From e4451cbe7eea81c397353e8d5649e52dcbd3b7d1 Mon Sep 17 00:00:00 2001
From: guyue <1721849008@qq.com>
Date: 星期五, 05 九月 2025 13:14:00 +0800
Subject: [PATCH] 平台筛选
---
src/main/java/com/linghu/controller/CollectController.java | 1344 ---------------------------------------------------------
1 files changed, 16 insertions(+), 1,328 deletions(-)
diff --git a/src/main/java/com/linghu/controller/CollectController.java b/src/main/java/com/linghu/controller/CollectController.java
index 1cafe88..5ccdb76 100644
--- a/src/main/java/com/linghu/controller/CollectController.java
+++ b/src/main/java/com/linghu/controller/CollectController.java
@@ -1,45 +1,23 @@
package com.linghu.controller;
-import java.time.Duration;
-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.servlet.http.HttpServletRequest;
+import javax.validation.Valid;
import com.fasterxml.jackson.core.JsonProcessingException;
import com.linghu.model.dto.*;
-import com.linghu.model.entity.*;
+
import com.linghu.service.*;
import lombok.extern.slf4j.Slf4j;
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.transaction.annotation.Transactional;
-import org.springframework.web.reactive.function.client.ExchangeStrategies;
-import org.springframework.web.reactive.function.client.WebClient;
-
-import com.baomidou.mybatisplus.core.conditions.query.LambdaQueryWrapper;
-import com.baomidou.mybatisplus.core.conditions.update.LambdaUpdateWrapper;
import com.linghu.model.common.ResponseResult;
-import com.linghu.utils.JwtUtils;
-
-import io.jsonwebtoken.lang.Collections;
import io.swagger.annotations.Api;
import io.swagger.annotations.ApiOperation;
-import reactor.core.publisher.Flux;
import reactor.core.publisher.Mono;
-
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;
+
+
@RestController
@RequestMapping("/collect")
@@ -47,1334 +25,49 @@
@Slf4j
public class CollectController {
- @Autowired
- private ReferenceService referenceService;
-
- @Value("${linghu.url}")
- private String baseUrl;
@Autowired
- private WebClient webClient;
+ private CollectionService collectionService;
- @Autowired
- private JwtUtils jwtUtils;
- @Autowired
- private KeywordService keywordService;
- @Autowired
- private QuestionService questionService;
- @Autowired
- private KeywordTaskService keywordTaskService;
- @Autowired
- 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<>();
-
-// private static boolean isProcessing = false;
- private static volatile boolean isProcessing = false; // 添加 volatile
@PostMapping("/search")
@ApiOperation(value = "开始采集")
public Mono<ResponseResult<?>> createSearchTask(
- @RequestBody SearchTaskRequest searchTaskRequest,
+ @Valid @RequestBody SearchTaskRequest searchTaskRequest,
HttpServletRequest request) throws JsonProcessingException {
-
// 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());
-
- 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));
- }
-
- // 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 -> {
- 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()));
- });
+ return collectionService.getResponseResultMono(searchTaskRequest);
}
- private void processNextTaskInQueue() {
- // 设置为正在处理
- isProcessing = true;
-
- // 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, keywordId,searchTaskRequest.getIs_first());
-
-
- // 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
-
- if (batchQueue == null || batchQueue.isEmpty()) {
- // 清理资源
- batchQueues.remove(keywordId);
- return Mono.just(ResponseResult.success("所有批次已完成"));
- }
-
- 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);
-
- return createSingleBatchTask(batchRequest)
- .flatMap(taskResponse -> {
- if (taskResponse != null && taskResponse.getTask_id() != null) {
- // 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"));
- }
- })
- .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);
- }
- });
- }
-
- 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) {
- return webClient.get()
- .uri(baseUrl + "/api/v1/tasks/" + taskId)
- .accept(MediaType.APPLICATION_JSON)
- .retrieve()
- .onStatus(HttpStatus::isError, response -> response.bodyToMono(TaskStatusResponse.class)
- .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody.getDetail()))))
- .bodyToMono(TaskStatusResponse.class)
- .onErrorResume(e -> {
- // 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);
- });
- }
-
- // 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/search")
- .contentType(MediaType.APPLICATION_JSON)
- .bodyValue(batchRequest)
- .retrieve()
- .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(taskResponse);
- });
+ public Mono<TaskStatusResponse> getTaskStatus(@RequestParam(value = "taskId" )String taskId) {
+ return collectionService.getError(taskId);
}
@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())
- )
- ));
- });
+ public Mono<ResponseResult<TaskCancelResponse>> cancelTask(@PathVariable Integer keywordId) {
+ return collectionService.getResponseResult(keywordId);
}
- // 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) {
- 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()
- .onStatus(HttpStatus::is4xxClientError, response -> {
- if (response.statusCode() == HttpStatus.NOT_FOUND) {
- return response.bodyToMono(String.class)
- .flatMap(errorBody -> Mono.error(new RuntimeException("任务不存在")));
- } else if (response.statusCode() == HttpStatus.BAD_REQUEST) {
- return response.bodyToMono(String.class)
- .flatMap(errorBody -> Mono.error(new RuntimeException("任务未完成，无法获取结果")));
- }
- return response.createException().flatMap(Mono::error);
- })
- .bodyToMono(TaskResultResponse.class)
- .flatMap(responseResult -> {
- TaskResultResponse result = responseResult;
-
- if (result != null && result.getResults() != null) {
- return updateQuestionAndReference(result)
- .thenReturn(responseResult);
- }
- return Mono.just(responseResult);
- })
- .onErrorResume(e -> {
- System.out.println("获取任务结果失败");
- 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);
- });
- }
-
- /**
- * 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());
-//
-// }
-// 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);
-//
-// }
-//
-// 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.更新订单状态为待处理 查询该订单下的所有关键词，更新订单状态（有取消）
-// 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.更新订单状态为完成 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<>();
-// queryWrapper.eq(Question::getKeyword_id, keyword.getKeyword_id());
-// List<Question> questions = questionService.list(queryWrapper);
-//
-// // 构建问题映射表，用于快速查找
-// Map<String, Question> questionMap = questions.stream()
-// .collect(Collectors.toMap(Question::getQuestion, q -> q));
-//
-// // 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<>();
-//
-// // 遍历账号
-// 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.setResponse(questionResult.getResponse());
-// question.setExtracted_count(questionResult.getExtracted_count());
-//// question.setError(questionResult.getError());
-// question.setKeyword_id(keyword.getKeyword_id());
-//
-// if (questionResult.getTimestamp() != null) {
-// DateTimeFormatter formatter = DateTimeFormatter
-// .ofPattern("yyyy-MM-dd'T'HH:mm:ss.SSSSSS");
-// question.setTimestamp(
-// LocalDateTime.parse(questionResult.getTimestamp(), formatter));
-// }
-//
-// 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();
-// }
-//
-// // 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());
-// }
-//
-// // 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
-// System.out.println(questionsToUpdate);
-// if (!questionsToUpdate.isEmpty()) {
-// questionService.updateBatchById(questionsToUpdate);
-// System.out.println("成功批量更新 " + questionsToUpdate.size() + " 个问题");
-// }
-//
-// } catch (Exception e) {
-// log.error("更新问题和引用数据失败: " ,e.getMessage(), e);
-// 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";
-// }
-
- private Mono<Void> updateQuestionAndReference(TaskResultResponse result) {
- return Mono.fromRunnable(() -> doUpdateQuestionAndReference(result))
- .onErrorResume(e -> {
- log.error("处理任务结果失败", e);
- return Mono.error(e); // 传播异常，触发事务回滚
- }).then();
- }
-
- // 核心业务逻辑，添加事务注解保证原子性
- @Transactional(rollbackFor = Exception.class)
- public void doUpdateQuestionAndReference(TaskResultResponse result) {
- try {
- // 1. 查询关键词任务并更新状态
- LambdaQueryWrapper<KeywordTask> keywordTaskWrapper = new LambdaQueryWrapper<>();
- keywordTaskWrapper.eq(KeywordTask::getTask_id, result.getTask_id());
- KeywordTask keywordTask = keywordTaskService.getOne(keywordTaskWrapper);
- if (keywordTask == null) {
- throw new Exception("未找到关键词任务，task_id: " + result.getTask_id());
- }
- keywordTask.setStatus("completed");
- keywordTaskService.updateById(keywordTask);
-
- // 2. 查询关键词信息
- Keyword keyword = keywordService.getById(keywordTask.getKeyword_id());
- if (keyword == null) {
- throw new Exception("未找到关联的关键词，task_id: " + result.getTask_id());
- }
-
- // 3. 更新关键词状态（基于关联任务状态）
- updateKeywordStatus(keyword);
-
- // 4. 更新订单状态（基于关键词状态）
- updateOrderStatus(keyword);
-
- // 5. 预查询问题列表（一次查询，内存映射）
- LambdaQueryWrapper<Question> questionWrapper = new LambdaQueryWrapper<>();
- questionWrapper.eq(Question::getKeyword_id, keyword.getKeyword_id());
- List<Question> questions = questionService.list(questionWrapper);
- Map<String, Question> questionMap = questions.stream()
- .collect(Collectors.toMap(Question::getQuestion, q -> q));
-
- // 6. 收集批量保存的数据（避免循环内保存）
- List<QuestionResultList> questionResultsToSave = new ArrayList<>(); // 批量保存问题结果
- List<Reference> allReferences = new ArrayList<>(); // 收集所有引用，后续统一处理
-
- // 7. 遍历结果处理问题和引用
- for (UserResult userResult : result.getResults()) {
- // 7.1 更新用户状态（失败处理）
- handleUserStatus(userResult, keyword);
-
- // 7.2 处理问题结果
- for (QuestionResult questionResult : userResult.getQuestions_results()) {
- Question question = questionMap.get(questionResult.getQuestion());
- if (question == null) {
- log.warn("未找到问题记录: {}", questionResult.getQuestion());
- continue;
- }
-
- // 7.2.1 构建问题结果并加入批量列表
- QuestionResultList questionResultList = buildQuestionResultList(questionResult, keyword, result);
- questionResultsToSave.add(questionResultList);
-
- // 7.2.2 处理引用数据（仅收集，不立即保存）
- List<Reference> references = buildReferences(questionResult, question, keyword, result);
- allReferences.addAll(references);
- }
- }
-
- // 8. 批量保存问题结果（一次数据库交互）
- if (!questionResultsToSave.isEmpty()) {
- questionResultService.saveBatch(questionResultsToSave);
- log.info("批量保存问题结果 {} 条", questionResultsToSave.size());
- }
-
- // 9. 批量处理引用数据（去重+更新重复次数+批量保存）
- if (!allReferences.isEmpty()) {
- handleReferencesInBatch(allReferences, keyword);
- }
-
- // 10. 批量更新问题状态（基于最终结果）
- updateQuestionsStatusInBatch(questions, keyword);
-
- } catch (Exception e) {
- log.error("更新数据失败", e);
- throw new RuntimeException("更新数据失败", e); // 触发事务回滚
- }
- }
-
- // 更新关键词状态
- private void updateKeywordStatus(Keyword keyword) {
- LambdaQueryWrapper<KeywordTask> taskWrapper = new LambdaQueryWrapper<>();
- taskWrapper.eq(KeywordTask::getKeyword_id, keyword.getKeyword_id());
- List<KeywordTask> keywordTasks = keywordTaskService.list(taskWrapper);
-
- boolean allCompletedOrFailed = keywordTasks.stream()
- .allMatch(task -> "completed".equals(task.getStatus())
- || "false".equals(task.getStatus())
- || "cancelled".equals(task.getStatus())
- || "canceled".equals(task.getStatus())
- || "nonentity".equals(task.getStatus()));
-
- if (allCompletedOrFailed) {
- keyword.setStatus("completed");
- keywordService.updateById(keyword);
- }
- }
-
- // 更新订单状态
- private void updateOrderStatus(Keyword keyword) {
- String orderId = keyword.getOrder_id();
- if (orderId == null || orderId.isEmpty()) {
- log.info("关键词[{}]未关联订单，跳过订单更新", keyword.getKeyword_id());
- return;
- }
-
- LambdaQueryWrapper<Keyword> orderKeywordsWrapper = new LambdaQueryWrapper<>();
- orderKeywordsWrapper.eq(Keyword::getOrder_id, orderId);
- List<Keyword> orderKeywords = keywordService.list(orderKeywordsWrapper);
- if (orderKeywords.isEmpty()) {
- log.info("订单[{}]无关键词，跳过状态更新", orderId);
- return;
- }
-
- boolean allValid2 = orderKeywords.stream()
- .allMatch(k -> "completed".equals(k.getStatus())
- || "false".equals(k.getStatus())
- || "cancelled".equals(k.getStatus()));
- if (allValid2) {
- updateOrderStatus(orderId, 1, "所有关键词采集完成或取消");
- }
-
- boolean allValid = orderKeywords.stream()
- .allMatch(k -> "completed".equals(k.getStatus()) || "false".equals(k.getStatus()));
- if (allValid) {
- updateOrderStatus(orderId, 3, "所有关键词状态符合条件");
- }
- }
-
- // 封装订单状态更新
- private void updateOrderStatus(String orderId, Integer status, String logMsg) {
- Orders orders = orderService.getById(orderId);
- if (orders != null) {
- orders.setStatus(status);
- orderService.updateById(orders);
- log.info("订单[{}]{}，已更新状态为{}", orderId, logMsg, status);
- } else {
- log.warn("未找到订单[{}]，无法更新状态", orderId);
- }
- }
-
- // 构建问题结果对象
- private QuestionResultList buildQuestionResultList(QuestionResult questionResult, Keyword keyword, TaskResultResponse result) {
- 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));
- }
- return questionResultList;
- }
-
- // 构建引用数据列表
- private List<Reference> buildReferences(QuestionResult questionResult, Question question, Keyword keyword, TaskResultResponse result) {
- List<Reference> references = new ArrayList<>();
- List<TaskResultResponse.Reference> originalReferences = questionResult.getReferences();
- if (originalReferences == null) {
- return references;
- }
-
- for (TaskResultResponse.Reference ref : originalReferences) {
- // 过滤无效引用（标题、URL、域名不能为空）
- if (ref.getTitle() == null || ref.getUrl() == null || ref.getDomain() == null) {
- log.warn("引用数据字段缺失，跳过：title={}, url={}, domain={}",
- ref.getTitle(), ref.getUrl(), ref.getDomain());
- continue;
- }
-
- Reference reference = new Reference();
- 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 (ref.getPublish_time() != null) {
- reference.setCreate_time(ref.getPublish_time().atStartOfDay());
- }
-
- Platform platform = getOrCreatePlatform(ref.getDomain(), ref.getPlatform_name());
- // 校验平台信息非空
- if (platform == null || platform.getPlatform_id() == null) {
- log.warn("平台信息无效，跳过引用：domain={}", ref.getDomain());
- continue;
- }
- reference.setPlatform_id(platform.getPlatform_id());
- reference.setType_id(platform.getType_id());
- references.add(reference);
- }
- return references;
- }
-
- // 批量处理引用数据（去重+更新重复次数）
- private void handleReferencesInBatch(List<Reference> allReferences, Keyword keyword) {
- // 1. 过滤原始列表中的 null 元素
- List<Reference> validReferences = allReferences.stream()
- .filter(Objects::nonNull)
- .collect(Collectors.toList());
-
- if (validReferences.isEmpty()) {
- log.info("无有效引用数据，跳过批量保存");
- return;
- }
-
- // 2. 查询数据库中已存在的引用并过滤 null
- LambdaQueryWrapper<Reference> dbRefWrapper = new LambdaQueryWrapper<>();
- dbRefWrapper.eq(Reference::getKeyword_id, keyword.getKeyword_id())
- .eq(Reference::getNum, keyword.getNum());
- List<Reference> dbReferences = referenceService.list(dbRefWrapper);
- List<Reference> validDbReferences = dbReferences.stream()
- .filter(Objects::nonNull)
- .collect(Collectors.toList());
-
- // 3. 合并并去重
- Map<String, List<Reference>> compositeKeyMap = new HashMap<>();
- validReferences.forEach(ref -> addToCompositeMap(compositeKeyMap, ref));
- validDbReferences.forEach(ref -> addToCompositeMap(compositeKeyMap, ref));
-
- // 4. 处理重复次数
- List<Reference> referencesToSave = new ArrayList<>();
- compositeKeyMap.forEach((key, refGroup) -> {
- List<Reference> validRefGroup = refGroup.stream()
- .filter(Objects::nonNull)
- .collect(Collectors.toList());
- if (validRefGroup.isEmpty()) return;
-
- Optional<Reference> existingRef = validRefGroup.stream()
- .filter(ref -> ref.getReference_id() != null)
- .findFirst();
-
- Reference finalRef = existingRef.orElse(validRefGroup.get(0));
- int repetitionNum = (finalRef.getRepetition_num() == null ? 1 : finalRef.getRepetition_num())
- + (validRefGroup.size() - 1);
- finalRef.setRepetition_num(repetitionNum);
- referencesToSave.add(finalRef);
- });
-
- // 5. 最终校验并保存
- List<Reference> finalSaveList = referencesToSave.stream()
- .filter(Objects::nonNull)
- .collect(Collectors.toList());
-
- // 关键校验：列表非空且元素有效
- if (finalSaveList.isEmpty()) {
- log.info("处理后无有效引用数据可保存");
- return;
- }
- if (finalSaveList.stream().anyMatch(ref -> !(ref instanceof Reference))) {
- log.error("引用数据类型异常，无法保存");
- return;
- }
-
- // 执行保存
- try {
- referenceService.saveOrUpdateBatch(finalSaveList);
- log.info("批量保存引用数据成功，数量：{}", finalSaveList.size());
- } catch (Exception e) {
- log.error("批量保存引用数据失败", e);
- throw new RuntimeException("保存引用数据失败", e);
- }
- }
-
- // 辅助方法：将引用添加到复合键Map
- private void addToCompositeMap(Map<String, List<Reference>> map, Reference ref) {
- // 再次校验引用的核心字段非空
- if (ref.getTitle() == null || ref.getUrl() == null || ref.getDomain() == null) {
- log.warn("引用核心字段为空，跳过映射：{}", ref);
- return;
- }
- String key = ref.getQuestion_id() + "|" + ref.getTitle() + "|" + ref.getUrl() + "|" + ref.getDomain();
- map.computeIfAbsent(key, k -> new ArrayList<>()).add(ref);
- }
-
- // 批量更新问题状态
- private void updateQuestionsStatusInBatch(List<Question> questions, Keyword keyword) {
- if (questions.isEmpty()) {
- return;
- }
-
- // 一次性查询所有问题结果（基于关键词+轮次）
- LambdaQueryWrapper<QuestionResultList> resultWrapper = new LambdaQueryWrapper<>();
- resultWrapper.eq(QuestionResultList::getKeyword_id, keyword.getKeyword_id())
- .eq(QuestionResultList::getNum, keyword.getNum());
- List<QuestionResultList> allQuestionResults = questionResultService.list(resultWrapper);
-
- // 按问题分组，便于查询
- Map<String, List<QuestionResultList>> questionResultsMap = allQuestionResults.stream()
- .collect(Collectors.groupingBy(QuestionResultList::getQuestion));
-
- // 批量更新问题状态
- List<Question> questionsToUpdate = new ArrayList<>();
- questions.forEach(question -> {
- List<QuestionResultList> results = questionResultsMap.getOrDefault(question.getQuestion(), Collections.emptyList());
- String finalStatus = determineFinalStatus(results);
-
- 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");
- }
- // 更新其他字段（响应、时间戳等）
- results.stream().findFirst().ifPresent(result -> {
- question.setResponse(result.getResponse());
- question.setExtracted_count(result.getExtracted_count());
- question.setTimestamp(result.getTimestamp());
- });
- questionsToUpdate.add(question);
- });
-
- if (!questionsToUpdate.isEmpty()) {
- questionService.updateBatchById(questionsToUpdate);
- log.info("批量更新问题状态 {} 条", questionsToUpdate.size());
- }
- }
-
- // 处理用户状态异常
- private void handleUserStatus(UserResult userResult, Keyword keyword) {
- if ("failed".equals(userResult.getStatus())) {
- LambdaUpdateWrapper<User> userWrapper = new LambdaUpdateWrapper<>();
- userWrapper.eq(User::getUser_email, userResult.getUser_email());
-
- if (userResult.getError().contains("登录失败")) {
- userWrapper.set(User::getStatus, "无法登录");
- userService.update(userWrapper);
- // 批量更新问题状态为失败
- 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")) {
- userWrapper.set(User::getStatus, "信息错误");
- userService.update(userWrapper);
- }
- }
- }
-
- // 原方法：判断最终状态（复用）
- private String determineFinalStatus(List<QuestionResultList> results) {
- if (results.isEmpty()) {
- return "no_results";
- }
-
- int totalCount = results.size();
- int emptyResponseCount = 0;
- int systemBusyCount = 0;
-
- for (QuestionResultList result : results) {
- if (result.getExtracted_count() == 0) {
- emptyResponseCount++;
- }
- if ("success".equals(result.getStatus()) &&
- (result.getResponse() == null || result.getResponse().isEmpty()
- || result.getResponse().contains("WebDriver连接中断")
- || result.getResponse().contains("响应超时"))) {
- systemBusyCount++;
- }
- }
-
- if (systemBusyCount == totalCount) {
- return "busyness";
- }
- if (emptyResponseCount == totalCount) {
- return "no_results";
- }
- return "success";
+ return collectionService.getTaskResultResponseMono(taskId);
}
@GetMapping("/tasks/all")
@ApiOperation(value = "获取所有任务列表")
public Mono<TaskListResponse> getAllTasks() {
- return webClient.get()
- .uri(baseUrl + "/api/v1/tasks")
- .accept(MediaType.APPLICATION_JSON)
- .retrieve()
- .bodyToMono(new ParameterizedTypeReference<TaskListResponse>() {
- })
- .onErrorResume(e -> {
- TaskListResponse response = new TaskListResponse();
- response.setDetail("获取任务列表失败: " + e.getMessage());
- return Mono.just(response);
- });
+ return collectionService.getTaskListResponseMono();
}
@GetMapping("/health")
@ApiOperation("健康检查")
public Mono<HealthResponse> checkThirdPartyHealth() {
- return webClient.get()
- .uri(baseUrl + "/health") // 假设第三方健康检查接口路径为/health
- .retrieve()
- .bodyToMono(HealthResponse.class)
- .onErrorResume(e -> Mono.just(
- new HealthResponse("unhealthy", null, "", e.getMessage())));
+ return collectionService.getHealthResponseMono();
}
/**
@@ -1383,12 +76,7 @@
@GetMapping("/server/resource")
@ApiOperation(value = "查询服务器资源")
public Mono<ServerResourceResponse> getServerResource() {
- return webClient.get()
- .uri(baseUrl + "/api/v1/system/resources")
- .retrieve()
- .bodyToMono(ServerResourceResponse.class)
- .onErrorResume(e -> Mono.just(
- new ServerResourceResponse( e.getMessage())));
+ return collectionService.getServerResourceResponseMono();
}
--
Gitblit v1.7.1