Examples with Dataset bio.terra.service.dataset.Dataset used on opensource projects

Example 1 with Dataset

use of bio.terra.service.dataset.Dataset in project jade-data-repo by DataBiosphere.

the class IngestSetupStep method doStep.

@Override
public StepResult doStep(FlightContext context) throws InterruptedException {
    if (configService.testInsertFault(ConfigEnum.TABLE_INGEST_LOCK_CONFLICT_STOP_FAULT)) {
        logger.info("TABLE_INGEST_LOCK_CONFLICT_STOP_FAULT");
        while (!configService.testInsertFault(ConfigEnum.TABLE_INGEST_LOCK_CONFLICT_CONTINUE_FAULT)) {
            logger.info("Sleeping for CONTINUE FAULT");
            TimeUnit.SECONDS.sleep(5);
        }
        logger.info("TABLE_INGEST_LOCK_CONFLICT_CONTINUE_FAULT");
    }
    IngestRequestModel ingestRequestModel = IngestUtils.getIngestRequestModel(context);
    // We don't actually care about the output here since BQ takes the raw "gs://" string as input.
    // As long as parsing succeeds, we're good to move forward.
    IngestUtils.parseBlobUri(ingestRequestModel.getPath());
    Dataset dataset = IngestUtils.getDataset(context, datasetService);
    IngestUtils.putDatasetName(context, dataset.getName());
    DatasetTable targetTable = IngestUtils.getDatasetTable(context, dataset);
    String sgName = DatasetUtils.generateAuxTableName(targetTable, "st");
    IngestUtils.putStagingTableName(context, sgName);
    return StepResult.getStepResultSuccess();
}

Example 2 with Dataset

use of bio.terra.service.dataset.Dataset in project jade-data-repo by DataBiosphere.

the class IngestValidateRefsStep method doStep.

@Override
public StepResult doStep(FlightContext context) throws InterruptedException {
    Dataset dataset = IngestUtils.getDataset(context, datasetService);
    Table table = IngestUtils.getDatasetTable(context, dataset);
    String stagingTableName = IngestUtils.getStagingTableName(context);
    // For each fileref column, scan the staging table and build an array of file ids
    // Then probe the file system to validate that the file exists and is part
    // of this dataset. We check all ids and return one complete error.
    List<String> invalidRefIds = new ArrayList<>();
    for (Column column : table.getColumns()) {
        if (StringUtils.equalsIgnoreCase(column.getType(), "FILEREF")) {
            List<String> refIdArray = bigQueryPdao.getRefIds(dataset, stagingTableName, column);
            List<String> badRefIds = fileDao.validateRefIds(dataset, refIdArray);
            if (badRefIds != null) {
                invalidRefIds.addAll(badRefIds);
            }
        }
    }
    int invalidIdCount = invalidRefIds.size();
    if (invalidIdCount != 0) {
        // Made a string buffer to appease findbugs; it saw + in the loop and said "bad!"
        StringBuffer errorMessage = new StringBuffer("Invalid file ids found during ingest (");
        List<String> errorDetails = new ArrayList<>();
        int count = 0;
        for (String badId : invalidRefIds) {
            errorDetails.add(badId);
            count++;
            if (count > MAX_ERROR_REF_IDS) {
                errorMessage.append(MAX_ERROR_REF_IDS + "out of ");
                break;
            }
        }
        errorMessage.append(invalidIdCount + " returned in details)");
        throw new InvalidFileRefException(errorMessage.toString(), errorDetails);
    }
    return StepResult.getStepResultSuccess();
}

Example 3 with Dataset

use of bio.terra.service.dataset.Dataset in project jade-data-repo by DataBiosphere.

the class CreateDatasetAssetStep method doStep.

@Override
public StepResult doStep(FlightContext context) {
    // TODO: Asset columns and tables need to match things in the dataset schema
    Dataset dataset = getDataset(context);
    FlightMap map = context.getWorkingMap();
    // get the dataset assets that already exist --asset name needs to be unique
    AssetSpecification newAssetSpecification = getNewAssetSpec(context, dataset);
    // add a fault that forces an exception to make sure the undo works
    try {
        configService.fault(ConfigEnum.CREATE_ASSET_FAULT, () -> {
            throw new RuntimeException("fault insertion");
        });
    } catch (Exception e) {
        throw new RuntimeException(e);
    }
    try {
        assetDao.create(newAssetSpecification, dataset.getId());
    } catch (InvalidAssetException e) {
        FlightUtils.setErrorResponse(context, e.getMessage(), HttpStatus.BAD_REQUEST);
        map.put(DatasetWorkingMapKeys.ASSET_NAME_COLLISION, true);
        return new StepResult(StepStatus.STEP_RESULT_FAILURE_FATAL, e);
    }
    map.put(JobMapKeys.STATUS_CODE.getKeyName(), HttpStatus.CREATED);
    return StepResult.getStepResultSuccess();
}

Example 4 with Dataset

use of bio.terra.service.dataset.Dataset in project jade-data-repo by DataBiosphere.

the class CreateDatasetAuthzBqJobUserStep method doStep.

@Override
public StepResult doStep(FlightContext context) throws InterruptedException {
    FlightMap workingMap = context.getWorkingMap();
    UUID datasetId = workingMap.get(DatasetWorkingMapKeys.DATASET_ID, UUID.class);
    Map<IamRole, String> policies = workingMap.get(DatasetWorkingMapKeys.POLICY_EMAILS, Map.class);
    Dataset dataset = datasetService.retrieve(datasetId);
    DatasetModel datasetModel = datasetService.retrieveModel(dataset);
    // The underlying service provides retries so we do not need to retry this operation
    resourceService.grantPoliciesBqJobUser(datasetModel.getDataProject(), policies.values());
    return StepResult.getStepResultSuccess();
}

Example 5 with Dataset

use of bio.terra.service.dataset.Dataset in project jade-data-repo by DataBiosphere.

the class CreateDatasetMetadataStep method doStep.

@Override
public StepResult doStep(FlightContext context) {
    try {
        Dataset newDataset = DatasetUtils.convertRequestWithGeneratedNames(datasetRequest);
        UUID datasetId = datasetDao.createAndLock(newDataset, context.getFlightId());
        FlightMap workingMap = context.getWorkingMap();
        workingMap.put(DatasetWorkingMapKeys.DATASET_ID, datasetId);
        DatasetSummaryModel datasetSummary = DatasetJsonConversion.datasetSummaryModelFromDatasetSummary(newDataset.getDatasetSummary());
        workingMap.put(JobMapKeys.RESPONSE.getKeyName(), datasetSummary);
        return StepResult.getStepResultSuccess();
    } catch (InvalidDatasetException idEx) {
        return new StepResult(StepStatus.STEP_RESULT_FAILURE_FATAL, idEx);
    } catch (Exception ex) {
        return new StepResult(StepStatus.STEP_RESULT_FAILURE_FATAL, new InvalidDatasetException("Cannot create dataset: " + datasetRequest.getName(), ex));
    }
}