OpenAI 集成

Java突击队大约 6 分钟

OpenAI是AI领域的领导者，提供了多个强大的模型和服务。

5.1.1 OpenAI模型家族

OpenAI提供了多个模型，每个都有其特定的用途：

GPT系列：文本生成和对话
- gpt-4：最强大的模型
- gpt-4-turbo：高速版本
- gpt-3.5-turbo：经济型选择
DALL-E系列：图像生成
- DALL-E 3：最新图像生成模型
Embeddings：文本嵌入
- text-embedding-ada-002：通用嵌入模型

OpenAI模型家族

5.1.2 API功能

Chat Completions：对话生成
Completions：文本补全
Images：图像生成
Embeddings：文本嵌入
Audio：语音处理
Fine-tuning：模型微调

5.2 配置OpenAI客户端

5.2.1 添加依赖

在pom.xml中添加OpenAI依赖：

<dependency>
    <groupId>org.springframework.ai</groupId>
    <artifactId>spring-ai-openai-spring-boot-starter</artifactId>
</dependency>

5.2.2 配置API密钥

在application.properties中配置OpenAI API密钥：

# OpenAI配置
spring.ai.openai.api-key=your-openai-api-key
spring.ai.openai.chat.options.model=gpt-4
spring.ai.openai.chat.options.temperature=0.7
spring.ai.openai.chat.options.max-tokens=1000

5.2.3 编程配置

@Configuration
public class OpenAiConfig {

    @Bean
    public ChatClient openAiChatClient(ChatClient.Builder builder) {
        return builder
                .defaultSystem("You are a helpful assistant")
                .defaultOptions(OpenAiChatOptions.builder()
                        .withModel("gpt-4-turbo")
                        .withTemperature(0.8f)
                        .withMaxTokens(2000)
                        .build())
                .build();
    }
}

5.3 ChatGPT应用开发

5.3.1 基础对话

@RestController
public class OpenAiController {

    private final ChatClient openAiChatClient;

    public OpenAiController(ChatClient openAiChatClient) {
        this.openAiChatClient = openAiChatClient;
    }

    @GetMapping("/openai/chat")
    public String chat(@RequestParam String message) {
        return openAiChatClient.prompt()
                .user(message)
                .call()
                .content();
    }
}

5.3.2 系统提示和角色设定

@GetMapping("/openai/role-play")
public String rolePlay(
        @RequestParam String role,
        @RequestParam String message) {

    return openAiChatClient.prompt()
            .system("You are " + role)
            .user(message)
            .call()
            .content();
}

5.3.3 多轮对话管理

@RestController
public class ConversationController {

    private final ChatClient openAiChatClient;

    public ConversationController(ChatClient openAiChatClient) {
        this.openAiChatClient = openAiChatClient;
    }

    @PostMapping("/openai/conversation")
    public ConversationResponse conversation(@RequestBody ConversationRequest request) {
        ChatPrompt chatPrompt = new ChatPrompt();

        // 添加系统提示
        chatPrompt.add(new SystemMessage(request.getSystemPrompt()));

        // 添加历史消息
        for (Message message : request.getHistory()) {
            chatPrompt.add(message);
        }

        // 添加当前消息
        chatPrompt.add(new UserMessage(request.getCurrentMessage()));

        ChatResponse response = openAiChatClient.prompt(chatPrompt).call();

        return new ConversationResponse(
            response.getContent(),
            response.getResults().get(0).getGenerationMetadata()
        );
    }
}

5.4 DALL-E图像生成

5.4.1 基础图像生成

@RestController
public class DalleController {

    private final ImageClient imageClient;

    public DalleController(ImageClient imageClient) {
        this.imageClient = imageClient;
    }

    @GetMapping("/openai/image")
    public String generateImage(@RequestParam String prompt) {
        ImagePrompt imagePrompt = new ImagePrompt(prompt);
        ImageResponse response = imageClient.generate(imagePrompt);

        return response.getResult().getOutput().getUrl();
    }
}

5.4.2 高级图像生成选项

@GetMapping("/openai/advanced-image")
public String generateAdvancedImage(
        @RequestParam String prompt,
        @RequestParam(defaultValue = "1024x1024") String size,
        @RequestParam(defaultValue = "1") int n) {

    ImagePrompt imagePrompt = new ImagePrompt(prompt);

    ImageOptions imageOptions = ImageOptions.builder()
            .withModel("dall-e-3")
            .withN(n)
            .withSize(size)
            .build();

    ImageResponse response = imageClient.generate(imagePrompt, imageOptions);

    return response.getResult().getOutput().getUrl();
}

5.4.3 图像编辑

@PostMapping("/openai/edit-image")
public String editImage(
        @RequestParam MultipartFile image,
        @RequestParam String prompt) {

    ImagePrompt imagePrompt = new ImagePrompt(image, prompt);
    ImageResponse response = imageClient.generate(imagePrompt);

    return response.getResult().getOutput().getUrl();
}

5.5 高级功能

5.5.1 函数调用

OpenAI的函数调用功能非常强大：

@RestController
public class OpenAiFunctionController {

    private final ChatClient openAiChatClient;

    public OpenAiFunctionController(ChatClient openAiChatClient) {
        this.openAiChatClient = openAiChatClient;
    }

    @GetMapping("/openai/function")
    public String functionCall(@RequestParam String question) {
        FunctionCallback weatherFunction = new FunctionCallback(
            "getWeather",
            "Get current weather for a location",
            new Schema(SchemaType.OBJECT)
                .addProperty("location", SchemaProperty.of(SchemaType.STRING, "City name"))
                .addRequired("location"),
            this::getWeather
        );

        FunctionCallback calculatorFunction = new FunctionCallback(
            "calculate",
            "Perform mathematical calculations",
            new Schema(SchemaType.OBJECT)
                .addProperty("expression", SchemaProperty.of(SchemaType.STRING, "Mathematical expression"))
                .addRequired("expression"),
            this::calculate
        );

        return openAiChatClient.prompt()
                .user(question)
                .functions(weatherFunction, calculatorFunction)
                .call()
                .content();
    }

    private String getWeather(Map<String, Object> arguments) {
        String location = (String) arguments.get("location");
        return "The current weather in " + location + " is 25°C and sunny";
    }

    private String calculate(Map<String, Object> arguments) {
        String expression = (String) arguments.get("expression");
        return "Result: " + evaluateExpression(expression);
    }
}

5.5.2 嵌入向量

@RestController
public class EmbeddingController {

    private final EmbeddingClient embeddingClient;

    public EmbeddingController(EmbeddingClient embeddingClient) {
        this.embeddingClient = embeddingClient;
    }

    @GetMapping("/openai/embedding")
    public List<Float> getEmbedding(@RequestParam String text) {
        EmbeddingResponse embeddingResponse = embeddingClient.call(
            new EmbeddingRequest(List.of(text))
        );

        return embeddingResponse.getResults().get(0).getOutput();
    }
}

5.5.3 语音转文本

@RestController
public class SpeechToTextController {

    private final AudioClient audioClient;

    public SpeechToTextController(AudioClient audioClient) {
        this.audioClient = audioClient;
    }

    @PostMapping("/openai/speech-to-text")
    public String speechToText(@RequestParam MultipartFile audio) {
        AudioRequest audioRequest = new AudioRequest(audio);
        AudioResponse response = audioClient.generate(audioRequest);

        return response.getResult().getOutput();
    }
}

5.6 实际应用示例

5.6.1 智能写作助手

@Service
public class WritingAssistant {

    private final ChatClient openAiChatClient;

    public WritingAssistant(ChatClient openAiChatClient) {
        this.openAiChatClient = openAiChatClient;
    }

    public WritingResponse improveWriting(String text, String style) {
        return openAiChatClient.prompt()
                .system("You are a professional writer specializing in " + style + " writing")
                .user("Improve the following text: " + text)
                .call()
                .entity(WritingResponse.class);
    }
}

public class WritingResponse {
    private String improvedText;
    private List<String> suggestions;
    private double score;
    // getters and setters
}

5.6.2 代码审查工具

@RestController
public class CodeReviewController {

    private final ChatClient openAiChatClient;

    public CodeReviewController(ChatClient openAiChatClient) {
        this.openAiChatClient = openAiChatClient;
    }

    @PostMapping("/openai/code-review")
    public CodeReviewResult reviewCode(@RequestBody CodeReviewRequest request) {
        String prompt = "Review the following code and provide suggestions:\n" + request.getCode();

        return openAiChatClient.prompt()
                .user(prompt)
                .call()
                .entity(CodeReviewResult.class);
    }
}

public class CodeReviewResult {
    private List<String> issues;
    private List<String> suggestions;
    private String overallRating;
    // getters and setters
}

5.7 性能优化

5.7.1 缓存策略

@Configuration
@EnableCaching
public class OpenAiCacheConfig {

    @Bean
    public CacheManager cacheManager() {
        SimpleCacheManager cacheManager = new SimpleCacheManager();
        cacheManager.setCaches(Arrays.asList(
            new ConcurrentMapCache("openai-responses"),
            new ConcurrentMapCache("openai-embeddings")
        ));
        return cacheManager;
    }
}

5.7.2 异步处理

@RestController
public class AsyncOpenAiController {

    private final ChatClient openAiChatClient;

    public AsyncOpenAiController(ChatClient openAiChatClient) {
        this.openAiChatClient = openAiChatClient;
    }

    @GetMapping("/openai/async")
    public CompletableFuture<String> asyncChat(@RequestParam String message) {
        return openAiChatClient.prompt()
                .user(message)
                .callAsync()
                .thenApply(ChatResponse::getContent);
    }
}

5.8 错误处理

5.8.1 异常处理

@RestControllerAdvice
public class OpenAiExceptionHandler {

    @ExceptionHandler(OpenAiApiException.class)
    public ResponseEntity<ErrorResponse> handleOpenAiError(OpenAiApiException ex) {
        ErrorResponse error = new ErrorResponse(
            "OPENAI_API_ERROR",
            "OpenAI API调用失败: " + ex.getMessage()
        );
        return new ResponseEntity<>(error, HttpStatus.SERVICE_UNAVAILABLE);
    }
}

5.8.2 速率限制处理

@Configuration
public class OpenAiRateLimitConfig {

    @Bean
    public ChatClient rateLimitedChatClient(ChatClient chatClient) {
        return new ChatClient() {
            @Override
            public ChatResponse call(Prompt prompt) {
                try {
                    return chatClient.call(prompt);
                } catch (OpenAiRateLimitException ex) {
                    // 处理速率限制
                    try {
                        Thread.sleep(1000); // 等待1秒后重试
                        return chatClient.call(prompt);
                    } catch (InterruptedException e) {
                        Thread.currentThread().interrupt();
                        throw new RuntimeException("Rate limit retry failed", e);
                    }
                }
            }

            // 实现其他方法...
        };
    }
}

5.9 监控与日志

5.9.1 指标监控

@RestController
public class OpenAiMetricsController {

    private final MeterRegistry meterRegistry;

    public OpenAiMetricsController(MeterRegistry meterRegistry) {
        this.meterRegistry = meterRegistry;
    }

    @GetMapping("/openai/metrics")
    public Map<String, Object> getMetrics() {
        return Map.of(
            "totalCalls", meterRegistry.counter("openai.total.calls").count(),
            "averageDuration", meterRegistry.timer("openai.duration").mean(),
            "errorRate", meterRegistry.counter("openai.errors").count() /
                       (meterRegistry.counter("openai.total.calls").count() + 1.0)
        );
    }
}

5.9.2 请求日志

@Configuration
public class OpenAiLogging {

    @Bean
    public ChatClient loggingChatClient(ChatClient chatClient) {
        return new ChatClient() {
            @Override
            public ChatResponse call(Prompt prompt) {
                long startTime = System.currentTimeMillis();
                try {
                    ChatResponse response = chatClient.call(prompt);
                    long duration = System.currentTimeMillis() - startTime;
                    log.info("OpenAI调用 - 耗时: {}ms, Prompt: {}", duration, prompt);
                    return response;
                } catch (Exception ex) {
                    log.error("OpenAI调用失败", ex);
                    throw ex;
                }
            }

            // 实现其他方法...
        };
    }
}

5.10 小结

本章详细介绍了OpenAI模型的集成和使用：

API概览：了解OpenAI的模型家族和功能
客户端配置：掌握OpenAI客户端的配置方法
ChatGPT应用：学习对话生成和多轮对话管理
图像生成：掌握DALL-E图像生成和编辑
高级功能：了解函数调用、嵌入向量等高级特性
实际应用：学习智能写作、代码审查等实际应用场景
性能优化：掌握缓存、异步等优化技术
错误处理：了解异常处理和速率限制处理
监控日志：掌握指标监控和请求日志

在下一章中，我们将学习文本处理与生成的高级技术。点击这里👇🏻获取：100万QPS短链系统、复杂的商城微服务系统、智能翻译助手AI Agent、SaaS点餐系统、刷题吧小程序、商城系统、秒杀系统、AI项目、代码生成神器、苏三demo项目、智能天气播报AI Agent、智能代码审查AI Agent等 10 个项目的：项目源代码、开发教程和技术答疑

OpenAI 集成

第五章：集成OpenAI

5.1 OpenAI API概览

5.1.1 OpenAI模型家族

5.1.2 API功能

5.2 配置OpenAI客户端

5.2.1 添加依赖

5.2.2 配置API密钥

5.2.3 编程配置

5.3 ChatGPT应用开发

5.3.1 基础对话

5.3.2 系统提示和角色设定

5.3.3 多轮对话管理

5.4 DALL-E图像生成

5.4.1 基础图像生成

5.4.2 高级图像生成选项

5.4.3 图像编辑

5.5 高级功能

5.5.1 函数调用

5.5.2 嵌入向量

5.5.3 语音转文本

5.6 实际应用示例

5.6.1 智能写作助手

5.6.2 代码审查工具

5.7 性能优化

5.7.1 缓存策略

5.7.2 异步处理

5.8 错误处理

5.8.1 异常处理

5.8.2 速率限制处理

5.9 监控与日志

5.9.1 指标监控

5.9.2 请求日志

5.10 小结