From a9287c6b562da327587e2a4bac92df14eb7e2b01 Mon Sep 17 00:00:00 2001
From: guyue <1721849008@qq.com>
Date: 星期六, 26 七月 2025 19:16:14 +0800
Subject: [PATCH] 增加获取结果缓冲区的上限
---
src/main/java/com/linghu/controller/CollectController.java | 1030 +++++++++++++++++++++++++++++++++-----------------------
1 files changed, 603 insertions(+), 427 deletions(-)
diff --git a/src/main/java/com/linghu/controller/CollectController.java b/src/main/java/com/linghu/controller/CollectController.java
index da004b9..23cf78b 100644
--- a/src/main/java/com/linghu/controller/CollectController.java
+++ b/src/main/java/com/linghu/controller/CollectController.java
@@ -4,6 +4,8 @@
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;
@@ -16,7 +18,9 @@
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.reactive.function.client.ExchangeStrategies;
import org.springframework.web.reactive.function.client.WebClient;
import com.baomidou.mybatisplus.core.conditions.query.LambdaQueryWrapper;
@@ -35,10 +39,6 @@
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;
-import java.util.stream.Stream;
@RestController
@RequestMapping("/collect")
@@ -71,129 +71,189 @@
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;
+ return createSingleBatchTask(batchRequest)
+ .flatMap(taskResponse -> {
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);
+ // 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 waitForTaskCompletion(taskResponse.getTask_id())
- .then(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);
+ }
});
}
- 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>
- }*/
+ 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 {
@@ -208,103 +268,42 @@
return 0.0;
}
}
- @PostMapping("/search")
- @ApiOperation(value = "开始采集")
- public Mono<ResponseResult<?>> createSearchTask(
- @RequestBody SearchTaskRequest searchTaskRequest,
- HttpServletRequest request) throws JsonProcessingException {
- // 首先检查服务器资源
- return getServerResource()
- .flatMap(resourceResponse -> {
- // 将字符串类型的使用率转换为double类型
- double cpuUsage = parseUsage(resourceResponse.getCpu_usage_percent());
- double memoryUsage = parseUsage(resourceResponse.getMemory_usage_percent());
- // 检查CPU和内存使用率
- if (cpuUsage >= 90.0 || memoryUsage >= 90.0) {
- String errorMsg = String.format("服务器资源不足：CPU使用率 %.1f%%，内存使用率 %.1f%%",
- resourceResponse.getCpu_usage_percent(), resourceResponse.getMemory_usage_percent());
- log.warn(errorMsg);
- return Mono.just(ResponseResult.error(503, errorMsg));
- }
- 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);
-
- 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
- )
- .map(responses -> ResponseResult.success(responses)) // 使用ResponseResult包装结果
- .onErrorResume(e -> {
- log.error("创建搜索任务失败: {}", e.getMessage(), e);
- return Mono.just(ResponseResult.error("创建搜索任务失败: " + e.getMessage()));
- });
- })
- .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()));
- });
- }
-
- 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());
- keywordTask.setStatus("pending");
- 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");
- // 添加异常处理
- Mono.error( 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,Integer keywordId) {
+ private List<List<UserDto>> splitUsersIntoBatches(List<UserDto> users, int batchSize,Integer keywordId,Boolean isFirst) {
Keyword keyword = keywordService.getById(keywordId);
- if (null==keyword.getNum()){
- keyword.setNum(0);
+ if (isFirst){
+ keyword.setNum(1);
+ }else {
+ keyword.setNum(keyword.getNum()+1);
}
- 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)
@@ -312,12 +311,24 @@
.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");
@@ -325,15 +336,36 @@
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"));
- //设置订单进入采集状态
- LambdaUpdateWrapper<Orders> updateOrderWrapper = new LambdaUpdateWrapper<>();
- updateOrderWrapper.eq(Orders::getOrder_id, keyword.getOrder_id()) // 确保字段名正确
- .set(Orders::getStatus, 2); // 直接设置状态值
+ }
+ //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);
- boolean success = orderService.update(updateOrderWrapper);
- log.info("订单状态更新结果: {}", success ? "成功" : "失败");
-
+ }
+ }
}).subscribeOn(Schedulers.boundedElastic()) // 在弹性线程池执行
.thenReturn(taskResponse);
}
@@ -341,99 +373,191 @@
});
}
- // 移除原来的waitForTaskCompletion方法，不再需要同步等待
- @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)
- .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());
+ @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
- // 包装成响应式操作
- return Mono.fromCallable(() -> {
- questionService.updateBatchById(updateQuestions);
- return result;
- });
+ // 2. 从批次队列移除任务 (新增逻辑)
+ int removedBatchQueue = removeBatchTasksByKeywordId(keywordId);
- }
- return Mono.just(result);
- })
- .onErrorResume(e -> {
- // 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());
+ // 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)
+ );
- return Mono.just(errorResponse);
+ // 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())
+ )
+ ));
});
}
- @PostMapping("/cancel/{taskId}")
- @ApiOperation(value = "取消任务")
- public Mono<ResponseResult<TaskCancelResponse>> cancelTask(@PathVariable String taskId) {
+ // 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(Collections.emptyMap()) // 添加空请求体
+ .bodyValue(requestBody)
.retrieve()
- .onStatus(HttpStatus::isError, response -> response.bodyToMono(TaskCancelResponse.class)
- .flatMap(errorBody -> Mono.error(new RuntimeException(errorBody.getDetail()))))
- .bodyToMono(TaskCancelResponse.class)
- .flatMap(cancelResponse -> {
- // 更新关键词状态
- Mono<Void> updateKeyword = Mono.fromRunnable(() -> {
- LambdaUpdateWrapper<Keyword> updateWrapper = new LambdaUpdateWrapper<>();
- updateWrapper.eq(Keyword::getTask_id, taskId);
- updateWrapper.set(Keyword::getStatus, "canceled"); // 统一使用"canceled"
- keywordService.update(updateWrapper);
- })
- .subscribeOn(Schedulers.boundedElastic())
- .then();
-
- // 更新关键词任务状态
- Mono<Void> updateKeywordTask = Mono.fromRunnable(() -> {
- LambdaUpdateWrapper<KeywordTask> updateWrapper = new LambdaUpdateWrapper<>();
- updateWrapper.eq(KeywordTask::getTask_id, taskId);
- updateWrapper.set(KeywordTask::getStatus, "canceled"); // 统一使用"canceled"
- keywordTaskService.update(updateWrapper);
- })
- .subscribeOn(Schedulers.boundedElastic())
- .then();
-
- // 并行执行两个更新操作，并在完成后返回cancelResponse
- return Mono.when(updateKeyword, updateKeywordTask)
- .thenReturn(cancelResponse);
- })
- .map(data -> ResponseResult.success(data))
- .onErrorResume(e -> {
- if (e.getMessage().contains("任务不存在")) {
- return Mono.just(ResponseResult.error(200, e.getMessage()));
- } else if (e.getMessage().contains("无法取消")) {
- return Mono.just(ResponseResult.error(200, e.getMessage()));
- }
- return Mono.just(ResponseResult.error(500, e.getMessage()));
- });
+ .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()
@@ -450,6 +574,7 @@
.bodyToMono(TaskResultResponse.class)
.flatMap(responseResult -> {
TaskResultResponse result = responseResult;
+
if (result != null && result.getResults() != null) {
return updateQuestionAndReference(result)
.thenReturn(responseResult);
@@ -458,81 +583,64 @@
})
.onErrorResume(e -> {
System.out.println("获取任务结果失败");
- if (e.getMessage().contains("登陆失败")){
-
- }
+ 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);
});
}
-// 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 {
@@ -543,36 +651,28 @@
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;
+
}
LambdaQueryWrapper<KeywordTask> keywordTaskWrapper2 = new LambdaQueryWrapper<>();
keywordTaskWrapper2.eq(KeywordTask::getKeyword_id, keyword.getKeyword_id());
List<KeywordTask> keywordTasks = keywordTaskService.list(keywordTaskWrapper2);
-// 定义状态优先级：canceled > false > completed
- String finalStatus = "completed"; // 默认状态为 completed
-
- for (KeywordTask task : keywordTasks) {
- String status = task.getStatus();
- if ("canceled".equals(status)) {
- finalStatus = "canceled";
- break; // 遇到 canceled 直接跳出循环，因为优先级最高
- } else if ("false".equals(status)) {
- finalStatus = "false";
- // 不跳出循环，继续检查是否存在 canceled
- }
- }
-// 更新关键词状态
- if (!finalStatus.equals(keyword.getStatus())) {
- keyword.setStatus(finalStatus);
+ //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");
@@ -588,7 +688,18 @@
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()));
@@ -604,7 +715,8 @@
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
@@ -620,77 +732,116 @@
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();
+ }
- 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(keyword.getNum());
- 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) {
- //平台为空 创建平台 类型为“默认”
- Type type = typeService.getOne(new LambdaQueryWrapper<Type>().eq(Type::getType_name,"默认"));
- if (type == null) {
- Type newType = new Type();
- newType.setType_name("默认");
- typeService.save(newType);
- type = newType;
- }
- Platform platform1 = new Platform();
- platform1.setDomain(reference.getDomain());
- platform1.setPlatform_name(reference.getDomain());
- platform1.setType_id(type.getType_id());
- platformService.save(platform1);
+ // 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());
+ }
- reference.setType_id(type.getType_id());
- reference.setPlatform_id(platform1.getPlatform_id());
-
- }
- else {
- reference.setPlatform_id(platform.getPlatform_id());
- Type type = typeService.getById(platform.getType_id());
- if (type != null){
- reference.setType_id(type.getType_id());
- }
- }
- 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);
@@ -725,18 +876,18 @@
int repetitionCount = refGroup.size() - 1;
// 3.3 决定最终保留的记录
- Reference recordToSave;
+ 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 ? 0 : recordToSave.getRepetition_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(repetitionCount);
+ recordToSave.setRepetition_num(1+repetitionCount);
}
resultList.add(recordToSave);
@@ -748,6 +899,8 @@
System.out.println("处理问题结果失败: " + e.getMessage());
}
}
+
+
}
// 4. 批量更新问题
@@ -757,28 +910,55 @@
System.out.println("成功批量更新 " + questionsToUpdate.size() + " 个问题");
}
- // 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() {
@@ -792,8 +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()));
});
}
@@ -821,9 +999,7 @@
.onErrorResume(e -> Mono.just(
new ServerResourceResponse( e.getMessage())));
}
- /**
- * 传入orderid查所有关键词id以及关键词下面的所有任务id，轮询所有任务状态，如果状态为completed，则循环调用获取结果接口，处理结果
- */
+
}
--
Gitblit v1.7.1