优化清洗重试机制,优化清洗进度展示,修复模板无法展示参数 (#113)

* bugfix: 模板无法展示参数

* bugfix: 优化清洗进度展示

* bugfix: 优化清洗重试机制
This commit is contained in:
hhhhsc701
2025-11-28 15:28:10 +08:00
committed by GitHub
parent f1bffdcd61
commit 07029d07ff
11 changed files with 93 additions and 21 deletions

View File

@@ -41,6 +41,7 @@ import java.util.*;
import java.util.concurrent.atomic.AtomicReference;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import java.util.stream.Collectors;
import java.util.stream.Stream;
@Slf4j
@@ -116,7 +117,7 @@ public class CleaningTaskService {
prepareTask(task, request.getInstance());
scanDataset(taskId, request.getSrcDatasetId());
executeTask(taskId);
taskScheduler.executeTask(taskId);
return task;
}
@@ -170,6 +171,11 @@ public class CleaningTaskService {
}
public void executeTask(String taskId) {
List<CleaningResultDto> failed = cleaningResultRepo.findByInstanceId(taskId, "FAILED");
Set<String> failedSet = failed.stream().map(CleaningResultDto::getSrcFileId).collect(Collectors.toSet());
CleaningTaskDto task = cleaningTaskRepo.findTaskById(taskId);
scanDataset(taskId, task.getSrcDatasetId(), failedSet);
cleaningResultRepo.deleteByInstanceId(taskId, "FAILED");
taskScheduler.executeTask(taskId);
}
@@ -226,6 +232,29 @@ public class CleaningTaskService {
} while (pageNumber < datasetFiles.getTotalPages());
}
private void scanDataset(String taskId, String srcDatasetId, Set<String> failedFiles) {
int pageNumber = 0;
int pageSize = 500;
PagingQuery pageRequest = new PagingQuery(pageNumber, pageSize);
PagedResponse<DatasetFile> datasetFiles;
do {
datasetFiles = datasetFileService.getDatasetFiles(srcDatasetId, null, null,null, pageRequest);
if (datasetFiles.getContent().isEmpty()) {
break;
}
List<Map<String, Object>> files = datasetFiles.getContent().stream()
.filter(content -> failedFiles.contains(content.getId()))
.map(content -> Map.of("fileName", (Object) content.getFileName(),
"fileSize", content.getFileSize(),
"filePath", content.getFilePath(),
"fileType", content.getFileType(),
"fileId", content.getId()))
.toList();
writeListMapToJsonlFile(files, FLOW_PATH + "/" + taskId + "/dataset.jsonl");
pageNumber += 1;
} while (pageNumber < datasetFiles.getTotalPages());
}
private void writeListMapToJsonlFile(List<Map<String, Object>> mapList, String fileName) {
ObjectMapper objectMapper = new ObjectMapper();

View File

@@ -6,8 +6,13 @@ import com.datamate.cleaning.domain.repository.OperatorInstanceRepository;
import com.datamate.cleaning.infrastructure.validator.CleanTaskValidator;
import com.datamate.cleaning.interfaces.dto.*;
import com.datamate.cleaning.domain.model.entity.TemplateWithInstance;
import com.datamate.common.infrastructure.exception.BusinessException;
import com.datamate.operator.application.OperatorService;
import com.datamate.operator.domain.repository.OperatorViewRepository;
import com.datamate.operator.infrastructure.exception.OperatorErrorCode;
import com.datamate.operator.interfaces.dto.OperatorDto;
import com.fasterxml.jackson.core.JsonProcessingException;
import com.fasterxml.jackson.databind.ObjectMapper;
import lombok.RequiredArgsConstructor;
import org.apache.commons.lang3.StringUtils;
import org.springframework.stereotype.Service;
@@ -31,6 +36,10 @@ public class CleaningTemplateService {
private final CleanTaskValidator cleanTaskValidator;
private final OperatorService operatorService;
private final ObjectMapper objectMapper = new ObjectMapper();
public List<CleaningTemplateDto> getTemplates(String keywords) {
List<OperatorDto> allOperators =
operatorViewRepo.findOperatorsByCriteria(null, null, null, null, null);
@@ -50,7 +59,12 @@ public class CleaningTemplateService {
.map(v -> {
OperatorDto operator = operatorsMap.get(v.getOperatorId());
if (StringUtils.isNotBlank(v.getSettingsOverride())) {
operator.setSettings(v.getSettingsOverride());
try {
operator.setOverrides(objectMapper.readValue(v.getSettingsOverride(), Map.class));
} catch (JsonProcessingException e) {
throw BusinessException.of(OperatorErrorCode.SETTINGS_PARSE_FAILED, e.getMessage());
}
operatorService.overrideSettings(operator);
}
return operator;
}).toList());

View File

@@ -10,7 +10,11 @@ import java.util.List;
public interface CleaningResultRepository extends IRepository<CleaningResult> {
void deleteByInstanceId(String instanceId);
void deleteByInstanceId(String instanceId, String status);
int[] countByInstanceId(String instanceId);
List<CleaningResultDto> findByInstanceId(String instanceId);
List<CleaningResultDto> findByInstanceId(String instanceId, String status);
}

View File

@@ -22,8 +22,14 @@ public class CleaningResultRepositoryImpl extends CrudRepository<CleaningResultM
@Override
public void deleteByInstanceId(String instanceId) {
deleteByInstanceId(instanceId, null);
}
@Override
public void deleteByInstanceId(String instanceId, String status) {
LambdaQueryWrapper<CleaningResult> queryWrapper = new LambdaQueryWrapper<>();
queryWrapper.eq(CleaningResult::getInstanceId, instanceId);
queryWrapper.eq(CleaningResult::getInstanceId, instanceId)
.eq(StringUtils.isNotBlank(status), CleaningResult::getStatus, status);
mapper.delete(queryWrapper);
}
@@ -40,8 +46,13 @@ public class CleaningResultRepositoryImpl extends CrudRepository<CleaningResultM
}
public List<CleaningResultDto> findByInstanceId(String instanceId) {
return findByInstanceId(instanceId, null);
}
public List<CleaningResultDto> findByInstanceId(String instanceId, String status) {
LambdaQueryWrapper<CleaningResult> queryWrapper = new LambdaQueryWrapper<>();
queryWrapper.eq(CleaningResult::getInstanceId, instanceId);
queryWrapper.eq(CleaningResult::getInstanceId, instanceId)
.eq(StringUtils.isNotBlank(status), CleaningResult::getStatus, status);
return CleaningResultConverter.INSTANCE.convertEntityToDto(mapper.selectList(queryWrapper));
}
}

View File

@@ -128,7 +128,7 @@ public class OperatorService {
return operatorBasePath + File.separator + "extract" + File.separator + fileName;
}
private void overrideSettings(OperatorDto operatorDto) {
public void overrideSettings(OperatorDto operatorDto) {
if (StringUtils.isBlank(operatorDto.getSettings()) || MapUtils.isEmpty(operatorDto.getOverrides())) {
return;
}