.from(currentLevel3Content.toString());
-                        segments.add(segment);
-                        currentLevel3Content = new StringBuilder();
-                    }
-                    // 开始新的三级标题内容
-                    currentLevel3Content.append(paraText).append("\n");
-                } else if (pLevel == 2 && !paraText.equals(content.split("\n")[0].trim())) {
-                    // 下一个二级标题，结束当前二级标题的处理
-                    break;
-                } else {
-                    // 普通内容或当前二级标题
-                    if (currentLevel3Content.length() != 0 || pLevel == 2) {
-                        currentLevel3Content.append(paraText).append("\n");
-                    }
-                }
-            }
-        }
-
-        // 保存最后一个三级标题内容
-        if (currentLevel3Content.length() != 0) {
-            TextSegment segment = TextSegment.from(currentLevel3Content.toString());
-            segments.add(segment);
-        }
-    }
-
-    /**
-     * 检索知识库
-     */
-    private List<TextSegment> splitDocument(File transferFile) throws Exception {
-        // 加载文档
-        Document document;
-        InputStream fileInputStream = new FileInputStream(transferFile);
-        String filename = transferFile.getName().toLowerCase();
-        if (filename.endsWith(".docx")) {
-            // 使用XWPFDocument处理DOCX文件，按段落层级分割
-            List<TextSegment> segments = new ArrayList<>();
-            try (XWPFDocument xwpfDocument = new XWPFDocument(fileInputStream)) {
-                StringBuilder currentLevel2Content = new StringBuilder();
-                boolean inLevel2Section = false;
-
-                for (XWPFParagraph paragraph : xwpfDocument.getParagraphs()) {
-                    String text = paragraph.getText().trim();
-                    if (text.isEmpty()) continue;
-
-                    int paraLevel = getDocxOutlineLevel(paragraph, xwpfDocument);
-
-                    // 二级标题（大纲级别2）开始新的分段
-                    if (paraLevel == 2) {
-                        // 保存当前二级标题下的内容
-                        if (currentLevel2Content.length() != 0) {
-                            // 检查当前二级标题内容的字节数
-                            int level2Length = currentLevel2Content.toString().getBytes().length;
-                            if (level2Length <= 65535) {
-                                // 字数不大于65535，按二级标题分割
-                                TextSegment segment = TextSegment.from(currentLevel2Content.toString());
-                                segments.add(segment);
-                            } else {
-                                // 字数大于65535，按三级标题分割
-                                splitDocxByLevel3(currentLevel2Content.toString(), xwpfDocument, segments);
-                            }
-                            currentLevel2Content = new StringBuilder();
-                        }
-                        // 开始新的二级标题内容
-                        currentLevel2Content.append(text).append("\n");
-                        inLevel2Section = true;
-                    }
-                    // 其他层级（包括三级标题）
-                    else {
-                        if (inLevel2Section) {
-                            currentLevel2Content.append(text).append("\n");
-                        }
-                    }
-                }
-
-                // 保存最后一个二级标题内容
-                if (currentLevel2Content.length() != 0) {
-                    // 检查字节数
-                    int level2Length = currentLevel2Content.toString().getBytes().length;
-                    if (level2Length <= 65535) {
-                        // 字数不大于65535，按二级标题分割
-                        TextSegment segment = TextSegment.from(currentLevel2Content.toString());
-                        segments.add(segment);
-                    } else {
-                        // 字数大于65535，按三级标题分割
-                        splitDocxByLevel3(currentLevel2Content.toString(), xwpfDocument, segments);
-                    }
-                }
-            }
-            return segments;
-        }
-        else if (filename.endsWith(".doc")) {
-            // 使用HWPFDocument处理DOC文件，按段落层级分割
-            List<TextSegment> segments = new ArrayList<>();
-            try (HWPFDocument hwpfDocument = new HWPFDocument(fileInputStream)) {
-                StringBuilder currentLevel2Content = new StringBuilder();
-                boolean inLevel2Section = false;
-
-                Range range = hwpfDocument.getRange();
-                for (int i = 0; i < range.numParagraphs(); i++) {
-                    Paragraph paragraph = range.getParagraph(i);
-                    String text = paragraph.text().trim();
-                    if (text.isEmpty()) continue;
-
-                    int paraLevel = getDocOutlineLevel(paragraph);
-
-                    if (paraLevel == 2) {
-                        // 二级标题，开始新的分段
-                        // 保存当前二级标题下的内容
-                        if (currentLevel2Content.length() != 0) {
-                            // 检查当前二级标题内容的字节数
-                            int level2Length = currentLevel2Content.toString().getBytes().length;
-                            if (level2Length <= 65535) {
-                                // 字数不大于65535，按二级标题分割
-                                TextSegment segment = TextSegment.from(currentLevel2Content.toString());
-                                segments.add(segment);
-                            } else {
-                                // 字数大于65535，按三级标题分割
-                                splitDocByLevel3(currentLevel2Content.toString(), hwpfDocument, segments);
-                            }
-                            currentLevel2Content = new StringBuilder();
-                        }
-                        // 开始新的二级标题内容
-                        currentLevel2Content.append(text).append("\n");
-                        inLevel2Section = true;
-                    } else {
-                        // 非二级标题或普通内容段落
-                        if (inLevel2Section) {
-                            currentLevel2Content.append(text).append("\n");
-                        }
-                    }
-                }
-
-                // 保存最后一个二级标题内容
-                if (currentLevel2Content.length() != 0) {
-                    // 检查字节数
-                    int level2Length = currentLevel2Content.toString().getBytes().length;
-                    if (level2Length <= 65535) {
-                        // 字数不大于65535，按二级标题分割
-                        TextSegment segment = TextSegment.from(currentLevel2Content.toString());
-                        segments.add(segment);
-                    } else {
-                        // 字数大于65535，按三级标题分割
-                        splitDocByLevel3(currentLevel2Content.toString(), hwpfDocument, segments);
-                    }
-                }
-            }
-            return segments;
-        }
-        else if (filename.endsWith(".pdf")) {
-            document = new ApachePdfBoxDocumentParser().parse(fileInputStream);
-        }
-        else if (filename.endsWith(".txt")) {
-            document = new TextDocumentParser().parse(fileInputStream);
-        }
-        else {
-            throw new UnsupportedOperationException("不支持文件类型: " + filename);
-        }
-        DocumentByParagraphSplitter splitter = new DocumentByParagraphSplitter(300,50);
-        return splitter.split(document);
-    }
-
-}

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/impl/MilvusServiceImpl.java

-package com.ruoyi.web.llm.service.impl;
-
-import com.alibaba.fastjson2.JSONArray;
-import com.alibaba.fastjson2.JSONObject;
-import com.ruoyi.web.llm.service.IMilvusService;
-import io.milvus.v2.client.ConnectConfig;
-import io.milvus.v2.client.MilvusClientV2;
-import io.milvus.v2.common.DataType;
-import io.milvus.v2.common.IndexParam;
-import io.milvus.v2.service.collection.request.*;
-import io.milvus.v2.service.collection.response.DescribeCollectionResp;
-import io.milvus.v2.service.collection.response.ListCollectionsResp;
-import io.milvus.v2.service.vector.request.DeleteReq;
-import io.milvus.v2.service.vector.request.QueryReq;
-import io.milvus.v2.service.vector.response.QueryResp;
-import org.springframework.beans.factory.annotation.Value;
-import org.springframework.stereotype.Service;
-
-import javax.annotation.PostConstruct;
-import javax.annotation.PreDestroy;
-import java.text.SimpleDateFormat;
-import java.util.*;
-import java.util.stream.Collectors;
-
-@Service
-public class MilvusServiceImpl implements IMilvusService {
-
-    @Value("${cmc.milvusService.url}")
-    private String milvusServiceUrl;
-
-    private MilvusClientV2 milvusClient;
-
-    @PostConstruct
-    public void initMilvusClient() {
-        if (milvusServiceUrl == null || milvusServiceUrl.isEmpty()) {
-            throw new IllegalArgumentException("milvusServiceUrl 配置不能为空");
-        }
-        milvusClient = new MilvusClientV2(
-                ConnectConfig.builder()
-                        .uri(milvusServiceUrl)
-                        .build());
-    }
-
-    @PreDestroy
-    public void destroyMilvusClient() {
-        if (milvusClient != null) {
-            try {
-                // 关闭 Milvus 客户端，释放 gRPC 通道
-                milvusClient.close();
-                System.out.println("Milvus client closed successfully");
-            } catch (Exception e) {
-                System.err.println("Error closing Milvus client: " + e.getMessage());
-                e.printStackTrace();
-            }
-        }
-    }
-
-    /**
-     * 新建知识库Collection（含Schema、Field、Index）
-     */
-    @Override
-    public void createCollection(String collectionName, String description, int dimension) {
-        CreateCollectionReq.CollectionSchema schema = MilvusClientV2.CreateSchema();
-
-        schema.addField(AddFieldReq.builder()
-                .fieldName("id")
-                .dataType(DataType.Int64)
-                .isPrimaryKey(true)
-                .autoID(true)
-                .build());
-
-        schema.addField(AddFieldReq.builder()
-                .fieldName("file_name")
-                .dataType(DataType.VarChar)
-                .maxLength(256)
-                .build());
-
-        schema.addField(AddFieldReq.builder()
-                .fieldName("title")
-                .dataType(DataType.VarChar)
-                .maxLength(256)
-                .build());
-
-        schema.addField(AddFieldReq.builder()
-                .fieldName("file_type")
-                .dataType(DataType.VarChar)
-                .maxLength(10)
-                .build());
-
-        schema.addField(AddFieldReq.builder()
-                .fieldName("content")
-                .dataType(DataType.VarChar)
-                .maxLength(65535)
-                .build());
-
-        schema.addField(AddFieldReq.builder()
-                .fieldName("embedding")
-                .dataType(DataType.FloatVector)
-                .dimension(dimension)
-                .build());
-
-        // 创建索引
-        Map<String, Object> extraParams = new HashMap<>();
-        extraParams.put("nlist", 64);
-        IndexParam indexParam = IndexParam.builder()
-                .fieldName("embedding")
-                .indexType(IndexParam.IndexType.IVF_FLAT)
-                .metricType(IndexParam.MetricType.COSINE)
-                .extraParams(extraParams)
-                .build();
-
-        CreateCollectionReq createCollectionParam = CreateCollectionReq.builder()
-                .collectionName(collectionName)
-                .description(description)
-                .collectionSchema(schema)
-                .indexParam(indexParam)
-                .build();
-
-        milvusClient.createCollection(createCollectionParam);
-    }
-
-    /**
-     * 查询知识库Collection
-     */
-    @Override
-    public JSONArray getCollectionNames() {
-        JSONArray jsonArray = new JSONArray();
-        ListCollectionsResp listResponse = milvusClient.listCollections();
-        if (listResponse != null) {
-            List<String> collectionNames = listResponse.getCollectionNames();
-            for (String collectionName : collectionNames) {
-                JSONObject jsonObject = new JSONObject();
-                DescribeCollectionReq request = DescribeCollectionReq.builder()
-                        .collectionName(collectionName)
-                        .build();
-                DescribeCollectionResp describeResponse = milvusClient.describeCollection(request);
-                jsonObject.put("collectionId", describeResponse.getCollectionID());
-                jsonObject.put("collectionName", collectionName);
-                SimpleDateFormat beijingFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
-                beijingFormat.setTimeZone(TimeZone.getTimeZone("Asia/Shanghai"));
-                String beijingTime = beijingFormat.format(describeResponse.getCreateUtcTime());
-                jsonObject.put("createdTime", beijingTime);
-                jsonObject.put("description", describeResponse.getDescription());
-                jsonArray.add(jsonObject);
-            }
-        }
-        return jsonArray;
-    }
-
-    /**
-     * 根据名称查询知识库Collection
-     * 返回指定名称的集合信息
-     */
-    @Override
-    public JSONArray listKnowLedgeByCollectionName(String collectionName) {
-        JSONArray jsonArray = new JSONArray();
-        ListCollectionsResp listResponse = milvusClient.listCollections();
-
-        if (listResponse != null) {
-            List<String> allCollectionNames = listResponse.getCollectionNames();
-            for (String name : allCollectionNames) {
-                JSONObject jsonObject = new JSONObject();
-                DescribeCollectionReq request = DescribeCollectionReq.builder()
-                        .collectionName(name)
-                        .build();
-                DescribeCollectionResp describeResponse = milvusClient.describeCollection(request);
-                if (describeResponse.getDescription().contains(collectionName)) {
-                    jsonObject.put("collectionId", describeResponse.getCollectionID());
-                    jsonObject.put("collectionName", name);
-                    SimpleDateFormat beijingFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
-                    beijingFormat.setTimeZone(TimeZone.getTimeZone("Asia/Shanghai"));
-                    String beijingTime = beijingFormat.format(describeResponse.getCreateUtcTime());
-                    jsonObject.put("createdTime", beijingTime);
-                    jsonObject.put("description", describeResponse.getDescription());
-                    jsonArray.add(jsonObject);
-                }
-            }
-        }
-        return jsonArray;
-    }
-
-    /**
-     * 修改知识库Collection
-     */
-    @Override
-    public void collectionRename(String collectionName, String newCollectionName) {
-        RenameCollectionReq renameCollectionReq = RenameCollectionReq.builder()
-                .collectionName(collectionName)
-                .newCollectionName(newCollectionName)
-                .build();
-
-        milvusClient.renameCollection(renameCollectionReq);
-    }
-
-    /**
-     * 删除知识库Collection
-     */
-    @Override
-    public void deleteCollectionName(String collectionName) {
-        DropCollectionReq dropCollectionReq = DropCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.dropCollection(dropCollectionReq);
-    }
-
-    /**
-     * 查询知识库文件
-     */
-    @Override
-    public List<String> listDocument(String collectionName, String fileType) {
-        List<String> documentList = new ArrayList<>();
-        LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.loadCollection(loadCollectionReq);
-        QueryReq queryParam = QueryReq.builder()
-                .collectionName(collectionName)
-                .filter("id > 0")
-                .outputFields(Arrays.asList("file_type", "file_name"))
-                .build();
-        if (fileType != null && !fileType.equals(""))
-            queryParam = QueryReq.builder()
-                    .collectionName(collectionName)
-                    .filter(String.format("file_type == \"%s\"", fileType))
-                    .outputFields(Arrays.asList("file_type", "file_name"))
-                    .build();
-        QueryResp queryResp = milvusClient.query(queryParam);
-        List<QueryResp.QueryResult> rowRecordList;
-        if (queryResp != null) {
-            rowRecordList = queryResp.getQueryResults();
-            for (QueryResp.QueryResult rowRecord : rowRecordList) {
-                documentList.add(rowRecord.getEntity().get("file_name").toString());
-            }
-        }
-        ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.releaseCollection(releaseCollectionReq);
-        return documentList.stream().distinct().collect(Collectors.toList());
-    }
-
-    /**
-     * 删除知识库文件
-     */
-    @Override
-    public void removeDocument(String collectionName, String fileName) {
-        LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.loadCollection(loadCollectionReq);
-        DeleteReq deleteReq = DeleteReq.builder()
-                .collectionName(collectionName)
-                .filter(String.format("file_name == \"%s\"", fileName))
-                .build();
-        milvusClient.delete(deleteReq);
-    }
-
-    /**
-     * 删除知识库所有文件
-     */
-    @Override
-    public void removeAllDocument(String collectionName) {
-        LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.loadCollection(loadCollectionReq);
-        DeleteReq deleteReq = DeleteReq.builder()
-                .collectionName(collectionName)
-                .filter("id > 0")
-                .build();
-        milvusClient.delete(deleteReq);
-    }
-
-    /**
-     * 列出所有的title
-     */
-    @Override
-    public List<String> listTiles(String collectionName) {
-        List<String> titleList = new ArrayList<>();
-        LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.loadCollection(loadCollectionReq);
-        QueryReq queryParam = QueryReq.builder()
-                .collectionName(collectionName)
-                .filter("id > 0 && title != \"\"")
-                .outputFields(Arrays.asList("title"))
-                .build();
-        QueryResp queryResp = milvusClient.query(queryParam);
-        List<QueryResp.QueryResult> rowRecordList;
-        if (queryResp != null) {
-            rowRecordList = queryResp.getQueryResults();
-            for (QueryResp.QueryResult rowRecord : rowRecordList) {
-                Object title = rowRecord.getEntity().get("title");
-                if (title != null) {
-                    titleList.add(title.toString());
-                }
-            }
-        }
-        ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.releaseCollection(releaseCollectionReq);
-        return titleList.stream().distinct().collect(Collectors.toList());
-    }
-
-    /**
-     * 根据文件名获取标题列表
-     */
-    @Override
-    public List<String> listTilesByFile(String collectionName, String fileName) {
-        List<String> titleList = new ArrayList<>();
-        LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.loadCollection(loadCollectionReq);
-        QueryReq queryParam = QueryReq.builder()
-                .collectionName(collectionName)
-                .filter(String.format("id > 0 && title != \"\" && file_name == \"%s\"", fileName))
-                .outputFields(Arrays.asList("title"))
-                .build();
-        QueryResp queryResp = milvusClient.query(queryParam);
-        List<QueryResp.QueryResult> rowRecordList;
-        if (queryResp != null) {
-            rowRecordList = queryResp.getQueryResults();
-            for (QueryResp.QueryResult rowRecord : rowRecordList) {
-                Object title = rowRecord.getEntity().get("title");
-                if (title != null) {
-                    titleList.add(title.toString());
-                }
-            }
-        }
-        ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.releaseCollection(releaseCollectionReq);
-        return titleList.stream().distinct().collect(Collectors.toList());
-    }
-
-    /**
-     * 根据title名查询相关content
-     */
-    @Override
-    public JSONArray listByTitle(String collectionName, String title) {
-        JSONArray resultArray = new JSONArray();
-        LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.loadCollection(loadCollectionReq);
-        QueryReq queryParam = QueryReq.builder()
-                .collectionName(collectionName)
-                .filter(String.format("title == \"%s\"", title))
-                .outputFields(Arrays.asList("content", "file_name"))
-                .build();
-        QueryResp queryResp = milvusClient.query(queryParam);
-        List<QueryResp.QueryResult> rowRecordList;
-        if (queryResp != null) {
-            rowRecordList = queryResp.getQueryResults();
-            for (QueryResp.QueryResult rowRecord : rowRecordList) {
-                JSONObject item = new JSONObject();
-                Object content = rowRecord.getEntity().get("content");
-                Object fileName = rowRecord.getEntity().get("file_name");
-                if (content != null) {
-                    item.put("content", content.toString());
-                }
-                if (fileName != null) {
-                    item.put("file_name", fileName.toString());
-                }
-                resultArray.add(item);
-            }
-        }
-        ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
-                .collectionName(collectionName)
-                .build();
-        milvusClient.releaseCollection(releaseCollectionReq);
-        return resultArray;
-    }
-
-}

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/impl/SessionServiceImpl.java

-package com.ruoyi.web.llm.service.impl;
-
-import com.alibaba.fastjson2.JSONObject;
-import com.ruoyi.llm.domain.CmcChat;
-import com.ruoyi.llm.service.ICmcChatService;
-import com.ruoyi.web.llm.service.ILangChainMilvusService;
-import com.ruoyi.web.llm.service.ISessionService;
-import org.noear.solon.ai.chat.ChatModel;
-import org.noear.solon.ai.chat.ChatResponse;
-import org.noear.solon.ai.chat.ChatSession;
-import org.noear.solon.ai.chat.message.ChatMessage;
-import org.noear.solon.ai.chat.prompt.Prompt;
-import org.noear.solon.ai.chat.session.InMemoryChatSession;
-import org.noear.solon.ai.mcp.McpChannel;
-import org.noear.solon.ai.mcp.client.McpClientProvider;
-import org.springframework.beans.factory.annotation.Autowired;
-import org.springframework.beans.factory.annotation.Value;
-import org.springframework.stereotype.Service;
-import reactor.core.publisher.Flux;
-
-import java.util.ArrayList;
-import java.util.List;
-
-@Service
-public class SessionServiceImpl implements ISessionService {
-
-    @Autowired
-    private ICmcChatService cmcChatService;
-
-    @Autowired
-    private ILangChainMilvusService langChainMilvusService;
-
-    @Value("${cmc.llmService.url}")
-    private String llmServiceUrl;
-
-    @Value("${cmc.mcpService.url}")
-    private String mcpServiceUrl;
-
-    @Override
-    public Flux<String> answer(String topicId, String question) {
-        McpClientProvider clientProvider = McpClientProvider.builder()
-                .channel(McpChannel.STREAMABLE_STATELESS)
-                .url(mcpServiceUrl)
-                .build();
-        ChatModel chatModel = ChatModel.of(llmServiceUrl)
-                .model("Qwen")
-                .defaultToolAdd(clientProvider)
-                .build();
-
-        List<ChatMessage> messages = new ArrayList<>();
-        CmcChat cmcChat = new CmcChat();
-        cmcChat.setTopicId(topicId);
-        List<CmcChat> cmcChatList = cmcChatService.selectCmcChatList(cmcChat);
-        for (CmcChat chat : cmcChatList) {
-            messages.add(ChatMessage.ofUser(chat.getInput()));
-            messages.add(ChatMessage.ofAssistant(chat.getOutput()));
-        }
-
-        // 第一步：调用 GetAllTableNames 获取所有表名
-        String step1Prompt = "你是一个MySQL专家。请调用 GetAllTableNames 工具获取数据库中所有表名列表。";
-        messages.add(ChatMessage.ofUser(step1Prompt));
-        ChatSession session1 = InMemoryChatSession.builder().messages(messages).build();
-        Prompt prompt1 = Prompt.of(step1Prompt).attrPut("session", session1);
-
-        Flux<ChatResponse> chatResponse = chatModel.prompt(prompt1).stream();
-        Flux<String> contentFlux = chatResponse
-                .concatMap(resp -> {
-                    try {
-                        // 执行第一步：获取所有表名
-                        String toolCall1 = "{\"name\": \"GetAllTableNames\", \"arguments\": {}}";
-                        String toolResult1 = clientProvider.callTool("GetAllTableNames", new JSONObject()).getContent();
-
-                        // 第二步：模型分析表名，找到相关表
-                        List<ChatMessage> messages2 = new ArrayList<>(messages);
-                        messages2.add(ChatMessage.ofAssistant("<tool_call>" + toolCall1 + "</tool_call>"));
-                        messages2.add(ChatMessage.ofUser(toolResult1));
-
-                        String step2Prompt = "请分析以下表名列表，找出与" + question + "相关的表。\n\n" +
-                        "**输出格式要求**：\n" +
-                        "- 如果需要数据库查询，请以固定格式输出：以\"相关表：\" 开头，中间用英文逗号隔开，以英文句号结尾\n" +
-                        "- 如果不需要进行数据库查询，请直接回答。\n\n" + toolResult1;
-                        messages2.add(ChatMessage.ofUser(step2Prompt));
-                        ChatSession session2 = InMemoryChatSession.builder().messages(messages2).build();
-
-                        // 生成第二步的响应
-                        return chatModel.prompt(Prompt.of(step2Prompt).attrPut("session", session2))
-                                .stream()
-                                .concatMap(resp1 -> {
-                                    try {
-                                        String content2 = resp1.getContent();
-
-                                        // 提取相关表名（从模型响应中提取）
-                                        String relevantTables = extractRelevantTables(content2);
-
-                                        // 检查是否需要继续执行（如果包含工具调用或明确需要数据库查询）
-                                        if (relevantTables.equals("")) {
-                                            // 不需要数据库查询，直接返回最终回答
-                                            return Flux.just(content2)
-                                                    .map(this::toSseDataFrame)
-                                                    .concatWith(Flux.just("data: [DONE]\n\n"));
-                                        }
-
-                                        // 第三步：调用 GetTableStructure 获取相关表的结构
-                                        List<ChatMessage> messages3 = new ArrayList<>(messages2);
-                                        messages3.add(ChatMessage.ofAssistant(content2));
-
-                                        // 执行第三步：获取表结构
-                                        StringBuilder tableStructures = new StringBuilder();
-                                        for (String tableName : relevantTables.split(",")) {
-                                            tableName = tableName.trim();
-                                            if (!tableName.isEmpty()) {
-                                                JSONObject args = new JSONObject();
-                                                args.put("tableName", tableName);
-                                                String toolResult3 = clientProvider.callTool("GetTableStructure", args)
-                                                        .getContent();
-                                                tableStructures.append("表：").append(tableName).append("\n")
-                                                        .append(toolResult3)
-                                                        .append("\n\n");
-                                            }
-                                        }
-
-                                        // 第四步：模型生成 SQL 查询
-                                        List<ChatMessage> messages4 = new ArrayList<>(messages3);
-                                        messages4.add(ChatMessage.ofAssistant(
-                                                "<tool_call>{\"name\": \"GetTableStructure\", \"arguments\": {\"tableName\": \"employee\"}}</tool_call>"));
-                                        messages4.add(ChatMessage.ofUser(tableStructures.toString()));
-
-                                        String step4Prompt = "请根据以下表结构，生成" + question + "的SQL语句：\n\n"
-                                                + tableStructures.toString();
-                                        messages4.add(ChatMessage.ofUser(step4Prompt));
-
-                                        // 生成第四步的响应
-                                        return chatModel
-                                                .prompt(Prompt.of(step4Prompt).attrPut("session",
-                                                        InMemoryChatSession.builder().messages(messages4).build()))
-                                                .stream()
-                                                .concatMap(resp2 -> {
-                                                    try {
-                                                        String content4 = resp2.getContent();
-
-                                                        // 提取 SQL 语句
-                                                        String sqlQuery = extractSqlFromContent(content4);
-
-                                                        // 第五步：调用 SQLQuery 执行查询
-                                                        List<ChatMessage> messages5 = new ArrayList<>(messages4);
-                                                        messages5.add(ChatMessage.ofAssistant(content4));
-
-                                                        // 执行第五步：执行 SQL 查询
-                                                        JSONObject sqlArgs = new JSONObject();
-                                                        sqlArgs.put("sqlString", sqlQuery);
-                                                        String toolResult5 = clientProvider
-                                                                .callTool("SQLQuery", sqlArgs)
-                                                                .getContent();
-
-                                                        // 第六步：模型生成最终回答
-                                                        List<ChatMessage> messages6 = new ArrayList<>(messages5);
-                                                        messages6.add(ChatMessage.ofAssistant(
-                                                                "<tool_call>{\"name\": \"SQLQuery\", \"arguments\": {\"sqlString\": \""
-                                                                        + sqlQuery + "\"}}</tool_call>"));
-                                                        messages6.add(ChatMessage.ofUser(toolResult5));
-
-                                                        String step6Prompt = "请根据以下查询结果，生成关于" + question + "的最终回答：\n\n"
-                                                                + toolResult5;
-                                                        messages6.add(ChatMessage.ofUser(step6Prompt));
-
-                                                        // 生成第六步的响应（流式）
-                                                        return chatModel
-                                                                .prompt(Prompt.of(step6Prompt).attrPut("session",
-                                                                        InMemoryChatSession.builder()
-                                                                                .messages(messages6)
-                                                                                .build()))
-                                                                .stream()
-                                                                .map(resp3 -> resp3.getContent())
-                                                                .filter(answer -> answer != null && !answer.isEmpty());
-                                                    } catch (Exception e) {
-                                                        return Flux.just("抱歉，系统执行查询时遇到问题，请稍后重试。")
-                                                                .map(this::toSseDataFrame)
-                                                                .concatWith(Flux.just("data: [DONE]\n\n"));
-                                                    }
-                                                });
-                                    } catch (Exception e) {
-                                        return Flux.just("抱歉，系统执行查询时遇到问题，请稍后重试。")
-                                                .map(this::toSseDataFrame)
-                                                .concatWith(Flux.just("data: [DONE]\n\n"));
-                                    }
-                                });
-                    } catch (Exception e) {
-                        return Flux.just("抱歉，系统执行查询时遇到问题，请稍后重试。")
-                                .map(this::toSseDataFrame)
-                                .concatWith(Flux.just("data: [DONE]\n\n"));
-                    }
-                });
-        // 手工拼接 SSE 帧，绕过 ServerSentEvent 编码差异；并在末尾发送 [DONE]
-        Flux<String> sseFlux = contentFlux
-                .map(this::toSseDataFrame)
-                .concatWith(Flux.just("data: [DONE]\n\n"));
-        return sseFlux;
-    }
-
-    /**
-     * 从模型响应中提取相关表名
-     */
-    private String extractRelevantTables(String content) {
-        // 简化处理，实际应该根据模型输出格式提取
-        // 假设模型会返回类似 "相关表：employee, department"
-        if (content.contains("相关表：")) {
-            int start = content.indexOf("相关表：") + 4;
-            int end = content.indexOf(".", start);
-            if (end > start) {
-                String tables = content.substring(start, end).trim();
-                // 限制相关表最多只保留10个
-                String[] tableNames = tables.split(",");
-                if (tableNames.length > 10) {
-                    StringBuilder limitedTables = new StringBuilder();
-                    for (int i = 0; i < 10; i++) {
-                        if (i > 0) {
-                            limitedTables.append(",");
-                        }
-                        limitedTables.append(tableNames[i].trim());
-                    }
-                    return limitedTables.toString();
-                }
-                return tables;
-            }
-        }
-        // 默认返回可能相关的表
-        return "";
-    }
-
-    /**
-     * 从模型响应中提取 SQL 语句
-     */
-    private String extractSqlFromContent(String content) {
-        // 简化处理，实际应该根据模型输出格式提取
-        if (content.contains("SELECT")) {
-            int start = content.indexOf("SELECT");
-            int end = content.indexOf(";", start);
-            if (end > start) {
-                return content.substring(start, end + 1);
-            }
-        }
-        return "";
-    }
-
-    @Override
-    public Flux<String> answerWithDocument(String topicId, String chatId, String question) throws Exception {
-        return langChainMilvusService.generateAnswerWithDocument(topicId, chatId, question);
-    }
-
-    private String toSseDataFrame(String data) {
-        if (data == null) {
-            return "data: \n\n";
-        }
-        // SSE 要求每一行都以 data: 开头
-        String normalized = data.replace("\r\n", "\n");
-        StringBuilder sb = new StringBuilder(normalized.length() + 16);
-        sb.append("data: ");
-        int start = 0;
-        while (true) {
-            int idx = normalized.indexOf('\n', start);
-            if (idx < 0) {
-                sb.append(normalized.substring(start));
-                break;
-            }
-            sb.append(normalized, start, idx);
-            sb.append("\n");
-            sb.append("data: ");
-            start = idx + 1;
-        }
-        sb.append("\n\n");
-        return sb.toString();
-    }
-
-}

oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/domain/AnswerStreamResponse.java

+package com.ruoyi.llm.domain;
+
+/**
+ * 回答流式响应
+ */
+public class AnswerStreamResponse {
+
+    /**
+     * 回答内容
+     */
+    private String content;
+
+    /**
+     * 是否是最后一块内容
+     */
+    private boolean isLast;
+
+    /**
+     * 是否发生错误
+     */
+    private boolean error;
+
+    /**
+     * 错误信息
+     */
+    private String errorMessage;
+
+    /**
+     * 提示词token数量
+     */
+    private long promptTokens;
+
+    /**
+     * 完成token数量
+     */
+    private long completionTokens;
+
+    /**
+     * 总token数量
+     */
+    private long totalTokens;
+
+    public AnswerStreamResponse() {
+    }
+
+    /**
+     * 创建内容响应
+     */
+    public static AnswerStreamResponse content(String content, boolean isLast) {
+        AnswerStreamResponse response = new AnswerStreamResponse();
+        response.content = content;
+        response.isLast = isLast;
+        response.error = false;
+        return response;
+    }
+
+    /**
+     * 创建错误响应
+     */
+    public static AnswerStreamResponse error(String errorMessage) {
+        AnswerStreamResponse response = new AnswerStreamResponse();
+        response.error = true;
+        response.errorMessage = errorMessage;
+        return response;
+    }
+
+    // Getters and Setters
+    public String getContent() {
+        return content;
+    }
+
+    public void setContent(String content) {
+        this.content = content;
+    }
+
+    public boolean isLast() {
+        return isLast;
+    }
+
+    public void setLast(boolean last) {
+        isLast = last;
+    }
+
+    public boolean isError() {
+        return error;
+    }
+
+    public void setError(boolean error) {
+        this.error = error;
+    }
+
+    public String getErrorMessage() {
+        return errorMessage;
+    }
+
+    public void setErrorMessage(String errorMessage) {
+        this.errorMessage = errorMessage;
+    }
+
+    public long getPromptTokens() {
+        return promptTokens;
+    }
+
+    public void setPromptTokens(long promptTokens) {
+        this.promptTokens = promptTokens;
+    }
+
+    public long getCompletionTokens() {
+        return completionTokens;
+    }
+
+    public void setCompletionTokens(long completionTokens) {
+        this.completionTokens = completionTokens;
+    }
+
+    public long getTotalTokens() {
+        return totalTokens;
+    }
+
+    public void setTotalTokens(long totalTokens) {
+        this.totalTokens = totalTokens;
+    }
+}

oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/domain/ChapterStreamResponse.java

      */
     private boolean completed;
 
+    /**
+     * 提示词token数量
+     */
+    private long promptTokens;
+
+    /**
+     * 完成token数量
+     */
+    private long completionTokens;
+
+    /**
+     * 总token数量
+     */
+    private long totalTokens;
+
     public ChapterStreamResponse() {
     }
 
         return response;
     }
 
+    public static ChapterStreamResponse progress(String progressMessage) {
+        ChapterStreamResponse response = new ChapterStreamResponse();
+        response.content = progressMessage;
+        response.completed = false;
+        response.isLast = false;
+        return response;
+    }
+
     // Getters and Setters
     public String getTitle() {
         return title;
         this.completed = completed;
     }
 
+    public long getPromptTokens() {
+        return promptTokens;
+    }
+
+    public void setPromptTokens(long promptTokens) {
+        this.promptTokens = promptTokens;
+    }
+
+    public long getCompletionTokens() {
+        return completionTokens;
+    }
+
+    public void setCompletionTokens(long completionTokens) {
+        this.completionTokens = completionTokens;
+    }
+
+    public long getTotalTokens() {
+        return totalTokens;
+    }
+
+    public void setTotalTokens(long totalTokens) {
+        this.totalTokens = totalTokens;
+    }
+
     @Override
     public String toString() {
         return "ChapterStreamResponse{" +

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/ILangChainMilvusService.java → oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/service/ILangChainMilvusService.java

-package com.ruoyi.web.llm.service;
+package com.ruoyi.llm.service;
 
 import com.alibaba.fastjson2.JSONObject;
 import org.springframework.web.multipart.MultipartFile;
-import reactor.core.publisher.Flux;
+import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
 
 import java.io.IOException;
 import java.util.List;
      * 调用LLM生成回答
      * @return
      */
-    public Flux<String> generateAnswer(String topicId, String question);
+    public SseEmitter generateAnswer(String topicId, String question);
 
     /**
      * 调用LLM+RAG（知识库）生成回答
      * @return
      */
-    public Flux<String> generateAnswerWithCollection(String topicId, String question, List<JSONObject> contexts);
+    public SseEmitter generateAnswerWithCollection(String topicId, String question, List<JSONObject> contexts);
 
     /**
      * 调用LLM+RAG（外部文件）生成回答
      * @return
      */
-    public Flux<String> generateAnswerWithDocument(String topicId, String chatId, String question) throws Exception;
+    public SseEmitter generateAnswerWithDocument(String topicId, String chatId, String question) throws Exception;
 
     /**
      * 调用LLM+RAG（外部文件+知识库）生成回答
      * @return
      */
-    public Flux<String> generateAnswerWithDocumentAndCollection(String topicId, String question,  List<JSONObject> requests) throws Exception;
+    public SseEmitter generateAnswerWithDocumentAndCollection(String topicId, String question,  List<JSONObject> requests) throws Exception;
 
     /**
      * 获取二级标题下三级标题列表

oa-back/ruoyi-llm/src/main/java/com/ruoyi/web/llm/service/IMilvusService.java → oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/service/IMilvusService.java

-package com.ruoyi.web.llm.service;
+package com.ruoyi.llm.service;
 
 import com.alibaba.fastjson2.JSONArray;
 

oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/service/ISessionService.java

+package com.ruoyi.llm.service;
+
+import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
+
+public interface ISessionService {
+
+    /**
+     * 生成回答
+     */
+    SseEmitter answer(String topicId, String question);
+
+    /**
+     * 调用LLM+RAG（外部文件）生成回答
+     */
+    SseEmitter answerWithDocument(String topicId, String chatId, String question) throws Exception;
+
+}

oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/service/impl/MilvusServiceImpl.java

+package com.ruoyi.llm.service.impl;
+
+import com.alibaba.fastjson2.JSONArray;
+import com.alibaba.fastjson2.JSONObject;
+import com.ruoyi.common.utils.milvus.MilvusConnectionPool;
+import com.ruoyi.llm.service.IMilvusService;
+import io.milvus.v2.client.MilvusClientV2;
+import io.milvus.v2.common.DataType;
+import io.milvus.v2.common.IndexParam;
+import io.milvus.v2.service.collection.request.*;
+import io.milvus.v2.service.collection.response.DescribeCollectionResp;
+import io.milvus.v2.service.collection.response.ListCollectionsResp;
+import io.milvus.v2.service.vector.request.DeleteReq;
+import io.milvus.v2.service.vector.request.QueryReq;
+import io.milvus.v2.service.vector.response.QueryResp;
+import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.stereotype.Service;
+
+import java.text.SimpleDateFormat;
+import java.util.*;
+import java.util.stream.Collectors;
+
+@Service
+public class MilvusServiceImpl implements IMilvusService {
+
+    @Autowired
+    private MilvusConnectionPool milvusConnectionPool;
+
+    /**
+     * 新建知识库Collection（含Schema、Field、Index）
+     */
+    @Override
+    public void createCollection(String collectionName, String description, int dimension) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            CreateCollectionReq.CollectionSchema schema = MilvusClientV2.CreateSchema();
+
+            schema.addField(AddFieldReq.builder()
+                    .fieldName("id")
+                    .dataType(DataType.Int64)
+                    .isPrimaryKey(true)
+                    .autoID(true)
+                    .build());
+
+            schema.addField(AddFieldReq.builder()
+                    .fieldName("file_name")
+                    .dataType(DataType.VarChar)
+                    .maxLength(256)
+                    .build());
+
+            schema.addField(AddFieldReq.builder()
+                    .fieldName("title")
+                    .dataType(DataType.VarChar)
+                    .maxLength(256)
+                    .build());
+
+            schema.addField(AddFieldReq.builder()
+                    .fieldName("file_type")
+                    .dataType(DataType.VarChar)
+                    .maxLength(10)
+                    .build());
+
+            schema.addField(AddFieldReq.builder()
+                    .fieldName("content")
+                    .dataType(DataType.VarChar)
+                    .maxLength(65535)
+                    .build());
+
+            schema.addField(AddFieldReq.builder()
+                    .fieldName("embedding")
+                    .dataType(DataType.FloatVector)
+                    .dimension(dimension)
+                    .build());
+
+            // 创建索引
+            Map<String, Object> extraParams = new HashMap<>();
+            extraParams.put("nlist", 64);
+            IndexParam indexParam = IndexParam.builder()
+                    .fieldName("embedding")
+                    .indexType(IndexParam.IndexType.IVF_FLAT)
+                    .metricType(IndexParam.MetricType.COSINE)
+                    .extraParams(extraParams)
+                    .build();
+
+            CreateCollectionReq createCollectionParam = CreateCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .description(description)
+                    .collectionSchema(schema)
+                    .indexParam(indexParam)
+                    .build();
+
+            milvusClient.createCollection(createCollectionParam);
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 查询知识库Collection
+     */
+    @Override
+    public JSONArray getCollectionNames() {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            JSONArray jsonArray = new JSONArray();
+            ListCollectionsResp listResponse = milvusClient.listCollections();
+            if (listResponse != null) {
+                List<String> collectionNames = listResponse.getCollectionNames();
+                for (String collectionName : collectionNames) {
+                    JSONObject jsonObject = new JSONObject();
+                    DescribeCollectionReq request = DescribeCollectionReq.builder()
+                            .collectionName(collectionName)
+                            .build();
+                    DescribeCollectionResp describeResponse = milvusClient.describeCollection(request);
+                    jsonObject.put("collectionId", describeResponse.getCollectionID());
+                    jsonObject.put("collectionName", collectionName);
+                    SimpleDateFormat beijingFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
+                    beijingFormat.setTimeZone(TimeZone.getTimeZone("Asia/Shanghai"));
+                    String beijingTime = beijingFormat.format(describeResponse.getCreateUtcTime());
+                    jsonObject.put("createdTime", beijingTime);
+                    jsonObject.put("description", describeResponse.getDescription());
+                    jsonArray.add(jsonObject);
+                }
+            }
+            return jsonArray;
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 根据名称查询知识库Collection
+     * 返回指定名称的集合信息
+     */
+    @Override
+    public JSONArray listKnowLedgeByCollectionName(String collectionName) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            JSONArray jsonArray = new JSONArray();
+            ListCollectionsResp listResponse = milvusClient.listCollections();
+
+            if (listResponse != null) {
+                List<String> allCollectionNames = listResponse.getCollectionNames();
+                for (String name : allCollectionNames) {
+                    JSONObject jsonObject = new JSONObject();
+                    DescribeCollectionReq request = DescribeCollectionReq.builder()
+                            .collectionName(name)
+                            .build();
+                    DescribeCollectionResp describeResponse = milvusClient.describeCollection(request);
+                    if (describeResponse.getDescription().contains(collectionName)) {
+                        jsonObject.put("collectionId", describeResponse.getCollectionID());
+                        jsonObject.put("collectionName", name);
+                        SimpleDateFormat beijingFormat = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
+                        beijingFormat.setTimeZone(TimeZone.getTimeZone("Asia/Shanghai"));
+                        String beijingTime = beijingFormat.format(describeResponse.getCreateUtcTime());
+                        jsonObject.put("createdTime", beijingTime);
+                        jsonObject.put("description", describeResponse.getDescription());
+                        jsonArray.add(jsonObject);
+                    }
+                }
+            }
+            return jsonArray;
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 修改知识库Collection
+     */
+    @Override
+    public void collectionRename(String collectionName, String newCollectionName) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            RenameCollectionReq renameCollectionReq = RenameCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .newCollectionName(newCollectionName)
+                    .build();
+
+            milvusClient.renameCollection(renameCollectionReq);
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 删除知识库Collection
+     */
+    @Override
+    public void deleteCollectionName(String collectionName) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            DropCollectionReq dropCollectionReq = DropCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.dropCollection(dropCollectionReq);
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 查询知识库文件
+     */
+    @Override
+    public List<String> listDocument(String collectionName, String fileType) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            List<String> documentList = new ArrayList<>();
+            LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.loadCollection(loadCollectionReq);
+            QueryReq queryParam = QueryReq.builder()
+                    .collectionName(collectionName)
+                    .filter("id > 0")
+                    .outputFields(Arrays.asList("file_type", "file_name"))
+                    .build();
+            if (fileType != null && !fileType.equals(""))
+                queryParam = QueryReq.builder()
+                        .collectionName(collectionName)
+                        .filter(String.format("file_type == \"%s\"", fileType))
+                        .outputFields(Arrays.asList("file_type", "file_name"))
+                        .build();
+            QueryResp queryResp = milvusClient.query(queryParam);
+            List<QueryResp.QueryResult> rowRecordList;
+            if (queryResp != null) {
+                rowRecordList = queryResp.getQueryResults();
+                for (QueryResp.QueryResult rowRecord : rowRecordList) {
+                    documentList.add(rowRecord.getEntity().get("file_name").toString());
+                }
+            }
+            ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.releaseCollection(releaseCollectionReq);
+            return documentList.stream().distinct().collect(Collectors.toList());
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 删除知识库文件
+     */
+    @Override
+    public void removeDocument(String collectionName, String fileName) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.loadCollection(loadCollectionReq);
+            DeleteReq deleteReq = DeleteReq.builder()
+                    .collectionName(collectionName)
+                    .filter(String.format("file_name == \"%s\"", fileName))
+                    .build();
+            milvusClient.delete(deleteReq);
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 删除知识库所有文件
+     */
+    @Override
+    public void removeAllDocument(String collectionName) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.loadCollection(loadCollectionReq);
+            DeleteReq deleteReq = DeleteReq.builder()
+                    .collectionName(collectionName)
+                    .filter("id > 0")
+                    .build();
+            milvusClient.delete(deleteReq);
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 列出所有的title
+     */
+    @Override
+    public List<String> listTiles(String collectionName) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            List<String> titleList = new ArrayList<>();
+            LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.loadCollection(loadCollectionReq);
+            QueryReq queryParam = QueryReq.builder()
+                    .collectionName(collectionName)
+                    .filter("id > 0")
+                    .outputFields(Arrays.asList("title"))
+                    .build();
+            QueryResp queryResp = milvusClient.query(queryParam);
+            List<QueryResp.QueryResult> rowRecordList;
+            if (queryResp != null) {
+                rowRecordList = queryResp.getQueryResults();
+                for (QueryResp.QueryResult rowRecord : rowRecordList) {
+                    Object title = rowRecord.getEntity().get("title");
+                    if (title != null) {
+                        titleList.add(title.toString());
+                    }
+                }
+            }
+            ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.releaseCollection(releaseCollectionReq);
+            return titleList.stream().distinct().collect(Collectors.toList());
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 根据文件名获取标题列表
+     */
+    @Override
+    public List<String> listTilesByFile(String collectionName, String fileName) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            List<String> titleList = new ArrayList<>();
+            LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.loadCollection(loadCollectionReq);
+            QueryReq queryParam = QueryReq.builder()
+                    .collectionName(collectionName)
+                    .filter(String.format("id > 0 && file_name == \"%s\"", fileName))
+                    .outputFields(Arrays.asList("title"))
+                    .build();
+            QueryResp queryResp = milvusClient.query(queryParam);
+            List<QueryResp.QueryResult> rowRecordList;
+            if (queryResp != null) {
+                rowRecordList = queryResp.getQueryResults();
+                for (QueryResp.QueryResult rowRecord : rowRecordList) {
+                    Object title = rowRecord.getEntity().get("title");
+                    if (title != null) {
+                        titleList.add(title.toString());
+                    }
+                }
+            }
+            ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.releaseCollection(releaseCollectionReq);
+            return titleList.stream().distinct().collect(Collectors.toList());
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+
+    /**
+     * 根据title名查询相关content
+     */
+    @Override
+    public JSONArray listByTitle(String collectionName, String title) {
+        MilvusClientV2 milvusClient = null;
+        try {
+            milvusClient = milvusConnectionPool.getClient();
+            JSONArray resultArray = new JSONArray();
+            LoadCollectionReq loadCollectionReq = LoadCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.loadCollection(loadCollectionReq);
+            QueryReq queryParam = QueryReq.builder()
+                    .collectionName(collectionName)
+                    .filter(String.format("title == \"%s\"", title))
+                    .outputFields(Arrays.asList("content", "file_name"))
+                    .build();
+            QueryResp queryResp = milvusClient.query(queryParam);
+            List<QueryResp.QueryResult> rowRecordList;
+            if (queryResp != null) {
+                rowRecordList = queryResp.getQueryResults();
+                for (QueryResp.QueryResult rowRecord : rowRecordList) {
+                    JSONObject item = new JSONObject();
+                    Object content = rowRecord.getEntity().get("content");
+                    Object fileName = rowRecord.getEntity().get("file_name");
+                    if (content != null) {
+                        item.put("content", content.toString());
+                    }
+                    if (fileName != null) {
+                        item.put("file_name", fileName.toString());
+                    }
+                    resultArray.add(item);
+                }
+            }
+            ReleaseCollectionReq releaseCollectionReq = ReleaseCollectionReq.builder()
+                    .collectionName(collectionName)
+                    .build();
+            milvusClient.releaseCollection(releaseCollectionReq);
+            return resultArray;
+        } catch (InterruptedException e) {
+            Thread.currentThread().interrupt();
+            throw new RuntimeException("获取Milvus连接失败", e);
+        } finally {
+            if (milvusClient != null) {
+                milvusConnectionPool.returnClient(milvusClient);
+            }
+        }
+    }
+}

oa-back/ruoyi-system/src/main/java/com/ruoyi/llm/service/impl/SessionServiceImpl.java

+package com.ruoyi.llm.service.impl;
+
+import com.alibaba.fastjson2.JSONArray;
+import com.alibaba.fastjson2.JSONObject;
+import com.ruoyi.llm.domain.CmcChat;
+import com.ruoyi.llm.service.ICmcChatService;
+import com.ruoyi.llm.domain.AnswerStreamResponse;
+import com.ruoyi.llm.service.ILangChainMilvusService;
+import com.ruoyi.llm.service.ISessionService;
+import com.fasterxml.jackson.databind.ObjectMapper;
+import org.noear.solon.ai.chat.ChatModel;
+import org.noear.solon.ai.chat.message.ChatMessage;
+import org.noear.solon.ai.chat.prompt.Prompt;
+import org.noear.solon.ai.chat.session.InMemoryChatSession;
+import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.beans.factory.annotation.Value;
+import org.springframework.jdbc.core.JdbcTemplate;
+import org.springframework.stereotype.Service;
+import org.springframework.web.servlet.mvc.method.annotation.SseEmitter;
+
+import java.io.IOException;
+import java.sql.Connection;
+import java.sql.DatabaseMetaData;
+import java.sql.PreparedStatement;
+import java.sql.ResultSet;
+import java.sql.ResultSetMetaData;
+import java.sql.SQLException;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Map;
+import java.util.concurrent.*;
+import java.util.concurrent.atomic.AtomicBoolean;
+import java.util.concurrent.atomic.AtomicLong;
+
+@Service
+public class SessionServiceImpl implements ISessionService {
+
+    @Autowired
+    private ICmcChatService cmcChatService;
+
+    @Autowired
+    private ILangChainMilvusService langChainMilvusService;
+
+    @Autowired
+    private JdbcTemplate jdbcTemplate;
+
+    @Value("${cmc.llmService.url}")
+    private String llmServiceUrl;
+
+    /**
+     * 获取所有表名列表（只包含cmc、sys和files开头的表）
+     */
+    private String getAllTableNames() {
+        JSONArray tableNames = new JSONArray();
+        try {
+            String sql = "SELECT table_name, table_comment FROM information_schema.tables WHERE table_schema = DATABASE() "
+                    +
+                    "AND (table_name LIKE 'cmc%' OR table_name LIKE 'sys%' OR table_name LIKE 'files%')";
+            List<Map<String, Object>> tables = jdbcTemplate.queryForList(sql);
+
+            for (Map<String, Object> table : tables) {
+                JSONObject tableInfo = new JSONObject();
+                tableInfo.put("tableName", table.get("table_name"));
+                tableInfo.put("tableComment", table.get("table_comment") != null ? table.get("table_comment") : "");
+                tableNames.add(tableInfo);
+            }
+        } catch (Exception e) {
+            throw new RuntimeException("获取表名列表失败", e);
+        }
+        return tableNames.toJSONString();
+    }
+
+    /**
+     * 获取指定表的结构信息
+     */
+    private String getTableStructure(String tableName) {
+        JSONObject result = new JSONObject();
+
+        jdbcTemplate.execute((Connection conn) -> {
+            try {
+                DatabaseMetaData metaData = conn.getMetaData();
+
+                ResultSet columns = metaData.getColumns(null, null, tableName, null);
+                JSONArray columnsInfo = new JSONArray();
+
+                while (columns.next()) {
+                    JSONObject column = new JSONObject();
+                    column.put("columnName", columns.getString("COLUMN_NAME"));
+                    column.put("dataType", columns.getString("TYPE_NAME"));
+                    column.put("columnSize", columns.getString("COLUMN_SIZE"));
+                    column.put("isNullable", columns.getString("IS_NULLABLE"));
+                    column.put("columnComment", columns.getString("REMARKS"));
+                    columnsInfo.add(column);
+                }
+                columns.close();
+
+                ResultSet primaryKeys = metaData.getPrimaryKeys(null, null, tableName);
+                JSONArray primaryKeyColumns = new JSONArray();
+
+                while (primaryKeys.next()) {
+                    primaryKeyColumns.add(primaryKeys.getString("COLUMN_NAME"));
+                }
+                primaryKeys.close();
+
+                result.put("tableName", tableName);
+                result.put("columns", columnsInfo);
+                result.put("primaryKey", primaryKeyColumns);
+
+            } catch (SQLException e) {
+                throw new RuntimeException("获取表结构失败: " + tableName, e);
+            }
+            return null;
+        });
+
+        return result.toJSONString();
+    }
+
+    /**
+     * 执行SQL查询
+     */
+    private String executeSQL(String sqlString) {
+        String cleanedSqlString = sqlString.replace("[", "").replace("]", "");
+
+        JSONArray result = new JSONArray();
+
+        jdbcTemplate.execute((Connection conn) -> {
+            try (PreparedStatement stmt = conn.prepareStatement(cleanedSqlString)) {
+                ResultSet rs = stmt.executeQuery();
+                ResultSetMetaData metaData = rs.getMetaData();
+                int columnCount = metaData.getColumnCount();
+
+                while (rs.next()) {
+                    JSONObject row = new JSONObject();
+                    for (int i = 1; i <= columnCount; i++) {
+                        String columnName = metaData.getColumnLabel(i);
+                        Object value = rs.getObject(i);
+                        row.put(columnName, value);
+                    }
+                    result.add(row);
+                }
+                rs.close();
+            } catch (SQLException e) {
+                throw new RuntimeException("SQL执行失败", e);
+            }
+            return null;
+        });
+
+        return result.toJSONString();
+    }
+
+    @Override
+    public SseEmitter answer(String topicId, String question) {
+        // 设置SSE超时时间为10分钟
+        SseEmitter emitter = new SseEmitter(600000L);
+
+        // 设置回调
+        emitter.onCompletion(() -> {
+        });
+        emitter.onTimeout(() -> {
+            emitter.complete();
+        });
+        emitter.onError(e -> System.err.println("SSE连接错误: " + e.getMessage()));
+
+        // 使用独立线程池处理，避免阻塞主线程
+        ExecutorService executor = Executors.newSingleThreadExecutor();
+
+        executor.execute(() -> {
+            ObjectMapper objectMapper = new ObjectMapper();
+            try {
+                // 用于累计token用量
+                AtomicLong[] tokenUsage = new AtomicLong[3];
+                tokenUsage[0] = new AtomicLong(0); // promptTokens
+                tokenUsage[1] = new AtomicLong(0); // completionTokens
+                tokenUsage[2] = new AtomicLong(0); // totalTokens
+                System.out.println("llmServiceUrl: " + llmServiceUrl);
+
+                ChatModel chatModel = ChatModel.of(llmServiceUrl).model("Qwen")
+                        .build();
+                // ChatModel chatModel =
+                // ChatModel.of("https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions")
+                // .model("qwen3-vl-32b-instruct")
+                // .apiKey("sk-750a17cc723847f28b31fa1bc17c255c")
+                // .build();
+
+                System.out.println("成功");
+                List<ChatMessage> messages = new ArrayList<>();
+                CmcChat cmcChat = new CmcChat();
+                cmcChat.setTopicId(topicId);
+                List<CmcChat> cmcChatList = cmcChatService.selectCmcChatList(cmcChat);
+                for (CmcChat chat : cmcChatList) {
+                    messages.add(ChatMessage.ofUser(chat.getInput()));
+                    messages.add(ChatMessage.ofAssistant(chat.getOutput()));
+                }
+
+                // 第一步：调用 getAllTableNames 获取所有表名（使用本地方法替代MCP）
+                String toolResult1 = getAllTableNames();
+                System.out.println(toolResult1);
+
+                // 第二步：模型分析表名，找到相关表
+                String step2Prompt = "请分析以下表名列表，找出与" + question + "相关的表。\n\n" +
+                        "**输出格式要求**：\n" +
+                        "- 如果需要数据库查询，请以固定格式输出：以\"相关表：\" 开头，中间用英文逗号隔开，以英文句号结尾\n" +
+                        "- 如果不需要进行数据库查询，请直接回答。\n\n" + toolResult1;
+
+                // 使用 CountDownLatch 等待生成完成
+                CountDownLatch latch = new CountDownLatch(1);
+                AtomicBoolean cancelled = new AtomicBoolean(false);
+                StringBuilder step2ContentBuilder = new StringBuilder();
+
+                // 生成第二步的响应
+                Prompt step3Prompt = Prompt.of(step2Prompt).attrPut("session",
+                        InMemoryChatSession.builder().messages(messages).build());
+                chatModel.prompt(step3Prompt)
+                        .stream()
+                        .subscribe(
+                                resp1 -> {
+                                    if (cancelled.get()) {
+                                        return;
+                                    }
+                                    // 收集流式响应内容
+                                    String chunk = resp1.getContent();
+                                    if (chunk != null && !chunk.isEmpty()) {
+                                        step2ContentBuilder.append(chunk);
+                                    }
+                                },
+                                error -> {
+                                    if (cancelled.getAndSet(true)) {
+                                        return;
+                                    }
+                                    try {
+                                        emitter.send(SseEmitter.event()
+                                                .name("error")
+                                                .data("生成回答时发生错误: " + error.getMessage()));
+                                    } catch (Exception sendEx) {
+                                    }
+                                    latch.countDown();
+                                },
+                                () -> {
+                                    if (cancelled.get()) {
+                                        return;
+                                    }
+                                    try {
+                                        String content2 = step2ContentBuilder.toString();
+                                        System.out.println(content2);
+                                        // 提取相关表名（从完整的模型响应中提取）
+                                        String relevantTables = extractRelevantTables(content2);
+                                        System.out.println(relevantTables);
+
+                                        // 检查是否需要继续执行（如果包含工具调用或明确需要数据库查询）
+                                        if (relevantTables.equals("")) {
+                                            // 不需要数据库查询，直接返回最终回答
+                                            // 估算token用量
+                                            StringBuilder fullPrompt = new StringBuilder();
+                                            for (ChatMessage msg : messages) {
+                                                fullPrompt.append(msg.getContent());
+                                            }
+                                            fullPrompt.append(step2Prompt);
+                                            tokenUsage[0].set((long) fullPrompt.length());
+                                            tokenUsage[1].set((long) content2.length());
+                                            tokenUsage[2].set(tokenUsage[0].get() + tokenUsage[1].get());
+
+                                            AnswerStreamResponse response = AnswerStreamResponse.content(content2,
+                                                    false);
+                                            response.setPromptTokens(tokenUsage[0].get());
+                                            response.setCompletionTokens(tokenUsage[1].get());
+                                            response.setTotalTokens(tokenUsage[2].get());
+                                            emitter.send(SseEmitter.event()
+                                                    .name("answer")
+                                                    .data(objectMapper.writeValueAsString(response)));
+
+                                            AnswerStreamResponse completeResponse = AnswerStreamResponse
+                                                    .content("[DONE]", true);
+                                            completeResponse.setPromptTokens(tokenUsage[0].get());
+                                            completeResponse.setCompletionTokens(tokenUsage[1].get());
+                                            completeResponse.setTotalTokens(tokenUsage[2].get());
+                                            emitter.send(SseEmitter.event()
+                                                    .name("completed")
+                                                    .data(objectMapper.writeValueAsString(completeResponse)));
+                                            latch.countDown();
+                                            return;
+                                        }
+
+                                        // 第三步：调用 getTableStructure 获取相关表的结构（使用本地方法替代MCP）
+                                        // 执行第三步：获取表结构
+                                        StringBuilder tableStructures = new StringBuilder();
+                                        for (String tableName : relevantTables.split(",")) {
+                                            tableName = tableName.trim();
+                                            if (!tableName.isEmpty()) {
+                                                String toolResult3 = getTableStructure(tableName);
+                                                tableStructures.append("表：").append(tableName).append("\n")
+                                                        .append(toolResult3)
+                                                        .append("\n\n");
+                                            }
+                                        }
+
+                                        // 第四步：模型生成 SQL 查询
+                                        String step4Prompt = "请根据以下表结构，生成" + question + "的SQL语句：\n\n"
+                                                + tableStructures.toString();
+
+                                        // 生成第四步的响应 - 同样需要收集完整响应后再提取SQL
+                                        StringBuilder step4ContentBuilder = new StringBuilder();
+
+                                        Prompt step5Prompt = Prompt.of(step4Prompt).attrPut("session",
+                                                InMemoryChatSession.builder().messages(messages).build());
+                                        chatModel.prompt(step5Prompt)
+                                                .stream()
+                                                .subscribe(
+                                                        resp2 -> {
+                                                            if (cancelled.get()) {
+                                                                return;
+                                                            }
+                                                            // 收集流式响应内容
+                                                            String chunk = resp2.getContent();
+                                                            if (chunk != null && !chunk.isEmpty()) {
+                                                                step4ContentBuilder.append(chunk);
+                                                            }
+                                                        },
+                                                        step4Error -> {
+                                                            if (cancelled.getAndSet(true)) {
+                                                                return;
+                                                            }
+                                                            try {
+                                                                emitter.send(SseEmitter.event()
+                                                                        .name("error")
+                                                                        .data("生成SQL时发生错误: "
+                                                                                + step4Error.getMessage()));
+                                                            } catch (Exception sendEx) {
+                                                            }
+                                                            latch.countDown();
+                                                        },
+                                                        () -> {
+                                                            if (cancelled.get()) {
+                                                                return;
+                                                            }
+                                                            try {
+                                                                String content4 = step4ContentBuilder.toString();
+
+                                                                // 提取 SQL 语句
+                                                                String sqlQuery = extractSqlFromContent(content4);
+
+                                                                // 第五步：调用 executeSQL 执行查询（使用本地方法替代MCP）
+                                                                // 执行第五步：执行 SQL 查询
+                                                                String toolResult5 = executeSQL(sqlQuery);
+
+                                                                // 第六步：模型生成最终回答
+                                                                String step6Prompt = "请根据以下查询结果，生成关于" + question
+                                                                        + "的最终回答：\n\n"
+                                                                        + toolResult5;
+
+                                                                // 估算prompt的token数量（包含历史对话）
+                                                                StringBuilder fullPrompt = new StringBuilder();
+                                                                for (ChatMessage msg : messages) {
+                                                                    fullPrompt.append(msg.getContent());
+                                                                }
+                                                                fullPrompt.append(step6Prompt);
+                                                                tokenUsage[0].set((long) fullPrompt.length());
+
+                                                                // 生成第六步的响应（流式）
+                                                                Prompt step7Prompt = Prompt.of(step6Prompt).attrPut(
+                                                                        "session",
+                                                                        InMemoryChatSession.builder().messages(messages)
+                                                                                .build());
+                                                                chatModel.prompt(step7Prompt)
+                                                                        .stream()
+                                                                        .map(resp -> {
+                                                                            String content = resp.getContent();
+                                                                            // 累计completion tokens
+                                                                            if (content != null) {
+                                                                                tokenUsage[1].addAndGet(
+                                                                                        (long) content.length());
+                                                                                tokenUsage[2].set(tokenUsage[0].get()
+                                                                                        + tokenUsage[1].get());
+                                                                            }
+                                                                            return content;
+                                                                        })
+                                                                        .subscribe(
+                                                                                content -> {
+                                                                                    if (cancelled.get()) {
+                                                                                        return;
+                                                                                    }
+                                                                                    try {
+                                                                                        if (content != null
+                                                                                                && !content.isEmpty()) {
+                                                                                            AnswerStreamResponse response = AnswerStreamResponse
+                                                                                                    .content(content,
+                                                                                                            false);
+                                                                                            response.setPromptTokens(
+                                                                                                    tokenUsage[0]
+                                                                                                            .get());
+                                                                                            response.setCompletionTokens(
+                                                                                                    tokenUsage[1]
+                                                                                                            .get());
+                                                                                            response.setTotalTokens(
+                                                                                                    tokenUsage[2]
+                                                                                                            .get());
+                                                                                            emitter.send(SseEmitter
+                                                                                                    .event()
+                                                                                                    .name("answer")
+                                                                                                    .data(objectMapper
+                                                                                                            .writeValueAsString(
+                                                                                                                    response)));
+                                                                                        }
+                                                                                    } catch (Exception e) {
+                                                                                        cancelled.set(true);
+                                                                                    }
+                                                                                },
+                                                                                error -> {
+                                                                                    if (cancelled.getAndSet(true)) {
+                                                                                        return;
+                                                                                    }
+                                                                                    try {
+                                                                                        AnswerStreamResponse errorResponse = AnswerStreamResponse
+                                                                                                .error("生成回答时发生错误: "
+                                                                                                        + error.getMessage());
+                                                                                        emitter.send(SseEmitter.event()
+                                                                                                .name("error")
+                                                                                                .data(objectMapper
+                                                                                                        .writeValueAsString(
+                                                                                                                errorResponse)));
+                                                                                    } catch (Exception sendEx) {
+                                                                                    }
+                                                                                    latch.countDown();
+                                                                                },
+                                                                                () -> {
+                                                                                    if (cancelled.getAndSet(true)) {
+                                                                                        return;
+                                                                                    }
+                                                                                    try {
+                                                                                        // 发送完成标记（包含最终token用量）
+                                                                                        AnswerStreamResponse completeResponse = AnswerStreamResponse
+                                                                                                .content("[DONE]",
+                                                                                                        true);
+                                                                                        completeResponse
+                                                                                                .setPromptTokens(
+                                                                                                        tokenUsage[0]
+                                                                                                                .get());
+                                                                                        completeResponse
+                                                                                                .setCompletionTokens(
+                                                                                                        tokenUsage[1]
+                                                                                                                .get());
+                                                                                        completeResponse.setTotalTokens(
+                                                                                                tokenUsage[2].get());
+                                                                                        emitter.send(SseEmitter.event()
+                                                                                                .name("completed")
+                                                                                                .data(objectMapper
+                                                                                                        .writeValueAsString(
+                                                                                                                completeResponse)));
+                                                                                    } catch (Exception e) {
+                                                                                    }
+                                                                                    latch.countDown();
+                                                                                });
+                                                            } catch (Exception e) {
+                                                                if (cancelled.getAndSet(true)) {
+                                                                    return;
+                                                                }
+                                                                try {
+                                                                    emitter.send(SseEmitter.event()
+                                                                            .name("error")
+                                                                            .data("抱歉，系统执行查询时遇到问题，请稍后重试。"));
+                                                                } catch (Exception sendEx) {
+                                                                }
+                                                                latch.countDown();
+                                                            }
+                                                        });
+                                    } catch (Exception e) {
+                                        if (cancelled.getAndSet(true)) {
+                                            return;
+                                        }
+                                        try {
+                                            emitter.send(SseEmitter.event()
+                                                    .name("error")
+                                                    .data("抱歉，系统执行查询时遇到问题，请稍后重试。"));
+                                        } catch (Exception sendEx) {
+                                        }
+                                        latch.countDown();
+                                    }
+                                });
+
+                // 等待生成完成
+                latch.await();
+
+                emitter.complete();
+
+            } catch (Exception e) {
+                try {
+                    emitter.send(SseEmitter.event()
+                            .name("error")
+                            .data("生成过程中发生错误: " + e.getMessage()));
+                } catch (IOException sendEx) {
+                    System.err.println("发送错误消息失败: " + sendEx.getMessage());
+                }
+                emitter.completeWithError(e);
+            } finally {
+                executor.shutdown();
+            }
+        });
+
+        return emitter;
+    }
+
+    /**
+     * 从模型响应中提取相关表名
+     */
+    private String extractRelevantTables(String content) {
+        if (content == null || content.trim().isEmpty()) {
+            return "";
+        }
+
+        // 尝试匹配 "相关表：" 或 "相关表\n" 格式
+        String[] markers = { "相关表：", "相关表", "相关表名：", "相关表名" };
+        int start = -1;
+        String matchedMarker = "";
+
+        for (String marker : markers) {
+            int idx = content.indexOf(marker);
+            if (idx != -1 && (start == -1 || idx < start)) {
+                start = idx;
+                matchedMarker = marker;
+            }
+        }
+
+        if (start != -1) {
+            start += matchedMarker.length();
+            // 跳过可能的换行符
+            while (start < content.length() && (content.charAt(start) == '\n' || content.charAt(start) == '\r'
+                    || content.charAt(start) == ' ' || content.charAt(start) == '：')) {
+                start++;
+            }
+
+            // 找到表名结束位置（换行、句号或字符串结尾）
+            int end = start;
+            while (end < content.length() && content.charAt(end) != '\n' && content.charAt(end) != '\r'
+                    && content.charAt(end) != '.') {
+                end++;
+            }
+
+            if (end > start) {
+                String tables = content.substring(start, end).trim();
+                // 限制相关表最多只保留10个
+                String[] tableNames = tables.split(",");
+                if (tableNames.length > 10) {
+                    StringBuilder limitedTables = new StringBuilder();
+                    for (int i = 0; i < 10; i++) {
+                        if (i > 0) {
+                            limitedTables.append(",");
+                        }
+                        limitedTables.append(tableNames[i].trim());
+                    }
+                    return limitedTables.toString();
+                }
+                return tables;
+            }
+        }
+        // 默认返回可能相关的表
+        return "";
+    }
+
+    /**
+     * 从模型响应中提取 SQL 语句
+     */
+    private String extractSqlFromContent(String content) {
+        // 简化处理，实际应该根据模型输出格式提取
+        if (content.contains("SELECT")) {
+            int start = content.indexOf("SELECT");
+            int end = content.indexOf(";", start);
+            if (end > start) {
+                return content.substring(start, end + 1);
+            }
+        }
+        return "";
+    }
+
+    @Override
+    public SseEmitter answerWithDocument(String topicId, String chatId, String question) throws Exception {
+        return langChainMilvusService.generateAnswerWithDocument(topicId, chatId, question);
+    }
+
+}

oa-back/ruoyi-system/src/main/java/com/ruoyi/oa/domain/CmcTransfer.java

 
     public SysPost getAfterPost()
     {
-        return beforePost;
+        return afterPost;
     }
     public void setBeforeDept(SysDept beforeDept)
     {
 
     public SysDept getAfterDept()
     {
-        return beforeDept;
+        return afterDept;
     }
     public void setApplierUser(SysUser applierUser)
     {
 
     public SysUser getAfterDeptUser()
     {
-        return beforeDeptUser;
+        return afterDeptUser;
     }
     public void setZhUser(SysUser zhUser)
     {

oa-back/ruoyi-system/src/main/resources/mapper/oa/CmcContractMapper.xml

     </select>
 
     <select id="selectCmcContractListByRange" parameterType="CmcContract" resultMap="CmcContractResult">
-        select distinct t1.contract_id, t1.contract_code, t1.contract_name, t1.contract_number, t1.tender_id, t1.t_project_name,t1.party_a_id, t1.party_a_name, t1.a_person,
-        t1.a_phone, t1.agent, t1.agent_person,t1.paid_amount, t1.paid_percentage, t1.invoice_amount, t1.invoice_percentage, p.project_number, p.project_name, p.project_source,
-        t1.agent_phone, t1.amount, t1.deposit, t1.contract_document, t1.drafter, t1.draft_nick_name, t1.draft_time, t1.remark, t1.sign_date, t1.sign_remark, t1.sign_scan,
-        t1.comment_type, t1.manager_comment, t1.manager_user_id, t1.manager_nick_name, t1.manager_time, t1.gm_user_id, t1.gm_nick_name, t1.gm_time, t1.gm_comment from
-        (select distinct c.contract_id, c.contract_code, c.contract_name, c.contract_number, c.tender_id, t.project_name as t_project_name,c.party_a_id, pa.party_a_name,
+        select distinct t3.*, p.project_number, p.project_name, p.project_source from
+        (select t1.*, t2.invoice_amount, t2.invoice_percentage from (select distinct c.contract_id, c.contract_code, c.contract_name, c.contract_number, c.tender_id, t.project_name as t_project_name,c.party_a_id, pa.party_a_name,
         t.a_person as a_person, t.a_phone as a_phone, t.agent as agent, t.agent_person as agent_person, t.agent_phone as agent_phone,
         sum(cp.paid_amount) as paid_amount, case c.amount when 0 then 0 else round(sum(cp.paid_amount) / c.amount * 100, 2) end as paid_percentage,
-        sum(ci.invoice_amount) as invoice_amount, case c.amount when 0 then 0 else round(sum(ci.invoice_amount) / c.amount * 100, 2) end as invoice_percentage,
         c.amount, c.deposit, c.contract_document, c.drafter, u.nick_name as draft_nick_name, c.draft_time, c.remark, c.sign_date, c.sign_remark, c.sign_scan,
         c.comment_type, c.manager_comment, c.manager_user_id, u1.nick_name as manager_nick_name, c.manager_time, c.gm_user_id, u2.nick_name as gm_nick_name, c.gm_time, c.gm_comment from cmc_contract as c
         left join sys_user as u on u.user_id = c.drafter
         left join cmc_tender as t on t.tender_id = c.tender_id
         left join cmc_party_a as pa on pa.party_a_id = c.party_a_id
         left join cmc_contract_paid as cp on cp.contract_id = c.contract_id
+        group by c.contract_id) as t1
+        left join
+        (select distinct c.contract_id, 
+        sum(ci.invoice_amount) as invoice_amount, case c.amount when 0 then 0 else round(sum(ci.invoice_amount) / c.amount * 100, 2) end as invoice_percentage
+        from cmc_contract as c
         left join cmc_contract_invoice as ci on ci.contract_id = c.contract_id
-        group by c.contract_id)
-        as t1
-        left join cmc_project_contract as pc on pc.contract_id = t1.contract_id
+        group by c.contract_id) as t2
+        on t1.contract_id = t2.contract_id)
+        as t3
+        left join cmc_project_contract as pc on pc.contract_id = t3.contract_id
         left join cmc_project as p on pc.project_id = p.project_id
         <where>
-            <if test="contractId != null  and contractId != ''"> and t1.contract_id like concat('%', #{contractId}, '%')</if>
-            <if test="contractName!= null  and contractName != ''"> and t1.contract_name like concat('%', #{contractName}, '%')</if>
-            <if test="contractCode!= null  and contractCode != ''"> and t1.contract_code like concat('%', #{contractCode}, '%')</if>
-            <if test="tenderId != null  and tenderId != ''"> and t1.tender_id = #{tenderId}</if>
-            <if test="contractNumber != null  and contractNumber != ''"> and t1.contract_number = #{contractNumber}</if>
-            <if test="partyAId != null  and partyAId != ''"> and t1.party_a_id = #{partyAId}</if>
-            <if test="amount != null "> and t1.amount = #{amount}</if>
-            <if test="deposit != null "> and t1.deposit = #{deposit}</if>
-            <if test="contractDocument != null  and contractDocument != ''"> and t1.contract_document = #{contractDocument}</if>
-            <if test="drafter != null "> and t1.drafter = #{drafter}</if>
-            <if test="draftTime != null "> and t1.draft_time = #{draftTime}</if>
-            <if test="signDate != null"> and YEAR(t1.sign_date) = YEAR(#{signDate})</if>
-            <if test="signRemark != null  and signRemark != ''"> and t1.sign_remark = #{signRemark}</if>
-            <if test="signScan != null  and signScan != ''"> and t1.sign_scan = #{signScan}</if>
-            <if test="commentType != null  and commentType != ''"> and t1.comment_type = #{commentType}</if>
+            <if test="contractId != null  and contractId != ''"> and t3.contract_id like concat('%', #{contractId}, '%')</if>
+            <if test="contractName!= null  and contractName != ''"> and t3.contract_name like concat('%', #{contractName}, '%')</if>
+            <if test="contractCode!= null  and contractCode != ''"> and t3.contract_code like concat('%', #{contractCode}, '%')</if>
+            <if test="tenderId != null  and tenderId != ''"> and t3.tender_id = #{tenderId}</if>
+            <if test="contractNumber != null  and contractNumber != ''"> and t3.contract_number = #{contractNumber}</if>
+            <if test="partyAId != null  and partyAId != ''"> and t3.party_a_id = #{partyAId}</if>
+            <if test="amount != null "> and t3.amount = #{amount}</if>
+            <if test="deposit != null "> and t3.deposit = #{deposit}</if>
+            <if test="contractDocument != null  and contractDocument != ''"> and t3.contract_document = #{contractDocument}</if>
+            <if test="drafter != null "> and t3.drafter = #{drafter}</if>
+            <if test="draftTime != null "> and t3.draft_time = #{draftTime}</if>
+            <if test="signDate != null"> and YEAR(t3.sign_date) = YEAR(#{signDate})</if>
+            <if test="signRemark != null  and signRemark != ''"> and t3.sign_remark = #{signRemark}</if>
+            <if test="signScan != null  and signScan != ''"> and t3.sign_scan = #{signScan}</if>
+            <if test="commentType != null  and commentType != ''"> and t3.comment_type = #{commentType}</if>
             <if test="projectSource != null  and projectSource != ''"> and p.project_source = #{projectSource}</if>
             <if test="projectName != null  and projectName != ''"> and p.project_name like concat('%', #{projectName}, '%')</if>
             <if test="partyAName != null  and partyAName != ''"> and pa.party_a_name like concat('%', #{partyAName}, '%')</if>
             <if test="params.beginTime != null and params.beginTime != ''"><!-- 开始时间检索 -->
-                and date_format(t1.sign_date,'%y%m%d') &gt;= date_format(#{params.beginTime},'%y%m%d')
+                and date_format(t3.sign_date,'%y%m%d') &gt;= date_format(#{params.beginTime},'%y%m%d')
             </if>
             <if test="params.endTime != null and params.endTime != ''"><!-- 结束时间检索 -->
-                and date_format(t1.sign_date,'%y%m%d') &lt;= date_format(#{params.endTime},'%y%m%d')
+                and date_format(t3.sign_date,'%y%m%d') &lt;= date_format(#{params.endTime},'%y%m%d')
             </if>
         </where>
-        group by t1.contract_id
-        order by t1.draft_time desc
+        group by t3.contract_id
+        order by t3.draft_time desc
     </select>
 
     <select id="selectCmcContractStatistic" resultMap="CmcContractResult" >

oa-back/ruoyi-system/src/main/resources/mapper/oa/CmcSubContractMapper.xml

     </select>
 
     <select id="selectCmcSubContractListByRange" parameterType="CmcSubContract" resultMap="CmcSubContractResult">
-        select distinct t1.sub_contract_id, t1.sub_contract_name, t1.sub_amount, t1.partner_id, t1.partner_name, t1.contact_person, t1.telephone, t1.contract_document, t1.drafter,
-        t1.paid_amount, t1.paid_percentage, t1.invoice_amount, t1.invoice_percentage, t1.draft_nick_name, t1.draft_time, t1.remark, t1.sign_date, t1.sign_remark, t1.sign_scan,
-        t1.comment_type, t1.manager_comment, t1.manager_user_id, t1.manager_nick_name, t1.manager_time, t1.gm_user_id, t1.gm_nick_name, t1.gm_time, t1.gm_comment, p.project_number, p.project_name, p.project_source from
-        (select distinct sc.sub_contract_id, sc.sub_contract_name, sc.sub_amount, sc.partner_id, pa.partner_name, sc.contact_person, sc.telephone, sc.contract_document, sc.drafter,
+        select distinct t3.*, p.project_number, p.project_name, p.project_source from
+        (select t1.*, t2.invoice_amount, t2.invoice_percentage from (select distinct sc.sub_contract_id, sc.sub_contract_name, sc.sub_amount, sc.partner_id, pa.partner_name, sc.contact_person, sc.telephone, sc.contract_document, sc.drafter,
         sum(cp.paid_amount) as paid_amount, case sc.sub_amount when 0 then 0 else round(sum(cp.paid_amount) / sc.sub_amount * 100, 2) end as paid_percentage,
-        sum(ci.invoice_amount) as invoice_amount, case sc.sub_amount when 0 then 0 else round(sum(ci.invoice_amount) / sc.sub_amount * 100, 2) end as invoice_percentage,
         u.nick_name as draft_nick_name, sc.draft_time, sc.remark, sc.sign_date, sc.sign_remark, sc.sign_scan, sc.comment_type, sc.manager_comment,
         sc.manager_user_id, u1.nick_name as manager_nick_name, sc.manager_time, sc.gm_user_id, u2.nick_name as gm_nick_name, sc.gm_time, sc.gm_comment from cmc_sub_contract as sc
         left join sys_user as u on u.user_id = sc.drafter
         left join sys_user as u2 on u2.user_id = sc.gm_user_id
         left join cmc_partner as pa on pa.partner_id = sc.partner_id
         left join cmc_contract_paid as cp on cp.contract_id = sc.sub_contract_id
+        group by sc.sub_contract_id) as t1
+        left join
+        (select distinct sc.sub_contract_id, 
+        sum(ci.invoice_amount) as invoice_amount, case sc.sub_amount when 0 then 0 else round(sum(ci.invoice_amount) / sc.sub_amount * 100, 2) end as invoice_percentage
+        from cmc_sub_contract as sc
         left join cmc_contract_invoice as ci on ci.contract_id = sc.sub_contract_id
-        group by sc.sub_contract_id)
-        as t1
-        left join cmc_project_sub_contract as psc on psc.sub_contract_id = t1.sub_contract_id
+        group by sc.sub_contract_id) as t2
+        on t1.sub_contract_id = t2.sub_contract_id)
+        as t3
+        left join cmc_project_sub_contract as psc on psc.sub_contract_id = t3.sub_contract_id
         left join cmc_project as p on psc.project_id = p.project_id
         <where>
-            <if test="subContractName != null  and subContractName != ''"> and t1.sub_contract_name like concat('%', #{subContractName}, '%')</if>
-            <if test="subAmount != null "> and t1.sub_amount = #{subAmount}</if>
-            <if test="partnerId != null  and partnerId != ''"> and t1.partner_id = #{partnerId}</if>
-            <if test="contactPerson != null  and contactPerson != ''"> and t1.contact_person = #{contactPerson}</if>
-            <if test="telephone != null  and telephone != ''"> and t1.telephone = #{telephone}</if>
-            <if test="contractDocument != null  and contractDocument != ''"> and t1.contract_document = #{contractDocument}</if>
-            <if test="drafter != null "> and t1.drafter = #{drafter}</if>
-            <if test="draftTime != null "> and t1.draft_time = #{draftTime}</if>
-            <if test="signDate != null "> and YEAR(t1.sign_date) = YEAR(#{signDate})</if>
-            <if test="signRemark != null  and signRemark != ''"> and t1.sign_remark = #{signRemark}</if>
-            <if test="signScan != null  and signScan != ''"> and t1.sign_scan = #{signScan}</if>
-            <if test="commentType != null  and commentType != ''"> and t1.comment_type = #{commentType}</if>
+            <if test="subContractName != null  and subContractName != ''"> and t3.sub_contract_name like concat('%', #{subContractName}, '%')</if>
+            <if test="subAmount != null "> and t3.sub_amount = #{subAmount}</if>
+            <if test="partnerId != null  and partnerId != ''"> and t3.partner_id = #{partnerId}</if>
+            <if test="contactPerson != null  and contactPerson != ''"> and t3.contact_person = #{contactPerson}</if>
+            <if test="telephone != null  and telephone != ''"> and t3.telephone = #{telephone}</if>
+            <if test="contractDocument != null  and contractDocument != ''"> and t3.contract_document = #{contractDocument}</if>
+            <if test="drafter != null "> and t3.drafter = #{drafter}</if>
+            <if test="draftTime != null "> and t3.draft_time = #{draftTime}</if>
+            <if test="signDate != null "> and YEAR(t3.sign_date) = YEAR(#{signDate})</if>
+            <if test="signRemark != null  and signRemark != ''"> and t3.sign_remark = #{signRemark}</if>
+            <if test="signScan != null  and signScan != ''"> and t3.sign_scan = #{signScan}</if>
+            <if test="commentType != null  and commentType != ''"> and t3.comment_type = #{commentType}</if>
             <if test="projectSource != null  and projectSource != ''"> and p.project_source = #{projectSource}</if>
             <if test="params.beginTime != null and params.beginTime != ''"><!-- 开始时间检索 -->
-                and date_format(t1.sign_date,'%y%m%d') &gt;= date_format(#{params.beginTime},'%y%m%d')
+                and date_format(t3.sign_date,'%y%m%d') &gt;= date_format(#{params.beginTime},'%y%m%d')
             </if>
             <if test="params.endTime != null and params.endTime != ''"><!-- 结束时间检索 -->
-                and date_format(t1.sign_date,'%y%m%d') &lt;= date_format(#{params.endTime},'%y%m%d')
+                and date_format(t3.sign_date,'%y%m%d') &lt;= date_format(#{params.endTime},'%y%m%d')
             </if>
         </where>
-        group by t1.sub_contract_id
-        order by t1.draft_time desc
+        group by t3.sub_contract_id
+        order by t3.draft_time desc
     </select>
 
     <select id="selectCmcSubContractBySubContractId" parameterType="String" resultMap="CmcSubContractResult">

oa-back/ruoyi-system/src/main/resources/mapper/oa/CmcWageMapper.xml

                w.deduct_total, w.social_security_unit, w.individual_income_tax, w.paid_wage, w.pay_day, w.pay_month, w.remark, w.performance_id, u.nick_name from cmc_wage as w
         left join sys_user as u on u.user_id = w.user_id
         left join sys_dept as d on d.dept_id = u.dept_id
+        left join cmc_post_salary ps on u.post_level = ps.post_level and u.salary_level = ps.salary_level
     </sql>
 
     <select id="selectCmcWageList" parameterType="CmcWage" resultMap="CmcWageResult">
         <include refid="selectCmcWageVo"/>
-        <where>  
+        <where>
+            and u.del_flag = '0'
             <if test="userId != null "> and w.user_id = #{userId}</if>
             <if test="deptId != null "> and u.dept_id = #{deptId}</if>
             <if test="baseSalary != null "> and w.base_salary = #{baseSalary}</if>
             <if test="payDay != null "> and w.pay_day = #{payDay}</if>
             <if test="payMonth != null "> and MONTH(w.pay_month) = MONTH(#{payMonth})</if>
             <if test="performanceId != null and performanceId != ''"> and w.performance_id = #{performanceId}</if>
-        </where>
-        order by w.user_id, w.pay_month
+        </where>        
+		order by u.dept_id, ps.post_level desc, ps.salary_level desc, u.user_id, w.pay_month
     </select>
     
     <select id="selectCmcWageByWageId" parameterType="Integer" resultMap="CmcWageResult">

oa-ui/src/api/llm/rag.js

   })
 }
 
+/**
+ * 解析SSE事件 - 优化版
+ * 正确处理多行data和事件类型
+ */
 function parseSseEvents(buffer) {
-  // SSE 事件以空行分隔：\n\n（兼容 \r\n）
   const events = []
-  const normalized = buffer.replace(/\r\n/g, '\n')
-  const parts = normalized.split('\n\n')
-  // 最后一段可能是不完整事件，留给下次拼接
-  const rest = parts.pop() ?? ''
-
-  for (const part of parts) {
-    if (!part.trim()) continue
-    const lines = part.split('\n')
-    const dataLines = []
-    for (const line of lines) {
-      if (line.startsWith('data:')) {
-        dataLines.push(line.slice(5).trimStart())
+  const lines = buffer.split(/\r?\n/)
+  let currentEvent = null
+  let currentData = []
+  let remaining = []
+  let i = 0
+  
+  for (; i < lines.length; i++) {
+    const line = lines[i]
+    
+    // 空行表示事件结束
+    if (line === '') {
+      if (currentData.length > 0) {
+        const data = currentData.join('\n')
+        events.push({
+          event: currentEvent || 'message',
+          data: data
+        })
+        currentEvent = null
+        currentData = []
       }
+      continue
+    }
+    
+    // 解析事件类型
+    if (line.startsWith('event:')) {
+      currentEvent = line.slice(6).trim()
+      continue
+    }
+    
+    // 解析数据行
+    if (line.startsWith('data:')) {
+      const dataValue = line.slice(5).trimStart()
+      currentData.push(dataValue)
+      continue
     }
-    const data = dataLines.join('\n')
-    if (data !== '') events.push(data)
   }
-
-  return { events, rest }
+  
+  // 保存剩余未完成的数据
+  if (i < lines.length) {
+    remaining = lines.slice(i).join('\n')
+  } else if (currentData.length > 0) {
+    // 如果最后没有空行，但数据已完整，也要保留
+    remaining = lines.slice(i - currentData.length - (currentEvent ? 1 : 0)).join('\n')
+  } else {
+    remaining = ''
+  }
+  
+  return { events, rest: remaining }
 }
 
+/**
+ * 规范化SSE数据
+ */
 function normalizeSseData(data) {
-  if (!data) return ''
-  const trimmed = data.trim()
-  // 移除可能的JSON包装（如果有的话）
-  if (trimmed.startsWith('{') && trimmed.endsWith('}')) {
-    try {
-      const parsed = JSON.parse(trimmed)
-      if (parsed.content) return parsed.content
-      if (parsed.resultContent) return parsed.resultContent
-    } catch (e) {
-      // 不是JSON，直接返回
-    }
+  if (data == null) return ''
+  let s = String(data)
+  // 移除可能的 "data:" 前缀
+  while (s.startsWith('data:')) {
+    s = s.slice(5)
+    if (s.startsWith(' ')) s = s.slice(1)
   }
-  return trimmed
+  return s
 }
 
 function stripToolCallStream(text, state) {
   }
 }
 
+/**
+ * 流式请求SSE - 优化版
+ * @param {string} url - 请求URL
+ * @param {Function} onMessage - 收到消息时的回调，接收解析后的JSON对象和事件类型
+ * @param {Function} onError - 错误回调
+ * @param {Function} onComplete - 完成回调
+ * @returns {AbortController} 用于取消请求的控制器
+ */
 function streamFetchSse(url, onMessage, onError, onComplete) {
   const controller = new AbortController()
-  const toolCallState = { inToolCall: false }
+
   fetch(url, {
     method: 'GET',
     headers: {
         if (done) break
 
         buffer += decoder.decode(value, { stream: true })
-        const parsed = parseSseEvents(buffer)
-        buffer = parsed.rest
+        
+        // 解析SSE事件
+        const { events, rest } = parseSseEvents(buffer)
+        buffer = rest
 
-        for (const data of parsed.events) {
-          const normalized = normalizeSseData(data)
-          if (normalized === '[DONE]') {
-            onComplete()
+        for (const event of events) {
+          const normalizedData = normalizeSseData(event.data)
+          
+          // 检查是否结束
+          if (normalizedData === '[DONE]') {
+            onComplete && onComplete()
             controller.abort()
             return
           }
-          const visible = stripToolCallStream(normalized, toolCallState)
-          if (visible !== '') onMessage(visible)
-        }
-      }
-
-      // 兜底：流结束但没收到 [DONE]
-      if (buffer.trim()) {
-        const parsed = parseSseEvents(buffer + '\n\n')
-        for (const data of parsed.events) {
-          const normalized = normalizeSseData(data)
-          if (normalized !== '' && normalized !== '[DONE]') {
-            const visible = stripToolCallStream(normalized, toolCallState)
-            if (visible !== '') onMessage(visible)
+          
+          try {
+            const jsonData = JSON.parse(normalizedData)
+            // 根据事件名称分发，传递JSON对象和事件类型
+            onMessage && onMessage(jsonData, event.event)
+          } catch (e) {
+            console.warn('解析JSON失败，作为普通文本处理:', normalizedData)
+            onMessage && onMessage({ content: normalizedData }, event.event)
           }
         }
       }
-      onComplete()
+      
+      onComplete && onComplete()
     })
     .catch((error) => {
-      if (error.name === 'AbortError') return
+      if (error.name === 'AbortError') {
+        console.log('请求已取消')
+        return
+      }
       console.error('流式请求错误:', error)
-      onError(error)
+      onError && onError(error)
     })
 
   return controller
 }
 
-// 流式回答API（SSE）
+/**
+ * 流式回答API（SSE）
+ * @param {string} question - 问题
+ * @param {string} collectionName - 知识库名称
+ * @param {Function} onMessage - 收到消息时的回调，接收参数：(jsonData, eventType)
+ *                              - jsonData: 解析后的JSON对象，包含content、tokenUsage等字段
+ *                              - eventType: 事件类型（如'answer'、'error'、'completed'）
+ * @param {Function} onError - 错误回调
+ * @param {Function} onComplete - 完成回调
+ * @returns {AbortController} 用于取消请求的控制器
+ */
 export function getAnswerStream(question, collectionName, onMessage, onError, onComplete) {
   const baseURL = process.env.VUE_APP_BASE_API
   const url = `${baseURL}/llm/rag/answer?question=${encodeURIComponent(question)}&collectionName=${encodeURIComponent(collectionName)}`

oa-ui/src/api/llm/session.js

   })
 }
 
+/**
+ * 解析SSE事件 - 优化版
+ * 正确处理多行data和事件类型
+ */
 function parseSseEvents(buffer) {
-  // SSE 事件以空行分隔：\n\n（兼容 \r\n）
   const events = []
-  const normalized = buffer.replace(/\r\n/g, '\n')
-  const parts = normalized.split('\n\n')
-  // 最后一段可能是不完整事件，留给下次拼接
-  const rest = parts.pop() ?? ''
-
-  for (const part of parts) {
-    if (!part.trim()) continue
-    const lines = part.split('\n')
-    const dataLines = []
-    for (const line of lines) {
-      if (line.startsWith('data:')) {
-        dataLines.push(line.slice(5).trimStart())
+  const lines = buffer.split(/\r?\n/)
+  let currentEvent = null
+  let currentData = []
+  let remaining = []
+  let i = 0
+  
+  for (; i < lines.length; i++) {
+    const line = lines[i]
+    
+    // 空行表示事件结束
+    if (line === '') {
+      if (currentData.length > 0) {
+        const data = currentData.join('\n')
+        events.push({
+          event: currentEvent || 'message',
+          data: data
+        })
+        currentEvent = null
+        currentData = []
       }
+      continue
+    }
+    
+    // 解析事件类型
+    if (line.startsWith('event:')) {
+      currentEvent = line.slice(6).trim()
+      continue
+    }
+    
+    // 解析数据行
+    if (line.startsWith('data:')) {
+      const dataValue = line.slice(5).trimStart()
+      currentData.push(dataValue)
+      continue
     }
-    const data = dataLines.join('\n')
-    if (data !== '') events.push(data)
   }
-
-  return { events, rest }
+  
+  // 保存剩余未完成的数据
+  if (i < lines.length) {
+    remaining = lines.slice(i).join('\n')
+  } else if (currentData.length > 0) {
+    // 如果最后没有空行，但数据已完整，也要保留
+    remaining = lines.slice(i - currentData.length - (currentEvent ? 1 : 0)).join('\n')
+  } else {
+    remaining = ''
+  }
+  
+  return { events, rest: remaining }
 }
 
+/**
+ * 规范化SSE数据
+ */
 function normalizeSseData(data) {
   if (data == null) return ''
   let s = String(data)
-  // 有些后端会把 payload 自己也带上 "data:"，这里做兼容剥离
-  // 例如：data:data: xxx  ->  xxx
+  // 移除可能的 "data:" 前缀
   while (s.startsWith('data:')) {
     s = s.slice(5)
     if (s.startsWith(' ')) s = s.slice(1)
   return out
 }
 
+/**
+ * 流式请求SSE - 优化版
+ * @param {string} url - 请求URL
+ * @param {Function} onMessage - 收到消息时的回调，接收解析后的JSON对象和事件类型
+ * @param {Function} onError - 错误回调
+ * @param {Function} onComplete - 完成回调
+ * @returns {AbortController} 用于取消请求的控制器
+ */
 function streamFetchSse(url, onMessage, onError, onComplete) {
   const controller = new AbortController()
-  const toolCallState = { inToolCall: false }
+
   fetch(url, {
     method: 'GET',
     headers: {
         if (done) break
 
         buffer += decoder.decode(value, { stream: true })
-        const parsed = parseSseEvents(buffer)
-        buffer = parsed.rest
-
-        for (const data of parsed.events) {
-          const normalized = normalizeSseData(data)
-          if (normalized === '[DONE]') {
-            onComplete()
+        
+        // 解析SSE事件
+        const { events, rest } = parseSseEvents(buffer)
+        buffer = rest
+
+        for (const event of events) {
+          const normalizedData = normalizeSseData(event.data)
+          
+          // 检查是否结束
+          if (normalizedData === '[DONE]') {
+            onComplete && onComplete()
             controller.abort()
             return
           }
-          const visible = stripToolCallStream(normalized, toolCallState)
-          if (visible !== '') onMessage(visible)
-        }
-      }
-
-      // 兜底：流结束但没收到 [DONE]
-      if (buffer.trim()) {
-        const parsed = parseSseEvents(buffer + '\n\n')
-        for (const data of parsed.events) {
-          const normalized = normalizeSseData(data)
-          if (normalized !== '' && normalized !== '[DONE]') {
-            const visible = stripToolCallStream(normalized, toolCallState)
-            if (visible !== '') onMessage(visible)
+          
+          try {
+            const jsonData = JSON.parse(normalizedData)
+            // 根据事件名称分发，传递JSON对象和事件类型
+            onMessage && onMessage(jsonData, event.event)
+          } catch (e) {
+            console.warn('解析JSON失败，作为普通文本处理:', normalizedData)
+            onMessage && onMessage({ content: normalizedData }, event.event)
           }
         }
       }
-      onComplete()
+      
+      onComplete && onComplete()
     })
     .catch((error) => {
-      if (error.name === 'AbortError') return
+      if (error.name === 'AbortError') {
+        console.log('请求已取消')
+        return
+      }
       console.error('流式请求错误:', error)
-      onError(error)
+      onError && onError(error)
     })
 
   return controller
 }
 
-// 流式回答API（SSE）
+/**
+ * 流式回答API（SSE）
+ * @param {Object} params - 参数对象
+ * @param {string} params.topicId - 话题ID
+ * @param {string} params.question - 问题
+ * @param {Function} onMessage - 收到消息时的回调，接收参数：(jsonData, eventType)
+ *                              - jsonData: 解析后的JSON对象，包含content、tokenUsage等字段
+ *                              - eventType: 事件类型（如'answer'、'error'、'completed'）
+ * @param {Function} onError - 错误回调
+ * @param {Function} onComplete - 完成回调
+ * @returns {AbortController} 用于取消请求的控制器
+ */
 export function getAnswerStream(params, onMessage, onError, onComplete) {
   const baseURL = process.env.VUE_APP_BASE_API
   const url = `${baseURL}/llm/session/answer?topicId=${params.topicId}&question=${encodeURIComponent(params.question)}`
   return streamFetchSse(url, onMessage, onError, onComplete)
 }
 
-// 流式回答API（带文档）- 简化版
+/**
+ * 流式回答API（带文档）
+ * @param {Object} params - 参数对象
+ * @param {string} params.topicId - 话题ID
+ * @param {string} params.chatId - 聊天ID
+ * @param {string} params.question - 问题
+ * @param {Function} onMessage - 收到消息时的回调，接收参数：(jsonData, eventType)
+ *                              - jsonData: 解析后的JSON对象，包含content、tokenUsage等字段
+ *                              - eventType: 事件类型（如'answer'、'error'、'completed'）
+ * @param {Function} onError - 错误回调
+ * @param {Function} onComplete - 完成回调
+ * @returns {AbortController} 用于取消请求的控制器
+ */
 export function getAnswerWithDocumentStream(params, onMessage, onError, onComplete) {
   const baseURL = process.env.VUE_APP_BASE_API
   const url = `${baseURL}/llm/session/answerWithDocument?topicId=${params.topicId}&chatId=${params.chatId}&question=${encodeURIComponent(params.question)}`

oa-ui/src/views/llm/agent/AgentDetail.vue

                           :before-upload="beforeUpload" :show-file-list="true" :limit="1"
                           :disabled="bidFileList.length > 0">
                         </el-upload>
-                        <el-button type="primary" @click="generateBid" :disabled="!currentFilename">生成标书</el-button>
-                        <el-button type="primary" icon="el-icon-download" @click="exportWord" :disabled="!bidResult">
+                        <el-button type="primary" @click="isStreaming ? stopGeneration() : generateBid()"
+                          :disabled="!currentFilename && !isStreaming">
+                          {{ isStreaming ? '停止生成' : '生成标书' }}
+                        </el-button>
+                        <el-button type="primary" icon="el-icon-download" @click="exportWord"
+                          :disabled="isStreaming">
                           导出Word文档
                         </el-button>
                       </div>
                   <div v-if="directoryTree.length > 0" class="chapter-section">
                     <div class="chapter-content-wrapper">
                       <div class="directory-tree">
-                        <div class="directory-actions">
-                          <el-button type="success" size="small" @click="saveDirectory" icon="el-icon-check"
-                            title="保存目录">保存目录</el-button>
-                        </div>
                         <div class="tree-container">
-                          <div v-for="(item, index) in directoryTree" :key="index" class="directory-item level-1">
+                          <div v-for="(item, index) in directoryTree" :key="index" class="directory-item level-1"
+                            :class="{ 'selected': selectedNode === item }">
                             <div class="directory-node">
-                              <el-input v-model="item.title" placeholder="请输入目录名称" size="small" />
+                              <el-input v-model="item.title" placeholder="请输入目录名称" size="small"
+                                @click.native.stop="selectNode(item)" />
                               <div class="node-actions">
                                 <el-button type="primary" size="mini" icon="el-icon-plus" circle
                                   @click.stop="addChildNode(item)" title="添加子节点" />
                             </div>
                             <div v-if="item.children && item.children.length > 0">
                               <div v-for="(child, childIndex) in item.children" :key="childIndex"
-                                class="directory-item level-2">
+                                class="directory-item level-2" :class="{ 'selected': selectedNode === child }">
                                 <div class="directory-node">
-                                  <el-input v-model="child.title" placeholder="请输入子目录名称" size="small" />
+                                  <el-input v-model="child.title" placeholder="请输入子目录名称" size="small"
+                                    @click.native.stop="selectNode(child)" />
                                   <div class="node-actions">
                                     <el-button type="primary" size="mini" icon="el-icon-plus" circle
                                       @click.stop="addChildNode(child)" title="添加子节点" />
                         </div>
                       </div>
                       <div class="bid-content-area">
-                        <!-- 在 bid-content-area 内部，bid-result-content 之前添加 -->
+                        <!-- 生成进度提示（仅在生成章节内容时显示） -->
                         <div v-if="isStreaming" class="generation-progress"
                           style="padding: 8px 12px; background: #ecf5ff; border-radius: 4px; margin-bottom: 12px;">
                           <i class="el-icon-loading" style="margin-right: 8px;"></i>
                           <span>正在生成中... {{ currentChapterIndex }}/{{ totalChapters }} 章</span>
                         </div>
-                        <div v-if="bidResult" class="bid-result-content">
+                        <!-- 章节内容（生成中显示，生成完成后不再显示） -->
+                        <div v-if="isStreaming && bidResult" class="bid-result-content">
+                          <div v-html="bidResult"></div>
+                        </div>
+                        <!-- 已选中章节的预览内容（非生成状态且不是完成状态） -->
+                        <div v-if="!isStreaming && !generationCompleted && bidResult" class="bid-result-content">
                           <div v-html="bidResult"></div>
                         </div>
-                        <div v-else class="empty-bid-content">
+                        <!-- 生成完成提示（居中显示，不与章节内容同时出现） -->
+                        <div v-if="generationCompleted" class="empty-bid-content generation-completed"
+                          style="color: #67c23a;">
+                          <i class="el-icon-success" style="font-size: 64px; margin-bottom: 20px; opacity: 0.8;"></i>
+                          <p style="font-size: 18px; font-weight: 500;">生成标书完成</p>
+                        </div>
+                        <!-- 初始空状态（未选中章节且未开始生成） -->
+                        <div v-if="!isStreaming && !generationCompleted && !bidResult" class="empty-bid-content">
                           <i class="el-icon-file-text"></i>
                           <p>{{ selectedNode ? '已选中章节: ' + selectedNode.title : '请选择章节或直接生成全部内容' }}</p>
                         </div>
                       </div>
                     </div>
+                    <div class="directory-toolbar">
+                      <el-button type="success" size="small" @click="saveDirectory" icon="el-icon-check"
+                        title="保存目录">保存目录</el-button>
+                      <div class="token-info-container">
+                        <div class="token-info-row">
+                          <span class="token-label">提示词:</span>
+                          <span class="token-value">{{ tokenUsage.promptTokens }}</span>
+                        </div>
+                        <div class="token-info-row">
+                          <span class="token-label">生成:</span>
+                          <span class="token-value">{{ tokenUsage.completionTokens }}</span>
+                        </div>
+                        <div class="token-info-row">
+                          <span class="token-label">总计:</span>
+                          <span class="token-value">{{ tokenUsage.totalTokens }}</span>
+                        </div>
+                      </div>
+                    </div>
                   </div>
                 </div>
               </div>
       loading: false,
       agentInfo: null,
       openingMessage: '',
-      scoringTabLabel: '评审标准',
+      scoringTabLabel: '评分标准',
       chatMessages: [],
       inputMessage: '',
       isTyping: false,
       currentStreamingChapter: null, // 当前正在流式生成的章节标题
       streamingContent: '', // 当前流式生成的内容
       isStreaming: false, // 是否正在流式生成
+      generationCompleted: false, // 标书生成是否已完成
+      abortController: null, // 用于取消流式请求
       currentChapterIndex: 0, // 当前章节索引
-      totalChapters: 0 // 总章节数
+      totalChapters: 0, // 总章节数
+
+      // token用量相关
+      tokenUsage: {
+        promptTokens: 0,
+        completionTokens: 0,
+        totalTokens: 0
+      },
+      chapterTokenUsage: {} // 存储每个章节的token用量
     }
   },
   computed: {
         const response = await getAgent(agentId)
         this.agentInfo = response.data
 
-        // 设置评审标准/应价人须知标签
+        // 设置评分标准/应价人须知标签
         if (this.agentInfo?.agentName) {
           if (this.agentInfo.agentName.includes('技术') && this.agentInfo.agentName.includes('询价')) {
             this.scoringTabLabel = '应价人须知'
           } else {
-            this.scoringTabLabel = '评审标准'
+            this.scoringTabLabel = '评分标准'
           }
         }
 
 
     // 选择话题（对话模式）
     async selectTopic(topic) {
-      console.log('选择话题:', topic)
       this.currentTopicId = topic.topicId
       this.chatTitle = topic.topic
 
     parseAnalysisResults(output) {
       if (!output) return
 
-      // 使用正则匹配【项目概况】【评审标准】【应价人须知】【详细目录】
-      // 匹配项目概况（结束标记可以是评审标准或应价人须知）
-      const overviewMatch = output.match(/【项目概况】\n([\s\S]*?)(?=\n\n【评审标准】|\n\n【应价人须知】)/)
-      // 匹配评审标准或应价人须知
-      const scoringMatch = output.match(/【评审标准】\n([\s\S]*?)(?=\n\n【详细目录】)/)
+      // 使用正则匹配【项目概况】【评分标准】【应价人须知】【详细目录】
+      // 匹配项目概况（结束标记可以是评分标准或应价人须知）
+      const overviewMatch = output.match(/【项目概况】\n([\s\S]*?)(?=\n\n【评分标准】|\n\n【应价人须知】)/)
+      // 匹配评分标准或应价人须知
+      const scoringMatch = output.match(/【评分标准】\n([\s\S]*?)(?=\n\n【详细目录】)/)
       const biddingMatch = output.match(/【应价人须知】\n([\s\S]*?)(?=\n\n【详细目录】)/)
       const directoryMatch = output.match(/【详细目录】\n([\s\S]*)/)
 
           this.currentTopicId = null
           this.chatMessages = []
           this.chatTitle = '智能体新对话'
+
+          // 任务模式下清空相关数据
+          if (this.currentMode === 'task') {
+            // 清空文档分析内容
+            this.analysisResults = {
+              overview: '',
+              scoring: '',
+              directory: ''
+            }
+            // 清空标书内容相关数据
+            this.directoryTree = []
+            this.currentFilename = ''
+            this.bidResult = ''
+            this.bidFilePath = ''
+            this.uploadedDocumentPath = ''
+            this.bidFileList = []
+          }
         }
 
         // 重新加载话题列表
           question: message
         })
         let content = JSON.parse(response.resultContent).content;
-        console.log(content);
 
         // 检查是否是默认的失败回复
         const defaultFailureMessage = '抱歉，我暂时无法回答这个问题。';
 
         // 重置流式生成状态
         this.isStreaming = true
+        this.generationCompleted = false
         this.streamingContent = ''
         this.currentStreamingChapter = null
         this.chapterContents = {}
         this.bidResult = ''
 
+        // 重置token用量
+        this.tokenUsage = {
+          promptTokens: 0,
+          completionTokens: 0,
+          totalTokens: 0
+        }
+        this.chapterTokenUsage = {}
+
         // 创建FormData
         const formData = new FormData()
         formData.append('topicId', this.currentTopicId)
      * 使用API进行流式请求
      */
     fetchStreamGeneration(formData, bidTitle) {
-      console.log('开始流式请求:', process.env.VUE_APP_BASE_API + '/llm/agent/streamGenerate')
-
       const onMessage = (data, eventType) => {
         this.handleStreamMessage(data, eventType)
       }
         console.error('流式请求失败:', error)
         this.$message.error('生成标书失败: ' + error.message)
         this.isStreaming = false
+        this.abortController = null
         this.$nextTick(() => {
           this.scrollToBottom()
         })
       }
 
       const onComplete = () => {
-        console.log('流式请求完成')
         this.isStreaming = false
-        this.$message.success('标书生成完成')
-
-        // 如果有生成的内容，保存到历史记录
-        if (this.currentTopicId && this.streamingContent) {
-          this.saveStreamingResult()
-        }
+        this.abortController = null
+        // 完成消息由轮询方法在文档写入完成后显示
       }
 
-      // 使用封装的API
-      streamGenerateChapters(formData, onMessage, onError, onComplete)
+      // 使用封装的API，并保存controller用于取消
+      this.abortController = streamGenerateChapters(formData, onMessage, onError, onComplete)
+    },
+
+    /**
+     * 停止生成标书
+     */
+    stopGeneration() {
+      if (this.abortController) {
+        this.abortController.abort()
+        this.abortController = null
+        this.isStreaming = false
+        this.generationCompleted = false
+        this.$message.warning('已停止生成')
+      }
     },
 
     /**
      * 处理流式消息 - 支持逐字输出
      */
     handleStreamMessage(data, eventType) {
-      console.log('收到消息:', data, '事件类型:', eventType)
-
       // 处理完成事件
       if (data.completed) {
         this.isStreaming = false
+        this.generationCompleted = true
         this.currentStreamingChapter = null
+        this.bidResult = ''
+
+        // 设置标书文件路径，使导出功能立即可用
+        if (this.currentFilename) {
+          this.bidFilePath = process.env.VUE_APP_BASE_API + '/profile/upload/agent/' + this.agentInfo.agentName + '/' + this.currentFilename
+        }
+
+        this.$message.success('标书生成完成')
         return
       }
 
       const chapterIndex = data.chapterIndex
       const totalChapters = data.totalChapters
 
+      // 获取token用量信息
+      const promptTokens = data.promptTokens || 0
+      const completionTokens = data.completionTokens || 0
+      const totalTokens = data.totalTokens || 0
+
       // 保存总章节数
       if (totalChapters) {
         this.totalChapters = totalChapters
       if (content) {
         this.streamingContent += content
 
-        // 实时更新显示内容
-        const formattedContent = this.streamingContent.replace(/\n/g, '<br>')
+        // 使用 marked 进行 Markdown 格式化
+        const formattedContent = marked(this.streamingContent)
         this.bidResult = `<h3 style="color: #409EFF; margin: 10px 0;">【${this.currentStreamingChapter}】</h3>
                         <div style="line-height: 1.8;">${formattedContent}</div>`
 
       }
       this.chapterContents[title] += content || ''
 
-      // 如果是当前章节的最后一块内容，显示进度
-      if (isLast) {
-        console.log(`章节 "${title}" 生成完成，共 ${this.chapterContents[title].length} 字符`)
+      // 实时更新并显示token用量（每个章节发送的是当前章节的累计值）
+      if (promptTokens > 0 || completionTokens > 0 || totalTokens > 0) {
+        // 保存当前章节的token用量（如果值更大则更新，避免重复累加）
+        const currentChapterTokens = this.chapterTokenUsage[title] || { promptTokens: 0, completionTokens: 0, totalTokens: 0 }
 
-        // 显示进度提示
-        const progressMsg = `已完成: ${this.currentChapterIndex}/${this.totalChapters} 章`
-        const progressElement = document.querySelector('.generation-progress')
-        if (progressElement) {
-          progressElement.textContent = progressMsg
+        // 只在有新值时更新
+        if (promptTokens > currentChapterTokens.promptTokens ||
+          completionTokens > currentChapterTokens.completionTokens ||
+          totalTokens > currentChapterTokens.totalTokens) {
+
+          this.chapterTokenUsage[title] = {
+            promptTokens,
+            completionTokens,
+            totalTokens
+          }
+
+          // 重新计算总token用量
+          this.recalculateTotalTokenUsage()
         }
       }
     },
 
+    /**
+     * 重新计算总token用量
+     */
+    recalculateTotalTokenUsage() {
+      let totalPrompt = 0
+      let totalCompletion = 0
+      let totalAll = 0
+
+      // 遍历所有章节的token用量
+      for (const chapterTokens of Object.values(this.chapterTokenUsage)) {
+        totalPrompt += chapterTokens.promptTokens || 0
+        totalCompletion += chapterTokens.completionTokens || 0
+        totalAll += chapterTokens.totalTokens || 0
+      }
+
+      // 更新总token用量
+      this.tokenUsage.promptTokens = totalPrompt
+      this.tokenUsage.completionTokens = totalCompletion
+      this.tokenUsage.totalTokens = totalAll
+    },
+
     /**
      * 滚动到底部（用于bid-content-area区域）
      */
       })
     },
 
-    /**
-     * 保存流式生成结果
-     */
-    async saveStreamingResult() {
-      try {
-        // 构建完整的标书内容
-        let fullContent = ''
-        for (const [title, content] of Object.entries(this.chapterContents)) {
-          fullContent += `【${title}】\n${content}\n\n`
-        }
-
-        // 保存到聊天记录
-        const { addChat } = await import('@/api/llm/chat')
-        await addChat({
-          topicId: this.currentTopicId,
-          input: '生成技术文件',
-          output: fullContent,
-          outputTime: parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}')
-        })
-
-        console.log('流式生成结果已保存')
-      } catch (error) {
-        console.error('保存流式生成结果失败:', error)
-      }
-    },
-
-    // 在模板中需要添加一个进度显示（可选）
-    // 可以在 bid-content-area 顶部添加：
-    // <div v-if="isStreaming" class="generation-progress">正在生成中...</div>
-
     // 在目录树中选中指定标题的节点
     selectDirectoryNode(title) {
       const findNode = (nodes) => {
 
     // 导出Word
     exportWord() {
-      console.log(this.bidFilePath)
       if (!this.bidFilePath) {
         this.$message.warning('暂无文件可导出')
         return
           if (this.selectedNode && this.selectedNode.title === title) {
             this.selectedChapterContent = content
           }
-          // 可以根据需要进行其他处理，比如存储到某个数据结构中
-          console.log(`章节: ${title}`, content)
         }
       }
     },
     // 删除节点
     deleteNode(node) {
       this.removeNode(this.directoryTree, node)
+      // 删除后重新编号
+      this.renumberDirectory(node)
     },
 
     // 递归删除节点
       return false
     },
 
+    // 重新编号目录树
+    /**
+ * 重新编号目录树，并可选删除指定节点
+ * @param {string|null} deleteTarget - 需要删除的节点编号，不传则仅重新编号
+ */
+    renumberDirectory(deleteTarget = null) {
+      // directoryTree 结构: [{ title: '6 技术文件', children: [二级标题...] }]
+      if (!this.directoryTree || this.directoryTree.length === 0) return
+
+      const rootNode = this.directoryTree[0]
+      if (!rootNode || !rootNode.children) return
+
+      // 提取根标题的前缀（如 "6"）
+      const rootMatch = rootNode.title.match(/^(\d+)/)
+      const rootPrefix = rootMatch ? rootMatch[1] : '6'
+
+      // 递归重新编号节点
+      const renumberNodes = (nodes, parentPrefix) => {
+        // 1. 如果指定了删除目标，则倒序遍历删除目标节点及其子节点
+        if (deleteTarget) {
+          for (let i = nodes.length - 1; i >= 0; i--) {
+            // 提取当前节点原始的编号部分
+            const numMatch = nodes[i].title.match(/^(\d+(?:\.\d+)*)/)
+            const originalNum = numMatch ? numMatch[1] : ''
+
+            // 如果原始编号等于目标，或者是目标的子级（以 目标. 开头），则移除
+            if (originalNum === deleteTarget || originalNum.startsWith(`${deleteTarget}.`)) {
+              nodes.splice(i, 1)
+            }
+          }
+        }
+
+        // 2. 遍历剩余节点，进行重新编号
+        nodes.forEach((node, index) => {
+          const num = index + 1
+          const newPrefix = `${parentPrefix}.${num}`
+
+          // 从当前标题中提取文本内容（匹配任意层级的编号格式）
+          const numberPattern = /^\d+(?:\.\d+)*\s+(.+)/
+          const titleMatch = node.title.match(numberPattern)
+          const text = titleMatch ? titleMatch[1].trim() : node.title.replace(/^\d+(?:\.\d+)*\s*/, '').trim()
+
+          // 更新标题
+          node.title = `${newPrefix} ${text}`
+
+          // 递归处理子节点
+          if (node.children && node.children.length > 0) {
+            renumberNodes(node.children, newPrefix)
+          }
+        })
+      }
+
+      // 从二级标题开始重新编号
+      renumberNodes(rootNode.children, rootPrefix)
+    },
+
     // 选择节点
     async selectNode(node) {
       this.selectedNode = node
+      // 重置生成状态，确保 bidResult 能正常显示
+      this.isStreaming = false
+      this.generationCompleted = false
       // 更新文本域内容为已选中章节标题
       if (node && node.title) {
-        // 先查询数据库中是否已有该节点的章节内容（只有存在topicId时才查询）
-        if (this.currentTopicId) {
+        console.log(node.title)
+        // 判断是否为叶子节点（没有 children 或 children 为空），只有叶子节点才查询数据库
+        const isLeafNode = !node.children || node.children.length === 0
+        if (isLeafNode && this.currentTopicId) {
           try {
-            const chatRes = await listChat({ topicId: this.currentTopicId })
+            const chatRes = await listChat({ topicId: this.currentTopicId, input: node.title })
             if (chatRes.rows && chatRes.rows.length > 0) {
-              // 查找input字段等于选中节点标题的记录
-              const existingChat = chatRes.rows.find(chat => chat.input === node.title)
-              if (existingChat && existingChat.output) {
-                // 如果找到了，显示已保存的内容
-                this.bidResult = `<h3 style="color: #409EFF; margin: 10px 0;">【${node.title}】（已保存）</h3>
-                                 <div style="line-height: 1.8;">${existingChat.output.replace(/\n/g, '<br>')}</div>`
-                this.inputMessage = '已选中章节: ' + node.title + '（已有内容）'
-                return
-              }
+              // 如果找到了，显示已保存的内容
+              this.bidResult = `<h3 style="color: #409EFF; margin: 10px 0;">【${node.title}】（已保存）</h3>
+                                 <div style="line-height: 1.8;">${marked(chatRes.rows[0].output)}</div>`
+              this.inputMessage = '已选中章节: ' + node.title + '（已有内容）'
+              return
             }
           } catch (error) {
             console.error('查询章节内容失败:', error)
           }
         }
-        // 如果没有找到已保存的内容，显示选中标题和暂无内容提示
+        // 如果没有找到已保存的内容，或非叶子节点，显示选中标题和暂无内容提示
         this.inputMessage = '已选中章节: ' + node.title
         this.bidResult = `<h3 style="color: #409EFF; margin: 10px 0;">【${node.title}】</h3>
                           <div class="empty-bid-content" style="padding: 40px; text-align: center; color: #999;">
-                           <p>暂无内容</p>
-                           <p style="font-size: 12px; margin-top: 8px;">点击"生成标书"按钮生成该章节内容</p>
+                           <p>${isLeafNode ? '暂无内容' : '该节点为目录节点，请选择子章节'}</p>
+                           <p style="font-size: 12px; margin-top: 8px;">${isLeafNode ? '点击"生成标书"按钮生成该章节内容' : '请展开并选择具体章节'}</p>
                          </div>`
       }
     },
     gap: 15px;
   }
 
+  .directory-toolbar {
+    display: flex;
+    align-items: center;
+    justify-content: space-between;
+    margin-bottom: 10px;
+  }
+
+  .token-info-container {
+    display: flex;
+    flex-direction: row;
+    align-items: center;
+    gap: 15px;
+    padding: 8px 12px;
+    background: #f5f7fa;
+    border-radius: 4px;
+    font-size: 13px;
+  }
+
+  .token-info-row {
+    display: flex;
+    align-items: center;
+    gap: 5px;
+  }
+
+  .token-label {
+    color: #666;
+    font-weight: 500;
+  }
+
+  .token-value {
+    color: #409EFF;
+    font-weight: 600;
+    min-width: 60px;
+    text-align: right;
+  }
+
   .chapter-section {
     display: flex;
     flex-direction: column;

oa-ui/src/views/llm/agent/index.vue

       dialogTitle: '',
       isModifyAgent: false,
       // 模型列表
-      modelList: [{name: "Qwen2.5-7B-Instruct"}]
+      modelList: [{name: "Qwen3-VL-8B-Instruct"}]
     }
   },
   mounted() {

oa-ui/src/views/llm/chat/index.vue

                   <div class="message-bubble" :class="msg.type">
                     <div class="message-text" v-html="formatMessage(msg.text)"></div>
                     <div class="message-time">{{ formatMessageTime(msg.time) }}</div>
+                    <!-- AI消息的token用量统计 -->
+                    <div v-if="msg.type === 'ai' && msg.tokenUsage && msg.tokenUsage.totalTokens > 0"
+                      class="token-usage">
+                      <span class="token-label">消耗Token:</span>
+                      <span class="token-value">{{ msg.tokenUsage.totalTokens }}</span>
+                      <span class="token-breakdown">(输入: {{ msg.tokenUsage.promptTokens }}, 输出: {{ msg.tokenUsage.completionTokens }})</span>
+                    </div>
                   </div>
                   <!-- 用户消息的文件列表 -->
                   <div v-if="msg.type === 'user' && msg.chatId && msg.fileList && msg.fileList.length > 0"
 <script>
 import { parseTime } from "@/utils/ruoyi";
 import { Message as ElMessage, MessageBox as ElMessageBox } from 'element-ui'
-import { listTopic, getTopic, delTopic, addTopic, updateTopic } from "@/api/llm/topic";
-import { listChat, addChat, updateChat } from "@/api/llm/chat";
+import { listTopic, delTopic, addTopic, updateTopic } from "@/api/llm/topic";
+import { listChat, addChat } from "@/api/llm/chat";
 import { listDocument, uploadDocument } from "@/api/llm/document";
-import { getAnswer, getAnswerWithDocument, getAnswerStream, getAnswerWithDocumentStream } from "@/api/llm/session";
+import { getAnswerStream, getAnswerWithDocumentStream } from "@/api/llm/session";
 import logoImg from '@/assets/images/logo.png'
 import { marked } from 'marked';
 
-function createTypewriter(appender, options = {}) {
-  const intervalMs = typeof options.intervalMs === 'number' ? options.intervalMs : 25 // 约 40 字/秒
-  const maxCharsPerTick = typeof options.maxCharsPerTick === 'number' ? options.maxCharsPerTick : 1
-
-  let queue = ''
-  let timer = null
-  let ended = false
-  let onDrained = null
-
-  const tick = () => {
-    if (!queue) {
-      if (ended) {
-        if (timer) clearInterval(timer)
-        timer = null
-        if (onDrained) onDrained()
-      }
-      return
-    }
-
-    const n = Math.min(maxCharsPerTick, queue.length)
-    const chunk = queue.slice(0, n)
-    queue = queue.slice(n)
-    appender(chunk)
-  }
-
-  return {
-    push(text) {
-      if (!text) return
-      queue += text
-      if (!timer) timer = setInterval(tick, intervalMs)
-    },
-    end(cb) {
-      ended = true
-      onDrained = cb
-      if (!timer) timer = setInterval(tick, intervalMs)
-    },
-    stop() {
-      ended = true
-      queue = ''
-      if (timer) clearInterval(timer)
-      timer = null
-    }
-  }
-}
-
-function trimToLastSentenceEnd(text) {
-  const s = String(text || '')
-  // 句末标点：中文/英文句号、问号、叹号，或换行
-  const last = Math.max(
-    s.lastIndexOf('。'),
-    s.lastIndexOf('！'),
-    s.lastIndexOf('？'),
-    s.lastIndexOf('.'),
-    s.lastIndexOf('!'),
-    s.lastIndexOf('?'),
-    s.lastIndexOf('\n')
-  )
-  if (last === -1) return s.trim()
-  return s.slice(0, last + 1).trim()
-}
-
 export default {
   name: 'ChatView',
   data() {
             text: msg.output,
             time: msg.outputTime,
             id: msg.chatId ? msg.chatId + '_ai' : (msg.topicId + '_output_' + msg.outputTime),
+            tokenUsage: msg.tokenUsage
           })
         }
       }
         input: '',
         output: '',
         topicId: this.currentTopicId,
-        outputTime: this.parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}')
+        outputTime: this.parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}'),
+        tokenUsage: {
+          promptTokens: 0,
+          completionTokens: 0,
+          totalTokens: 0
+        }
       };
       this.chatMessages.push(aiMessage);
 
 
       this.isLoading = true;
 
-      const typewriter = createTypewriter((chunk) => {
-        aiMessage.output += chunk
-        aiMessage.outputTime = this.parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}')
-        this.$nextTick(() => this.scrollToBottom())
-      }, {
-        intervalMs: 25,
-        maxCharsPerTick: 1
-      })
-
       const streamParams = {
         topicId: this.currentTopicId,
         question: userMessage.input
               return;
             }
 
-            typewriter.push(String(content))
+            // 直接追加内容，不使用打字机效果
+            aiMessage.output += String(content);
+            aiMessage.outputTime = that.parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}');
+            that.$nextTick(() => that.scrollToBottom());
 
             if (window.responseTimeout) {
               clearTimeout(window.responseTimeout);
               if (that.isLoading) {
                 console.log('=== 响应超时强制结束 ===');
                 that.isLoading = false;
-                typewriter.stop()
                 if (window.responseTimeout) {
                   clearTimeout(window.responseTimeout);
                   window.responseTimeout = null;
               aiMessage.output += '\n\n[回答生成中断]';
             }
             that.isLoading = false;
-            typewriter.stop()
 
             that.$nextTick(() => {
               that.scrollToBottom();
               window.responseTimeout = null;
             }
 
-            typewriter.end(async () => {
-              aiMessage.output = trimToLastSentenceEnd(aiMessage.output)
-              try {
-                // 保存用户消息
-                const savedUserMessage = await addChat({
+            try {
+              // 保存用户消息
+              const savedUserMessage = await addChat({
+                ...that.chatMessages[userMessageIndex],
+                topicId: that.currentTopicId
+              });
+
+              // 保存AI消息
+              const savedMessage = await addChat({
+                ...aiMessage,
+                topicId: that.currentTopicId
+              });
+
+              if (savedUserMessage && savedUserMessage.chatId) {
+                that.$set(that.chatMessages, userMessageIndex, {
                   ...that.chatMessages[userMessageIndex],
-                  topicId: that.currentTopicId
+                  chatId: savedUserMessage.chatId
                 });
+              }
 
-                // 保存AI消息
-                const savedMessage = await addChat({
-                  ...aiMessage,
-                  topicId: that.currentTopicId
+              if (savedMessage && savedMessage.chatId) {
+                that.$set(that.chatMessages, messageIndex, {
+                  ...that.chatMessages[messageIndex],
+                  chatId: savedMessage.chatId
                 });
+              }
 
-                if (savedUserMessage && savedUserMessage.chatId) {
-                  that.$set(that.chatMessages, userMessageIndex, {
-                    ...that.chatMessages[userMessageIndex],
-                    chatId: savedUserMessage.chatId
-                  });
-                }
-
-                if (savedMessage && savedMessage.chatId) {
-                  that.$set(that.chatMessages, messageIndex, {
-                    ...that.chatMessages[messageIndex],
-                    chatId: savedMessage.chatId
-                  });
-                }
+              if (uploadedFileId) {
+                try {
+                  const fileResponse = await listDocument({ chatId: uploadedFileId });
+                  if (fileResponse.rows && fileResponse.rows.length > 0) {
+                    const messageChatId = that.chatMessages[messageIndex].chatId;
+                    const keyToUse = messageChatId || uploadedFileId;
+                    that.messageFileMap.set(keyToUse, fileResponse.rows);
 
-                if (uploadedFileId) {
-                  try {
-                    const fileResponse = await listDocument({ chatId: uploadedFileId });
-                    if (fileResponse.rows && fileResponse.rows.length > 0) {
-                      const messageChatId = that.chatMessages[messageIndex].chatId;
-                      const keyToUse = messageChatId || uploadedFileId;
-                      that.messageFileMap.set(keyToUse, fileResponse.rows);
-
-                      if (!that.chatMessages[messageIndex].chatId) {
-                        that.$set(that.chatMessages[messageIndex], 'chatId', uploadedFileId);
-                      }
+                    if (!that.chatMessages[messageIndex].chatId) {
+                      that.$set(that.chatMessages[messageIndex], 'chatId', uploadedFileId);
                     }
-                  } catch (error) {
-                    console.error('Failed to load files for new message:', error);
                   }
+                } catch (error) {
+                  console.error('Failed to load files for new message:', error);
                 }
-              } catch (error) {
-                console.error('保存消息失败:', error);
               }
+            } catch (error) {
+              console.error('保存消息失败:', error);
+            }
 
-              that.isLoading = false;
-              that.$nextTick(() => {
-                that.scrollToBottom();
-              });
-            })
+            that.isLoading = false;
+            that.$nextTick(() => {
+              that.scrollToBottom();
+            });
           }
         );
       } else {
         streamController = getAnswerStream(
           streamParams,
-          (content) => {
+          (jsonData, eventType) => {
             const that = this;
             if (!that.streamingStarted) that.streamingStarted = true;
 
-            if (!content || !String(content).trim()) {
+            // 根据事件类型处理不同的响应
+            if (eventType === 'error') {
+              // 错误事件
+              console.error('收到错误事件:', jsonData);
+              if (jsonData.errorMessage) {
+                aiMessage.output = jsonData.errorMessage;
+              }
               return;
             }
 
+            if (eventType === 'completed') {
+              // 完成事件，更新最终的token用量
+              if (jsonData.promptTokens !== undefined || jsonData.completionTokens !== undefined || jsonData.totalTokens !== undefined) {
+                aiMessage.tokenUsage.promptTokens = jsonData.promptTokens || aiMessage.tokenUsage.promptTokens;
+                aiMessage.tokenUsage.completionTokens = jsonData.completionTokens || aiMessage.tokenUsage.completionTokens;
+                aiMessage.tokenUsage.totalTokens = jsonData.totalTokens || aiMessage.tokenUsage.totalTokens;
+              }
+              return;
+            }
+
+            // 默认处理 answer 事件
+            if (!jsonData || !jsonData.content) {
+              return;
+            }
+
+            // 更新token用量
+            if (jsonData.promptTokens !== undefined || jsonData.completionTokens !== undefined || jsonData.totalTokens !== undefined) {
+              aiMessage.tokenUsage.promptTokens = jsonData.promptTokens || aiMessage.tokenUsage.promptTokens;
+              aiMessage.tokenUsage.completionTokens = jsonData.completionTokens || aiMessage.tokenUsage.completionTokens;
+              aiMessage.tokenUsage.totalTokens = jsonData.totalTokens || aiMessage.tokenUsage.totalTokens;
+            }
+
+            const content = jsonData.content;
+
             // 过滤不完整的句子（以逗号、顿号等结尾）
             const trimmed = String(content).trim();
             if (trimmed.length > 0 && trimmed.length < 30) {
               }
             }
 
-            typewriter.push(trimmed)
+            // 直接追加内容，不使用打字机效果
+            aiMessage.output += trimmed;
+            aiMessage.outputTime = that.parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}');
+            that.$nextTick(() => that.scrollToBottom());
 
             if (window.responseTimeout) {
               clearTimeout(window.responseTimeout);
               if (that.isLoading) {
                 console.log('=== 响应超时强制结束 ===');
                 that.isLoading = false;
-                typewriter.stop()
                 if (window.responseTimeout) {
                   clearTimeout(window.responseTimeout);
                   window.responseTimeout = null;
               aiMessage.output += '\n\n[回答生成中断]';
             }
             that.isLoading = false;
-            typewriter.stop()
 
             that.$nextTick(() => {
               that.scrollToBottom();
               window.responseTimeout = null;
             }
 
-            typewriter.end(async () => {
-              aiMessage.output = trimToLastSentenceEnd(aiMessage.output)
-              try {
-                // 保存用户消息
-                const savedUserMessage = await addChat({
+            try {
+              // 保存用户消息
+              const savedUserMessage = await addChat({
+                ...that.chatMessages[userMessageIndex],
+                topicId: that.currentTopicId
+              });
+
+              // 保存AI消息
+              const savedMessage = await addChat({
+                ...aiMessage,
+                topicId: that.currentTopicId
+              });
+
+              if (savedUserMessage && savedUserMessage.chatId) {
+                that.$set(that.chatMessages, userMessageIndex, {
                   ...that.chatMessages[userMessageIndex],
-                  topicId: that.currentTopicId
+                  chatId: savedUserMessage.chatId
                 });
+              }
 
-                // 保存AI消息
-                const savedMessage = await addChat({
-                  ...aiMessage,
-                  topicId: that.currentTopicId
+              if (savedMessage && savedMessage.chatId) {
+                that.$set(that.chatMessages, messageIndex, {
+                  ...that.chatMessages[messageIndex],
+                  chatId: savedMessage.chatId
                 });
-
-                if (savedUserMessage && savedUserMessage.chatId) {
-                  that.$set(that.chatMessages, userMessageIndex, {
-                    ...that.chatMessages[userMessageIndex],
-                    chatId: savedUserMessage.chatId
-                  });
-                }
-
-                if (savedMessage && savedMessage.chatId) {
-                  that.$set(that.chatMessages, messageIndex, {
-                    ...that.chatMessages[messageIndex],
-                    chatId: savedMessage.chatId
-                  });
-                }
-              } catch (error) {
-                console.error('保存消息失败:', error);
               }
+            } catch (error) {
+              console.error('保存消息失败:', error);
+            }
 
-              that.isLoading = false;
-              that.$nextTick(() => {
-                that.scrollToBottom();
-              });
-            })
+            that.isLoading = false;
+            that.$nextTick(() => {
+              that.scrollToBottom();
+            });
           }
         );
       }
 
       window.currentChatController = streamController;
     },
-    //简化formatMessage方法，不再需要复杂的过滤
+    //格式化消息内容
     formatMessage(content) {
-      // 直接使用marked解析markdown内容，不需要额外过滤
-      // 因为后端和session.js已经处理干净了
+      // 直接使用marked解析markdown内容
       return marked(content);
     },
 
       }
     },
 
-    formatMessage(content) {
-      // 使用marked.js解析markdown内容
-      return marked(content);
-    },
-
     formatMessageTime(time) {
       if (!time) return '';
-      const date = this.parseTime(new Date(time), '{y}-{m}-{d}');
+      const date = this.parseTime(new Date(time), '{y}-{m}-{d} {h}:{i}:{s}');
       return date;
     },
 
               max-width: 100%;
 
               .message-text {
-                margin-bottom: 4px;
-                white-space: pre-wrap;
-                overflow-wrap: break-word;
-              }
+                  margin-bottom: 4px;
+                  white-space: pre-wrap;
+                  overflow-wrap: break-word;
+                  line-height: 1.6;
+
+                  /* 表格样式 */
+                  table {
+                    width: 100%;
+                    border-collapse: collapse;
+                    margin: 8px 0;
+                    font-size: 14px;
+                  }
+
+                  th,
+                  td {
+                    border: 1px solid #ddd;
+                    padding: 8px 12px;
+                    text-align: left;
+                  }
+
+                  th {
+                    background-color: #f5f5f5;
+                    font-weight: 600;
+                  }
+
+                  tr:nth-child(even) {
+                    background-color: #f9f9f9;
+                  }
+
+                  /* 列表样式 */
+                  ul,
+                  ol {
+                    padding-left: 24px;
+                    margin: 8px 0;
+                  }
+
+                  li {
+                    margin-bottom: 4px;
+                  }
+
+                  /* 引用样式 */
+                  blockquote {
+                    border-left: 4px solid #1890ff;
+                    padding: 8px 12px;
+                    margin: 8px 0;
+                    color: #666;
+                    background-color: #f8f9fa;
+                    border-radius: 0 4px 4px 0;
+                  }
+
+                  /* 标题样式 */
+                  h1, h2, h3, h4, h5, h6 {
+                    margin: 12px 0 8px;
+                    font-weight: 600;
+                    color: #333;
+                  }
+
+                  h1 { font-size: 20px; }
+                  h2 { font-size: 18px; }
+                  h3 { font-size: 16px; }
+                  h4, h5, h6 { font-size: 14px; }
+
+                  /* 链接样式 */
+                  a {
+                    color: #1890ff;
+                    text-decoration: none;
+                  }
+
+                  a:hover {
+                    text-decoration: underline;
+                  }
+
+                  /* 粗体和斜体 */
+                  strong {
+                    font-weight: 600;
+                  }
+
+                  em {
+                    font-style: italic;
+                  }
+
+                  /* 行内代码 */
+                  code {
+                    background-color: #f4f4f4;
+                    padding: 2px 6px;
+                    border-radius: 4px;
+                    font-family: 'Consolas', 'Monaco', 'Courier New', monospace;
+                    font-size: 14px;
+                    color: #e06c75;
+                  }
+                }
 
               .message-time {
                 font-size: 12px;
                 opacity: 0.7;
                 text-align: right;
               }
+
+              .token-usage {
+                margin-top: 8px;
+                padding-top: 8px;
+                border-top: 1px dashed #e0e0e0;
+                font-size: 12px;
+                text-align: right;
+                color: #666;
+
+                .token-label {
+                  margin-right: 4px;
+                }
+
+                .token-value {
+                  color: #1890ff;
+                  font-weight: 500;
+                  margin-right: 4px;
+                }
+
+                .token-breakdown {
+                  color: #999;
+                  font-size: 11px;
+                }
+              }
             }
 
             .message-files {
   }
 }
 
-/* Markdown样式 */
-.message-bubble>>>.message-text {
-
-  /* 标题样式 */
-  h1 {
-    font-size: 24px;
-    font-weight: 700;
-    margin: 16px 0 12px 0;
-    color: #333;
-  }
-
-  h2 {
-    font-size: 20px;
-    font-weight: 600;
-    margin: 14px 0 10px 0;
-    color: #333;
-  }
-
-  h3 {
-    font-size: 18px;
-    font-weight: 600;
-    margin: 12px 0 8px 0;
-    color: #333;
-  }
-
-  h4,
-  h5,
-  h6 {
-    font-size: 16px;
-    font-weight: 600;
-    margin: 10px 0 6px 0;
-    color: #333;
-  }
-
-  /* 段落样式 */
-  p {
-    margin: 8px 0;
-    line-height: 1.6;
-  }
-
-  /* 列表样式 */
-  ul,
-  ol {
-    margin: 8px 0;
-    padding-left: 24px;
-  }
-
-  li {
-    margin: 4px 0;
-    line-height: 1.6;
-  }
-
-  /* 代码块样式 */
-  pre {
-    background-color: #f5f5f5;
-    border-radius: 6px;
-    padding: 12px;
-    overflow-x: auto;
-    margin: 12px 0;
-    font-family: 'Monaco', 'Menlo', 'Ubuntu Mono', monospace;
-    font-size: 14px;
-    line-height: 1.5;
-  }
-
-  code {
-    background-color: #f0f0f0;
-    border-radius: 3px;
-    padding: 2px 6px;
-    font-family: 'Monaco', 'Menlo', 'Ubuntu Mono', monospace;
-    font-size: 14px;
-  }
-
-  pre code {
-    background-color: transparent;
-    padding: 0;
-  }
-
-  /* 引用样式 */
-  blockquote {
-    border-left: 4px solid #007bff;
-    padding: 8px 12px;
-    margin: 12px 0;
-    background-color: #f8f9fa;
-    color: #666;
-    border-radius: 0 6px 6px 0;
-  }
-
-  /* 表格样式 */
-  table {
-    border-collapse: collapse;
-    width: 100%;
-    margin: 12px 0;
-  }
-
-  th,
-  td {
-    border: 1px solid #e9ecef;
-    padding: 8px 12px;
-    text-align: left;
-  }
-
-  th {
-    background-color: #f8f9fa;
-    font-weight: 600;
-  }
-
-  /* 链接样式 */
-  a {
-    color: #007bff;
-    text-decoration: none;
-    transition: color 0.2s ease;
-  }
-
-  a:hover {
-    color: #0056b3;
-    text-decoration: underline;
-  }
-
-  /* 粗体和斜体样式 */
-  strong {
-    font-weight: 600;
-  }
-
-  em {
-    font-style: italic;
-  }
-
-  /* 水平线样式 */
-  hr {
-    border: none;
-    border-top: 1px solid #e9ecef;
-    margin: 20px 0;
-  }
-}
 </style>

oa-ui/src/views/llm/knowledge/index.vue

                     </div>
 
                     <div class="message-time">{{ message.time }}</div>
-                  </div>
-                </div>
-              </div>
 
-              <!-- 加载状态：仅在尚未收到首段流内容时显示，避免出现两个AI气泡 -->
-              <div v-if="isSending && !streamingStarted" class="message-item ai">
-                <div class="message-avatar">
-                  <div class="ai-avatar">
-                    🤖
-                  </div>
-                </div>
-                <div class="message-content">
-                  <div class="message-bubble ai">
-                    <div class="loading-dots">
-                      <span></span><span></span><span></span>
+                    <!-- Token用量统计 -->
+                    <div v-if="message.type === 'ai' && message.tokenUsage"
+                      class="token-usage">
+                      <div class="token-info">
+                        <span class="token-label">消耗Token:</span>
+                        <span class="token-value">{{ message.tokenUsage.totalTokens || 0 }}</span>
+                        <span class="token-breakdown">(输入: {{ message.tokenUsage.promptTokens || 0 }}, 输出: {{ message.tokenUsage.completionTokens || 0 }})</span>
+                      </div>
                     </div>
                   </div>
                 </div>
               </div>
+
+              <!-- 注意：加载状态已整合到AI消息的typing样式中，不需要单独的加载div -->
             </div>
           </div>
 
 import { getAnswer, getAnswerStream, getContextFile } from '@/api/llm/rag';
 import { marked } from 'marked';
 
-function createTypewriter(appender, options = {}) {
-  const intervalMs = typeof options.intervalMs === 'number' ? options.intervalMs : 25 // 约 40 字/秒
-  const maxCharsPerTick = typeof options.maxCharsPerTick === 'number' ? options.maxCharsPerTick : 1
-
-  let queue = ''
-  let timer = null
-  let ended = false
-  let onDrained = null
-
-  const tick = () => {
-    if (!queue) {
-      if (ended) {
-        if (timer) clearInterval(timer)
-        timer = null
-        if (onDrained) onDrained()
-      }
-      return
-    }
-
-    const n = Math.min(maxCharsPerTick, queue.length)
-    const chunk = queue.slice(0, n)
-    queue = queue.slice(n)
-    appender(chunk)
-  }
-
-  return {
-    push(text) {
-      if (!text) return
-      queue += text
-      if (!timer) timer = setInterval(tick, intervalMs)
-    },
-    end(cb) {
-      ended = true
-      onDrained = cb
-      if (!timer) timer = setInterval(tick, intervalMs)
-    },
-    stop() {
-      ended = true
-      queue = ''
-      if (timer) clearInterval(timer)
-      timer = null
-    }
-  }
-}
-
-function trimToLastSentenceEnd(text) {
-  const s = String(text || '')
-  // 句末标点：中文/英文句号、问号、叹号，或换行
-  const last = Math.max(
-    s.lastIndexOf('。'),
-    s.lastIndexOf('！'),
-    s.lastIndexOf('？'),
-    s.lastIndexOf('.'),
-    s.lastIndexOf('!'),
-    s.lastIndexOf('?'),
-    s.lastIndexOf('\n')
-  )
-  if (last === -1) return s.trim()
-  return s.slice(0, last + 1).trim()
-}
-
 export default {
   name: 'KnowledgeManager',
   // 处理引用文件的显示状态
         this.stopGenerating();
       }
 
+      // 立即设置为发送状态，防止重复发送
+      this.isSending = true;
+
+      // 如果用户快速发送多条消息，取消之前的请求（放在发送新请求之前）
+      if (window.currentController) {
+        window.currentController.abort();
+        window.currentController = null;
+      }
+      if (window.responseTimeout) {
+        clearTimeout(window.responseTimeout);
+        window.responseTimeout = null;
+      }
+
       const userMessage = {
         type: 'user',
         content: this.chatInput.trim(),
       this.chatMessages.push(userMessage);
       const currentInput = this.chatInput;
       this.chatInput = '';
-      this.isSending = true;
       this.streamingStarted = false;
 
       // 滚动到底部
         content: '',
         time: parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}'),
         references: [], // 引用的文件信息
-        showAllReferences: false // 是否显示所有引用文件
+        showAllReferences: false, // 是否显示所有引用文件
+        tokenUsage: {
+          promptTokens: 0,
+          completionTokens: 0,
+          totalTokens: 0
+        }
       };
       this.chatMessages.push(aiMessage);
 
-      const typewriter = createTypewriter((chunk) => {
-        aiMessage.content += chunk
-        aiMessage.time = parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}')
-        this.$nextTick(() => this.scrollToBottom())
-      }, {
-        intervalMs: 25,
-        maxCharsPerTick: 1
-      })
-
       // 使用流式API获取回答
       const eventSource = getAnswerStream(
         currentInput,
         this.selectedKnowledge.collectionName,
         // onMessage: 接收到每个字符时的回调
-        (content) => {
+        (jsonData, eventType) => {
           const that = this;
           if (!that.streamingStarted) that.streamingStarted = true;
 
-          if (!content || !String(content).trim()) {
+          // 根据事件类型处理不同的响应
+          if (eventType === 'error') {
+            // 错误事件
+            console.error('收到错误事件:', jsonData);
+            if (jsonData.errorMessage) {
+              aiMessage.content = jsonData.errorMessage;
+            }
             return;
           }
 
-          typewriter.push(String(content))
+          if (eventType === 'completed') {
+            // 完成事件，更新最终的token用量
+            if (jsonData.promptTokens !== undefined || jsonData.completionTokens !== undefined || jsonData.totalTokens !== undefined) {
+              const index = that.chatMessages.length - 1;
+              if (that.chatMessages[index] && that.chatMessages[index].tokenUsage) {
+                that.$set(that.chatMessages[index].tokenUsage, 'promptTokens', jsonData.promptTokens || 0);
+                that.$set(that.chatMessages[index].tokenUsage, 'completionTokens', jsonData.completionTokens || 0);
+                that.$set(that.chatMessages[index].tokenUsage, 'totalTokens', jsonData.totalTokens || 0);
+              }
+            }
+            return;
+          }
+
+          // 默认处理 answer 事件
+          if (!jsonData) {
+            return;
+          }
+
+          // 更新token用量
+          if (jsonData.promptTokens !== undefined || jsonData.completionTokens !== undefined || jsonData.totalTokens !== undefined) {
+            const index = that.chatMessages.length - 1;
+            if (that.chatMessages[index] && that.chatMessages[index].tokenUsage) {
+              that.$set(that.chatMessages[index].tokenUsage, 'promptTokens', jsonData.promptTokens || that.chatMessages[index].tokenUsage.promptTokens);
+              that.$set(that.chatMessages[index].tokenUsage, 'completionTokens', jsonData.completionTokens || that.chatMessages[index].tokenUsage.completionTokens);
+              that.$set(that.chatMessages[index].tokenUsage, 'totalTokens', jsonData.totalTokens || that.chatMessages[index].tokenUsage.totalTokens);
+            }
+          }
+
+          // 获取内容
+          const content = jsonData.content;
+          if (content && content !== '[DONE]') {
+            // 直接追加内容，不使用打字机效果
+            aiMessage.content += String(content);
+            aiMessage.time = parseTime(new Date(), '{y}-{m}-{d} {h}:{i}:{s}');
+            that.$nextTick(() => that.scrollToBottom());
+          }
 
           if (window.responseTimeout) {
             clearTimeout(window.responseTimeout);
             if (that.isSending) {
               console.log('=== 响应超时强制结束 ===');
               that.isSending = false;
-              typewriter.stop()
               if (window.responseTimeout) {
                 clearTimeout(window.responseTimeout);
                 window.responseTimeout = null;
             aiMessage.content += '\n\n[回答生成中断]';
           }
           that.isSending = false;
-          typewriter.stop()
 
           that.$nextTick(() => {
             that.scrollToBottom();
           });
         },
         // onComplete: 回答完成时的回调
-        () => {
+        async () => {
           const that = this;
           console.log('=== 回答完成 ===');
 
             window.responseTimeout = null;
           }
 
-          typewriter.end(async () => {
-            aiMessage.content = trimToLastSentenceEnd(aiMessage.content)
-            try {
-              // 获取上下文引用文件
-              const response = await getContextFile(currentInput, that.selectedKnowledge.collectionName);
-              console.log('=== 上下文文件 ===', response)
-              if (response && Array.isArray(response)) {
-                aiMessage.references = response.map(item => ({
-                  fileName: item.file_name,
-                  similarity: item.score,
-                  content: item.content
-                }));
-                that.$forceUpdate(); // 强制更新以显示引用文件
-              }
-            } catch (error) {
-              console.error('获取上下文文件失败:', error);
+          try {
+            // 获取上下文引用文件
+            const response = await getContextFile(currentInput, that.selectedKnowledge.collectionName);
+            console.log('=== 上下文文件 ===', response)
+            if (response && Array.isArray(response)) {
+              aiMessage.references = response.map(item => ({
+                fileName: item.file_name,
+                similarity: item.score,
+                content: item.content
+              }));
+              that.$forceUpdate(); // 强制更新以显示引用文件
             }
+          } catch (error) {
+            console.error('获取上下文文件失败:', error);
+          }
 
-            that.isSending = false;
-            that.$nextTick(() => {
-              that.scrollToBottom();
-            });
-          })
+          that.isSending = false;
+          that.$nextTick(() => {
+            that.scrollToBottom();
+          });
         }
       );
-      // 如果用户快速发送多条消息，取消之前的请求
-      if (window.currentController) {
-        window.currentController.abort();
-      }
-      if (window.responseTimeout) {
-        clearTimeout(window.responseTimeout);
-      }
 
       window.currentController = eventSource;
     },
         window.currentController.abort();
         window.currentController = null;
       }
+
+      // 如果最后一条消息是空的AI消息，移除它（防止重复消息）
+      const lastMessage = this.chatMessages[this.chatMessages.length - 1];
+      if (lastMessage && lastMessage.type === 'ai' && !lastMessage.content) {
+        this.chatMessages.pop();
+      }
+
       this.isSending = false;
     },
 
     font-size: 12px;
     opacity: 0.7;
   }
+
+  .token-usage {
+    margin-top: 8px;
+    padding-top: 8px;
+    border-top: 1px dashed #e0e0e0;
+  }
+
+  .token-info {
+    display: flex;
+    align-items: center;
+    font-size: 12px;
+    color: #666;
+  }
+
+  .token-label {
+    margin-right: 4px;
+    color: #999;
+  }
+
+  .token-value {
+    font-weight: 600;
+    color: #409eff;
+    margin-right: 8px;
+  }
+
+  .token-breakdown {
+    color: #999;
+    font-size: 11px;
+  }
 }
 
-// 打字机效果的光标 - 只在正在生成时显示
+// 打字机效果的光标 - 只在正在生成且有内容时显示
 .message-bubble.ai.typing .message-text:after {
   content: '|';
   animation: blink 1s infinite;
   color: #409eff;
 }
 
+// 加载状态：内容为空时显示加载动画
+.message-bubble.ai.typing:empty::before {
+  content: '';
+  display: inline-block;
+  width: 60px;
+  height: 20px;
+  background: url("data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 60 20'%3E%3Ccircle cx='10' cy='10' r='4' fill='%23409eff'%3E%3Canimate attributeName='opacity' values='1;0.3;1' dur='1s' repeatCount='indefinite'/%3E%3C/circle%3E%3Ccircle cx='30' cy='10' r='4' fill='%23409eff'%3E%3Canimate attributeName='opacity' values='0.3;1;0.3' dur='1s' begin='0.2s' repeatCount='indefinite'/%3E%3C/circle%3E%3Ccircle cx='50' cy='10' r='4' fill='%23409eff'%3E%3Canimate attributeName='opacity' values='0.3;1;0.3' dur='1s' begin='0.4s' repeatCount='indefinite'/%3E%3C/circle%3E%3C/svg%3E") no-repeat center;
+  background-size: contain;
+}
+
+.message-bubble.ai.typing:empty .message-text {
+  display: none;
+}
+
 @keyframes blink {
 
   0%,