直接使用大模型会话

llm-back/cmc-agent/pom.xml

             <version>0.35.0</version>
         </dependency>
 
-        <!-- 解析PDF -->
+        <!-- LangChain4j pdfParser 集成 -->
         <dependency>
-            <groupId>org.apache.pdfbox</groupId>
-            <artifactId>pdfbox</artifactId>
-            <version>3.0.3</version>
+            <groupId>dev.langchain4j</groupId>
+            <artifactId>langchain4j-document-parser-apache-pdfbox</artifactId>
+            <version>0.35.0</version>
         </dependency>
 
-
     </dependencies>
 
     <build>

llm-back/cmc-agent/src/main/java/com/cmc/agent/controller/KnowLedgeController.java

     public AjaxResult createKnowLedgeBase(String collectionName)
     {
         LangChainMilvusService langChainMilvusService = new LangChainMilvusService(
-                "localhost",
+                "192.168.28.188",
                 19530,
                 collectionName,
                 embeddingModel);
     @PostMapping("/insert")
     public AjaxResult insertKnowledge(MultipartFile file, String collectionName) throws IOException {
         LangChainMilvusService langChainMilvusService = new LangChainMilvusService(
-                "localhost",
+                "192.168.28.188",
                 19530,
                 collectionName,
                 embeddingModel);

llm-back/cmc-agent/src/main/java/com/cmc/agent/controller/RagController.java

      * 增强检索生成回答
      */
     @PostMapping("/answer")
-    public AjaxResult answerWithRAG(String question, String collectionName) throws IOException {
+    public AjaxResult answer(String question, String collectionName) throws IOException {
         LangChainMilvusService langChainMilvusService = new LangChainMilvusService(
-                "localhost",
+                "192.168.28.188",
                 19530,
                 collectionName,
                 embeddingModel);
 
-        // 2. Milvus检索
+        // 1. Milvus检索
         List<String> contexts = langChainMilvusService.retrieveFromMilvus(question, 3);
 
-        // 3. 调用本地LLM或HTTP服务
-        return success(langChainMilvusService.generateAnswer(question, contexts));
+        // 2. 调用本地LLM或HTTP服务
+        return success(langChainMilvusService.generateAnswerWithRag(question, contexts));
     }
 
 }

llm-back/cmc-agent/src/main/java/com/cmc/agent/controller/SessionController.java

+package com.cmc.agent.controller;
+
+import com.google.gson.Gson;
+import com.ruoyi.common.core.controller.BaseController;
+import com.ruoyi.common.core.domain.AjaxResult;
+import okhttp3.*;
+import org.springframework.web.bind.annotation.PostMapping;
+import org.springframework.web.bind.annotation.RequestMapping;
+import org.springframework.web.bind.annotation.RestController;
+
+import java.io.IOException;
+import java.util.HashMap;
+import java.util.Map;
+
+/**
+ * cmc知识库Controller
+ * 
+ * @author cmc
+ * @date 2025-04-08
+ */
+@RestController
+@RequestMapping("/llm/session")
+public class SessionController extends BaseController
+{
+    /**
+     * 生成回答
+     */
+    @PostMapping("/answer")
+    public AjaxResult answer(String question) throws IOException {
+
+        // 1. 调用本地LLM或HTTP服务
+        return success(generateAnswer(question));
+    }
+
+    // 调用LLM生成回答
+    public String generateAnswer(String question) throws IOException {
+        Gson gson = new Gson();
+        Map<String, String> hashMap = new HashMap<>();
+        hashMap.put("prompt", question);
+        hashMap.put("max_tokens", "512");
+        RequestBody body = RequestBody.create(
+                MediaType.parse("application/json"),
+                new Gson().toJson(hashMap));
+
+        Request request = new Request.Builder()
+                .url("http://192.168.28.188:8080/generate")
+                .post(body)
+                .build();
+
+        try (Response response = new OkHttpClient().newCall(request).execute()) {
+            return gson.fromJson(response.body().string(), Map.class).get("generated_text").toString();
+        }
+    }
+
+}

llm-back/cmc-agent/src/main/java/com/cmc/agent/service/LangChainMilvusService.java

 import java.util.stream.Collectors;
 
 public class LangChainMilvusService {
-    private static final String LLM_SERVICE_URL = "http://localhost:8000/generate";
-    private final MilvusServiceClient milvusClient;
-    private final String collectionName;
-    private  EmbeddingModel embeddingModel;
+    private String LLM_SERVICE_URL;
+    private MilvusServiceClient milvusClient;
+    private String collectionName;
+    private EmbeddingModel embeddingModel;
 
     /**
      * 连接milvus知识库，加入langchain自带emdding
                         .withPort(port)
                         .build()
         );
+        this.LLM_SERVICE_URL = "http://" + host + ":8000/generate";
         this.collectionName = collectionName;
         this.embeddingModel = embeddingModel;
     }
 
+    /**
+     * 不使用向量数据库
+     */
+    public LangChainMilvusService(String host, int port) {
+        this.milvusClient = new MilvusServiceClient(
+                ConnectParam.newBuilder()
+                        .withHost(host)
+                        .withPort(port)
+                        .build()
+        );
+        this.LLM_SERVICE_URL = "http://" + host + ":8000/generate";
+    }
+
     /**
      * 新建知识库Collection（含Schema、Field、Index）
      */
                 .collect(Collectors.toList());
     }
 
-    // 调用LLM生成回答
-    public String generateAnswer(String question, List<String> contexts) throws IOException {
+    // 调用LLM+RAG生成回答
+    public String generateAnswerWithRag(String question, List<String> contexts) throws IOException {
         String prompt = buildPrompt(question, contexts);
         Gson gson = new Gson();
         Map<String, String> hashMap = new HashMap<>();
                 .build();
 
         try (Response response = new OkHttpClient().newCall(request).execute()) {
-            return gson.fromJson(response.body().string(), Map.class).get("text").toString();
+            return gson.fromJson(response.body().string(), Map.class).get("generated_text").toString();
         }
     }