当前位置： 首页 > news >正文

SpringAI从入门到精通 (2)

news 2025/10/13 7:30:42

“不积跬步，无以至千里。”

Spring AI 进阶功能实现

章节二：对话上下文管理

在实际应用中，我们需要保持多轮对话的上下文，让 AI 能够"记住"之前的对话内容。

1. 创建会话管理器

创建 ConversationMemory.java：

package com.example.springaiproject.service;import org.springframework.ai.chat.messages.Message;
import org.springframework.ai.chat.messages.UserMessage;
import org.springframework.ai.chat.messages.AssistantMessage;
import org.springframework.stereotype.Component;import java.util.*;
import java.util.concurrent.ConcurrentHashMap;/*** 会话记忆管理器*/
@Component
public class ConversationMemory {// 使用 Map 存储每个会话的消息历史private final Map<String, List<Message>> conversationHistory = new ConcurrentHashMap<>();// 最大保留消息数量private static final int MAX_MESSAGES = 10;/*** 添加用户消息*/public void addUserMessage(String sessionId, String content) {List<Message> messages = conversationHistory.computeIfAbsent(sessionId, k -> new ArrayList<>());messages.add(new UserMessage(content));trimMessages(sessionId);}/*** 添加 AI 回复消息*/public void addAssistantMessage(String sessionId, String content) {List<Message> messages = conversationHistory.get(sessionId);if (messages != null) {messages.add(new AssistantMessage(content));trimMessages(sessionId);}}/*** 获取会话历史*/public List<Message> getHistory(String sessionId) {return conversationHistory.getOrDefault(sessionId, new ArrayList<>());}/*** 清空会话历史*/public void clearHistory(String sessionId) {conversationHistory.remove(sessionId);}/*** 限制消息数量，避免上下文过长*/private void trimMessages(String sessionId) {List<Message> messages = conversationHistory.get(sessionId);if (messages != null && messages.size() > MAX_MESSAGES) {conversationHistory.put(sessionId, new ArrayList<>(messages.subList(messages.size() - MAX_MESSAGES, messages.size())));}}
}

2. 更新 AIService

修改 AIService.java，添加上下文管理功能：

package com.example.springaiproject.service;import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.messages.Message;
import org.springframework.stereotype.Service;import java.util.List;@Service
public class AIService {private final ChatClient chatClient;private final ConversationMemory conversationMemory;public AIService(ChatClient.Builder chatClientBuilder, ConversationMemory conversationMemory) {this.chatClient = chatClientBuilder.build();this.conversationMemory = conversationMemory;}/*** 无上下文的简单对话*/public String generateText(String prompt) {return this.chatClient.prompt().user(prompt).call().content();}/*** 带上下文的对话*/public String chat(String sessionId, String userMessage) {// 添加用户消息到历史conversationMemory.addUserMessage(sessionId, userMessage);// 获取历史消息List<Message> history = conversationMemory.getHistory(sessionId);// 发送请求（包含历史上下文）String response = this.chatClient.prompt().messages(history).call().content();// 保存 AI 回复到历史conversationMemory.addAssistantMessage(sessionId, response);return response;}/*** 清空对话历史*/public void clearHistory(String sessionId) {conversationMemory.clearHistory(sessionId);}
}

3. 更新 Controller

在 AIController.java 中添加新的端点：

package com.example.springaiproject.controller;import com.example.springaiproject.service.AIService;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.bind.annotation.*;import java.util.Map;@RestController
@RequestMapping("/api/ai")
public class AIController {private final AIService aiService;@Autowiredpublic AIController(AIService aiService) {this.aiService = aiService;}@PostMapping("/generate")public String generateText(@RequestBody String prompt) {return aiService.generateText(prompt);}/*** 带会话的聊天接口*/@PostMapping("/chat")public Map<String, String> chat(@RequestParam(defaultValue = "default") String sessionId,@RequestBody Map<String, String> request) {String message = request.get("message");String response = aiService.chat(sessionId, message);return Map.of("response", response, "sessionId", sessionId);}/*** 清空会话历史*/@DeleteMapping("/chat/{sessionId}")public Map<String, String> clearHistory(@PathVariable String sessionId) {aiService.clearHistory(sessionId);return Map.of("message", "History cleared for session: " + sessionId);}
}

4. 测试对话上下文

# 第一轮对话
curl -X POST "http://localhost:8080/api/ai/chat?sessionId=user123" \-H "Content-Type: application/json" \-d '{"message": "我的名字是张三"}'# 第二轮对话（AI 会记住你的名字）
curl -X POST "http://localhost:8080/api/ai/chat?sessionId=user123" \-H "Content-Type: application/json" \-d '{"message": "你还记得我的名字吗？"}'# 清空历史
curl -X DELETE "http://localhost:8080/api/ai/chat/user123"

章节三：检索增强生成（RAG）

RAG 允许 AI 从外部知识库中检索相关信息，生成更准确的答案。

1. 添加依赖

在 pom.xml 中添加向量存储和文档处理依赖：

<dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-pdf-document-reader</artifactId>
</dependency>
<dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-pgvector-store-spring-boot-starter</artifactId>
</dependency>
<!-- 或使用简单的内存向量存储 -->
<dependency><groupId>org.springframework.ai</groupId><artifactId>spring-ai-transformers-spring-boot-starter</artifactId>
</dependency>

2. 配置向量存储

更新 application.yml：

spring:ai:vectorstore:simple:enabled: trueembedding:transformer:enabled: true

3. 创建文档加载器

创建 DocumentLoader.java：

package com.example.springaiproject.service;import org.springframework.ai.document.Document;
import org.springframework.ai.reader.TextReader;
import org.springframework.ai.transformer.splitter.TokenTextSplitter;
import org.springframework.ai.vectorstore.VectorStore;
import org.springframework.core.io.Resource;
import org.springframework.stereotype.Service;import java.util.List;/*** 文档加载和向量化服务*/
@Service
public class DocumentLoader {private final VectorStore vectorStore;public DocumentLoader(VectorStore vectorStore) {this.vectorStore = vectorStore;}/*** 加载文档到向量存储*/public void loadDocument(Resource resource) {// 读取文档TextReader textReader = new TextReader(resource);List<Document> documents = textReader.get();// 分割文档为小块TokenTextSplitter textSplitter = new TokenTextSplitter();List<Document> splitDocuments = textSplitter.apply(documents);// 存储到向量数据库vectorStore.add(splitDocuments);}/*** 从文本加载*/public void loadText(String content, Map<String, Object> metadata) {Document document = new Document(content, metadata);TokenTextSplitter textSplitter = new TokenTextSplitter();List<Document> splitDocuments = textSplitter.apply(List.of(document));vectorStore.add(splitDocuments);}
}

4. 创建 RAG 服务

创建 RAGService.java：

package com.example.springaiproject.service;import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.client.advisor.QuestionAnswerAdvisor;
import org.springframework.ai.vectorstore.SearchRequest;
import org.springframework.ai.vectorstore.VectorStore;
import org.springframework.stereotype.Service;/*** RAG 检索增强生成服务*/
@Service
public class RAGService {private final ChatClient chatClient;private final VectorStore vectorStore;public RAGService(ChatClient.Builder chatClientBuilder, VectorStore vectorStore) {this.vectorStore = vectorStore;this.chatClient = chatClientBuilder.defaultAdvisors(new QuestionAnswerAdvisor(vectorStore, SearchRequest.defaults())).build();}/*** 基于知识库回答问题*/public String askWithContext(String question) {return this.chatClient.prompt().user(question).call().content();}/*** 手动检索相关文档并回答*/public String askWithManualRetrieval(String question, int topK) {// 检索相关文档var similarDocuments = vectorStore.similaritySearch(SearchRequest.query(question).withTopK(topK));// 构建上下文StringBuilder context = new StringBuilder("参考以下信息回答问题：\n\n");similarDocuments.forEach(doc -> {context.append(doc.getContent()).append("\n\n");});context.append("问题：").append(question);// 生成回答return this.chatClient.prompt().user(context.toString()).call().content();}
}

5. 添加 RAG Controller

package com.example.springaiproject.controller;import com.example.springaiproject.service.DocumentLoader;
import com.example.springaiproject.service.RAGService;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.bind.annotation.*;
import org.springframework.web.multipart.MultipartFile;import java.util.Map;@RestController
@RequestMapping("/api/rag")
public class RAGController {private final RAGService ragService;private final DocumentLoader documentLoader;@Autowiredpublic RAGController(RAGService ragService, DocumentLoader documentLoader) {this.ragService = ragService;this.documentLoader = documentLoader;}/*** 上传知识文档*/@PostMapping("/upload")public Map<String, String> uploadDocument(@RequestBody Map<String, String> request) {String content = request.get("content");documentLoader.loadText(content, Map.of("source", "user-upload"));return Map.of("message", "Document uploaded successfully");}/*** 基于知识库问答*/@PostMapping("/ask")public Map<String, String> ask(@RequestBody Map<String, String> request) {String question = request.get("question");String answer = ragService.askWithContext(question);return Map.of("answer", answer);}
}

6. 测试 RAG

# 1. 上传知识文档
curl -X POST "http://localhost:8080/api/rag/upload" \-H "Content-Type: application/json" \-d '{"content": "Spring AI 是 Spring 框架的 AI 集成模块，版本 1.0.1 支持 OpenAI、Azure OpenAI 等多种模型提供商。"}'# 2. 基于知识库提问
curl -X POST "http://localhost:8080/api/rag/ask" \-H "Content-Type: application/json" \-d '{"question": "Spring AI 支持哪些模型提供商？"}'

章节四：多模态输入输出

Spring AI 支持处理图像、音频等多种数据类型。

1. 添加多模态依赖

确保使用支持多模态的模型（如 GPT-4 Vision）：

spring:ai:openai:chat:options:model: gpt-4-vision-preview  # 支持图像输入

2. 创建多模态服务

创建 MultiModalService.java：

package com.example.springaiproject.service;import org.springframework.ai.chat.client.ChatClient;
import org.springframework.ai.chat.messages.Media;
import org.springframework.ai.chat.messages.UserMessage;
import org.springframework.ai.model.Media.MimeType;
import org.springframework.stereotype.Service;
import org.springframework.util.MimeTypeUtils;import java.io.IOException;
import java.net.URL;
import java.util.List;/*** 多模态 AI 服务*/
@Service
public class MultiModalService {private final ChatClient chatClient;public MultiModalService(ChatClient.Builder chatClientBuilder) {this.chatClient = chatClientBuilder.build();}/*** 分析图像*/public String analyzeImage(String imageUrl, String prompt) {try {var imageMedia = new Media(MimeTypeUtils.IMAGE_PNG, new URL(imageUrl));var userMessage = new UserMessage(prompt,List.of(imageMedia));return this.chatClient.prompt().messages(userMessage).call().content();} catch (Exception e) {return "Error analyzing image: " + e.getMessage();}}/*** 分析本地图像（Base64）*/public String analyzeImageBase64(String base64Image, String prompt) {try {// 去除 data:image/png;base64, 前缀（如果有）String imageData = base64Image.replaceFirst("^data:image/[^;]+;base64,", "");byte[] imageBytes = java.util.Base64.getDecoder().decode(imageData);var imageResource = new org.springframework.core.io.ByteArrayResource(imageBytes);var imageMedia = new Media(MimeTypeUtils.IMAGE_PNG, imageResource);var userMessage = new UserMessage(prompt,List.of(imageMedia));return this.chatClient.prompt().messages(userMessage).call().content();} catch (Exception e) {return "Error analyzing image: " + e.getMessage();}}/*** 批量分析多张图像*/public String analyzeMultipleImages(List<String> imageUrls, String prompt) {try {List<Media> mediaList = imageUrls.stream().map(url -> {try {return new Media(MimeTypeUtils.IMAGE_PNG, new URL(url));} catch (Exception e) {throw new RuntimeException(e);}}).toList();var userMessage = new UserMessage(prompt, mediaList);return this.chatClient.prompt().messages(userMessage).call().content();} catch (Exception e) {return "Error analyzing images: " + e.getMessage();}}
}

3. 创建多模态 Controller

package com.example.springaiproject.controller;import com.example.springaiproject.service.MultiModalService;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.bind.annotation.*;
import org.springframework.web.multipart.MultipartFile;import java.util.Base64;
import java.util.List;
import java.util.Map;@RestController
@RequestMapping("/api/multimodal")
public class MultiModalController {private final MultiModalService multiModalService;@Autowiredpublic MultiModalController(MultiModalService multiModalService) {this.multiModalService = multiModalService;}/*** 分析图像 URL*/@PostMapping("/analyze-image")public Map<String, String> analyzeImage(@RequestBody Map<String, String> request) {String imageUrl = request.get("imageUrl");String prompt = request.getOrDefault("prompt", "请描述这张图片的内容");String result = multiModalService.analyzeImage(imageUrl, prompt);return Map.of("result", result);}/*** 上传并分析图像*/@PostMapping("/upload-image")public Map<String, String> uploadAndAnalyzeImage(@RequestParam("file") MultipartFile file,@RequestParam(defaultValue = "请描述这张图片的内容") String prompt) {try {byte[] bytes = file.getBytes();String base64 = Base64.getEncoder().encodeToString(bytes);String base64WithPrefix = "data:image/png;base64," + base64;String result = multiModalService.analyzeImageBase64(base64WithPrefix, prompt);return Map.of("result", result);} catch (Exception e) {return Map.of("error", e.getMessage());}}/*** 批量分析图像*/@PostMapping("/analyze-multiple")public Map<String, String> analyzeMultipleImages(@RequestBody Map<String, Object> request) {@SuppressWarnings("unchecked")List<String> imageUrls = (List<String>) request.get("imageUrls");String prompt = (String) request.getOrDefault("prompt", "比较这些图片的异同");String result = multiModalService.analyzeMultipleImages(imageUrls, prompt);return Map.of("result", result);}
}

4. 测试多模态功能

# 分析图像 URL
curl -X POST "http://localhost:8080/api/multimodal/analyze-image" \-H "Content-Type: application/json" \-d '{"imageUrl": "https://example.com/image.jpg","prompt": "这张图片里有什么？"}'# 上传图像文件
curl -X POST "http://localhost:8080/api/multimodal/upload-image" \-F "file=@/path/to/image.jpg" \-F "prompt=描述图片中的主要元素"# 批量分析
curl -X POST "http://localhost:8080/api/multimodal/analyze-multiple" \-H "Content-Type: application/json" \-d '{"imageUrls": ["https://example.com/image1.jpg","https://example.com/image2.jpg"],"prompt": "比较这两张图片"}'

完整项目结构

spring-ai-demo/
├── src/main/java/com/example/springaiproject/
│   ├── SpringAiDemoApplication.java
│   ├── controller/
│   │   ├── AIController.java
│   │   ├── RAGController.java
│   │   └── MultiModalController.java
│   └── service/
│       ├── AIService.java
│       ├── ConversationMemory.java
│       ├── DocumentLoader.java
│       ├── RAGService.java
│       └── MultiModalService.java
└── src/main/resources/└── application.yml

总结

通过这三个章节，我们实现了：

  1. 对话上下文管理：支持多轮对话，AI 能记住之前的交互内容
  2. RAG 检索增强生成：集成向量数据库，从知识库中检索相关信息
  3. 多模态处理：支持图像分析、多图对比等视觉任务

这些功能是构建企业级 AI 应用坚实的基础。

🔗 参考资源

  • Spring AI RAG 指南
  • Spring AI 多模态支持
  • 向量数据库对比
http://www.dtcms.com/a/473928.html

相关文章：

  • Linux 12mybash的实现
  • K8s YAML 文件详解：从语法到实战编写指南
  • 社区版Idea怎么创建Spring Boot项目？Selected Java version 17 is not supported. 问题解决
  • 益阳市 网站建设电子商务网站建设的主要风险
  • SpringBootRemotePowershellAdmin：开箱即用的 Windows远程运维开源工具
  • 插槽vue/react
  • 对vue生命周期的理解
  • 2017民非单位年检那个网站做黄山旅游攻略景点必去
  • [笔记 自用]CAN总线通信配置
  • HTML 教程
  • 用自己服务器做网站用备案怎样在亚马逊网上开店
  • PHP操作elasticsearch7.8
  • 学校网站建设需求分析哪个小说网站可以做封面
  • 网站制作类软件推荐莆田网站格在哪里做
  • TypeScript 面试题及详细答案 100题 （21-30）-- 接口（Interface）
  • 承德网站新手怎么做网络推广
  • 6. 从0到上线：.NET 8 + ML.NET LTR 智能类目匹配实战--渐进式学习闭环：从反馈到再训练
  • 2.c++面向对象(五)
  • python中的一些运算符
  • 【嵌入式面试题】boss收集的11道，持续更新中
  • 保证样式稿高度还原
  • 网站建设 源码怎么注册公司名
  • [xboard] 34 buildroot 的overlay机制
  • 某公司站点的挖掘实战分享
  • 第三方和审核场景回调还是主动查询
  • Git基本命令的使用（超详细）
  • NC40 链表相加(二)
  • 网安面试题收集（3）
  • JetLinks设备接入的认识与理解
  • 从HashMap到ConcurrentHashMap深入剖析Java并发容器的演进与实战