山东大学软件学院项目实训-基于大模型的模拟面试系统-面试对话标题自动总结
面试对话标题自动总结
主要实现思路:每当AI回复用户之后,调用方法查看当前对话是否大于三条,如果大于则将用户的两条和AI回复的一条对话传给DeepSeek让其进行总结(后端),总结后调用updateChatTopic
进行更新标题,此外本次标题的更改还实现了仿打字机效果。
后端实现
首先,要在.env
文件中配置DEEPSEEK_API
,然后在application.yml
中添加:
deepseek:api:key: ${DEEPSEEK_API} # DeepSeek API Key,从环境变量中获取
之后创建DeesSeekService.java
package com.sdumagicode.backend.openai.service;import com.fasterxml.jackson.annotation.JsonInclude;
import com.fasterxml.jackson.databind.DeserializationFeature;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.fasterxml.jackson.databind.PropertyNamingStrategy;import okhttp3.*;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.stereotype.Service;import java.io.IOException;
import java.util.*;
import java.util.concurrent.TimeUnit;/*** DeepSeek API服务类* 用于与DeepSeek API进行交互*/
@Service
public class DeepSeekService {private static final Logger logger = LoggerFactory.getLogger(DeepSeekService.class);private static final String API_URL = "https://api.deepseek.com/v1/chat/completions";@Value("${deepseek.api.key}")private String apiKey;private final OkHttpClient client;private final ObjectMapper objectMapper;public DeepSeekService() {this.client = new OkHttpClient.Builder().connectTimeout(30, TimeUnit.SECONDS).readTimeout(60, TimeUnit.SECONDS).writeTimeout(30, TimeUnit.SECONDS).build();this.objectMapper = new ObjectMapper().configure(DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES, false).setSerializationInclusion(JsonInclude.Include.NON_NULL).setPropertyNamingStrategy(PropertyNamingStrategy.SNAKE_CASE);}/*** 发送聊天请求到DeepSeek API* * @param messages 消息列表* @param model 模型名称,默认为 "deepseek-chat"* @param temperature 温度参数,控制随机性* @return API响应的JSON字符串* @throws IOException 如果API请求失败*/public String chatCompletion(List<Map<String, String>> messages, String model, double temperature) throws IOException {Map<String, Object> requestBody = new HashMap<>();requestBody.put("messages", messages);requestBody.put("model", model);requestBody.put("temperature", temperature);String jsonBody = objectMapper.writeValueAsString(requestBody);RequestBody body = RequestBody.create(MediaType.parse("application/json"), jsonBody);Request request = new Request.Builder().url(API_URL).addHeader("Authorization", "Bearer " + apiKey).addHeader("Content-Type", "application/json").post(body).build();try (Response response = client.newCall(request).execute()) {if (!response.isSuccessful()) {throw new IOException("API请求失败: " + response.code() + " " + response.message());}return response.body().string();}}/*** 发送聊天请求到DeepSeek API(使用默认参数)* * @param messages 消息列表* @return API响应的JSON字符串* @throws IOException 如果API请求失败*/public String chatCompletion(List<Map<String, String>> messages) throws IOException {return chatCompletion(messages, "deepseek-chat", 0.7);}/*** 创建聊天消息* * @param role 角色 (system, user, assistant)* @param content 消息内容* @return 包含角色和内容的Map*/public Map<String, String> createMessage(String role, String content) {Map<String, String> message = new HashMap<>();message.put("role", role);message.put("content", content);return message;}/*** 流式聊天请求(SSE)* * @param messages 消息列表* @param model 模型名称* @param temperature 温度参数* @param callback 回调函数,用于处理每个SSE事件* @throws IOException 如果API请求失败*/public void streamingChatCompletion(List<Map<String, String>> messages, String model, double temperature, Callback callback) throws IOException {Map<String, Object> requestBody = new HashMap<>();requestBody.put("messages", messages);requestBody.put("model", model);requestBody.put("temperature", temperature);requestBody.put("stream", true);String jsonBody = objectMapper.writeValueAsString(requestBody);RequestBody body = RequestBody.create(MediaType.parse("application/json"), jsonBody);Request request = new Request.Builder().url(API_URL).addHeader("Authorization", "Bearer " + apiKey).addHeader("Content-Type", "application/json").post(body).build();client.newCall(request).enqueue(callback);}
}
该文件用于与DeepSeek AI API进行交互。主要功能包括:
-
基本配置:
- 使用OkHttpClient进行HTTP请求
- 配置ObjectMapper处理JSON序列化/反序列化
- 从配置文件读取API密钥
-
主要方法:
chatCompletion
:向DeepSeek API发送聊天请求,有两个版本:- 完整参数版:可指定消息、模型和温度
- 简化版:使用默认参数
createMessage
:创建聊天消息对象streamingChatCompletion
:流式聊天请求,支持SSE(服务器发送事件)listModels
:获取可用模型列表
-
技术特点:
- 使用Spring的
@Service
注解标记为服务 - 通过
@Value
注入API密钥 - 支持异步回调处理流式响应
- 包含完整的异常处理和日志记录
- 使用Spring的
这个服务类是后端与DeepSeek AI API通信的核心组件,负责处理所有AI聊天相关的请求。
DeepSeekController.java
package com.sdumagicode.backend.openai;import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.sdumagicode.backend.core.result.GlobalResult;
import com.sdumagicode.backend.core.result.GlobalResultGenerator;
import com.sdumagicode.backend.openai.service.DeepSeekService;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.bind.annotation.*;import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;/*** DeepSeek API控制器* 提供DeepSeek模型API的接口*/
@RestController
@RequestMapping("/api/v1/deepseek")
public class DeepSeekController {@Autowiredprivate DeepSeekService deepSeekService;private final ObjectMapper objectMapper = new ObjectMapper();/*** 发送聊天请求** @param requestBody 请求体,包含messages, model, temperature* @return 聊天响应*/@PostMapping("/chat")public GlobalResult<String> chatCompletion(@RequestBody Map<String, Object> requestBody) {try {@SuppressWarnings("unchecked")List<Map<String, String>> messages = (List<Map<String, String>>) requestBody.get("messages");String model = requestBody.containsKey("model") ? (String) requestBody.get("model") : "deepseek-chat";double temperature = requestBody.containsKey("temperature") ? Double.parseDouble(requestBody.get("temperature").toString()) : 0.7;String response = deepSeekService.chatCompletion(messages, model, temperature);return GlobalResultGenerator.genSuccessResult(response);} catch (Exception e) {return GlobalResultGenerator.genErrorResult("聊天请求失败:" + e.getMessage());}}/*** 生成对话摘要* * @param requestBody 包含对话消息和chatId的请求体* @return 生成的摘要*/@PostMapping("/summarize")public GlobalResult<String> summarizeConversation(@RequestBody Map<String, Object> requestBody) {try {@SuppressWarnings("unchecked")List<Map<String, Object>> messages = (List<Map<String, Object>>) requestBody.get("messages");// 准备系统提示词和用户消息List<Map<String, String>> promptMessages = new ArrayList<>();// 添加系统提示Map<String, String> systemPrompt = new HashMap<>();systemPrompt.put("role", "system");systemPrompt.put("content", "你是一个专业的面试对话摘要生成器。请根据以下面试对话生成一个简短的标题,标题应概括对话的主要内容。"+ "标题必须精炼,不超过20个字,不要加任何前缀和标点符号,直接输出标题文本。标题应突出面试的主要话题或技能领域。");promptMessages.add(systemPrompt);// 构建对话历史StringBuilder conversationBuilder = new StringBuilder();for (Map<String, Object> message : messages) {String role = (String) message.get("role");String content = (String) message.get("content");if (content == null || content.trim().isEmpty()) {continue;}conversationBuilder.append(role.equals("assistant") ? "面试官: " : "候选人: ");conversationBuilder.append(content).append("\n\n");}// 添加用户消息,包含对话内容Map<String, String> userMessage = new HashMap<>();userMessage.put("role", "user");userMessage.put("content", "以下是一段面试对话,请为其生成一个简短的标题:\n\n" + conversationBuilder.toString());promptMessages.add(userMessage);// 调用DeepSeek API生成摘要,使用较低的温度以获得更确定性的结果String response = deepSeekService.chatCompletion(promptMessages, "deepseek-chat", 0.3);// 解析响应,提取摘要文本JsonNode responseJson = objectMapper.readTree(response);String summary = responseJson.path("choices").get(0).path("message").path("content").asText();// 清理摘要文本,去除多余的引号、空格和换行符summary = summary.replaceAll("\"", "").trim();summary = summary.replaceAll("\\r?\\n", " ").trim();return GlobalResultGenerator.genSuccessResult(summary);} catch (Exception e) {return GlobalResultGenerator.genErrorResult("生成摘要失败:" + e.getMessage());}}
前端实现
自动总结标题功能在handlePollingCompleted
方法中实现,主要流程如下:
async handlePollingCompleted() {try {if (!this.activeChatRecord) return;// 检查是否有足够内容生成摘要且未生成过if (this.$refs.chatArea &&this.$refs.chatArea.messageListForShow &&!this.summarizedChatIds.has(this.activeChatRecord)) {const messages = this.$refs.chatArea.messageListForShow;// 至少需要3条消息才生成摘要if (messages.length >= 3) {// 准备对话数据const dialogData = messages.map(msg => ({role: msg.role,content: msg.content.text}));try {// 调用DeepSeek API生成摘要const summaryResponse = await axios.post('/api/deepseek/summarize', {messages: dialogData,chatId: this.activeChatRecord});if (summaryResponse.message) {const summary = summaryResponse.message;// 更新对话标题await axios.post('/api/chat/updateChatTopic', null, {params: {chatId: this.activeChatRecord,newTopic: summary}});// 记录已生成摘要的对话IDthis.summarizedChatIds.add(this.activeChatRecord);// 重新加载聊天记录显示新标题await this.loadChatRecords();// 启动打字机效果展示新标题this.startTypingAnimation(this.activeChatRecord, summary);}} catch (error) {console.warn('生成对话摘要失败:', error);// 即使失败也标记为已处理,避免重复尝试this.summarizedChatIds.add(this.activeChatRecord);}}}// 后续处理actions的代码...} catch (error) {console.error('获取actions失败:', error);}
}
关键设计要点:
- 触发时机:当轮询完成后自动触发,通过
handlePollingCompleted
方法 - 条件判断:
- 必须有活跃的聊天记录
- 对话消息数量至少3条
- 该对话ID未被记录在
summarizedChatIds
集合中
- 数据收集:从
messageListForShow
中提取对话内容并格式化 - API调用:通过
/api/deepseek/summarize
请求生成摘要 - 标题更新:通过
/api/chat/updateChatTopic
更新对话标题 - 状态管理:
- 使用
summarizedChatIds
集合跟踪已处理的对话 - 无论成功失败都标记为已处理,避免重复请求
- 使用
- 用户体验:使用打字机效果动态展示新标题
打字机效果
打字机效果在文件中的实现主要包含三个核心方法和相应的CSS样式,用于在聊天记录标题中实现逐字显示的动画效果。
数据结构
typingAnimation: {chatId: null, // 当前执行动画的对话IDoriginalText: '', // 完整文本displayText: '', // 当前显示的文本(逐渐增加)isActive: false, // 动画激活状态charIndex: 0, // 当前字符索引timerId: null // 定时器ID
}
核心方法
- 启动动画:
startTypingAnimation(chatId, text) {this.stopTypingAnimation(); // 先停止已有动画if (!text || !chatId || text.length < 3) return;this.typingAnimation = {chatId, originalText: text, displayText: '',isActive: true, charIndex: 0, timerId: null};setTimeout(() => this.animateNextChar(), 200); // 延迟启动
}
- 字符添加动画:
animateNextChar() {const { charIndex, originalText } = this.typingAnimation;if (charIndex <= originalText.length) {this.typingAnimation.displayText = originalText.substring(0, charIndex);this.typingAnimation.charIndex = charIndex + 1;// 随机速度模拟自然打字const baseSpeed = 70;const randomVariation = Math.random() * 100;const speed = baseSpeed + randomVariation;this.typingAnimation.timerId = setTimeout(() => {this.animateNextChar();}, speed);} else {this.stopTypingAnimation(true);}
}
- 停止动画:
stopTypingAnimation(completed = false) {if (this.typingAnimation.timerId) {clearTimeout(this.typingAnimation.timerId);}if (completed) {// 完成后短暂延迟setTimeout(() => {this.typingAnimation.isActive = false;}, 500);} else {// 立即重置this.typingAnimation = {chatId: null, originalText: '', displayText: '',isActive: false, charIndex: 0, timerId: null};}
}
视觉效果
- 闪烁光标: 使用CSS动画模拟打字光标闪烁
.cursor {width: 2px;height: 16px;background-color: #409eff;animation: blink 0.7s infinite;
}@keyframes blink {0%, 100% { opacity: 1; }50% { opacity: 0; }
}
- 调用时机:
- 生成聊天摘要后:
this.startTypingAnimation(this.activeChatRecord, summary);
- 重命名对话后:
this.startTypingAnimation(chatId, this.newTopicName);