增加获取结果缓冲区的上限

guyue

7 天以前 a9287c6b562da327587e2a4bac92df14eb7e2b01

 src/main/java/com/linghu/controller/CollectController.java

@@ -4,28 +4,24 @@
import java.time.LocalDateTime;
import java.time.format.DateTimeFormatter;
import java.util.*;
import java.util.concurrent.*;
import java.util.concurrent.locks.ReentrantLock;
import java.util.stream.Collectors;

import javax.annotation.Resource;
import javax.servlet.http.HttpServletRequest;

import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.JsonNode;
import com.fasterxml.jackson.databind.ObjectMapper;
import com.linghu.mapper.PlatformMapper;
import com.linghu.mapper.TypeMapper;
import com.linghu.model.dto.*;
import com.linghu.model.entity.*;
import com.linghu.service.*;
import lombok.extern.slf4j.Slf4j;
import org.springframework.beans.BeanUtils;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.beans.factory.annotation.Value;
import org.springframework.core.ParameterizedTypeReference;
import org.springframework.dao.DuplicateKeyException;
import org.springframework.http.*;
import org.springframework.web.client.RestTemplate;
import org.springframework.web.reactive.function.client.ExchangeStrategies;
import org.springframework.web.reactive.function.client.WebClient;
import org.springframework.http.client.HttpComponentsClientHttpRequestFactory;

import com.baomidou.mybatisplus.core.conditions.query.LambdaQueryWrapper;
import com.baomidou.mybatisplus.core.conditions.update.LambdaUpdateWrapper;
@@ -37,16 +33,12 @@
import io.swagger.annotations.ApiOperation;
import reactor.core.publisher.Flux;
import reactor.core.publisher.Mono;
import reactor.core.publisher.SignalType;

import org.springframework.web.bind.annotation.* ;
import org.springframework.http.HttpStatus;
import com.linghu.model.dto.TaskResultResponse.QuestionResult;
import com.linghu.model.dto.TaskResultResponse.UserResult;
import reactor.core.scheduler.Schedulers;

import java.util.stream.Collectors;
import java.util.stream.IntStream;

@RestController
@RequestMapping("/collect")
@@ -75,220 +67,173 @@
    private PlatformService platformService;
    @Autowired
    private TypeService typeService;
    @Autowired
    private UserService userService;
    @Autowired
    private OrderService orderService;
    @Autowired
    private QuestionResultService questionResultService;

    // 替换为线程安全队列
    private static final Queue<SearchTaskRequest> taskQueue = new ConcurrentLinkedQueue<>();
    // 全局映射：关键词ID -> 批次队列
    private static final ConcurrentMap<Integer, Queue<List<UserDto>>> batchQueues = new ConcurrentHashMap<>();

 /*   @PostMapping("/search")
//    private static boolean isProcessing = false;
    private static volatile boolean isProcessing = false; // 添加 volatile

    @PostMapping("/search")
    @ApiOperation(value = "开始采集")
    public Mono<SearchTaskResponse> createSearchTask(
    public Mono<ResponseResult<?>> createSearchTask(
            @RequestBody SearchTaskRequest searchTaskRequest,
            HttpServletRequest request) throws JsonProcessingException {

        return webClient.post()
                .uri(baseUrl + "/api/v1/search")
                .contentType(MediaType.APPLICATION_JSON)
                .bodyValue(searchTaskRequest)
                .retrieve()
                .onStatus(HttpStatus::is4xxClientError, response -> response.bodyToMono(String.class)
                        .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody))))
                .bodyToMono(new ParameterizedTypeReference<SearchTaskResponse>() {
                })
                .flatMap(responseResult -> {
                    // 提取任务ID
                    SearchTaskResponse taskResponse = responseResult;
                    if (taskResponse != null && taskResponse.getTask_id() != null) {
                        // 保存任务ID到关键词
                        LambdaUpdateWrapper<Keyword> updateWrapper = new LambdaUpdateWrapper<>();
                        updateWrapper.eq(Keyword::getKeyword_id, searchTaskRequest.getKeyword_id());
                        updateWrapper.set(Keyword::getStatus,"Submitted");
                        updateWrapper.set(Keyword::getTask_id, taskResponse.getTask_id());
                        keywordService.update(updateWrapper);
        // 首先检查服务器资源
        return getServerResource()
                .flatMap(resourceResponse -> {
                    double cpuUsage = parseUsage(resourceResponse.getCpu_usage_percent());
                    double memoryUsage = parseUsage(resourceResponse.getMemory_usage_percent());

                        // 可选：更新响应中的其他信息
                    if (cpuUsage >= 90.0 || memoryUsage >= 90.0) {

                            String errorMsg = String.format("服务器资源不足,请稍后再试");

                        log.warn(errorMsg);
                        return Mono.just(ResponseResult.error(503, errorMsg));
                    }
                    return Mono.just(taskResponse);

                    // 将新的任务请求加入队列
                    taskQueue.add(searchTaskRequest);

                    // 如果当前没有任务在处理中，则启动任务队列的处理
                    if (!isProcessing) {
                        processNextTaskInQueue();
                    }

                    // 返回响应，通知用户任务已开始
                    return Mono.just(ResponseResult.success("任务已加入队列，正在处理..."));
                })
                .onErrorResume(e -> {
                    // return Mono.just(ResponseResult.error("调用失败: " + e.getMessage()));
                    SearchTaskResponse task = new SearchTaskResponse();
                    task.setMessage("调用失败: " + e.getMessage());
                    return Mono.just(task);
                    log.error("检查服务器资源失败: {}", e.getMessage(), e);
                    return Mono.just(ResponseResult.error("检查服务器资源失败: " + e.getMessage()));
                });
    }*/
    }

//    public SearchTaskController(WebClient.Builder webClientBuilder, KeywordService keywordService) {
//        this.webClient = webClientBuilder.build();
//        this.keywordService = keywordService;
//    }
    private void processNextTaskInQueue() {
        // 设置为正在处理
        isProcessing = true;

   /* @PostMapping("/search")
    @ApiOperation(value = "开始采集")
    public Mono<SearchTaskResponse> createSearchTask(
            @RequestBody SearchTaskRequest searchTaskRequest,
            HttpServletRequest request) throws JsonProcessingException {
        // 从队列中取出下一个任务
        SearchTaskRequest nextTaskRequest = taskQueue.poll();

        if (nextTaskRequest != null) {
            Integer keywordId = nextTaskRequest.getKeyword_id();
            log.info("开始处理任务队列，keywordId: {}", keywordId);

            executeBatchTask(nextTaskRequest)
                    .doFinally(signal -> {
                        isProcessing = false;
                        if (!taskQueue.isEmpty()) {
                            processNextTaskInQueue();
                        }
                    })
                    .subscribe(
                            result -> log.info("任务处理完成，keywordId: {}", keywordId), // 成功日志
                            error -> { // 关键：添加错误处理
                                log.error("任务队列处理异常，keywordId: {}", keywordId, error);
                            }
                    );
        } else {
            isProcessing = false; // 无任务时重置状态
        }
    }
    private Mono<ResponseResult<String>> executeBatchTask(SearchTaskRequest searchTaskRequest) {
        Integer keywordId = searchTaskRequest.getKeyword_id();
        //
        int maxConcurrentUsers = searchTaskRequest.getConfig() != null ?
                searchTaskRequest.getConfig().getMax_concurrent_users() : 3;
        List<List<UserDto>> userBatches = splitUsersIntoBatches(searchTaskRequest.getUsers(), maxConcurrentUsers);
        List<List<UserDto>> userBatches = splitUsersIntoBatches(searchTaskRequest.getUsers(), maxConcurrentUsers, keywordId,searchTaskRequest.getIs_first());

        return processBatchesSequentially(userBatches, searchTaskRequest)
                .onErrorResume(e -> {
                    SearchTaskResponse task = new SearchTaskResponse();
                    task.setMessage("调用失败: " + e.getMessage());
                    return Mono.just(task);

        // 创建批次队列并存入全局映射
        Queue<List<UserDto>> batchQueue = new ConcurrentLinkedQueue<>(userBatches);
        batchQueues.put(keywordId, batchQueue); // 存储到全局映射

        return Mono.just(ResponseResult.success("第一个批次已开始"))
                .doOnTerminate(() -> {
                    executeBatchTask(batchQueue, searchTaskRequest, keywordId)
                            .subscribe(
                                    result -> log.info("批次任务启动成功，keywordId: {}", keywordId),
                                    error -> { // 处理批次执行异常
                                        log.error("批次任务执行异常，keywordId: {}", keywordId, error);
                                        // 可选：异常时清理资源
                                        batchQueues.remove(keywordId);
                                    }
                            );
                });
    }
    private Mono<ResponseResult<?>> executeBatchTask(Queue<List<UserDto>> batchQueue, SearchTaskRequest searchTaskRequest, Integer keywordId) {
        // 如果队列为空，说明所有批次已经完成

    private List<List<UserDto>> splitUsersIntoBatches(List<UserDto> users, int batchSize) {
        List<List<UserDto>> batches = new ArrayList<>();
        for (int i = 0; i < users.size(); i += batchSize) {
            batches.add(users.subList(i, Math.min(i + batchSize, users.size())));
        if (batchQueue == null || batchQueue.isEmpty()) {
            // 清理资源
            batchQueues.remove(keywordId);
            return Mono.just(ResponseResult.success("所有批次已完成"));
        }
        return batches;
    }

    private Mono<SearchTaskResponse> processBatchesSequentially(List<List<UserDto>> userBatches, SearchTaskRequest originalRequest) {
        Mono<SearchTaskResponse> resultMono = Mono.empty();
        for (List<UserDto> batch : userBatches) {
            SearchTaskRequest batchRequest = new SearchTaskRequest();
            batchRequest.setUsers(batch);
            batchRequest.setQuestions(originalRequest.getQuestions());
            batchRequest.setConfig(originalRequest.getConfig());
            batchRequest.setSave_to_database(originalRequest.getSave_to_database());
            batchRequest.setWebhook_url(originalRequest.getWebhook_url());
            batchRequest.setKeyword_id(originalRequest.getKeyword_id());
        List<UserDto> currentBatch = batchQueue.poll(); // 从队列中获取当前批次
        SearchTaskRequest batchRequest = new SearchTaskRequest();
        batchRequest.setUsers(currentBatch);
        batchRequest.setQuestions(searchTaskRequest.getQuestions());
        batchRequest.setConfig(searchTaskRequest.getConfig());
        batchRequest.setSave_to_database(searchTaskRequest.getSave_to_database());
        batchRequest.setWebhook_url(searchTaskRequest.getWebhook_url());
        batchRequest.setKeyword_id(keywordId);

            resultMono = resultMono.then(createSingleBatchTask(batchRequest));
        }
        return resultMono;
    }

    private Mono<SearchTaskResponse> createSingleBatchTask(SearchTaskRequest batchRequest) {
        return webClient.post()
                .uri(baseUrl + "/api/v1/search")
                .contentType(MediaType.APPLICATION_JSON)
                .bodyValue(batchRequest)
                .retrieve()
                .onStatus(HttpStatus::is4xxClientError, response -> response.bodyToMono(String.class)
                        .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody))))
                .bodyToMono(new ParameterizedTypeReference<SearchTaskResponse>() {
                })
                .flatMap(responseResult -> {
                    SearchTaskResponse taskResponse = responseResult;
                    if (taskResponse != null && taskResponse.getTask_id() != null) {
                        LambdaUpdateWrapper<Keyword> updateWrapper = new LambdaUpdateWrapper<>();
                        updateWrapper.eq(Keyword::getKeyword_id, batchRequest.getKeyword_id());
                        updateWrapper.set(Keyword::getStatus, "Submitted");
                        updateWrapper.set(Keyword::getTask_id, taskResponse.getTask_id());
                        keywordService.update(updateWrapper);
                    }
                    return waitForTaskCompletion(taskResponse.getTask_id())
                            .then(Mono.just(taskResponse));
                });
    }

    private Mono<Void> waitForTaskCompletion(String taskId) {
        return Flux.interval(Duration.ofSeconds(5)) // 每5秒执行一次
                .flatMap(tick -> webClient.get()
                        .uri(baseUrl + "/api/v1/tasks/" + taskId)
                        .retrieve()
                        .bodyToMono(TaskStatusResponse.class)
                )
                .filter(response -> "completed".equals(response.getStatus()))
                .next() // 找到第一个完成的响应后结束流
                .then(); // 转换为Mono<Void>
    }*/
   @PostMapping("/search")
   @ApiOperation(value = "开始采集")
   public Mono<List<SearchTaskResponse>> createSearchTask(
           @RequestBody SearchTaskRequest searchTaskRequest,
           HttpServletRequest request) throws JsonProcessingException {

       int maxConcurrentUsers = searchTaskRequest.getConfig() != null ?
               searchTaskRequest.getConfig().getMax_concurrent_users() : 3;
       List<List<UserDto>> userBatches = splitUsersIntoBatches(searchTaskRequest.getUsers(), maxConcurrentUsers);

       // 获取 keywordId
       Integer keywordId = searchTaskRequest.getKeyword_id();

       return Flux.fromIterable(userBatches)
               .flatMap(batch -> {
                   SearchTaskRequest batchRequest = new SearchTaskRequest();
                   batchRequest.setUsers(batch);
                   batchRequest.setQuestions(searchTaskRequest.getQuestions());
                   batchRequest.setConfig(searchTaskRequest.getConfig());
                   batchRequest.setSave_to_database(searchTaskRequest.getSave_to_database());
                   batchRequest.setWebhook_url(searchTaskRequest.getWebhook_url());
                   batchRequest.setKeyword_id(keywordId);

                   return createSingleBatchTask(batchRequest)
                           .delaySubscription(Duration.ofSeconds(2)); // 批次之间添加延迟
               }, 1) // 限制并发数为1，确保顺序执行
               .collectList() // 收集所有批次的响应
               .flatMap(responses ->

                       saveKeywordTasks(keywordId, responses) // 保存关联关系
                               .thenReturn(responses) // 返回原始响应
               );
   }

    private Mono<Void> saveKeywordTasks(Integer keywordId, List<SearchTaskResponse> taskResponses) {
        List<KeywordTask> keywordTasks = taskResponses.stream()
                .filter(response -> response.getTask_id() != null)
                .map(response -> {
                    KeywordTask keywordTask = new KeywordTask();
                    keywordTask.setKeyword_id(keywordId);
                    keywordTask.setTask_id(response.getTask_id());
                    return keywordTask;
                })
                .collect(Collectors.toList());

        // 将 MyBatis-Plus 的同步方法包装为 Mono<Void>
        return Mono.fromRunnable(() -> {
                    boolean success = keywordTaskService.saveOrUpdateBatch(keywordTasks);
                    if (!success) {
                        throw new RuntimeException("保存关键词任务关联失败");
                    }
                })
                .doFinally(signalType -> log.info("成功保存 {} 个关键词任务关联", keywordTasks.size()))
                .then();
    }

    private List<List<UserDto>> splitUsersIntoBatches(List<UserDto> users, int batchSize) {
        List<List<UserDto>> batches = new ArrayList<>();
        for (int i = 0; i < users.size(); i += batchSize) {
            batches.add(users.subList(i, Math.min(i + batchSize, users.size())));
        }
        return batches;
    }

    private Mono<SearchTaskResponse> createSingleBatchTask(SearchTaskRequest batchRequest) {
        return webClient.post()
                .uri(baseUrl + "/api/v1/search")
                .contentType(MediaType.APPLICATION_JSON)
                .bodyValue(batchRequest)
                .retrieve()
                .onStatus(HttpStatus::is4xxClientError, response -> response.bodyToMono(String.class)
                        .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody))))
                .bodyToMono(new ParameterizedTypeReference<SearchTaskResponse>() {})
        return createSingleBatchTask(batchRequest)
                .flatMap(taskResponse -> {
                    if (taskResponse != null && taskResponse.getTask_id() != null) {

                        // 使用 Reactor 的方式更新数据库
                        return Mono.fromRunnable(() -> {
                                    LambdaUpdateWrapper<Keyword> updateWrapper = new LambdaUpdateWrapper<>();
                                    updateWrapper.eq(Keyword::getKeyword_id, batchRequest.getKeyword_id());
                                    updateWrapper.set(Keyword::getStatus, "submitted");
                                    updateWrapper.set(Keyword::getTask_id, taskResponse.getTask_id());
                                    keywordService.update(updateWrapper);
                                }).subscribeOn(Schedulers.boundedElastic()) // 在弹性线程池执行
                                .thenReturn(taskResponse);
                        // 直接等待任务完成，不再保存任务关联信息
                        return waitForTaskCompletion(taskResponse.getTask_id(), batchQueue, searchTaskRequest, keywordId);
                    } else {
                        return Mono.just(ResponseResult.error("创建批次任务失败"));
                    }
                    return Mono.just(taskResponse);
                })
                .onErrorResume(e -> {
                    log.error("调用第三方接口失败: {}", e.getMessage(), e); // 关键日志
                    return Mono.error(new RuntimeException("调用第三方接口失败: " + e.getMessage()));
                })
                .doFinally(signal -> {
                    // 任务完成时清理资源
                    if (batchQueue.isEmpty()) {
                        batchQueues.remove(keywordId);
                    }
                });
    }

    // 移除原来的waitForTaskCompletion方法，不再需要同步等待
    private Mono<ResponseResult<?>> waitForTaskCompletion(String taskId, Queue<List<UserDto>> batchQueue, SearchTaskRequest searchTaskRequest, Integer keywordId) {
        // 查询任务状态
        return getTaskStatus(taskId)
                .flatMap(statusResponse -> {
                    // 检查任务是否被取消
                    if ("cancelled".equalsIgnoreCase(statusResponse.getStatus())) {
                        batchQueues.remove(keywordId); // 清理资源
                        return Mono.just(ResponseResult.success("任务已被取消"));
                    }
                    // 如果任务状态是"submitted"或"running"，继续轮询
                    if (!"completed".equalsIgnoreCase(statusResponse.getStatus()) && !"failed".equalsIgnoreCase(statusResponse.getStatus()) && !"cancelled".equalsIgnoreCase(statusResponse.getStatus()) && !("ERROR".equalsIgnoreCase(statusResponse.getStatus()) && statusResponse.getMessage().contains("Task not found")) ) {
                        return Mono.delay(Duration.ofSeconds(5))  // 延迟 5 秒后再次查询
                                .flatMap(aLong -> waitForTaskCompletion(taskId, batchQueue, searchTaskRequest, keywordId));  // 递归调用继续等待
                    } else {
                        // 如果状态为其他状态，则继续处理下一个批次
                        return executeBatchTask(batchQueue, searchTaskRequest, keywordId);
                    }
                })
                .onErrorResume(e -> {
                    // 处理查询任务状态时的错误
                    return Mono.just(ResponseResult.error("查询任务状态失败: " + e.getMessage()));
                });
    }
    @ApiOperation(value = "查询任务状态")
    @GetMapping("/status")
    public Mono<TaskStatusResponse> getTaskStatus(String taskId) {
@@ -299,63 +244,320 @@
                .onStatus(HttpStatus::isError, response -> response.bodyToMono(TaskStatusResponse.class)
                        .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody.getDetail()))))
                .bodyToMono(TaskStatusResponse.class)
                .flatMap(result -> {
                    TaskStatusResponse taskStatusResponse = result;
                    if (taskStatusResponse != null && taskStatusResponse.getStatus() != null) {
                        List<Question> updateQuestions = taskStatusResponse.getQuestions_status().stream()
                                .map(qs -> {
                                    Question question = new Question();
                                    question.setQuestion_id(qs.getQuestion_id());
                                    question.setStatus(qs.getStatus());
                                    return question;
                                }).collect(Collectors.toList());

                        // 包装成响应式操作
                        return Mono.fromCallable(() -> {
                            questionService.updateBatchById(updateQuestions);
                            return result;
                        });

                    }
                    return Mono.just(result);
                })
                .onErrorResume(e -> {
                    // 创建一个自定义的错误响应对象
                    // 处理错误，创建一个自定义的错误响应对象
                    TaskStatusResponse errorResponse = new TaskStatusResponse();
                    errorResponse.setStatus("ERROR");
                    errorResponse.setMessage(e.getMessage());
                    errorResponse.setDetail(e.getMessage());

                    return Mono.just(errorResponse);
                });
    }

    @PostMapping("/cancel/{taskId}")
    @ApiOperation(value = "取消任务")
    public Mono<ResponseResult<TaskCancelResponse>> cancelTask(@PathVariable String taskId) {
   // 添加一个辅助方法来安全地将字符串转换为double
   private double parseUsage(String usageStr) {
       try {
           if (usageStr != null) {
               // 移除可能存在的百分号
               usageStr = usageStr.replace("%", "").trim();
               return Double.parseDouble(usageStr);
           }
           return 0.0;
       } catch (NumberFormatException e) {
           log.error("解析资源使用率失败: {}", e.getMessage());
           return 0.0;
       }
   }

    private List<List<UserDto>> splitUsersIntoBatches(List<UserDto> users, int batchSize,Integer keywordId,Boolean isFirst) {

        Keyword keyword = keywordService.getById(keywordId);
        if (isFirst){
            keyword.setNum(1);
        }else {
            keyword.setNum(keyword.getNum()+1);
        }
        keywordService.updateById(keyword);

        List<List<UserDto>> batches = new ArrayList<>();
        for (int i = 0; i < users.size(); i += batchSize) {
            batches.add(users.subList(i, Math.min(i + batchSize, users.size())));

        }
        for (int i = 0; i < batches.size(); i++){
            // 创建 KeywordTask 关联，task_id 设置为 null，表示任务尚未开始
            KeywordTask keywordTask = new KeywordTask();
            keywordTask.setKeyword_id(keywordId);
            keywordTask.setTask_id(null);  // 任务ID为空

            keywordTask.setNum(keyword.getNum());
            keywordTaskService.save(keywordTask);  // 保存 KeywordTask
        }


        return batches;
    }

    private Mono<SearchTaskResponse> createSingleBatchTask(SearchTaskRequest batchRequest) {
        // 记录请求第三方的基本信息（便于排查）
        String thirdPartyUrl = baseUrl + "/api/v1/search";
        Integer keywordId = batchRequest.getKeyword_id();
        log.info("开始向第三方提交任务，keywordId: {}, URL: {}, 请求参数: {}",
                keywordId, thirdPartyUrl, batchRequest.toString()); // 打印请求参数（建议用工具类转JSON）
        return webClient.post()
                .uri(baseUrl + "/api/v1/tasks/" + taskId + "/cancel")
                .uri(baseUrl + "/api/v1/search")
                .contentType(MediaType.APPLICATION_JSON)
                .bodyValue(Collections.emptyMap()) // 添加空请求体
                .bodyValue(batchRequest)
                .retrieve()
                .onStatus(HttpStatus::isError, response -> response.bodyToMono(TaskCancelResponse.class)
                        .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody.getDetail()))))
                .bodyToMono(TaskCancelResponse.class)
                .map(data -> ResponseResult.success(data))
                .onErrorResume(e -> {
                    if (e.getMessage().contains("任务不存在")) {
                        return Mono.just(ResponseResult.error(200, "任务不存在"));
                    } else if (e.getMessage().contains("无法取消")) {
                        return Mono.just(ResponseResult.error(200, "任务已完成，无法取消"));
                .onStatus(HttpStatus::is4xxClientError, response -> response.bodyToMono(String.class)
                        .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody))))
                // 处理第三方返回的5xx服务器错误（如第三方服务异常）
                .onStatus(HttpStatus::is5xxServerError, response ->
                        response.bodyToMono(String.class)
                                .flatMap(errorBody -> {
                                    String errorMsg = String.format("第三方接口5xx错误，keywordId: %d, URL: %s, 状态码: %d, 错误详情: %s",
                                            keywordId, thirdPartyUrl, response.statusCode().value(), errorBody);
                                    log.error(errorMsg);
                                    return Mono.error(new RuntimeException(errorMsg));
                                })
                )
                .bodyToMono(new ParameterizedTypeReference<SearchTaskResponse>() {})
                .flatMap(taskResponse -> {
                    if (taskResponse != null && taskResponse.getTask_id() != null) {

                        // 使用 Reactor 的方式更新数据库
                        return Mono.fromRunnable(() -> {

                                     //更新关键词状态
                                    LambdaUpdateWrapper<Keyword> updateWrapper = new LambdaUpdateWrapper<>();
                                    updateWrapper.eq(Keyword::getKeyword_id, batchRequest.getKeyword_id());
                                    updateWrapper.set(Keyword::getStatus, "submitted");
                                    updateWrapper.set(Keyword::getTask_id, taskResponse.getTask_id());
                                    keywordService.update(updateWrapper);
                                    //设置轮数
                                    Keyword keyword = keywordService.getById(batchRequest.getKeyword_id());
                                    // 更新关键词任务与任务ID的关联
                                    // 获取与关键词相关的任务，task_id 为 null，确保只取一个任务
                                    List<KeywordTask> keywordTasks = keywordTaskService.list(new LambdaQueryWrapper<KeywordTask>()
                                            .eq(KeywordTask::getKeyword_id, keyword.getKeyword_id())
                                            .eq(KeywordTask::getNum, keyword.getNum())
                                            .isNull(KeywordTask::getTask_id));
                                    if (keywordTasks.size() > 0) {
                                        KeywordTask keywordTask = keywordTasks.get(0);
                                        keywordTask.setTask_id(taskResponse.getTask_id());
                                        keywordTask.setStatus("pending");
                                        keywordTaskService.updateById(keywordTask);
                                    }
                                    //将提问词列表的状态转为pending
                                    for (String questionName : batchRequest.getQuestions()) {
                                        questionService.update(new LambdaUpdateWrapper<Question>().eq(Question::getKeyword_id, keyword.getKeyword_id()).eq(Question::getQuestion,questionName).set(Question::getStatus, "pending"));

                                    }
                                    //所有关键词都在采集中或者已完成或者错误设置订单进入采集状态
                                    List<Keyword> orderKeywords = keywordService.list(new LambdaQueryWrapper<Keyword>()
                                            .eq(Keyword::getOrder_id, keyword.getOrder_id()));
                                    if (!orderKeywords.isEmpty() && orderKeywords.stream().allMatch(k ->
                                            "completed".equals(k.getStatus()) || "submitted".equals(k.getStatus())
                                    )) {
                                        Orders orders = orderService.getById(keyword.getOrder_id());
                                        if (orders != null) {
                                            orders.setStatus(2);
                                            orderService.updateById(orders);

                                        }
                                    }
                                }).subscribeOn(Schedulers.boundedElastic()) // 在弹性线程池执行
                                .thenReturn(taskResponse);
                    }
                    return Mono.just(ResponseResult.error(500, "取消任务失败: " + e.getMessage()));
                    return Mono.just(taskResponse);
                });
    }

    @PostMapping("/cancel/{keywordId}")
    @ApiOperation(value = "取消任务")
    public Mono<ResponseResult<TaskCancelResponse>> cancelTask(@PathVariable Integer keywordId) {
        // 1. 从主队列移除任务
        List<SearchTaskRequest> removedMainQueueTasks = removeTasksFromQueueByKeywordId(keywordId);
        int removedMainQueueCount = removedMainQueueTasks.size(); // 获取移除的任务数量

        // 2. 从批次队列移除任务 (新增逻辑)
        int removedBatchQueue = removeBatchTasksByKeywordId(keywordId);

        // 3. 查询所有与关键词相关的任务
        List<KeywordTask> tasks = keywordTaskService.list(
                new LambdaQueryWrapper<KeywordTask>().eq(KeywordTask::getKeyword_id, keywordId)
        );

        // 4. 筛选出需要远程取消的任务
        List<KeywordTask> tasksToCancelRemotely = tasks.stream()
                .filter(task -> task.getTask_id() != null && "pending".equalsIgnoreCase(task.getStatus()))
                .collect(Collectors.toList());

        return Flux.fromIterable(tasksToCancelRemotely)
                .flatMap(task -> {
                    // 创建状态更新和远程取消的组合操作
                    Mono<Void> updateStatus = updateTaskStatus(task.getTask_id(), "cancelled");
                    Mono<ResponseResult<?>> cancelOp = cancelRemoteTask(task.getTask_id())
                            .onErrorResume(e -> {
                                log.error("取消任务 {} 失败: {}", task.getTask_id(), e.getMessage());
                                return Mono.just(ResponseResult.error("取消任务失败: " + e.getMessage()));
                            });

                    return Mono.zip(cancelOp, updateStatus)
                            .thenReturn(true);
                }, 10)
                .collectList()
                .flatMap(canceledTasks -> {
                    return updateKeywordAndOrderStatus(keywordId)
                            .thenReturn(ResponseResult.success(
                                    new TaskCancelResponse(
                                            String.format("任务已取消: 主队列移除%d, 批次队列移除%d, 远程取消%d",
                                                    removedMainQueueCount ,
                                                    removedBatchQueue,
                                                    tasksToCancelRemotely.size())
                                    )
                            ));
                });
    }

    // 新增方法：移除批次队列
    private int removeBatchTasksByKeywordId(Integer keywordId) {
        Queue<List<UserDto>> batchQueue = batchQueues.remove(keywordId);
        if (batchQueue != null) {
            int count = batchQueue.size();
            batchQueue.clear();
            log.info("从批次队列中移除关键词 {} 的 {} 个批次任务", keywordId, count);
            return count;
        }
        return 0;
    }
    // 辅助方法：获取待取消任务
    private List<KeywordTask> getTasksToCancel(Integer keywordId) {
        return keywordTaskService.list(
                new LambdaQueryWrapper<KeywordTask>()
                        .eq(KeywordTask::getKeyword_id, keywordId)
                        .isNotNull(KeywordTask::getTask_id)
                        .eq(KeywordTask::getStatus, "pending")
        );
    }

    // 提取关键词和订单状态更新的逻辑为单独方法
    private Mono<Void> updateKeywordAndOrderStatus(Integer keywordId) {
        return Mono.fromRunnable(() -> {
            try {
                // 查询关键词
                Keyword keyword = keywordService.getById(keywordId);
                if (keyword == null) {
                    log.warn("未找到关键词，keywordId: {}", keywordId);
                    return;
                }

                //把任务id为空的删除
                LambdaUpdateWrapper<KeywordTask> updateWrapper = new LambdaUpdateWrapper<>();
                updateWrapper.eq(KeywordTask::getKeyword_id, keywordId);
                updateWrapper.isNull(KeywordTask::getTask_id);
                keywordTaskService.remove(updateWrapper);
                // 查询该关键词下的所有任务
                LambdaQueryWrapper<KeywordTask> keywordTaskWrapper = new LambdaQueryWrapper<>();
                keywordTaskWrapper.eq(KeywordTask::getKeyword_id, keywordId);

                List<KeywordTask> keywordTasks = keywordTaskService.list(keywordTaskWrapper);

                // 更新关键词状态
                keyword.setStatus("completed");
                keywordService.updateById(keyword);

                //更新提问词状态为取消
                questionService.update(new LambdaUpdateWrapper<Question>().eq(Question::getKeyword_id, keywordId).isNull(Question::getResponse).set(Question::getStatus, "cancelled"));

                // 更新订单状态
                String orderId = keyword.getOrder_id();
                if (orderId != null && !orderId.isEmpty()) {
                    // 查询订单下所有关键词
                    LambdaQueryWrapper<Keyword> orderKeywordsWrapper = new LambdaQueryWrapper<>();
                    orderKeywordsWrapper.eq(Keyword::getOrder_id, orderId);
                    List<Keyword> orderKeywords = keywordService.list(orderKeywordsWrapper);

                    // 所有关键词均已完成，则更新订单状态为3
                    if (!orderKeywords.isEmpty() && orderKeywords.stream().allMatch(k ->
                            "completed".equals(k.getStatus()) || "false".equals(k.getStatus())
                    )) {
                        Orders orders = orderService.getById(orderId);
                        if (orders != null) {
                            orders.setStatus(3);
                            orderService.updateById(orders);
                            log.info("订单 {} 所有关键词已完成，更新状态为3", orderId);
                        }
                    }
                    if (!orderKeywords.isEmpty() && orderKeywords.stream().allMatch(k ->
                            !"completed".equals(k.getStatus()) || "false".equals(k.getStatus()) || "cancelled".equals(k.getStatus())
                    )) {
                        Orders orders = orderService.getById(orderId);
                        if (orders != null) {
                            orders.setStatus(1);
                            orderService.updateById(orders);
                            log.info("订单 {} 所有关键词已完成或者取消，更新状态为1", orderId);
                        }
                    }
                }
            } catch (Exception e) {
                log.error("更新关键词和订单状态失败: {}", e.getMessage(), e);
            }
        });
    }
private List<SearchTaskRequest> removeTasksFromQueueByKeywordId(Integer keywordId) {
    List<SearchTaskRequest> removedTasks = new ArrayList<>();

        Iterator<SearchTaskRequest> iterator = taskQueue.iterator();
        while (iterator.hasNext()) {
            SearchTaskRequest task = iterator.next();
            if (task.getKeyword_id() != null && task.getKeyword_id().equals(keywordId)) {
                removedTasks.add(task);
                iterator.remove();
            }
        }


    log.info("从队列中移除了 {} 个与关键词ID {} 相关的任务", removedTasks.size(), keywordId);
    return removedTasks;
}

// 发送远程取消请求
    private Mono<ResponseResult<?>> cancelRemoteTask(String taskId) {
        // 使用Collections.singletonMap或手动创建Map
        Map<String, Object> requestBody = new HashMap<>();
        requestBody.put("status", "pending");

        return webClient.post()
                .uri(baseUrl + "/api/v1/tasks/" + taskId + "/cancel")
                .contentType(MediaType.APPLICATION_JSON)
                .bodyValue(requestBody)
                .retrieve()
                .onStatus(HttpStatus::isError, response -> response.bodyToMono(String.class)
                        .flatMap(errorBody -> Mono.error(new RuntimeException("取消失败: " + errorBody))))
                .bodyToMono(Void.class)
                .thenReturn(ResponseResult.success("任务已取消"));
    }

    // 更新单个任务状态
    private Mono<Void> updateTaskStatus(String taskId, String status) {
        return Mono.fromRunnable(() -> {
            LambdaUpdateWrapper<KeywordTask> updateWrapper = new LambdaUpdateWrapper<>();
            updateWrapper.eq(KeywordTask::getTask_id, taskId);
            updateWrapper.set(KeywordTask::getStatus, status);
            keywordTaskService.update(updateWrapper);
        }).subscribeOn(Schedulers.boundedElastic()).then();
    }
    @ApiOperation(value = "获取任务结果")
    @GetMapping("/tasks/{taskId}")
    public Mono<TaskResultResponse> getTaskResult(@PathVariable String taskId) {
        return webClient.get()
        WebClient webClient2 = WebClient.builder()
                .exchangeStrategies(ExchangeStrategies.builder()
                        .codecs(configurer -> configurer.defaultCodecs()
                                .maxInMemorySize(10 * 1024 * 1024)) // 10MB
                        .build())
                .build();
        return webClient2.get()
                .uri(baseUrl + "/api/v1/tasks/" + taskId + "/result")
                .accept(MediaType.APPLICATION_JSON)
                .retrieve()
@@ -372,6 +574,7 @@
                .bodyToMono(TaskResultResponse.class)
                .flatMap(responseResult -> {
                    TaskResultResponse result = responseResult;

                    if (result != null && result.getResults() != null) {
                        return updateQuestionAndReference(result)
                                .thenReturn(responseResult);
@@ -380,97 +583,141 @@
                })
                .onErrorResume(e -> {
                    System.out.println("获取任务结果失败");
                    TaskResultResponse result = new TaskResultResponse();
                    log.error("获取任务结果失败: {}", e.getMessage(), e);
                        TaskResultResponse result = new TaskResultResponse();
                    result.setDetail("获取任务结果失败: " + e.getMessage());
                    return Mono.just(result);
                });
    }

//    private Mono<Void> updateQuestionAndReference(TaskResultResponse result) {
//        return Mono.fromRunnable(() -> {
//            // 1. 更新关键词状态
//            LambdaUpdateWrapper<Keyword> keywordUpdate = new LambdaUpdateWrapper<>();
//            keywordUpdate.eq(Keyword::getTask_id, result.getTask_id())
//                    .set(Keyword::getStatus, "completed");
//            keywordService.update(keywordUpdate);
//
//            // 查询关键词ID
//            LambdaQueryWrapper<Keyword> keywordQuery = new LambdaQueryWrapper<>();
//            keywordQuery.eq(Keyword::getTask_id, result.getTask_id());
//            Keyword keyword = keywordService.getOne(keywordQuery);
//
//            if (keyword == null) {
//                System.out.println("未找到关联的关键词，task_id: " + result.getTask_id());
//                return;
//            }
//
//            // 2. 处理每个用户的问题结果
//            for (UserResult userResult : result.getResults()) {
//                for (QuestionResult questionResult : userResult.getQuestions_results()) {
//                    // 2.1 查询问题ID
//                    LambdaQueryWrapper<Question> queryWrapper = new LambdaQueryWrapper<>();
//                    queryWrapper.eq(Question::getQuestion, questionResult.getQuestion())
//                            .eq(Question::getKeyword_id, keyword.getKeyword_id());
//                    Question question = questionService.getOne(queryWrapper);
//
//                    if (question != null) {
//                        // 更新问题状态
//                        LambdaUpdateWrapper<Question> updateWrapper = new LambdaUpdateWrapper<>();
//                        updateWrapper.eq(Question::getQuestion_id, question.getQuestion_id())
//                                .set(Question::getStatus, questionResult.getStatus())
//                                .set(Question::getResponse, questionResult.getResponse())
//                                .set(Question::getExtracted_count, questionResult.getExtracted_count())
//                                .set(Question::getError, questionResult.getError())
//                                .set(Question::getTimestamp, LocalDateTime.parse(
//                                        questionResult.getTimestamp(),
//                                        DateTimeFormatter.ofPattern("yyyy-MM-dd'T'HH:mm:ss.SSSSSS")
//                                ));
//                        questionService.update(updateWrapper);
//
//                        // 2.2 保存引用数据
//                        List<Reference> references = questionResult.getReferences().stream()
//                                .map(ref -> {
//                                    Reference reference = new Reference();
//                                    reference.setQuestion_id(question.getQuestion_id());
//                                    reference.setTitle(ref.getTitle());
//                                    reference.setUrl(ref.getUrl());
//                                    reference.setDomain(ref.getDomain());
//                                    reference.setCreate_time(LocalDateTime.now());
//                                    return reference;
//                                })
//                                .collect(Collectors.toList());
//
//                        if (!references.isEmpty()) {
//                            referenceService.saveBatch(references);
//                        }
//                    } else {
//                        System.out.println("未找到匹配的问题，question " + question.getQuestion());
//
//                    }
//                }
//            }
//        });
//    }
    /**
     * 获取或创建平台（确保同一domain只创建一次）
     * @param domain 平台域名
     * @return 已存在或新创建的Platform
     */
    private Platform getOrCreatePlatform(String domain,String platformName) {
        // 1. 先尝试查询已存在的平台

        Platform platform = platformService.getPlatformByDomain(domain);
        if (platform != null) {
            return platform;
        }

        // 2. 若不存在，尝试创建（处理并发场景）
        try {
            // 2.1 获取或创建“默认”类型（Type也需避免重复，建议Type表的type_name也加唯一约束）
            Type defaultType = typeService.getOne(new LambdaQueryWrapper<Type>()
                    .eq(Type::getType_name, "默认"));
            if (defaultType == null) {
                defaultType = new Type();
                defaultType.setType_name("默认");
                typeService.save(defaultType); // 若Type可能重复，此处也需处理DuplicateKeyException
            }

            // 2.2 构建新平台对象
            Platform newPlatform = new Platform();
            newPlatform.setDomain(domain);
            if (platformName != null) {
                newPlatform.setPlatform_name(platformName);
            }else {
                newPlatform.setPlatform_name(domain);
            }
        // 平台名称默认使用域名，可根据实际需求调整
            newPlatform.setType_id(defaultType.getType_id());
            newPlatform.setCreate_time(LocalDateTime.now()); // 补充创建时间

            // 2.3 尝试保存，若因唯一约束冲突失败，则捕获异常
            platformService.save(newPlatform);
            return newPlatform; // 保存成功，返回新创建的平台

        } catch (DuplicateKeyException e) {
            // 3. 若捕获到重复键异常，说明并发创建了，重新查询即可（此时数据库中已存在该平台）
            log.warn("平台domain={}已存在，无需重复创建", domain, e);
            return platformService.getPlatformByDomain(domain); // 重新查询，一定能获取到
        } catch (Exception e) {
            // 处理其他异常（如数据库连接失败等）
            log.error("创建平台失败，domain={}", domain, e);
            throw new RuntimeException("创建平台失败", e);
        }
    }
    //更新提问词和引用数据
    private Mono<Void> updateQuestionAndReference(TaskResultResponse result) {
        return Mono.fromRunnable(() -> {
            try {
                //查看每个账号信息的status是否正常

                // 1. 根据KeywordTask更新关键词状态
                // 查询关键词ID
                LambdaQueryWrapper<KeywordTask> keywordTaskWrapper = new LambdaQueryWrapper<>();
                keywordTaskWrapper.eq(KeywordTask::getTask_id, result.getTask_id());
                KeywordTask keywordTask = keywordTaskService.getOne(keywordTaskWrapper);

                keywordTask.setStatus("completed");
                keywordTaskService.updateById(keywordTask);
                Keyword keyword = keywordService.getById(keywordTask.getKeyword_id());

                if (keyword == null) {
                    System.out.println("未找到关联的关键词，task_id: " + result.getTask_id());
                    //报错
                    throw new Exception("未找到关联的关键词，task_id: " + result.getTask_id());
//                    return;

                }
                keyword.setStatus("completed");
                keywordService.updateById(keyword);
                LambdaQueryWrapper<KeywordTask> keywordTaskWrapper2 = new LambdaQueryWrapper<>();
                keywordTaskWrapper2.eq(KeywordTask::getKeyword_id, keyword.getKeyword_id());
                List<KeywordTask> keywordTasks = keywordTaskService.list(keywordTaskWrapper2);

                //如果全部为completed 或者错误、取消、任务不存在 关键词也为completed  ，如果关联关系没有任务id，或者状态为running  ，关键词为submitted，
                if (keywordTasks.stream().allMatch(task -> "completed".equals(task.getStatus()) || "false".equals(task.getStatus()) || "cancelled".equals(task.getStatus()) ||"canceled".equals(task.getStatus()) || "nonentity".equals(task.getStatus())) ) {
                    keyword.setStatus("completed");
                    keywordService.updateById(keyword);

                }

// 更新关键词状态
                    String orderId = keyword.getOrder_id();
                    if (orderId == null || orderId.isEmpty()) {
                        System.out.println("关键词[" + keyword.getKeyword_id() + "]未关联订单，跳过订单状态更新");
                        return;
                    }

                    // 2. 查询该订单下的所有关键词
                    LambdaQueryWrapper<Keyword> orderKeywordsWrapper = new LambdaQueryWrapper<>();
                    orderKeywordsWrapper.eq(Keyword::getOrder_id, orderId);
                    List<Keyword> orderKeywords = keywordService.list(orderKeywordsWrapper);

                    if (orderKeywords.isEmpty()) {
                        System.out.println("订单[" + orderId + "]下无关键词，跳过状态更新");
                        return;
                    }
                boolean  allValid2 = orderKeywords.stream()
                        .allMatch(k -> "completed".equals(k.getStatus()) || "false".equals(k.getStatus()) || "cancelled".equals(k.getStatus()));
                if (allValid2) {
                    Orders orders = orderService.getById(orderId);
                    if (orders != null) {
                        orders.setStatus(1); // 假设Orders有Integer类型的status字段
                        orderService.updateById(orders);
                        System.out.println("订单[" + orderId + "]所有关键词采集完成或者取消，已更新状态为1");
                    } else {
                        System.out.println("未找到订单[" + orderId + "]，无法更新状态");
                    }
                }
                    // 3. 检查所有关键词的状态是否均为 completed 或 false
                    boolean allValid = orderKeywords.stream()
                            .allMatch(k -> "completed".equals(k.getStatus()) || "false".equals(k.getStatus()));

                    // 4. 若所有关键词状态均有效，更新订单状态为3
                    if (allValid) {
                        Orders orders = orderService.getById(orderId);
                        if (orders != null) {
                            orders.setStatus(3); // 假设Orders有Integer类型的status字段
                            orderService.updateById(orders);
                            System.out.println("订单[" + orderId + "]所有关键词状态符合条件，已更新状态为3");
                        } else {
                            System.out.println("未找到订单[" + orderId + "]，无法更新状态");
                        }
                    }


                Orders orders = orderService.getById(keyword.getOrder_id());

                // 2. 批量查询所有问题
                LambdaQueryWrapper<Question> queryWrapper = new LambdaQueryWrapper<>();
@@ -484,91 +731,176 @@
                // 3. 收集所有需要更新的问题和引用
                List<Question> questionsToUpdate = new ArrayList<>();
                List<Reference> allReferences = new ArrayList<>();
                List<Reference> resultList = new ArrayList<>();

                // 遍历结果
                for (UserResult userResult : result.getResults()) {
                    //更新账号状态
                    if ( "failed".equals(userResult.getStatus())){
                        if (userResult.getError().contains("登录失败")){
                            LambdaUpdateWrapper<User> userWrapper =  new LambdaUpdateWrapper<>();
                            userWrapper.eq(User::getUser_email, userResult.getUser_email());
                            userWrapper.set(User::getStatus, "无法登录");
                            userService.update(userWrapper);
                            //更新所有提问词的状态
                            questionService.update(new LambdaUpdateWrapper<Question>().eq(Question::getKeyword_id, keyword.getKeyword_id())
                                    .set(Question::getStatus, "failed")
                                    .set(Question::getError, "账户登录失败"));

                        }else if (userResult.getError().contains("信息错误")){
                            LambdaUpdateWrapper<User> userWrapper =  new LambdaUpdateWrapper<>();
                            userWrapper.eq(User::getUser_email, userResult.getUser_email());
                            userWrapper.set(User::getStatus, "信息错误");
                            userService.update(userWrapper);
                        }
                    }
                    for (QuestionResult questionResult : userResult.getQuestions_results()) {
                        try {
                            Question question = questionMap.get(questionResult.getQuestion());
                            if (question != null) {


                                //保存问题结果
                                QuestionResultList questionResultList = new QuestionResultList();
                                questionResultList.setKeyword_id(keyword.getKeyword_id());
                                questionResultList.setQuestion(questionResult.getQuestion());
                                questionResultList.setResponse(questionResult.getResponse());
                                questionResultList.setStatus(questionResult.getStatus());
                                questionResultList.setExtracted_count(questionResult.getExtracted_count());
                                questionResultList.setKeyword_task_id(result.getTask_id());
                                questionResultList.setError(questionResult.getError());
                                questionResultList.setNum(keyword.getNum());
                                if (questionResult.getTimestamp() != null) {
                                    DateTimeFormatter formatter = DateTimeFormatter
                                            .ofPattern("yyyy-MM-dd'T'HH:mm:ss.SSSSSS");
                                    questionResultList.setTimestamp(
                                            LocalDateTime.parse(questionResult.getTimestamp(), formatter));
                                }
                                // 保存问题结果列表（新增保存逻辑）
                                questionResultService.save(questionResultList);
                                // 查询当前轮次下该提问词的所有结果
                                List<QuestionResultList> allResults = questionResultService.list(
                                        new LambdaQueryWrapper<QuestionResultList>()
                                                .eq(QuestionResultList::getKeyword_id, keyword.getKeyword_id())
                                                .eq(QuestionResultList::getQuestion, question.getQuestion())
                                                .eq(QuestionResultList::getNum, keyword.getNum())
                                );

                                // 判断最终状态
                                String finalStatus = determineFinalStatus(allResults);
                                if ("success".equals(finalStatus)){
                                    question.setStatus("success");
                                    question.setError("");
                                }else if ("no_results".equals(finalStatus)){
                                    question.setStatus("success");
                                    question.setError("采集结果无引用数据");
                                }else if ("busyness".equals(finalStatus)){
                                    question.setStatus("failed");
                                    question.setError("DeepSeek繁忙，请稍后尝试");
                                }

                                // 更新问题对象
                                question.setStatus(questionResult.getStatus());
                                question.setResponse(questionResult.getResponse());
                                question.setExtracted_count(questionResult.getExtracted_count());
                                question.setError(questionResult.getError());
//                                question.setError(questionResult.getError());
                                question.setKeyword_id(keyword.getKeyword_id());

                                // 解析时间戳
                                if (questionResult.getTimestamp() != null) {
                                    DateTimeFormatter formatter = DateTimeFormatter
                                            .ofPattern("yyyy-MM-dd'T'HH:mm:ss.SSSSSS");
                                    question.setTimestamp(
                                            LocalDateTime.parse(questionResult.getTimestamp(), formatter));
                                }
                                //更新
//                                questionService.updateById(question);

                                questionsToUpdate.add(question);
                                // 初始化引用列表（避免null）
                                List<Reference> references = new ArrayList<>();
                                List<TaskResultResponse.Reference> originalReferences = questionResult.getReferences();
                                if (originalReferences == null) {
                                    originalReferences = Collections.emptyList();
                                }

                                //如果查询结果不为空查询num
                                Integer maxNumByKeywordId = referenceService.getMaxNumByKeywordId(keyword.getKeyword_id());
                               if (maxNumByKeywordId != null){
                                   maxNumByKeywordId++;
                               }else {
                                   maxNumByKeywordId = 1;
                               }
                                // 遍历原始引用列表，转换为Reference对象
                                for (TaskResultResponse.Reference ref : originalReferences) {
                                    Reference reference = new Reference();
                                    // 设置基本字段
                                    reference.setQuestion_id(question.getQuestion_id());
                                    reference.setTitle(ref.getTitle());
                                    reference.setUrl(ref.getUrl());
                                    reference.setDomain(ref.getDomain());
                                    reference.setNum(keyword.getNum());
                                    reference.setTask_id(result.getTask_id());
                                    reference.setKeyword_id(keyword.getKeyword_id());
                                    if (null!=ref.getPublish_time()) {
                                        reference.setCreate_time(ref.getPublish_time().atStartOfDay());
                                    }


                                // 收集引用数据，处理空集合情况
                                Integer finalMaxNumByKeywordId = maxNumByKeywordId;
                                List<Reference> references =
                                        Optional.ofNullable(questionResult.getReferences())
                                                .orElse(Collections.emptyList())
                                                .stream()
                                                .map(ref -> {
                                                    Reference reference = new Reference();
                                                    reference.setQuestion_id(question.getQuestion_id());
                                                    reference.setTitle(ref.getTitle());
                                                    reference.setUrl(ref.getUrl());
                                                    reference.setDomain(ref.getDomain());
                                                    reference.setNum(finalMaxNumByKeywordId);
                                                    reference.setTask_id(result.getTask_id());
                                                    reference.setKeyword_id(keyword.getKeyword_id());
                                                    //域名和平台id映射
                                                    reference.setCreate_time(LocalDateTime.now());
                                                    Platform platform = platformService.getPlatformByDomain(reference.getDomain());
//                                                    if (platform == null) {
//                                                        throw new RuntimeException("未找到对应的平台: " + reference.getDomain());
//                                                    }
                                                    if (platform != null){
                                                        reference.setPlatform_id(platform.getPlatform_id());
                                                        Type type = typeService.getById(platform.getType_id());
//                                                    if (type == null) {
//                                                        throw new RuntimeException("未找到对应的类型: " + reference.getDomain());
//                                                    }
                                                        if (type != null){
                                                            reference.setType_id(type.getType_id());
                                                        }
                                                    }


                                                    // 根据 domain 查询类型


                                                    return reference;
                                                })
                                                .collect(Collectors.toList());

                                    // 关键：使用优化后的方法获取平台，避免重复创建
                                    Platform platform = getOrCreatePlatform(ref.getDomain(),ref.getPlatform_name());
                                    reference.setPlatform_id(platform.getPlatform_id());
                                    reference.setType_id(platform.getType_id()); // 直接从平台获取类型ID，更可靠
                                    // 添加到结果列表
                                    references.add(reference);
                                }
                                // 添加到总引用列表
                                if (!references.isEmpty()) {
                                    allReferences.addAll(references);
                                }

                                //取数据库中当前关键词的当前轮次的当前问题id结果拿出来
                                List<Reference> dbList = referenceService.list(new LambdaQueryWrapper<Reference>().eq(Reference::getKeyword_id, keyword.getKeyword_id())
                                        .eq(Reference::getNum, keyword.getNum())
                                        .eq(Reference::getQuestion_id, question.getQuestion_id())
                                );

                                // 1. 合并两个列表
                                List<Reference> combinedList = new ArrayList<>();
                                combinedList.addAll(allReferences);
                                combinedList.addAll(dbList);

                                // 2. 创建复合键的Map，用于统计完全匹配的记录
                                Map<String, List<Reference>> compositeKeyMap = combinedList.stream()
                                        .collect(Collectors.groupingBy(
                                                ref -> ref.getTitle() + "|" + ref.getUrl() + "|" + ref.getDomain()
                                        ));

                                // 3. 处理每组重复记录

                                compositeKeyMap.forEach((key, refGroup) -> {
                                    // 3.1 找出组内有ID的记录（优先从dbList中获取）
                                    Optional<Reference> existingRecord = refGroup.stream()
                                            .filter(ref -> ref.getReference_id() != null)
                                            .findFirst();

                                    // 3.2 统计该组的重复次数（总数-1）
                                    int repetitionCount = refGroup.size() - 1;

                                    // 3.3 决定最终保留的记录
                                    Reference recordToSave = new Reference();
                                    if (existingRecord.isPresent()) {
                                        // 使用已有ID的记录并更新重复次数
                                        recordToSave = existingRecord.get();
                                        recordToSave.setRepetition_num(
                                                (recordToSave.getRepetition_num() == null ? 1 : recordToSave.getRepetition_num())
                                                        + repetitionCount
                                        );
                                    } else {
                                        // 没有ID记录则取第一条并设置重复次数
                                        recordToSave = refGroup.get(0);
                                        recordToSave.setRepetition_num(1+repetitionCount);
                                    }

                                    resultList.add(recordToSave);
                                });
                                referenceService.saveOrUpdateBatch(resultList);
                            }
                        } catch (Exception e) {
                            log.error(e.getMessage(), e);
                            System.out.println("处理问题结果失败: " + e.getMessage());
                        }
                    }


                }

                // 4. 批量更新问题
@@ -577,29 +909,56 @@
                    questionService.updateBatchById(questionsToUpdate);
                    System.out.println("成功批量更新 " + questionsToUpdate.size() + " 个问题");
                }
                referenceService.saveBatch(allReferences);
                // 5. 批量插入引用，使用流式分批处理
//                if (!allReferences.isEmpty()) {
//                    int batchSize = 1000;
//                    IntStream.iterate(0, i -> i + batchSize)
//                            .limit((allReferences.size() + batchSize - 1) / batchSize)
//                            .forEach(i -> {
//                                List<Reference> batch = allReferences.subList(
//                                        i, Math.min(i + batchSize, allReferences.size()));
//                                referenceService.saveBatch(batch);
//                            });
//                    System.out.println("成功批量插入 " + allReferences.size() + " 条引用数据");
//                }


            } catch (Exception e) {
                log.error("更新问题和引用数据失败: " ,e.getMessage(), e);
//                System.out.println("更新问题和引用数据失败: " + e.getMessage());
                throw new RuntimeException("更新问题和引用数据失败", e);
            }
        });
    }
    // 根据所有批次的结果判断最终状态
    private String determineFinalStatus(List<QuestionResultList> results) {
        if (results.isEmpty()) {
            return "no_results"; // 无结果
        }

        // 统计关键指标
        int totalCount = results.size();
        int emptyResponseCount = 0;
        int systemBusyCount = 0;

        for (QuestionResultList result : results) {
            // 判断回答是否为空
            if (result.getExtracted_count() == 0 ) {
                emptyResponseCount++;
            }

            // 判断是否为系统繁忙
            if ("success".equals(result.getStatus())  && (result.getResponse().isEmpty()|| result.getResponse().contains("WebDriver连接中断") || result.getResponse().contains("响应超时"))) {
                systemBusyCount++;
            }

        }

        // 全返回系统繁忙
        if (systemBusyCount == totalCount) {
            return "busyness";
        }
        // 全返回信息为空
        if (emptyResponseCount == totalCount) {
            return "no_results";
        }


        // 系统繁忙比例超过阈值（可配置，这里设为70%）
//        double busyRate = (double) systemBusyCount / totalCount;
//        if (busyRate >= 0.7) {
//            return "系统繁忙，请稍后尝试";
//        }

        // 其他情况返回成功
        return "success";
    }
    @GetMapping("/tasks/all")
    @ApiOperation(value = "获取所有任务列表")
    public Mono<TaskListResponse> getAllTasks() {
@@ -613,8 +972,6 @@
                    TaskListResponse response = new TaskListResponse();
                    response.setDetail("获取任务列表失败: " + e.getMessage());
                    return Mono.just(response);

                    // return Mono.just(ResponseResult.error("获取任务列表失败: " + e.getMessage()));
                });
    }

@@ -642,4 +999,7 @@
                .onErrorResume(e -> Mono.just(
                        new ServerResourceResponse( e.getMessage())));
    }



}