feat(knowledge): 实现Milvus向量库策略并重构配置管理

- 新增Milvus向量库策略实现类MilvusVectorStoreStrategy
- 重构向量库配置管理，使用VectorStoreProperties统一配置
- 修改AbstractVectorStoreStrategy抽象类依赖注入方式
- 更新Weaviate策略实现类适配新的配置方式
- 移除旧的ConfigService配置读取方式
- 添加向量库类型配置项，默认使用weaviate
- 实现Milvus集合创建、数据存储、向量搜索和删除功能
- 优化向量库策略工厂类VectorStoreStrategyFactory初始化逻辑
- 删除已废弃的Milvus实现指南文档
- 升级Milvus SDK版本并调整相关API调用方式

MILVUS_IMPLEMENTATION_GUIDE.md

@@ -1,237 +0,0 @@
-# Milvus向量库实现指南
-
-## 概述
-
-本项目已完成Milvus向量库的集成，基于Milvus SDK 2.6.4版本实现。Milvus是一个开源的向量数据库，专为AI应用和相似性搜索而设计。
-
-## 实现特性
-
-### ✅ 已实现功能
-
-1. **集合管理**
-   - 自动创建集合（Collection）
-   - 检查集合是否存在
-   - 删除集合
-
-2. **数据存储**
-   - 批量插入向量数据
-   - 支持文本、fid、kid、docId等元数据
-   - 自动生成向量嵌入
-
-3. **向量搜索**
-   - 基于相似性的向量搜索
-   - 支持TopK结果返回
-   - 返回相关文本内容
-
-4. **数据删除**
-   - 按文档ID删除
-   - 按片段ID删除
-   - 删除整个集合
-
-## 架构设计
-
-### 策略模式实现
-
-```
-AbstractVectorStoreStrategy (抽象基类)
-    ↓
-MilvusVectorStoreStrategy (Milvus实现)
-WeaviateVectorStoreStrategy (Weaviate实现)
-```
-
-### 核心类说明
-
-- **MilvusVectorStoreStrategy**: Milvus向量库策略实现
-- **VectorStoreStrategyFactory**: 向量库策略工厂，支持动态切换
-- **VectorStoreService**: 向量库服务接口
-
-## 配置说明
-
-### 必需配置项
-
-在系统配置中需要设置以下Milvus相关配置：
-
-```properties
-# Milvus服务地址
-milvus.url=http://localhost:19530
-
-# 集合名称前缀
-milvus.collectionname=LocalKnowledge
-
-# 向量库类型选择
-vector.store_type=milvus
-```
-
-### 集合Schema设计
-
-每个集合包含以下字段：
-
-| 字段名 | 类型 | 说明 |
-|--------|------|------|
-| id | Int64 | 主键，自动生成 |
-| text | VarChar(65535) | 文本内容 |
-| fid | VarChar(255) | 片段ID |
-| kid | VarChar(255) | 知识库ID |
-| docId | VarChar(255) | 文档ID |
-| vector | FloatVector(1024) | 向量数据 |
-
-### 索引配置
-
-- **索引类型**: IVF_FLAT
-- **距离度量**: L2 (欧几里得距离)
-- **参数**: nlist=1024
-
-## 使用示例
-
-### 1. 创建集合
-
-```java
-MilvusVectorStoreStrategy strategy = new MilvusVectorStoreStrategy(configService);
-strategy.createSchema("bge-large-zh-v1.5", "test001", "test-model");
-```
-
-### 2. 存储向量数据
-
-```java
-StoreEmbeddingBo storeEmbeddingBo = new StoreEmbeddingBo();
-storeEmbeddingBo.setVectorModelName("bge-large-zh-v1.5");
-storeEmbeddingBo.setKid("test001");
-storeEmbeddingBo.setDocId("doc001");
-storeEmbeddingBo.setChunkList(Arrays.asList("文本1", "文本2"));
-storeEmbeddingBo.setFids(Arrays.asList("fid001", "fid002"));
-
-strategy.storeEmbeddings(storeEmbeddingBo);
-```
-
-### 3. 查询向量数据
-
-```java
-QueryVectorBo queryVectorBo = new QueryVectorBo();
-queryVectorBo.setQuery("查询文本");
-queryVectorBo.setKid("test001");
-queryVectorBo.setMaxResults(5);
-
-List<String> results = strategy.getQueryVector(queryVectorBo);
-```
-
-### 4. 删除数据
-
-```java
-// 按文档ID删除
-strategy.removeByDocId("doc001", "test001");
-
-// 按片段ID删除
-strategy.removeByFid("fid001", "test001");
-
-// 删除整个集合
-strategy.removeById("test001", "model");
-```
-
-## 部署要求
-
-### Milvus服务部署
-
-1. **Docker部署** (推荐)
-```bash
-# 下载docker-compose文件
-wget https://github.com/milvus-io/milvus/releases/download/v2.6.4/milvus-standalone-docker-compose.yml -O docker-compose.yml
-
-# 启动Milvus
-docker-compose up -d
-```
-
-2. **验证部署**
-```bash
-# 检查服务状态
-docker-compose ps
-
-# 查看日志
-docker-compose logs milvus-standalone
-```
-
-### 系统要求
-
-- **内存**: 最少8GB，推荐16GB+
-- **存储**: SSD推荐，至少50GB可用空间
-- **CPU**: 4核心以上
-- **网络**: 确保19530端口可访问
-
-## 性能优化
-
-### 1. 索引优化
-
-根据数据量调整索引参数：
-- 小数据集(<100万): nlist=1024
-- 中等数据集(100万-1000万): nlist=4096
-- 大数据集(>1000万): nlist=16384
-
-### 2. 批量操作
-
-- 批量插入：建议每批1000-10000条记录
-- 批量查询：避免频繁的单条查询
-
-### 3. 内存管理
-
-```yaml
-# docker-compose.yml中的内存配置
-environment:
-  MILVUS_CONFIG_PATH: /milvus/configs/milvus.yaml
-volumes:
-  - ./milvus.yaml:/milvus/configs/milvus.yaml
-```
-
-## 故障排除
-
-### 常见问题
-
-1. **连接失败**
-   - 检查Milvus服务是否启动
-   - 验证网络连接和端口
-   - 确认配置中的URL正确
-
-2. **集合创建失败**
-   - 检查集合名称是否符合规范
-   - 验证字段定义是否正确
-   - 查看Milvus日志获取详细错误
-
-3. **插入数据失败**
-   - 确认向量维度与schema一致
-   - 检查数据格式是否正确
-   - 验证集合是否已加载
-
-4. **查询无结果**
-   - 确认集合中有数据
-   - 检查查询参数设置
-   - 验证向量化模型一致性
-
-### 日志调试
-
-启用详细日志：
-```properties
-logging.level.org.ruoyi.service.strategy.impl.MilvusVectorStoreStrategy=DEBUG
-logging.level.io.milvus=DEBUG
-```
-
-## 与Weaviate对比
-
-| 特性 | Milvus | Weaviate |
-|------|--------|----------|
-| 性能 | 高性能，专为大规模设计 | 中等性能 |
-| 部署 | 需要独立部署 | 可独立部署或云服务 |
-| 生态 | 专注向量搜索 | 集成更多AI功能 |
-| 学习成本 | 中等 | 较低 |
-| 扩展性 | 优秀 | 良好 |
-
-## 后续优化建议
-
-1. **连接池管理**: 实现MilvusClient连接池
-2. **异步操作**: 支持异步插入和查询
-3. **分片策略**: 大数据集的分片管理
-4. **监控告警**: 集成性能监控
-5. **备份恢复**: 数据备份和恢复机制
-
-## 参考资料
-
-- [Milvus官方文档](https://milvus.io/docs)
-- [Milvus Java SDK](https://github.com/milvus-io/milvus-sdk-java)
-- [向量数据库最佳实践](https://milvus.io/docs/performance_faq.md)

ruoyi-admin/src/main/resources/application.yml

@@ -328,3 +328,17 @@ spring:
           servers-configuration: classpath:mcp-server.json
         request-timeout: 300s
 
+--- # 向量库配置
+vector-store:
+  # 向量存储类型 (weaviate/milvus)
+  type: weaviate
+  # Weaviate配置
+  weaviate:
+    protocol: http
+    host: 127.0.0.1:6038
+    classname: LocalKnowledge
+  # Milvus配置
+  milvus:
+    url: http://localhost:19530
+    collectionname: LocalKnowledge
+

ruoyi-common/ruoyi-common-core/src/main/java/org/ruoyi/common/core/config/VectorStoreProperties.java

@@ -0,0 +1,62 @@
+package org.ruoyi.common.core.config;
+
+import lombok.Data;
+import org.springframework.boot.context.properties.ConfigurationProperties;
+import org.springframework.stereotype.Component;
+
+/**
+ * 向量库配置属性
+ *
+ * @author ageer
+ */
+@Data
+@Component
+@ConfigurationProperties(prefix = "vector-store")
+public class VectorStoreProperties {
+
+    /**
+     * 向量库类型
+     */
+    private String type = "weaviate";
+
+    /**
+     * Weaviate配置
+     */
+    private Weaviate weaviate = new Weaviate();
+
+    /**
+     * Milvus配置
+     */
+    private Milvus milvus = new Milvus();
+
+    @Data
+    public static class Weaviate {
+        /**
+         * 协议
+         */
+        private String protocol = "http";
+
+        /**
+         * 主机地址
+         */
+        private String host = "localhost:8080";
+
+        /**
+         * 类名
+         */
+        private String classname = "Document";
+    }
+
+    @Data
+    public static class Milvus {
+        /**
+         * 连接URL
+         */
+        private String url = "http://localhost:19530";
+
+        /**
+         * 集合名称
+         */
+        private String collectionname = "knowledge_base";
+    }
+}

ruoyi-modules-api/ruoyi-knowledge-api/src/main/java/org/ruoyi/service/VectorStoreService.java

@@ -1,6 +1,6 @@
 package org.ruoyi.service;
 
-import com.google.protobuf.ServiceException;
+import org.ruoyi.common.core.exception.ServiceException;
 import org.ruoyi.domain.bo.QueryVectorBo;
 import org.ruoyi.domain.bo.StoreEmbeddingBo;
 

ruoyi-modules-api/ruoyi-knowledge-api/src/main/java/org/ruoyi/service/impl/VectorStoreServiceImpl.java

@@ -2,7 +2,6 @@ package org.ruoyi.service.impl;
 
 import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
-import org.ruoyi.common.core.service.ConfigService;
 import org.ruoyi.domain.bo.QueryVectorBo;
 import org.ruoyi.domain.bo.StoreEmbeddingBo;
 import org.ruoyi.service.VectorStoreService;
@@ -13,7 +12,7 @@ import org.springframework.stereotype.Service;
 import java.util.List;
 
 /**
- * 向量库管理服务实现 - 使用策略模式
+ * 向量库服务实现
  *
  * @author ageer
  */
@@ -22,7 +21,6 @@ import java.util.List;
 @RequiredArgsConstructor
 public class VectorStoreServiceImpl implements VectorStoreService {
 
-    private final ConfigService configService;
     private final VectorStoreStrategyFactory strategyFactory;
 
 
@@ -30,11 +28,7 @@ public class VectorStoreServiceImpl implements VectorStoreService {
      * 获取当前配置的向量库策略
      */
     private VectorStoreStrategy getCurrentStrategy() {
-        String vectorStoreType = configService.getConfigValue("vector", "type");
+        return strategyFactory.getStrategy();
-        if (vectorStoreType == null || vectorStoreType.trim().isEmpty()) {
-            vectorStoreType = "weaviate"; // 默认使用weaviate
-        }
-        return strategyFactory.getStrategy(vectorStoreType);
     }
 
     @Override

ruoyi-modules-api/ruoyi-knowledge-api/src/main/java/org/ruoyi/service/strategy/AbstractVectorStoreStrategy.java

@@ -1,13 +1,13 @@
 package org.ruoyi.service.strategy;
 
-import com.google.protobuf.ServiceException;
+import org.ruoyi.common.core.exception.ServiceException;
 import dev.langchain4j.model.embedding.EmbeddingModel;
 import dev.langchain4j.model.ollama.OllamaEmbeddingModel;
 import dev.langchain4j.model.openai.OpenAiEmbeddingModel;
 import lombok.RequiredArgsConstructor;
 import lombok.SneakyThrows;
 import lombok.extern.slf4j.Slf4j;
-import org.ruoyi.common.core.service.ConfigService;
+import org.ruoyi.common.core.config.VectorStoreProperties;
 
 /**
  * 向量库策略抽象基类
@@ -19,7 +19,7 @@ import org.ruoyi.common.core.service.ConfigService;
 @RequiredArgsConstructor
 public abstract class AbstractVectorStoreStrategy implements VectorStoreStrategy {
 
-    protected final ConfigService configService;
+    protected final VectorStoreProperties vectorStoreProperties;
 
     /**
      * 获取向量模型

ruoyi-modules-api/ruoyi-knowledge-api/src/main/java/org/ruoyi/service/strategy/VectorStoreStrategyFactory.java

@@ -1,15 +1,15 @@
 package org.ruoyi.service.strategy;
 
+import jakarta.annotation.PostConstruct;
 import lombok.RequiredArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
-import org.ruoyi.common.core.service.ConfigService;
+import org.ruoyi.common.core.config.VectorStoreProperties;
-import org.springframework.beans.factory.annotation.Autowired;
+import org.ruoyi.service.strategy.impl.MilvusVectorStoreStrategy;
-import org.springframework.context.ApplicationContext;
+import org.ruoyi.service.strategy.impl.WeaviateVectorStoreStrategy;
-import org.springframework.context.ApplicationContextAware;
 import org.springframework.stereotype.Component;
 
+import java.util.HashMap;
 import java.util.Map;
-import java.util.concurrent.ConcurrentHashMap;
 
 /**
  * 向量库策略工厂
@@ -20,69 +20,55 @@ import java.util.concurrent.ConcurrentHashMap;
 @Slf4j
 @Component
 @RequiredArgsConstructor
-public class VectorStoreStrategyFactory implements ApplicationContextAware {
+public class VectorStoreStrategyFactory {
 
-    private final ConfigService configService;
+    private final VectorStoreProperties vectorStoreProperties;
-    private final Map<String, VectorStoreStrategy> strategyMap = new ConcurrentHashMap<>();
+    private final WeaviateVectorStoreStrategy weaviateStrategy;
-    private ApplicationContext applicationContext;
+    private final MilvusVectorStoreStrategy milvusStrategy;
 
-    @Override
+    private Map<String, VectorStoreStrategy> strategies;
-    public void setApplicationContext(ApplicationContext applicationContext) {
-        this.applicationContext = applicationContext;
-        initStrategies();
-    }
 
-    /**
+    @PostConstruct
-     * 初始化所有策略实现
+    public void init() {
-     */
+        strategies = new HashMap<>();
-    private void initStrategies() {
+        strategies.put("weaviate", weaviateStrategy);
-        Map<String, VectorStoreStrategy> strategies = applicationContext.getBeansOfType(VectorStoreStrategy.class);
+        strategies.put("milvus", milvusStrategy);
-        for (VectorStoreStrategy strategy : strategies.values()) {
+        log.info("向量库策略工厂初始化完成，支持的策略: {}", strategies.keySet());
-            if (strategy instanceof AbstractVectorStoreStrategy) {
-                AbstractVectorStoreStrategy abstractStrategy = (AbstractVectorStoreStrategy) strategy;
-                strategyMap.put(abstractStrategy.getVectorStoreType(), strategy);
-                log.info("注册向量库策略: {}", abstractStrategy.getVectorStoreType());
-            }
-        }
     }
 
     /**
      * 获取当前配置的向量库策略
      */
     public VectorStoreStrategy getStrategy() {
-        String vectorStoreType = configService.getConfigValue("vector", "store_type");
+        String vectorStoreType = vectorStoreProperties.getType();
-        if (vectorStoreType == null || vectorStoreType.isEmpty()) {
+        if (vectorStoreType == null || vectorStoreType.trim().isEmpty()) {
             vectorStoreType = "weaviate"; // 默认使用weaviate
         }
         
-        VectorStoreStrategy strategy = strategyMap.get(vectorStoreType);
+        VectorStoreStrategy strategy = strategies.get(vectorStoreType.toLowerCase());
         if (strategy == null) {
             log.warn("未找到向量库策略: {}, 使用默认策略: weaviate", vectorStoreType);
-            strategy = strategyMap.get("weaviate");
+            strategy = strategies.get("weaviate");
-        }
-        
-        if (strategy == null) {
-            throw new RuntimeException("未找到可用的向量库策略实现");
         }
         
+        log.debug("使用向量库策略: {}", vectorStoreType);
         return strategy;
     }
 
     /**
-     * 根据类型获取特定的向量库策略
+     * 根据类型获取向量库策略
      */
-    public VectorStoreStrategy getStrategy(String vectorStoreType) {
+    public VectorStoreStrategy getStrategy(String type) {
-        VectorStoreStrategy strategy = strategyMap.get(vectorStoreType);
+        if (type == null || type.trim().isEmpty()) {
-        if (strategy == null) {
+            return getStrategy();
-            throw new RuntimeException("未找到向量库策略: " + vectorStoreType);
-        }
-        return strategy;
         }
         
-    /**
+        VectorStoreStrategy strategy = strategies.get(type.toLowerCase());
-     * 获取所有可用的向量库类型
+        if (strategy == null) {
-     */
+            log.warn("未找到向量库策略: {}, 使用默认策略", type);
-    public String[] getAvailableTypes() {
+            return getStrategy();
-        return strategyMap.keySet().toArray(new String[0]);
+        }
+        
+        return strategy;
     }
 }

ruoyi-modules-api/ruoyi-knowledge-api/src/main/java/org/ruoyi/service/strategy/impl/MilvusVectorStoreStrategy.java

@@ -1,30 +1,23 @@
 package org.ruoyi.service.strategy.impl;
 
-import com.google.gson.Gson;
+import org.ruoyi.common.core.exception.ServiceException;
-import com.google.gson.JsonObject;
-import com.google.protobuf.ServiceException;
 import dev.langchain4j.data.embedding.Embedding;
 import dev.langchain4j.model.embedding.EmbeddingModel;
-import io.milvus.v2.client.ConnectConfig;
+import io.milvus.client.MilvusServiceClient;
-import io.milvus.v2.client.MilvusClientV2;
+import io.milvus.common.clientenum.ConsistencyLevelEnum;
-import io.milvus.v2.common.DataType;
+import io.milvus.grpc.*;
-import io.milvus.v2.common.IndexParam;
+import io.milvus.param.*;
-import io.milvus.v2.service.collection.request.AddFieldReq;
+import io.milvus.param.collection.*;
-import io.milvus.v2.service.collection.request.CreateCollectionReq;
+import io.milvus.param.dml.DeleteParam;
-import io.milvus.v2.service.collection.request.DescribeCollectionReq;
+import io.milvus.param.dml.InsertParam;
-import io.milvus.v2.service.collection.request.DropCollectionReq;
+import io.milvus.param.dml.SearchParam;
-import io.milvus.v2.service.collection.request.HasCollectionReq;
+import io.milvus.param.index.CreateIndexParam;
-import io.milvus.v2.service.collection.response.DescribeCollectionResp;
+import io.milvus.param.index.DescribeIndexParam;
-import io.milvus.v2.service.vector.request.DeleteReq;
+import io.milvus.response.DescCollResponseWrapper;
-import io.milvus.v2.service.vector.request.InsertReq;
+import io.milvus.response.SearchResultsWrapper;
-import io.milvus.v2.service.vector.request.SearchReq;
+import lombok.SneakyThrows;
-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.DeleteResp;
-import io.milvus.v2.service.vector.response.InsertResp;
-import io.milvus.v2.service.vector.response.SearchResp;
 import lombok.extern.slf4j.Slf4j;
-import org.ruoyi.common.core.service.ConfigService;
+import org.ruoyi.common.core.config.VectorStoreProperties;
 import org.ruoyi.domain.bo.QueryVectorBo;
 import org.ruoyi.domain.bo.StoreEmbeddingBo;
 import org.ruoyi.service.strategy.AbstractVectorStoreStrategy;
@@ -41,10 +34,10 @@ import java.util.*;
 @Component
 public class MilvusVectorStoreStrategy extends AbstractVectorStoreStrategy {
 
-    private MilvusClientV2 client;
+    private MilvusServiceClient milvusClient;
 
-    public MilvusVectorStoreStrategy(ConfigService configService) {
+    public MilvusVectorStoreStrategy(VectorStoreProperties vectorStoreProperties) {
-        super(configService);
+        super(vectorStoreProperties);
     }
 
     @Override
@@ -54,92 +47,109 @@ public class MilvusVectorStoreStrategy extends AbstractVectorStoreStrategy {
 
     @Override
     public void createSchema(String vectorModelName, String kid, String modelName) {
-        log.info("Milvus创建schema: vectorModelName={}, kid={}, modelName={}", vectorModelName, kid, modelName);
+        String url = vectorStoreProperties.getMilvus().getUrl();
+        String collectionName = vectorStoreProperties.getMilvus().getCollectionname() + kid;
         
-        // 1. 获取Milvus配置
+        // 创建Milvus客户端连接
-        String host = configService.getConfigValue("milvus", "url");
+        ConnectParam connectParam = ConnectParam.newBuilder()
-        String collectionName = configService.getConfigValue("milvus", "collectionname") + kid;
+                .withUri(url)
-        
-        ConnectConfig config = ConnectConfig.builder()
-                .uri(host)
                 .build();
-        client = new MilvusClientV2(config);
+        milvusClient = new MilvusServiceClient(connectParam);
 
-        // 2. 检查集合是否存在
+        // 检查集合是否存在
-        HasCollectionReq hasCollectionReq = HasCollectionReq.builder()
+        HasCollectionParam hasCollectionParam = HasCollectionParam.newBuilder()
-                .collectionName(collectionName)
+                .withCollectionName(collectionName)
                 .build();
         
-        Boolean hasCollection = client.hasCollection(hasCollectionReq);
+        R<Boolean> hasCollectionResponse = milvusClient.hasCollection(hasCollectionParam);
+        if (hasCollectionResponse.getStatus() != R.Status.Success.getCode()) {
+            log.error("检查集合是否存在失败: {}", hasCollectionResponse.getMessage());
+            return;
+        }
         
-        if (!hasCollection) {
+        if (!hasCollectionResponse.getData()) {
-            // 3. 创建集合schema
+            // 创建字段
-            CreateCollectionReq.CollectionSchema schema = CreateCollectionReq.CollectionSchema.builder()
+            List<FieldType> fields = new ArrayList<>();
+            
+            // ID字段 (主键)
+            fields.add(FieldType.newBuilder()
+                    .withName("id")
+                    .withDataType(DataType.Int64)
+                    .withPrimaryKey(true)
+                    .withAutoID(true)
+                    .build());
+            
+            // 文本字段
+            fields.add(FieldType.newBuilder()
+                    .withName("text")
+                    .withDataType(DataType.VarChar)
+                    .withMaxLength(65535)
+                    .build());
+            
+            // fid字段
+            fields.add(FieldType.newBuilder()
+                    .withName("fid")
+                    .withDataType(DataType.VarChar)
+                    .withMaxLength(255)
+                    .build());
+            
+            // kid字段
+            fields.add(FieldType.newBuilder()
+                    .withName("kid")
+                    .withDataType(DataType.VarChar)
+                    .withMaxLength(255)
+                    .build());
+            
+            // docId字段
+            fields.add(FieldType.newBuilder()
+                    .withName("docId")
+                    .withDataType(DataType.VarChar)
+                    .withMaxLength(255)
+                    .build());
+            
+            // 向量字段
+            fields.add(FieldType.newBuilder()
+                    .withName("vector")
+                    .withDataType(DataType.FloatVector)
+                    .withDimension(1024) // 根据实际embedding维度调整
+                    .build());
+
+            // 创建集合
+            CreateCollectionParam createCollectionParam = CreateCollectionParam.newBuilder()
+                    .withCollectionName(collectionName)
+                    .withDescription("Knowledge base collection for " + kid)
+                    .withShardsNum(2)
+                    .withFieldTypes(fields)
                     .build();
 
-            // 添加字段定义
+            R<RpcStatus> createCollectionResponse = milvusClient.createCollection(createCollectionParam);
-            schema.addField(AddFieldReq.builder()
+            if (createCollectionResponse.getStatus() != R.Status.Success.getCode()) {
-                    .fieldName("id")
+                log.error("创建集合失败: {}", createCollectionResponse.getMessage());
-                    .dataType(DataType.Int64)
+                return;
-                    .isPrimaryKey(true)
+            }
-                    .autoID(true)
-                    .build());
 
-            schema.addField(AddFieldReq.builder()
+            // 创建索引
-                    .fieldName("text")
+            CreateIndexParam createIndexParam = CreateIndexParam.newBuilder()
-                    .dataType(DataType.VarChar)
+                    .withCollectionName(collectionName)
-                    .maxLength(65535)
+                    .withFieldName("vector")
-                    .build());
+                    .withIndexType(IndexType.IVF_FLAT)
-
+                    .withMetricType(MetricType.L2)
-            schema.addField(AddFieldReq.builder()
+                    .withExtraParam("{\"nlist\":1024}")
-                    .fieldName("fid")
-                    .dataType(DataType.VarChar)
-                    .maxLength(255)
-                    .build());
-
-            schema.addField(AddFieldReq.builder()
-                    .fieldName("kid")
-                    .dataType(DataType.VarChar)
-                    .maxLength(255)
-                    .build());
-
-            schema.addField(AddFieldReq.builder()
-                    .fieldName("docId")
-                    .dataType(DataType.VarChar)
-                    .maxLength(255)
-                    .build());
-
-            schema.addField(AddFieldReq.builder()
-                    .fieldName("vector")
-                    .dataType(DataType.FloatVector)
-                    .dimension(1024) // 根据实际embedding维度调整
-                    .build());
-
-            // 4. 创建索引参数
-            List<IndexParam> indexParams = new ArrayList<>();
-            indexParams.add(IndexParam.builder()
-                    .fieldName("vector")
-                    .indexType(IndexParam.IndexType.IVF_FLAT)
-                    .metricType(IndexParam.MetricType.L2)
-                    .extraParams(Map.of("nlist", 1024))
-                    .build());
-
-            // 5. 创建集合
-            CreateCollectionReq createCollectionReq = CreateCollectionReq.builder()
-                    .collectionName(collectionName)
-                    .collectionSchema(schema)
-                    .indexParams(indexParams)
                     .build();
 
-            client.createCollection(createCollectionReq);
+            R<RpcStatus> createIndexResponse = milvusClient.createIndex(createIndexParam);
-            log.info("Milvus集合创建成功: {}", collectionName);
+            if (createIndexResponse.getStatus() != R.Status.Success.getCode()) {
+                log.error("创建索引失败: {}", createIndexResponse.getMessage());
+            } else {
+                log.info("Milvus集合和索引创建成功: {}", collectionName);
+            }
         } else {
             log.info("Milvus集合已存在: {}", collectionName);
         }
     }
 
     @Override
-    public void storeEmbeddings(StoreEmbeddingBo storeEmbeddingBo) throws ServiceException {
+    public void storeEmbeddings(StoreEmbeddingBo storeEmbeddingBo) {
         createSchema(storeEmbeddingBo.getVectorModelName(), storeEmbeddingBo.getKid(), storeEmbeddingBo.getVectorModelName());
         
         EmbeddingModel embeddingModel = getEmbeddingModel(storeEmbeddingBo.getEmbeddingModelName(),
@@ -149,12 +159,13 @@ public class MilvusVectorStoreStrategy extends AbstractVectorStoreStrategy {
         List<String> fidList = storeEmbeddingBo.getFids();
         String kid = storeEmbeddingBo.getKid();
         String docId = storeEmbeddingBo.getDocId();
-        String collectionName = configService.getConfigValue("milvus", "collectionname") + kid;
+        String collectionName = vectorStoreProperties.getMilvus().getCollectionname() + kid;
         
         log.info("Milvus向量存储条数记录: " + chunkList.size());
         long startTime = System.currentTimeMillis();
         
         // 准备批量插入数据
+        List<InsertParam.Field> fields = new ArrayList<>();
         List<String> textList = new ArrayList<>();
         List<String> fidListData = new ArrayList<>();
         List<String> kidList = new ArrayList<>();
@@ -178,31 +189,25 @@ public class MilvusVectorStoreStrategy extends AbstractVectorStoreStrategy {
             vectorList.add(vector);
         }
         
-        // 构建插入数据
+        // 构建字段数据
-        List<JsonObject> data = new ArrayList<>();
+        fields.add(new InsertParam.Field("text", textList));
-        Gson gson = new Gson();
+        fields.add(new InsertParam.Field("fid", fidListData));
-        for (int i = 0; i < textList.size(); i++) {
+        fields.add(new InsertParam.Field("kid", kidList));
-            JsonObject row = new JsonObject();
+        fields.add(new InsertParam.Field("docId", docIdList));
-            row.addProperty("text", textList.get(i));
+        fields.add(new InsertParam.Field("vector", vectorList));
-            row.addProperty("fid", fidListData.get(i));
-            row.addProperty("kid", kidList.get(i));
-            row.addProperty("docId", docIdList.get(i));
-            row.add("vector", gson.toJsonTree(vectorList.get(i)));
-            data.add(row);
-        }
         
         // 执行插入
-        InsertReq insertReq = InsertReq.builder()
+        InsertParam insertParam = InsertParam.newBuilder()
-                .collectionName(collectionName)
+                .withCollectionName(collectionName)
-                .data(data)
+                .withFields(fields)
                 .build();
         
-        InsertResp insertResp = client.insert(insertReq);
+        R<MutationResult> insertResponse = milvusClient.insert(insertParam);
-        if (insertResp.getInsertCnt() > 0) {
+        if (insertResponse.getStatus() != R.Status.Success.getCode()) {
-            log.info("Milvus向量存储成功，插入条数: {}", insertResp.getInsertCnt());
+            log.error("Milvus向量存储失败: {}", insertResponse.getMessage());
-        } else {
-            log.error("Milvus向量存储失败");
             throw new ServiceException("Milvus向量存储失败");
+        } else {
+            log.info("Milvus向量存储成功，插入条数: {}", insertResponse.getData().getInsertCnt());
         }
         
         long endTime = System.currentTimeMillis();
@@ -217,99 +222,116 @@ public class MilvusVectorStoreStrategy extends AbstractVectorStoreStrategy {
                 queryVectorBo.getApiKey(), queryVectorBo.getBaseUrl());
         
         Embedding queryEmbedding = embeddingModel.embed(queryVectorBo.getQuery()).content();
-        String collectionName = configService.getConfigValue("milvus", "collectionname") + queryVectorBo.getKid();
+        String collectionName = vectorStoreProperties.getMilvus().getCollectionname() + queryVectorBo.getKid();
         
         List<String> resultList = new ArrayList<>();
         
-        // 准备查询向量
+        // 加载集合到内存
-        List<BaseVector> searchVectors = new ArrayList<>();
+        LoadCollectionParam loadCollectionParam = LoadCollectionParam.newBuilder()
-        float[] queryVectorArray = new float[queryEmbedding.vector().length];
+                .withCollectionName(collectionName)
-        for (int i = 0; i < queryEmbedding.vector().length; i++) {
+                .build();
-            queryVectorArray[i] = queryEmbedding.vector()[i];
+        milvusClient.loadCollection(loadCollectionParam);
-        }
-        searchVectors.add(new FloatVec(queryVectorArray));
         
-        // 构建搜索请求
+        // 准备查询向量
-        SearchReq searchReq = SearchReq.builder()
+        List<List<Float>> searchVectors = new ArrayList<>();
-                .collectionName(collectionName)
+        List<Float> queryVector = new ArrayList<>();
-                .data(searchVectors)
+        for (float f : queryEmbedding.vector()) {
-                .topK(queryVectorBo.getMaxResults())
+            queryVector.add(f);
-                .outputFields(Arrays.asList("text", "fid", "kid", "docId"))
+        }
+        searchVectors.add(queryVector);
+        
+        // 构建搜索参数
+        SearchParam searchParam = SearchParam.newBuilder()
+                .withCollectionName(collectionName)
+                .withMetricType(MetricType.L2)
+                .withOutFields(Arrays.asList("text", "fid", "kid", "docId"))
+                .withTopK(queryVectorBo.getMaxResults())
+                .withVectors(searchVectors)
+                .withVectorFieldName("vector")
+                .withParams("{\"nprobe\":10}")
                 .build();
         
-        SearchResp searchResp = client.search(searchReq);
+        R<SearchResults> searchResponse = milvusClient.search(searchParam);
-        if (searchResp != null && searchResp.getSearchResults() != null) {
+        if (searchResponse.getStatus() != R.Status.Success.getCode()) {
-            List<List<SearchResp.SearchResult>> searchResults = searchResp.getSearchResults();
+            log.error("Milvus查询失败: {}", searchResponse.getMessage());
+            return resultList;
+        }
         
-            for (List<SearchResp.SearchResult> results : searchResults) {
+        SearchResultsWrapper wrapper = new SearchResultsWrapper(searchResponse.getData().getResults());
-                for (SearchResp.SearchResult result : results) {
+        
-                    Map<String, Object> entity = result.getEntity();
+        // 遍历搜索结果
-                    String text = (String) entity.get("text");
+        for (int i = 0; i < wrapper.getIDScore(0).size(); i++) {
-                    if (text != null) {
+            SearchResultsWrapper.IDScore idScore = wrapper.getIDScore(0).get(i);
-                        resultList.add(text);
+            
+            // 获取text字段数据
+            List<?> textFieldData = wrapper.getFieldData("text", 0);
+            if (textFieldData != null && i < textFieldData.size()) {
+                Object textObj = textFieldData.get(i);
+                if (textObj != null) {
+                    resultList.add(textObj.toString());
+                    log.debug("找到相似文本，ID: {}, 距离: {}, 内容: {}", 
+                            idScore.getLongID(), idScore.getScore(), textObj.toString());
                 }
             }
         }
-        } else {
-            log.error("Milvus查询失败或无结果");
-        }
         
         return resultList;
     }
 
     @Override
-    public void removeById(String id, String modelName) throws ServiceException {
+    @SneakyThrows
-        String collectionName = configService.getConfigValue("milvus", "collectionname") + id;
+    public void removeById(String id, String modelName) {
+        String collectionName = vectorStoreProperties.getMilvus().getCollectionname() + id;
         
         // 删除整个集合
-        DropCollectionReq dropCollectionReq = DropCollectionReq.builder()
+        DropCollectionParam dropCollectionParam = DropCollectionParam.newBuilder()
-                .collectionName(collectionName)
+                .withCollectionName(collectionName)
                 .build();
         
-        try {
+        R<RpcStatus> dropResponse = milvusClient.dropCollection(dropCollectionParam);
-            client.dropCollection(dropCollectionReq);
+        if (dropResponse.getStatus() != R.Status.Success.getCode()) {
+            log.error("Milvus集合删除失败: {}", dropResponse.getMessage());
+            throw new ServiceException("Milvus集合删除失败");
+        } else {
             log.info("Milvus集合删除成功: {}", collectionName);
-        } catch (Exception e) {
-            log.error("Milvus集合删除失败: {}", e.getMessage());
-            throw new ServiceException("Milvus集合删除失败: " + e.getMessage());
         }
     }
 
     @Override
-    public void removeByDocId(String docId, String kid) throws ServiceException {
+    public void removeByDocId(String docId, String kid) {
-        String collectionName = configService.getConfigValue("milvus", "collectionname") + kid;
+        String collectionName = vectorStoreProperties.getMilvus().getCollectionname() + kid;
         
         String expr = "docId == \"" + docId + "\"";
-        DeleteReq deleteReq = DeleteReq.builder()
+        DeleteParam deleteParam = DeleteParam.newBuilder()
-                .collectionName(collectionName)
+                .withCollectionName(collectionName)
-                .filter(expr)
+                .withExpr(expr)
                 .build();
         
-        try {
+        R<MutationResult> deleteResponse = milvusClient.delete(deleteParam);
-            DeleteResp deleteResp = client.delete(deleteReq);
+        if (deleteResponse.getStatus() != R.Status.Success.getCode()) {
-            log.info("Milvus成功删除 docId={} 的所有向量数据，删除条数: {}", docId, deleteResp.getDeleteCnt());
+            log.error("Milvus删除失败: {}", deleteResponse.getMessage());
-        } catch (Exception e) {
+            throw new ServiceException("Milvus删除失败");
-            log.error("Milvus删除失败: {}", e.getMessage());
+        } else {
-            throw new ServiceException(e.getMessage());
+            log.info("Milvus成功删除 docId={} 的所有向量数据，删除条数: {}", docId, deleteResponse.getData().getDeleteCnt());
         }
     }
 
     @Override
-    public void removeByFid(String fid, String kid) throws ServiceException {
+    public void removeByFid(String fid, String kid) {
-        String collectionName = configService.getConfigValue("milvus", "collectionname") + kid;
+        String collectionName = vectorStoreProperties.getMilvus().getCollectionname() + kid;
         
         String expr = "fid == \"" + fid + "\"";
-        DeleteReq deleteReq = DeleteReq.builder()
+        DeleteParam deleteParam = DeleteParam.newBuilder()
-                .collectionName(collectionName)
+                .withCollectionName(collectionName)
-                .filter(expr)
+                .withExpr(expr)
                 .build();
         
-        try {
+        R<MutationResult> deleteResponse = milvusClient.delete(deleteParam);
-            DeleteResp deleteResp = client.delete(deleteReq);
+        if (deleteResponse.getStatus() != R.Status.Success.getCode()) {
-            log.info("Milvus成功删除 fid={} 的所有向量数据，删除条数: {}", fid, deleteResp.getDeleteCnt());
+            log.error("Milvus删除失败: {}", deleteResponse.getMessage());
-        } catch (Exception e) {
+            throw new ServiceException("Milvus删除失败");
-            log.error("Milvus删除失败: {}", e.getMessage());
+        } else {
-            throw new ServiceException(e.getMessage());
+            log.info("Milvus成功删除 fid={} 的所有向量数据，删除条数: {}", fid, deleteResponse.getData().getDeleteCnt());
         }
     }
 }

ruoyi-modules-api/ruoyi-knowledge-api/src/main/java/org/ruoyi/service/strategy/impl/WeaviateVectorStoreStrategy.java

@@ -1,7 +1,7 @@
 package org.ruoyi.service.strategy.impl;
 
 import cn.hutool.json.JSONObject;
-import com.google.protobuf.ServiceException;
+import org.ruoyi.common.core.exception.ServiceException;
 import dev.langchain4j.data.embedding.Embedding;
 import dev.langchain4j.model.embedding.EmbeddingModel;
 import io.weaviate.client.Config;
@@ -17,7 +17,7 @@ import io.weaviate.client.v1.schema.model.Schema;
 import io.weaviate.client.v1.schema.model.WeaviateClass;
 import lombok.SneakyThrows;
 import lombok.extern.slf4j.Slf4j;
-import org.ruoyi.common.core.service.ConfigService;
+import org.ruoyi.common.core.config.VectorStoreProperties;
 import org.ruoyi.domain.bo.QueryVectorBo;
 import org.ruoyi.domain.bo.StoreEmbeddingBo;
 import org.ruoyi.service.strategy.AbstractVectorStoreStrategy;
@@ -35,8 +35,8 @@ public class WeaviateVectorStoreStrategy extends AbstractVectorStoreStrategy {
 
     private WeaviateClient client;
 
-    public WeaviateVectorStoreStrategy(ConfigService configService) {
+    public WeaviateVectorStoreStrategy(VectorStoreProperties vectorStoreProperties) {
-        super(configService);
+        super(vectorStoreProperties);
     }
 
     @Override
@@ -46,9 +46,9 @@ public class WeaviateVectorStoreStrategy extends AbstractVectorStoreStrategy {
 
     @Override
     public void createSchema(String vectorModelName, String kid, String modelName) {
-        String protocol = configService.getConfigValue("weaviate", "protocol");
+        String protocol = vectorStoreProperties.getWeaviate().getProtocol();
-        String host = configService.getConfigValue("weaviate", "host");
+        String host = vectorStoreProperties.getWeaviate().getHost();
-        String className = configService.getConfigValue("weaviate", "classname") + kid;
+        String className = vectorStoreProperties.getWeaviate().getClassname() + kid;
         // 创建 Weaviate 客户端
         client = new WeaviateClient(new Config(protocol, host));
         // 检查类是否存在，如果不存在就创建 schema
@@ -128,7 +128,7 @@ public class WeaviateVectorStoreStrategy extends AbstractVectorStoreStrategy {
             vectorStrings.add(String.valueOf(v));
         }
         String vectorStr = String.join(",", vectorStrings);
-        String className = configService.getConfigValue("weaviate", "classname");
+        String className = vectorStoreProperties.getWeaviate().getClassname();
         
         // 构建 GraphQL 查询
         String graphQLQuery = String.format(
@@ -176,9 +176,9 @@ public class WeaviateVectorStoreStrategy extends AbstractVectorStoreStrategy {
     @Override
     @SneakyThrows
     public void removeById(String id, String modelName) {
-        String protocol = configService.getConfigValue("weaviate", "protocol");
+        String protocol = vectorStoreProperties.getWeaviate().getProtocol();
-        String host = configService.getConfigValue("weaviate", "host");
+        String host = vectorStoreProperties.getWeaviate().getHost();
-        String className = configService.getConfigValue("weaviate", "classname");
+        String className = vectorStoreProperties.getWeaviate().getClassname();
         String finalClassName = className + id;
         WeaviateClient client = new WeaviateClient(new Config(protocol, host));
         Result<Boolean> result = client.schema().classDeleter().withClassName(finalClassName).run();
@@ -192,7 +192,7 @@ public class WeaviateVectorStoreStrategy extends AbstractVectorStoreStrategy {
 
     @Override
     public void removeByDocId(String docId, String kid) {
-        String className = configService.getConfigValue("weaviate", "classname") + kid;
+        String className = vectorStoreProperties.getWeaviate().getClassname() + kid;
         // 构建 Where 条件
         WhereFilter whereFilter = WhereFilter.builder()
                 .path("docId")
@@ -212,7 +212,7 @@ public class WeaviateVectorStoreStrategy extends AbstractVectorStoreStrategy {
 
     @Override
     public void removeByFid(String fid, String kid) {
-        String className = configService.getConfigValue("weaviate", "classname") + kid;
+        String className = vectorStoreProperties.getWeaviate().getClassname() + kid;
         // 构建 Where 条件
         WhereFilter whereFilter = WhereFilter.builder()
                 .path("fid")