Spring Boot项目想加AI功能？手把手教你用Spring AI 1.0 M5快速集成GPT-4o

Spring Boot项目快速集成GPT-4o实战指南当AI能力成为企业应用的标配Java开发者如何在不重构现有架构的前提下为Spring Boot项目注入智能本文将带你从零开始通过Spring AI 1.0 M5实现GPT-4o的深度集成完成从环境配置到生产级部署的全流程实战。1. 环境准备与基础配置在开始集成前确保你的开发环境满足以下条件JDK 17或更高版本Spring Boot 3.2.xMaven 3.8或Gradle 8.0有效的AI服务API密钥如OpenAI关键依赖配置dependency groupIdorg.springframework.ai/groupId artifactIdspring-ai-openai-spring-boot-starter/artifactId version1.0.0-M5/version /dependency配置文件application.yml的基础设置spring: ai: openai: api-key: ${OPENAI_API_KEY} chat: model: gpt-4o temperature: 0.7 max-tokens: 2000注意建议通过环境变量管理API密钥避免硬编码在配置文件中2. 核心功能实现2.1 基础聊天功能集成创建ChatService实现类Service public class AIChatService { private final ChatClient chatClient; Autowired public AIChatService(ChatClient chatClient) { this.chatClient chatClient; } public String generateResponse(String prompt) { PromptTemplate promptTemplate new PromptTemplate( 你是一位专业的技术顾问请用简洁明了的方式回答 {question} ); MapString, Object model Map.of(question, prompt); return chatClient.call(promptTemplate.render(model)); } }测试用例示例SpringBootTest class AIChatServiceTest { Autowired private AIChatService chatService; Test void shouldReturnTechnicalAnswer() { String response chatService.generateResponse( 解释Spring Boot自动配置原理); assertNotNull(response); System.out.println(response); } }2.2 高级RAG实现构建知识库增强的问答系统需要以下步骤文档预处理管道Bean public DocumentReader pdfReader() { return new PdfDocumentReader(new ClassPathResource(knowledge.pdf)); } Bean public TextSplitter textSplitter() { return new TokenTextSplitter(1000, 200); } Bean public EmbeddingClient embeddingClient() { return new OpenAiEmbeddingClient(); }向量存储配置Bean public VectorStore vectorStore(EmbeddingClient embeddingClient) { return new SimpleVectorStore(embeddingClient); } Bean public CommandLineRunner initVectorStore( VectorStore vectorStore, DocumentReader reader, TextSplitter splitter) { return args - { ListDocument documents reader.read() .stream() .flatMap(doc - splitter.split(doc).stream()) .collect(Collectors.toList()); vectorStore.add(documents); }; }RAG服务实现Service public class RagService { private final ChatClient chatClient; private final VectorStore vectorStore; public String enhancedQuery(String question) { ListDocument relevantDocs vectorStore.similaritySearch( SearchRequest.query(question).withTopK(3)); String context relevantDocs.stream() .map(Document::getContent) .collect(Collectors.joining(\n\n)); PromptTemplate promptTemplate new PromptTemplate( 基于以下上下文回答问题 {context} 问题{question} ); return chatClient.call( promptTemplate.create( Map.of(context, context, question, question))); } }3. 生产环境优化3.1 性能调优策略配置连接池和超时设置spring: ai: openai: client: connect-timeout: 10s read-timeout: 30s max-in-memory-size: 10MB缓存策略实现示例Cacheable(value aiResponses, key #prompt.hashCode()) public String getCachedResponse(String prompt) { return chatClient.call(prompt); }3.2 监控与日志自定义健康检查指标Component public class AiHealthIndicator implements HealthIndicator { Autowired private ChatClient chatClient; Override public Health health() { try { String response chatClient.call(Ping); return Health.up() .withDetail(model, GPT-4o) .build(); } catch (Exception e) { return Health.down(e).build(); } } }结构化日志配置Aspect Component Slf4j public class AiLoggingAspect { Around(execution(* com..AIChatService.*(..))) public Object logAiCall(ProceedingJoinPoint joinPoint) throws Throwable { String method joinPoint.getSignature().getName(); Object[] args joinPoint.getArgs(); long start System.currentTimeMillis(); Object result joinPoint.proceed(); long duration System.currentTimeMillis() - start; log.info(AI调用统计 - 方法: {}, 耗时: {}ms, 输入: {}, 输出: {}, method, duration, args[0], result); return result; } }4. 安全与异常处理4.1 内容过滤机制实现敏感词过滤器Component public class ContentFilter { private static final SetString BLACKLIST Set.of( 敏感词1, 敏感词2, 敏感词3); public String filter(String content) { for (String word : BLACKLIST) { if (content.contains(word)) { throw new ContentViolationException( 检测到违规内容); } } return content; } }增强的Prompt安全处理public String safeGenerateResponse(String userInput) { String sanitizedInput HtmlUtils.htmlEscape(userInput); contentFilter.filter(sanitizedInput); return chatClient.call( 你是一位经过安全训练的AI助手。请回答 sanitizedInput); }4.2 异常处理策略全局异常处理器ControllerAdvice public class AiExceptionHandler { ExceptionHandler(AiClientException.class) public ResponseEntityErrorResponse handleAiException( AiClientException ex) { ErrorResponse response new ErrorResponse( AI_SERVICE_ERROR, AI服务暂时不可用: ex.getMessage()); return ResponseEntity .status(HttpStatus.SERVICE_UNAVAILABLE) .body(response); } ExceptionHandler(ContentViolationException.class) public ResponseEntityErrorResponse handleContentViolation( ContentViolationException ex) { return ResponseEntity .badRequest() .body(new ErrorResponse( CONTENT_VIOLATION, ex.getMessage())); } }重试机制实现Retryable( value {AiRateLimitException.class, AiTimeoutException.class}, maxAttempts 3, backoff Backoff(delay 1000, multiplier 2)) public String reliableChatCall(String prompt) { return chatClient.call(prompt); }5. 高级功能扩展5.1 多模态处理图像分析集成示例public String analyzeImage(MultipartFile imageFile) { byte[] imageBytes imageFile.getBytes(); String base64Image Base64.getEncoder().encodeToString(imageBytes); Prompt prompt new Prompt( new UserMessage( List.of( new Media(MimeTypeUtils.IMAGE_PNG, base64Image), new Text(请描述这张图片的主要内容)))); return chatClient.call(prompt).getResult().getOutput().getContent(); }5.2 函数调用集成定义可调用函数FunctionDescription( name getWeatherInfo, description 获取指定城市的天气信息) public Weather getWeather( Parameter(description 城市名称) String city) { // 调用真实天气API return weatherService.fetch(city); }函数调用配置Bean public FunctionCallbackContext functionCallbackContext() { return new DefaultFunctionCallbackContext(); } Bean public ChatClient functionChatClient( OpenAiChatClient chatClient, FunctionCallbackContext context) { return new FunctionCallingChatClient( chatClient, List.of(getClass().getMethod(getWeather, String.class)), context); }6. 部署与持续集成6.1 Docker化部署Dockerfile配置示例FROM eclipse-temurin:17-jdk-jammy WORKDIR /app COPY target/ai-service-*.jar app.jar ENTRYPOINT [java,-jar,app.jar]Kubernetes部署配置apiVersion: apps/v1 kind: Deployment metadata: name: ai-service spec: replicas: 3 template: spec: containers: - name: ai-app image: your-registry/ai-service:1.0.0 env: - name: OPENAI_API_KEY valueFrom: secretKeyRef: name: ai-secrets key: openai-key resources: limits: cpu: 2 memory: 2Gi6.2 CI/CD集成GitLab CI示例配置stages: - build - test - deploy build-job: stage: build script: - mvn clean package -DskipTests test-job: stage: test script: - mvn test - docker build -t $CI_REGISTRY_IMAGE:$CI_COMMIT_SHA . deploy-job: stage: deploy environment: production script: - echo $KUBE_CONFIG kubeconfig.yaml - kubectl apply -f k8s/deployment.yaml --kubeconfigkubeconfig.yaml