DeepSeek API 调用 - Spring Boot 实现

DeepSeek API 调用 - Spring Boot 实现

1. 项目依赖

在pom.xml中添加以下依赖：

<dependencies> <dependency> <groupId>org.springframework.boot</groupId> <artifactId>spring-boot-starter-webflux</artifactId> </dependency> <dependency> <groupId>org.projectlombok</groupId> <artifactId>lombok</artifactId> </dependency> <dependency> <groupId>com.fasterxml.jackson.core</groupId> <artifactId>jackson-databind</artifactId> </dependency> </dependencies>

2. 项目结构

deepseek-project/ ├── src/main/java/com/example/deepseek/ │ ├── DeepSeekApplication.java │ ├── config/ │ │ └── DeepSeekConfig.java │ ├── model/ │ │ ├── ChatRequest.java │ │ ├── ChatResponse.java │ │ └── Message.java │ └── service/ │ └── DeepSeekService.java └── conversation.txt

3. 完整代码实现

3.1 配置类DeepSeekConfig.java

package com.example.deepseek.config; import org.springframework.beans.factory.annotation.Value; import org.springframework.context.annotation.Configuration; @Configuration @Getter public class DeepSeekConfig { @Value("${deepseek.api.url}") private String apiUrl; @Value("${deepseek.api.key}") private String apiKey; }

3.2 请求/响应模型

Message.java:

package com.example.deepseek.model; import lombok.Data; @Data public class Message { private String role; private String content; }

ChatRequest.java:

package com.example.deepseek.model; import lombok.Data; import java.util.List; @Data public class ChatRequest { private String model = "deepseek-ai/DeepSeek-V3"; private List<Message> messages; private boolean stream = true; private int max_tokens = 2048; private double temperature = 0.7; private double top_p = 0.7; private int top_k = 50; private double frequency_penalty = 0.5; private int n = 1; private ResponseFormat response_format = new ResponseFormat("text"); @Data public static class ResponseFormat { private String type; public ResponseFormat(String type) { this.type = type; } } }

ChatResponse.java:

package com.example.deepseek.model; import lombok.Data; import java.util.List; @Data public class ChatResponse { private List<Choice> choices; @Data public static class Choice { private Delta delta; } @Data public static class Delta { private String content; } }

3.3 服务类DeepSeekService.java

package com.example.deepseek.service; import com.example.deepseek.config.DeepSeekConfig; import com.example.deepseek.model.ChatRequest; import com.example.deepseek.model.ChatResponse; import com.example.deepseek.model.Message; import com.fasterxml.jackson.databind.ObjectMapper; import lombok.RequiredArgsConstructor; import org.springframework.stereotype.Service; import org.springframework.web.reactive.function.client.WebClient; import reactor.core.publisher.Flux; import java.io.FileWriter; import java.io.IOException; import java.io.PrintWriter; import java.time.LocalDateTime; import java.time.format.DateTimeFormatter; import java.util.Collections; import java.util.Scanner; @Service @RequiredArgsConstructor public class DeepSeekService { private final DeepSeekConfig config; private final WebClient.Builder webClientBuilder; private final ObjectMapper objectMapper = new ObjectMapper(); public void startInteractiveChat() { try (Scanner scanner = new Scanner(System.in); PrintWriter fileWriter = new PrintWriter(new FileWriter("conversation.txt", true))) { while (true) { System.out.print(" 请输入您的问题 (输入 q 退出): "); String question = scanner.nextLine().trim(); if ("q".equalsIgnoreCase(question)) { System.out.println("程序已退出"); break; } // 保存问题 saveToFile(fileWriter, question, true); // 发起对话请求 Flux<String> responseFlux = sendChatRequest(question); StringBuilder fullResponse = new StringBuilder(); responseFlux .doOnNext(chunk -> { System.out.print(chunk); fullResponse.append(chunk); }) .doOnComplete(() -> { // 保存完整回复 saveToFile(fileWriter, fullResponse.toString(), false); System.out.println(" ----------------------------------------"); fileWriter.println(" ----------------------------------------"); fileWriter.flush(); }) .blockLast(); } } catch (IOException e) { e.printStackTrace(); } } private Flux<String> sendChatRequest(String question) { ChatRequest request = new ChatRequest(); Message userMessage = new Message(); userMessage.setRole("user"); userMessage.setContent(question); request.setMessages(Collections.singletonList(userMessage)); return webClientBuilder.build() .post() .uri(config.getApiUrl()) .header("Authorization", "Bearer " + config.getApiKey()) .header("Content-Type", "application/json") .bodyValue(request) .retrieve() .bodyToFlux(String.class) .filter(line -> line.startsWith("data: ") && !line.equals("data: [DONE]")) .map(line -> { try { String jsonStr = line.substring(6); ChatResponse response = objectMapper.readValue(jsonStr, ChatResponse.class); return response.getChoices().get(0).getDelta().getContent(); } catch (Exception e) { return ""; } }) .filter(content -> !content.isEmpty()); } private void saveToFile(PrintWriter fileWriter, String content, boolean isQuestion) { String timestamp = LocalDateTime.now().format(DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss")); if (isQuestion) { fileWriter.printf(" [%s] Question: %s [%s] Answer: ", timestamp, content, timestamp); } else { fileWriter.print(content); } fileWriter.flush(); } }

3.4 主应用类DeepSeekApplication.java

package com.example.deepseek; import com.example.deepseek.service.DeepSeekService; import org.springframework.boot.SpringApplication; import org.springframework.boot.autoconfigure.SpringBootApplication; import org.springframework.context.ConfigurableApplicationContext; @SpringBootApplication public class DeepSeekApplication { public static void main(String[] args) { ConfigurableApplicationContext context = SpringApplication.run(DeepSeekApplication.class, args); DeepSeekService deepSeekService = context.getBean(DeepSeekService.class); deepSeekService.startInteractiveChat(); } }

3.5 配置文件application.properties

deepseek.api.url=https://api.siliconflow.cn/v1/chat/completions deepseek.api.key=YOUR_API_KEY

4. 代码详解

4.1 关键特性

1. 使用 Spring WebFlux 的响应式编程模型
2. 流式处理 API 响应
3. 文件记录对话
4. 错误处理和异常管理

4.2 主要组件

• DeepSeekConfig: 管理 API 配置
• DeepSeekService: 处理对话逻辑和 API 交互
• 模型类: 定义请求和响应结构

5. 使用方法

1. 替换application.properties中的YOUR_API_KEY
2. 运行DeepSeekApplication
3. 在控制台输入问题
4. 输入 ‘q’ 退出程序
5. 查看conversation.txt获取对话记录

6. 性能和可扩展性

• 使用响应式编程提高并发性能
• 灵活的配置管理
• 易于扩展和定制

7. 注意事项

• 确保正确配置 API Key
• 处理网络异常
• 注意内存使用

总结

Spring Boot 实现提供了一个健壮、可扩展的 DeepSeek API 调用方案，利用响应式编程提供高效的流式对话体验。

立即体验

快来体验 DeepSeek：https://cloud.siliconflow.cn/i/vnCCfVaQ