package com.linghu.controller;
|
|
import java.time.Duration;
|
import java.time.LocalDateTime;
|
import java.time.format.DateTimeFormatter;
|
import java.util.*;
|
import java.util.stream.Collectors;
|
|
import javax.servlet.http.HttpServletRequest;
|
|
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.http.*;
|
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.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")
|
@Api(value = "采集接口", tags = "采集管理")
|
@Slf4j
|
public class CollectController {
|
|
@Autowired
|
private ReferenceService referenceService;
|
|
@Value("${linghu.url}")
|
private String baseUrl;
|
|
@Autowired
|
private WebClient webClient;
|
|
@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;
|
private static final Queue<SearchTaskRequest> taskQueue = new LinkedList<>();
|
private static boolean isProcessing = false;
|
@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 -> {
|
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()));
|
});
|
}
|
|
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) {
|
// 处理任务
|
executeBatchTask(nextTaskRequest)
|
.doFinally(signal -> {
|
// 完成后，继续处理下一个任务
|
isProcessing = false;
|
if (!taskQueue.isEmpty()) {
|
processNextTaskInQueue(); // 继续处理队列中的下一个任务
|
}
|
})
|
.subscribe();
|
}
|
}
|
|
private Mono<ResponseResult<String>> executeBatchTask(SearchTaskRequest searchTaskRequest) {
|
log.info("开始处理任务：{}", searchTaskRequest);
|
log.info("++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++");
|
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);
|
|
Queue<List<UserDto>> batchQueue = new LinkedList<>(userBatches); // 用队列存储批次
|
|
return Mono.just(ResponseResult.success("第一个批次已开始"))
|
.doOnTerminate(() -> {
|
// 启动后台任务，继续处理批次
|
executeBatchTask(batchQueue, searchTaskRequest, keywordId)
|
.subscribe(); // 使用subscribe()启动后台任务
|
});
|
}
|
|
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.isEmpty()) {
|
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) {
|
// 保存任务关联到数据库
|
return saveKeywordTasks(keywordId, taskResponse)
|
.then(waitForTaskCompletion(taskResponse.getTask_id(), batchQueue, searchTaskRequest, keywordId));
|
} else {
|
return Mono.just(ResponseResult.error("创建批次任务失败"));
|
}
|
});
|
}
|
|
private Mono<Void> saveKeywordTasks(Integer keywordId, SearchTaskResponse taskResponse) {
|
if (taskResponse == null || taskResponse.getTask_id() == null) {
|
return Mono.error(new RuntimeException("任务响应无效或任务ID为空"));
|
}
|
|
KeywordTask keywordTask = new KeywordTask();
|
keywordTask.setKeyword_id(keywordId);
|
keywordTask.setTask_id(taskResponse.getTask_id());
|
keywordTask.setStatus("pending");
|
|
// 将 MyBatis-Plus 的同步方法包装为 Mono<Void>
|
return Mono.fromRunnable(() -> {
|
boolean success = keywordTaskService.saveOrUpdate(keywordTask);
|
if (!success) {
|
throw new RuntimeException("保存关键词任务关联失败");
|
}
|
})
|
.doFinally(signalType -> log.info("成功保存关键词任务关联: Task ID {}", taskResponse.getTask_id()))
|
.then();
|
}
|
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 -> {
|
// 如果任务状态是"submitted"或"running"，继续轮询
|
if ("submitted".equalsIgnoreCase(statusResponse.getStatus()) || "running".equalsIgnoreCase(statusResponse.getStatus())) {
|
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;
|
}
|
}
|
/*@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) {
|
|
Keyword keyword = keywordService.getById(keywordId);
|
if (null==keyword.getNum()){
|
keyword.setNum(0);
|
}
|
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())));
|
}
|
return batches;
|
}
|
|
private Mono<SearchTaskResponse> createSingleBatchTask(SearchTaskRequest batchRequest) {
|
return webClient.post()
|
.uri(baseUrl + "/api/v1/search")
|
.contentType(MediaType.APPLICATION_JSON)
|
.bodyValue(batchRequest)
|
.retrieve()
|
.onStatus(HttpStatus::is4xxClientError, response -> response.bodyToMono(String.class)
|
.flatMap(errorBody -> Mono.error(new RuntimeException(errorBody))))
|
.bodyToMono(new ParameterizedTypeReference<SearchTaskResponse>() {})
|
.flatMap(taskResponse -> {
|
if (taskResponse != null && taskResponse.getTask_id() != null) {
|
|
// 使用 Reactor 的方式更新数据库
|
return Mono.fromRunnable(() -> {
|
LambdaUpdateWrapper<Keyword> updateWrapper = new LambdaUpdateWrapper<>();
|
updateWrapper.eq(Keyword::getKeyword_id, batchRequest.getKeyword_id());
|
updateWrapper.set(Keyword::getStatus, "submitted");
|
updateWrapper.set(Keyword::getTask_id, taskResponse.getTask_id());
|
keywordService.update(updateWrapper);
|
//设置轮数
|
Keyword keyword = keywordService.getById(batchRequest.getKeyword_id());
|
|
//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
|
|
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);
|
}
|
return Mono.just(taskResponse);
|
});
|
}
|
|
// 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());
|
//
|
// // 包装成响应式操作
|
// return Mono.fromCallable(() -> {
|
// questionService.updateBatchById(updateQuestions);
|
// return result;
|
// });
|
//
|
// }
|
// return Mono.just(result);
|
// })
|
// .onErrorResume(e -> {
|
// // 创建一个自定义的错误响应对象
|
// TaskStatusResponse errorResponse = new TaskStatusResponse();
|
// errorResponse.setStatus("ERROR");
|
// errorResponse.setMessage(e.getMessage());
|
// errorResponse.setDetail(e.getMessage());
|
//
|
// return Mono.just(errorResponse);
|
// });
|
// }
|
|
@PostMapping("/cancel/{taskId}")
|
@ApiOperation(value = "取消任务")
|
public Mono<ResponseResult<TaskCancelResponse>> cancelTask(@PathVariable String taskId) {
|
return webClient.post()
|
.uri(baseUrl + "/api/v1/tasks/" + taskId + "/cancel")
|
.contentType(MediaType.APPLICATION_JSON)
|
.bodyValue(Collections.emptyMap()) // 添加空请求体
|
.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()));
|
});
|
}
|
|
@ApiOperation(value = "获取任务结果")
|
@GetMapping("/tasks/{taskId}")
|
public Mono<TaskResultResponse> getTaskResult(@PathVariable String taskId) {
|
return webClient.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");
|
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());
|
//
|
// }
|
// }
|
// }
|
// });
|
// }
|
|
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);
|
|
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);
|
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. 查询该订单下的所有关键词
|
LambdaQueryWrapper<Keyword> orderKeywordsWrapper = new LambdaQueryWrapper<>();
|
orderKeywordsWrapper.eq(Keyword::getOrder_id, orderId);
|
List<Keyword> orderKeywords = keywordService.list(orderKeywordsWrapper);
|
|
if (orderKeywords.isEmpty()) {
|
System.out.println("订单[" + orderId + "]下无关键词，跳过状态更新");
|
return;
|
}
|
|
// 3. 检查所有关键词的状态是否均为 completed 或 false
|
boolean allValid = orderKeywords.stream()
|
.allMatch(k -> "completed".equals(k.getStatus()) || "false".equals(k.getStatus()));
|
|
// 4. 若所有关键词状态均有效，更新订单状态为3
|
if (allValid) {
|
Orders orders = orderService.getById(orderId);
|
if (orders != null) {
|
orders.setStatus(3); // 假设Orders有Integer类型的status字段
|
orderService.updateById(orders);
|
System.out.println("订单[" + orderId + "]所有关键词状态符合条件，已更新状态为3");
|
} else {
|
System.out.println("未找到订单[" + orderId + "]，无法更新状态");
|
}
|
}
|
}
|
Orders orders = orderService.getById(keyword.getOrder_id());
|
|
// 2. 批量查询所有问题
|
LambdaQueryWrapper<Question> queryWrapper = new LambdaQueryWrapper<>();
|
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<>();
|
// 遍历结果
|
for (UserResult userResult : result.getResults()) {
|
for (QuestionResult questionResult : userResult.getQuestions_results()) {
|
try {
|
Question question = questionMap.get(questionResult.getQuestion());
|
if (question != null) {
|
// 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.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);
|
|
/* 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);
|
|
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());*/
|
// 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) { // 注意：需将“原引用类型”替换为实际类型（如QuestionResult中的引用类型）
|
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());
|
reference.setCreate_time(LocalDateTime.now());
|
|
// 处理平台和类型关联
|
Platform platform = platformService.getPlatformByDomain(reference.getDomain());
|
if (platform == null) {
|
// 平台不存在，创建新平台（类型默认为“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());
|
platform1.setCreate_time(LocalDateTime.now());
|
platformService.save(platform1);
|
|
// 关联新平台和类型
|
reference.setType_id(type.getType_id());
|
reference.setPlatform_id(platform1.getPlatform_id());
|
} else {
|
// 平台已存在，直接关联
|
reference.setPlatform_id(platform.getPlatform_id());
|
Type type = typeService.getById(platform.getType_id());
|
if (type != null) {
|
reference.setType_id(type.getType_id());
|
}
|
}
|
|
// 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());
|
}
|
}
|
//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);
|
|
}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);
|
}
|
}
|
|
}
|
|
// 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");
|
}
|
|
// 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);
|
}
|
});
|
}
|
|
@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 Mono.just(ResponseResult.error("获取任务列表失败: " + e.getMessage()));
|
});
|
}
|
|
@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())));
|
}
|
|
/**
|
* 查询服务器资源
|
*/
|
@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())));
|
}
|
/**
|
* 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
|
*/
|
|
|
}
|
