掘金 人工智能 9小时前
SpringAI(RAG+MCP)使用-未完结
index_new5.html
../../../zaker_core/zaker_tpl_static/wap/tpl_guoji1.html

 

本文详细介绍了如何使用Spring AI构建智能聊天机器人，重点阐述了集成OpenAI和DeepSeek模型、实现会话上下文记忆（包括JDBC持久化和自定义会话历史表）以及构建RAG（Retrieval-Augmented Generation）知识库的关键技术。文章提供了具体的依赖配置、Spring Bean的定义以及数据源和文档处理的示例，旨在帮助开发者快速上手Spring AI，打造功能更强大的AI应用。

✨ **模型集成与ChatClient配置**: Spring AI支持集成多种大语言模型，如OpenAI和DeepSeek。通过`ChatClient.builder()`可以方便地配置模型、系统提示（SystemPrompt）和拦截器（Advisors），实现与AI的灵活交互，并可为不同场景创建特定的ChatClient，如用于生成标题的`titleChatClient`。

🧠 **会话记忆持久化**: 文章详细介绍了两种会话记忆的实现方式：一是利用Spring AI自带的`JDBCChatMemoryRepository`，通过JDBC将对话记录持久化到`SPRING_AI_CHAT_MEMORY`表中，以`conversation_id`关联同一会话；二是创建自定义的`d_chat_type_history`表，用于存储不同用户或会话的额外信息，如标题和状态，并结合MyBatis-Plus进行管理，实现更精细化的会话追踪。

🛡️ **Advisors实现交互增强**: Spring AI的Advisors机制为聊天机器人提供了强大的扩展能力。文中展示了如何通过`SimpleLoggerAdvisor`进行日志记录，`ChatTypeHistoryAdvisor`记录会话ID到自定义表中，`ChatTypeTitleAdvisor`调用另一个ChatClient为会话生成标题，以及`MessageChatMemoryAdvisor`将对话保存到Spring AI的内存库中，这些Advisors通过责任链模式协同工作，增强了聊天机器人的功能。

📚 **RAG知识库构建**: 实现RAG（Retrieval-Augmented Generation）的关键在于向量数据库的接入和文档处理。文章以Pgvector为例，介绍了如何配置`PgVectorStore`，将其与PostgreSQL数据库和Embedding Model结合，用于存储和检索向量化的文档。同时，通过`MarkdownDocumentReader`等插件，可以将各种格式的文档转换为Spring AI的`Document`对象，为知识库的构建奠定基础。

🗄️ **多数据源管理**: 当项目同时使用MySQL和PostgreSQL时，为了避免`JdbcTemplate`的冲突，需要通过`@Configuration`和`@Bean`注解自定义数据源配置。文章提供了`DataSourceConfiguration`类，明确指定了MySQL为主数据源（`@Primary`），并为PostgreSQL创建了单独的数据源和`JdbcTemplate`实例，确保了多数据源的正确使用。

依赖

<properties>  <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>  <java.version>17</java.version>  <spring.boot.version>3.5.0</spring.boot.version>  <spring-ai.version>1.0.0</spring-ai.version>  <lombok.version>1.18.20</lombok.version>  <mybatis-plus.version>3.5.7</mybatis-plus.version>  <fastjson.version>1.2.83</fastjson.version>  <hutool.version>5.8.25</hutool.version></properties><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-starter-model-openai</artifactId></dependency><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-starter-model-deepseek</artifactId></dependency><!--持久化插件--><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-starter-model-chat-memory-repository-jdbc</artifactId></dependency><!--rag相关--><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-rag</artifactId></dependency><!--向量化存储，这里用Postgresql--><dependency>  <groupId>org.postgresql</groupId>  <artifactId>postgresql</artifactId>  <version>42.7.5</version>  <scope>runtime</scope></dependency><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-pgvector-store</artifactId></dependency><!--    pg--><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-starter-vector-store-pgvector</artifactId></dependency><!-- JDBC--><dependency>  <groupId>org.springframework.boot</groupId>  <artifactId>spring-boot-starter-jdbc</artifactId></dependency><dependency>  <groupId>com.mysql</groupId>  <artifactId>mysql-connector-j</artifactId></dependency><dependency>  <groupId>org.postgresql</groupId>  <artifactId>postgresql</artifactId>  <version>42.7.5</version>  <scope>runtime</scope></dependency><dependency>  <groupId>com.baomidou</groupId>  <artifactId>mybatis-plus-spring-boot3-starter</artifactId>  <version>${mybatis-plus.version}</version></dependency><!--文档解析插件--><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-markdown-document-reader</artifactId></dependency><dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-pdf-document-reader</artifactId></dependency><!--其他--><dependency>  <groupId>cn.hutool</groupId>  <artifactId>hutool-all</artifactId>  <version>${hutool.version}</version></dependency><dependency>  <groupId>org.springframework.boot</groupId>  <artifactId>spring-boot-starter-log4j2</artifactId></dependency>

配置文件

spring:    datasource:    mysql:      jdbc-url: jdbc:mysql://localhost:3306/scai?serverTimezone=Asia/Shanghai&useSSL=false&useUnicode=true&characterEncoding=utf-8&zeroDateTimeBehavior=convertToNull&transformedBitIsBoolean=true&tinyInt1isBit=false&allowPublicKeyRetrieval=true&allowMultiQueries=true&useServerPrepStmts=false      driver-class-name: com.mysql.cj.jdbc.Driver      username: root      password: root    postgres:      jdbc-url: jdbc:postgresql://localhost:5432/postgres      driver-class-name: org.postgresql.Driver  # 建议明确指定驱动类username: sats      password: 123456  ai:    vectorstore:      pgvector:#        datasource: postgresinde-type: HNSW        dimensions: 1024        distance-metric: COSINE_DISTANCE        max-document-batch-size: 10000    chat:      memory:        repository:        # jdbc持久方式，查找jdbcTemplate          jdbc:            initialize-schema: ALWAYS            platform: mariadb    deepseek:      api-key: ni DeepSeek Key      base-url: https://api.deepseek.com      chat:        options:          model: deepseek-chat    openai:      base-url: https://dashscope.aliyuncs.com/compatible-mode      api-key: 你的openaikey或者支持openai规范的也行，吧URL和model换了      chat:        options:          model: qwen-max-latest      embedding:        options:          model: text-embedding-v3          dimensions: 1024

数据源配置

 @Configuration@Slf4jpublic  class DataSourceConfiguration {    // 主数据源配置（MySQL） @Bean @Primary @ConfigurationProperties("spring.datasource.mysql")    public DataSource mysqlDataSource() {        return DataSourceBuilder.create().build();    }    // PostgreSQL数据源配置 @Bean @ConfigurationProperties(prefix = "spring.datasource.postgres")    public DataSource postgresqlDataSource() {        log.info("创建。。。。。。。 postgresql <UNK>");        return DataSourceBuilder.create().build();    }    @Bean @Primary  // mysql----> 直接当作name为jdbcTemplate public JdbcTemplate mysqlJdbcTemplate( @Qualifier("mysqlDataSource") DataSource dataSource) {        return  new JdbcTemplate(dataSource);    }    @Bean public JdbcTemplate postgresqlJdbcTemplate( @Qualifier("postgresqlDataSource") DataSource dataSource) {        log.info("创建。。。。。。。");        return  new JdbcTemplate(dataSource);    }}

持久化目标

    Spring AI自带持久化。记录会话内容

<dependency>  <groupId>org.springframework.ai</groupId>  <artifactId>spring-ai-starter-model-chat-memory-repository-jdbc</artifactId></dependency>

docs.spring.io/spring-ai/r…

 @Beanpublic ChatMemory chatMemory(ChatMemoryRepository chatMemoryRepository) {    return MessageWindowChatMemory.builder()            .chatMemoryRepository(chatMemoryRepository)            .maxMessages(20)            .build();}
-- auto-generated definitioncreate  table SPRING_AI_CHAT_MEMORY(    conversation_id varchar(36)                         not  null,    content         mediumtext not  null,    type varchar(10)                         not  null,    timestamp  timestamp  default  CURRENT_TIMESTAMP  not  null  on  update  CURRENT_TIMESTAMP);

    给不同用户、不同新会话建立关联

做到常见的网页端效果

在SPRING_AI_CHAT_MEMORY基础上，新建一个表

-- 建数据库create database if not  exists scai character  set utf8mb4;-- 创建表CREATE  TABLE `d_chat_type_history` (                                       `id` bigint NOT  NULL AUTO_INCREMENT COMMENT '主键id',                                       `type` int  NOT  NULL COMMENT '会话类型，详见ChatType枚举',                                       `chat_id` varchar(225) NOT  NULL COMMENT '会话id',                                       `title` varchar(512) DEFAULT  NULL COMMENT '标题',                                       `create_time` datetime DEFAULT  NULL COMMENT '创建时间',                                       `edit_time` datetime DEFAULT  NULL COMMENT '编辑时间',                                       `status` tinyint(1) DEFAULT '1' COMMENT '1:正常 0:删除',                                       PRIMARY KEY (`id`)) ENGINE=InnoDB AUTO_INCREMENT=1 DEFAULT CHARSET=utf8mb3 COMMENT='会话历史表';

图方便就没用userid了

模型AI交互的ChatClient

advisor就相当于和AI交互的拦截器

  @Bean public ChatClient chatClient ( @Qualifier("deepSeekChatModel") ChatModel deepSeekChatModel) { return ChatClient.builder(deepSeekChatModel).build();        }

配置一个助手ChatClient

助手目标：

实现步骤

 @Beanpublic ChatClient loveChatClient( @Qualifier("deepSeekChatModel") ChatModel deepSeekChatModel, ChatMemory chatMemory,                                 ChatTypeHistoryService chatTypeHistoryService, @Qualifier("titleChatClient")ChatClient titleChatClient) {    return ChatClient.builder(deepSeekChatModel)            .defaultSystem(SystemPrompt.SYSTEM_PROMPT) // 角色            .defaultAdvisors(                    new SimpleLoggerAdvisor(),                    ChatTypeHistoryAdvisor.builder(chatTypeHistoryService).type(ChatType.ASSISTANT.getCode())                            .order(998).build(),                    ChatTypeTitleAdvisor.builder(chatTypeHistoryService).type(ChatType.ASSISTANT.getCode())                            .chatClient(titleChatClient).chatMemory(chatMemory).order(999).build(),                    MessageChatMemoryAdvisor.builder(chatMemory).order(1000).build()            ) // 拦截器            .build();}
 @Bean public ChatClient titleChatClient( @Qualifier("deepSeekChatModel") ChatModel deepSeekChatModel) {        return ChatClient                .builder(deepSeekChatModel)                .defaultAdvisors(                        new SimpleLoggerAdvisor()                )                .build();    }

advisor的实现

docs.spring.io/spring-ai/r…

    实现CallAdvisor, StreamAdvisor接口, 重写方法
public ChatClientResponse adviseCall(ChatClientRequest chatClientRequest, CallAdvisorChain callAdvisorChain) {    this.logRequest(chatClientRequest);    ChatClientResponse chatClientResponse = callAdvisorChain.nextCall(chatClientRequest);    this.logResponse(chatClientResponse);    return chatClientResponse;}public Flux<ChatClientResponse> adviseStream(ChatClientRequest chatClientRequest, StreamAdvisorChain streamAdvisorChain) {    this.logRequest(chatClientRequest);    Flux<ChatClientResponse> chatClientResponses = streamAdvisorChain.nextStream(chatClientRequest);    return (new ChatClientMessageAggregator()).aggregateChatClientResponse(chatClientResponses, this::logResponse);}

2. 上面的ChatTypeHistoryAdvisor和ChatTypeTitleAdvisor，是基于BaseChatMemoryAdvisor实现的，BaseChatMemoryAdvisor继承BaseAdvisor继承了CallAdvisor, StreamAdvisor。

BaseChatMemoryAdvisor可以传递本次会话相关信息上下文，比如会话ID。定义了before，after方法适配链路

RAG向量知识库构建

Pgvector的接入: docs.spring.io/spring-ai/r…

VectorStore:docs.spring.io/spring-ai/r…

VectorStore

PgVectorStore的配置

 @Configurationpublic  class MyVectorStoreConfiguration {    @Bean public VectorStore pgVectorVectorStore( @Qualifier("postgresqlJdbcTemplate") JdbcTemplate postgresqlJdbcTemplate,                                           EmbeddingModel openAIEmbeddingModel) {        VectorStore vectorStore = PgVectorStore.builder(postgresqlJdbcTemplate, openAIEmbeddingModel)                .dimensions(1024)                    // Optional: defaults to model dimensions or 1024.distanceType(COSINE_DISTANCE)       // Optional: defaults to COSINE_DISTANCE.indexType(HNSW)                     // Optional: defaults to HNSW.initializeSchema(true)              // Optional: defaults to false.schemaName("public")                // Optional: defaults to "public".vectorTableName("vector_store")     // Optional: defaults to "vector_store".maxDocumentBatchSize(10000)         // Optional: defaults to 10000.build();        return vectorStore;    }}

向量数据库一旦创建，维度确定，更换维度需要重新创建。

文档处理

Spring AI提供了系列对应的文档处理插件，可以将知识处理成SpringAI 的 document

以处理markdown为例子

 @Component@Slf4jpublic  class MyDocumentLoader {    private  final ResourcePatternResolver resourcePatternResolver;    MyDocumentLoader(ResourcePatternResolver resourcePatternResolver) {        this.resourcePatternResolver = resourcePatternResolver;    }    public List<Document> loadMarkdowns() {        List<Document> allDocuments = new ArrayList<>();        try {            Resource[] resources = resourcePatternResolver.getResources("classpath:document/*.md");            for (Resource resource : resources) {                String fileName = resource.getFilename();                MarkdownDocumentReaderConfig config = MarkdownDocumentReaderConfig.builder()                        .withHorizontalRuleCreateDocument(true)                        .withIncludeCodeBlock(false)                        .withIncludeBlockquote(false)                        .withAdditionalMetadata("filename", fileName)                        .build();                MarkdownDocumentReader reader = new MarkdownDocumentReader(resource, config);                allDocuments.addAll(reader.get());            }        } catch (IOException e) {//            log.error("Markdown 文档加载失败", e);}        return allDocuments;    }}

待续

Fish AI Reader

Fish AI Reader

AI辅助创作，多种专业模板，深度分析，高质量内容生成。从观点提取到深度思考，FishAI为您提供全方位的创作支持。新版本引入自定义参数，让您的创作更加个性化和精准。

FishAI

FishAI

鱼阅，AI 时代的下一个智能信息助手，助你摆脱信息焦虑

联系邮箱 441953276@qq.com

相关标签

Spring AI 聊天机器人 RAG 大模型 Java
相关文章
Retrieval-Augmented Generation: A More Reliable Approach
Learn AI Together — Towards AI Community Newsletter #23
回顾我的 prompt 能力从小白到熟练的一些重要节点： 防杠叠甲： 1. 仅代表我自己的认知，没啥权威性。 2. 认为提示词那么简单至于搞那么复杂么的朋友，你对 3. ...
Databricks Announces Major Updates to Its AI Suite to Boost AI Model Accuracy
SAP大中华区总裁黄陈宏：SAP不做大模型
Are Vector DBs the Future Data Platform for AI? with Ed Anuff - #664
AI for Customer Service and Marketing at Aeromexico with Brian Gross - TWiML Talk #79
Building Conversational Application for Financial Services with Kenneth Conroy - TWiML Talk #61
李开复：现在对中国大模型创业公司盖棺定论为时尚早，ofo式的补贴逻辑不再适用于AI 2.0
寒武纪：新一代智能处理器微架构和指令集正在研发中