Spring AI 开发本地deepseek对话快速上手笔记

Spring AI

Spring AI是一个旨在推进生成式人工智能应用程序发展的项目，Spring AI的核心目标是提供高度抽象化的组件，作为开发AI应用程序的基础，使得开发者能够以最少的代码改动便捷地交换和优化功能模块‌

在开发之前先得引入大模型，这里选择deepseek

至于导入deepseek，咱这里选用ollama 大模型工具来进行本地化部署和管理

ollama下载与启动

进入ollama官网：Ollama

下载对应版本

直接install

deepseek模型下载

下载deepseek模型，这里选择的是r1:8b版本的，各位可以根据自己的电脑配置进行选择

执行ollama run deepseek-r1:8b

看到显示了success即运行成功

spring依赖引入

         <!-- WebFlux 响应式支持 -->
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-webflux</artifactId>
        </dependency>
          <!--ollama spring ai依赖-->
         <dependency>
            <groupId>org.springframework.ai</groupId>
            <artifactId>spring-ai-ollama-spring-boot-starter</artifactId>
            <version>1.0.0-M6</version>
        </dependency>
        <!-- Swagger3-knife4j依赖 -->
        <dependency>
            <groupId>com.github.xiaoymin</groupId>
            <artifactId>knife4j-openapi3-jakarta-spring-boot-starter</artifactId>
            <version>4.5.0</version>
        </dependency>

ChatbotController

简单对话直接返回

@Slf4j
@RestController
public class ChatBotController {
    //注入模型，配置文件中的模型，或者可以在方法中指定模型
    @Resource
    private OllamaChatModel model;
@GetMapping("/chat")
    public String chat(@RequestParam("message") String message){
        String call = model.call(message);
        System.out.println(call);
        return call;

    }

}

启动

postman请求

返回

流式对话响应

@Slf4j
@RestController
public class ChatBotController {
//注入模型，配置文件中的模型，或者可以在方法中指定模型
    @Resource
    private OllamaChatModel model;
@Resource
    private StringRedisTemplate stringRedisTemplate;
    //引入存储消息服务
    @Resource
    private ChatService chatService;


@GetMapping(value = "/streamChat", produces = "text/event-stream;charset=UTF-8")
    public Flux<String> streamChat(@RequestParam("message")String message,@RequestParam String sessionId){
        Long userId = UserHolder.getUser().getId();

        return Flux.concat(
                processAnswering(message, sessionId, userId)
        );}


/**
     * 处理回答阶段
     */
    private Flux<String> processAnswering(String message, String sessionId, Long userId) {
        return buildPromptWithContext(sessionId, message)
                .flatMapMany(prompt ->
                        model.stream(prompt)
                                .index()
                                .map(tuple -> {
                                    // 第一个消息添加标识
                                    if (tuple.getT1() == 0L) {
                                        return "[ANSWER] " + tuple.getT2();
                                    }
                                    return tuple.getT2();
                                })
                )
                .doOnNext(content -> saveMessage(sessionId, userId, message, content))
                .delayElements(Duration.ofMillis(30));
    }

/**
     * 保存消息到Redis和数据库（带事务）
     */
    @Transactional
    protected void saveMessage(String sessionId, Long userId, String question, String answer) {
//        // 保存用户问题
//        ChatContent userMsg = new ChatContent();
//        userMsg.setSessionId(sessionId);
//        userMsg.setMessage(question);
//        chatService.save(userMsg);

        // 保存AI回答
        ChatContent aiMsg = new ChatContent();
//        aiMsg.setSessionId(sessionId);
        aiMsg.setReceiveUserId(userId);
        aiMsg.setSendUserId(Long.valueOf(sessionId));
        aiMsg.setMessage(answer);
//        aiMsg.setType("ASSISTANT");
        chatService.save(aiMsg);

        // 更新Redis上下文
        stringRedisTemplate.executePipelined((RedisCallback<Object>) connection -> {
            connection.lPush((CONTEXT_PREFIX + sessionId).getBytes(),
                    question.getBytes(),
                    answer.getBytes()
            );
            connection.lTrim((CONTEXT_PREFIX + sessionId).getBytes(), 0, MAX_CONTEXT_LENGTH * 2 - 1);
            return null;
        });
    }
}

请求postman

返回

可以看到返回的数据为流式的

前端引入

UI

      <div class="bot-chat-container">
        <!-- 聊天消息区域 -->
        <div class="bot-chat-messages" ref="messagesContainer">
          <div v-for="message in bot_messages" :key="message.id"
               :class="['message', message.sender]">
            <div class="avatar">
              <img :src="message.sender === 'user' ? userAvatar : botAvatar" alt="avatar">
            </div>
            <div class="bubble">

              <div class="content"  v-html="renderMarkdown(message.content)"></div>
              <!--
                        <div class="content" v-else>{{ message.content }}</div>
              -->
              <div class="status">
                <span class="time">{{ message.timestamp }}</span>
                <span v-if="message.loading" class="typing-indicator">
              <span class="dot"></span>
              <span class="dot"></span>
              <span class="dot"></span>
            </span>
              </div>
            </div>
          </div>
        </div>

        <!-- 输入区域 -->
        <div class="bot-input-area">
      <textarea v-model="inputMessage"
                @keydown.enter.exact.prevent="sendMessage"
                placeholder="输入你的消息..."></textarea>
          <button @click="sendMessage" :disabled="isSending">
            <span v-if="!isSending">发送</span>
            <span v-else class="sending-indicator"></span>
          </button>
        </div>
      </div>

函数typescript

const sendMessage = async () => {
  if (!inputMessage.value.trim() || isSending.value) return

  // 用户消息
  const userMsg: ChatMessage = {
    id: Date.now().toString(),
    content: inputMessage.value.trim(),
    sender: 'user',
    timestamp: new Date().toLocaleTimeString()
  }
  bot_messages.push(userMsg)

  // 机器人响应占位
  const botMsg: ChatMessage = {
    id: `bot-${Date.now()}`,
    content: '',
    sender: 'bot',
    timestamp: new Date().toLocaleTimeString(),
    loading: true
  }
  bot_messages.push(botMsg)

  inputMessage.value = ''
  isSending.value = true
  // scrollToBottom()

  try {
    const sessionId = crypto.randomUUID()
    // const eventSource = new EventSource(`api/bot/streamChat?message=${encodeURIComponent(userMsg.content)}`)

    // 发起带有 Authorization 头的流式请求
    await fetchEventSource(`api/streamChat?message=${encodeURIComponent(userMsg.content)}&sessionId=333`, {
      method: 'GET',   // 或 POST（需服务端支持）
      headers: {
        'Authorization': sessionStorage.getItem("token"),  // 注入认证头 :ml-citation{ref="8" data="citationList"}
      },
      onopen(response) {
        if (response.ok) return;  // 连接成功
        throw new Error('连接失败');
      },
      onmessage(event) {
        // 处理流式数据（与原 EventSource 逻辑相同）
        const index = bot_messages.findIndex(m => m.id === botMsg.id)
        if (index !== -1) {
          bot_messages[index].content += event.data
          bot_messages[index].loading = false
          bot_messages[index].parsedContent = renderMarkdown(bot_messages[index].content)
          // scrollToBottom()
        }
      },
      onerror(err) {
        console.error('流式请求异常:', err);
      }
    });
    eventSource.onmessage = (event) => {
      const index = bot_messages.findIndex(m => m.id === botMsg.id)
      if (index !== -1) {
        bot_messages[index].content += event.data
        bot_messages[index].loading = false
        bot_messages[index].parsedContent = renderMarkdown(bot_messages[index].content)

        // scrollToBottom()
      }
    }

    eventSource.onerror = () => {
      eventSource.close()
      isSending.value = false
    }

  } catch (error) {
    console.error('Error:', error)
    isSending.value = false
  }
}

css

.bot-chat-container {
  display: flex;
  flex-direction: column;
  height: 100vh;
  background: #f5f5f5;
}

.bot-chat-messages {
  flex: 1;
  overflow-y: auto;
  padding: 20px;
  background: linear-gradient(180deg, #f0f2f5 0%, #ffffff 100%);
}

.message {
  display: flex;
  margin-bottom: 20px;
  gap: 12px;
}

.message.user {
  flex-direction: row-reverse;
}

.avatar img {
  width: 40px;
  height: 40px;
  border-radius: 50%;
  box-shadow: 0 2px 4px rgba(0,0,0,0.1);
}

.bubble {
  max-width: 70%;
  position: relative;
}

.bubble .content {
  padding: 12px 16px;
  border-radius: 12px;
  line-height: 1.5;
  font-size: 14px;
}

.message.bot .content {
  background: white;
  border: 1px solid #e5e7eb;
  border-radius: 12px 12px 12px 4px;
}

.message.user .content {
  background: #3875f6;
  color: white;
  border-radius: 12px 12px 4px 12px;
}

.status {
  display: flex;
  align-items: center;
  gap: 8px;
  margin-top: 4px;
  font-size: 12px;
  color: #666;
}

.message.user .status {
  justify-content: flex-end;
}

.typing-indicator {
  display: inline-flex;
  gap: 4px;
}

.dot {
  width: 6px;
  height: 6px;
  background: #999;
  border-radius: 50%;
  animation: bounce 1.4s infinite ease-in-out;
}

.dot:nth-child(2) {
  animation-delay: 0.2s;
}

.dot:nth-child(3) {
  animation-delay: 0.4s;
}

@keyframes bounce {
  0%, 80%, 100% { transform: translateY(0); }
  40% { transform: translateY(-4px); }
}

.bot-input-area {
  display: flex;
  gap: 12px;
  padding: 20px;
  border-top: 1px solid #e5e7eb;
  background: white;
}

textarea {
  flex: 1;
  padding: 12px;
  border: 1px solid #e5e7eb;
  border-radius: 8px;
  resize: none;
  min-height: 44px;
  max-height: 120px;
  font-family: inherit;
}

button {
  padding: 0 20px;
  background: #3875f6;
  color: white;
  border: none;
  border-radius: 8px;
  cursor: pointer;
  transition: opacity 0.2s;
}

button:disabled {
  background: #a0aec0;
  cursor: not-allowed;
}

.sending-indicator {
  display: inline-block;
  width: 20px;
  height: 20px;
  border: 2px solid #fff;
  border-top-color: transparent;
  border-radius: 50%;
  animation: spin 0.8s linear infinite;
}

@keyframes spin {
  to { transform: rotate(360deg); }
}
.bubble :deep() pre {
  background: #f8f8f8;
  padding: 12px;
  border-radius: 6px;
  overflow-x: auto;
}

.bubble :deep() code {
  font-family: 'JetBrains Mono', monospace;
  font-size: 14px;
}

.bubble :deep() ul,
.bubble :deep() ol {
  padding-left: 20px;
  margin: 8px 0;
}

.bubble :deep() blockquote {
  border-left: 4px solid #ddd;
  margin: 8px 0;
  padding-left: 12px;
  color: #666;
}

发送消息

至此简单的AI对话完成了

源码地址

后端：https://github.com/enjoykanyu/chat_serve

前端：https://github.com/enjoykanyu/kChat_web

觉得不错得话，可以帮点个star呀，感谢

若在执行部署过程中有任何问题，欢迎githup提issue