Examples with SchemaColumn org.finra.herd.model.api.xml.SchemaColumn used on opensource projects

Example 86 with SchemaColumn

use of org.finra.herd.model.api.xml.SchemaColumn in project herd by FINRAOS.

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlSubpartitionKeysHaveHyphens.

@Test
public void testGenerateBusinessObjectDataDdlSubpartitionKeysHaveHyphens() {
    // Prepare test data with subpartition using key values with hyphens instead of underscores.
    List<SchemaColumn> partitionColumns = schemaColumnDaoTestHelper.getTestPartitionColumns();
    String partitionKey = partitionColumns.get(0).getName();
    businessObjectDataServiceTestHelper.createDatabaseEntitiesForBusinessObjectDataDdlTesting(FileTypeEntity.TXT_FILE_TYPE, partitionKey, PARTITION_KEY_GROUP, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, UNSORTED_PARTITION_VALUES, SUBPARTITION_VALUES, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, SCHEMA_NULL_VALUE_BACKSLASH_N, schemaColumnDaoTestHelper.getTestSchemaColumns(), partitionColumns, true, CUSTOM_DDL_NAME, true, ALLOW_DUPLICATE_BUSINESS_OBJECT_DATA);
    // Retrieve business object data ddl.
    BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(UNSORTED_PARTITION_VALUES, CUSTOM_DDL_NAME);
    BusinessObjectDataDdl resultDdl = businessObjectDataService.generateBusinessObjectDataDdl(request);
    // Validate the results.  Please note that we expect hyphens in subpartition key values.
    String expectedDdl = businessObjectDataServiceTestHelper.getExpectedBusinessObjectDataDdl(PARTITION_COLUMNS.length, FIRST_COLUMN_NAME, FIRST_COLUMN_DATA_TYPE, ROW_FORMAT, Hive13DdlGenerator.TEXT_HIVE_FILE_FORMAT, FileTypeEntity.TXT_FILE_TYPE, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, STORAGE_1_AVAILABLE_PARTITION_VALUES, SUBPARTITION_VALUES, true, true, true);
    businessObjectDataServiceTestHelper.validateBusinessObjectDataDdl(request, expectedDdl, resultDdl);
}

Example 87 with SchemaColumn

use of org.finra.herd.model.api.xml.SchemaColumn in project herd by FINRAOS.

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlSuppressScanForUnregisteredSubPartitionsLatestFormatHasLessPartitionColumnsThenBusinessObjectData.

@Test
public void testGenerateBusinessObjectDataDdlSuppressScanForUnregisteredSubPartitionsLatestFormatHasLessPartitionColumnsThenBusinessObjectData() {
    // Build a list of schema columns.
    List<SchemaColumn> schemaColumns = new ArrayList<>();
    schemaColumns.add(new SchemaColumn(PARTITION_KEY, "DATE", NO_COLUMN_SIZE, COLUMN_REQUIRED, NO_COLUMN_DEFAULT_VALUE, NO_COLUMN_DESCRIPTION));
    schemaColumns.add(new SchemaColumn(COLUMN_NAME, "NUMBER", COLUMN_SIZE, NO_COLUMN_REQUIRED, COLUMN_DEFAULT_VALUE, COLUMN_DESCRIPTION));
    // Build two list of partition columns, so the second format version would have one less partition column defined.
    List<SchemaColumn> partitionColumns1 = schemaColumns.subList(0, 2);
    List<SchemaColumn> partitionColumns2 = schemaColumns.subList(0, 1);
    // Create an initial version of business object format with the schema having two partition columns.
    BusinessObjectFormatEntity businessObjectFormatEntity1 = businessObjectFormatDaoTestHelper.createBusinessObjectFormatEntity(NAMESPACE, BDEF_NAME, FORMAT_USAGE_CODE, FileTypeEntity.TXT_FILE_TYPE, INITIAL_FORMAT_VERSION, FORMAT_DESCRIPTION, NO_LATEST_VERSION_FLAG_SET, PARTITION_KEY, NO_PARTITION_KEY_GROUP, NO_ATTRIBUTES, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, SCHEMA_NULL_VALUE_BACKSLASH_N, schemaColumns, partitionColumns1);
    // Create a second version of business object format with the schema having only one partition column.
    businessObjectFormatDaoTestHelper.createBusinessObjectFormatEntity(NAMESPACE, BDEF_NAME, FORMAT_USAGE_CODE, FileTypeEntity.TXT_FILE_TYPE, SECOND_FORMAT_VERSION, FORMAT_DESCRIPTION, LATEST_VERSION_FLAG_SET, PARTITION_KEY, NO_PARTITION_KEY_GROUP, NO_ATTRIBUTES, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, SCHEMA_NULL_VALUE_BACKSLASH_N, schemaColumns, partitionColumns2);
    // Provide sub-partition column name and value.
    List<SchemaColumn> subPartitionColumns = partitionColumns1.subList(1, 2);
    List<String> subPartitionValues = Arrays.asList(SUBPARTITION_VALUES.get(0));
    // Create a business object data entity.
    BusinessObjectDataEntity businessObjectDataEntity = businessObjectDataDaoTestHelper.createBusinessObjectDataEntity(businessObjectFormatEntity1, PARTITION_VALUE, subPartitionValues, DATA_VERSION, LATEST_VERSION_FLAG_SET, BusinessObjectDataStatusEntity.VALID);
    // Create an S3 storage entity.
    // Add the "bucket name" attribute to the storage along with the key prefix velocity template.
    StorageEntity storageEntity = storageDaoTestHelper.createStorageEntity(STORAGE_NAME, StoragePlatformEntity.S3, Arrays.asList(new Attribute(configurationHelper.getProperty(ConfigurationValue.S3_ATTRIBUTE_NAME_BUCKET_NAME), S3_BUCKET_NAME), new Attribute(configurationHelper.getProperty(ConfigurationValue.S3_ATTRIBUTE_NAME_KEY_PREFIX_VELOCITY_TEMPLATE), S3_KEY_PREFIX_VELOCITY_TEMPLATE)));
    // Build an S3 key prefix according to the Data Management S3 naming convention.
    String s3KeyPrefix = getExpectedS3KeyPrefix(NAMESPACE, DATA_PROVIDER_NAME, BDEF_NAME, FORMAT_USAGE_CODE, FileTypeEntity.TXT_FILE_TYPE, INITIAL_FORMAT_VERSION, PARTITION_KEY, PARTITION_VALUE, subPartitionColumns.toArray(new SchemaColumn[subPartitionColumns.size()]), subPartitionValues.toArray(new String[subPartitionValues.size()]), DATA_VERSION);
    // Create a storage unit with a storage directory path.
    storageUnitDaoTestHelper.createStorageUnitEntity(storageEntity, businessObjectDataEntity, StorageUnitStatusEntity.ENABLED, s3KeyPrefix);
    // and latest format has less partition columns then the business object data is registered with.
    try {
        businessObjectDataService.generateBusinessObjectDataDdl(new BusinessObjectDataDdlRequest(NAMESPACE, BDEF_NAME, FORMAT_USAGE_CODE, FileTypeEntity.TXT_FILE_TYPE, null, Arrays.asList(new PartitionValueFilter(PARTITION_KEY, Arrays.asList(PARTITION_VALUE), NO_PARTITION_VALUE_RANGE, NO_LATEST_BEFORE_PARTITION_VALUE, NO_LATEST_AFTER_PARTITION_VALUE)), NO_STANDALONE_PARTITION_VALUE_FILTER, null, NO_STORAGE_NAMES, STORAGE_NAME, BusinessObjectDataDdlOutputFormatEnum.HIVE_13_DDL, TABLE_NAME, NO_CUSTOM_DDL_NAME, INCLUDE_DROP_TABLE_STATEMENT, INCLUDE_IF_NOT_EXISTS_OPTION, INCLUDE_DROP_PARTITIONS, NO_ALLOW_MISSING_DATA, NO_INCLUDE_ALL_REGISTERED_SUBPARTITIONS, SUPPRESS_SCAN_FOR_UNREGISTERED_SUBPARTITIONS));
        fail();
    } catch (IllegalArgumentException e) {
        assertEquals(String.format("Number of primary and sub-partition values (2) specified for the business object data is not equal to " + "the number of partition columns (1) defined in the schema of the business object format selected for DDL generation. " + "Business object data: {%s},  business object format: {%s}", businessObjectDataServiceTestHelper.getExpectedBusinessObjectDataKeyAsString(NAMESPACE, BDEF_NAME, FORMAT_USAGE_CODE, FileTypeEntity.TXT_FILE_TYPE, INITIAL_FORMAT_VERSION, PARTITION_VALUE, subPartitionValues, DATA_VERSION), businessObjectFormatServiceTestHelper.getExpectedBusinessObjectFormatKeyAsString(NAMESPACE, BDEF_NAME, FORMAT_USAGE_CODE, FileTypeEntity.TXT_FILE_TYPE, SECOND_FORMAT_VERSION)), e.getMessage());
    }
}

Example 88 with SchemaColumn

use of org.finra.herd.model.api.xml.SchemaColumn in project herd by FINRAOS.

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlNoCustomDdlEscapeBackslashInRowFormat.

@Test
public void testGenerateBusinessObjectDataDdlNoCustomDdlEscapeBackslashInRowFormat() {
    // Prepare test data without custom ddl.
    List<SchemaColumn> partitionColumns = schemaColumnDaoTestHelper.getTestPartitionColumns();
    String partitionKey = partitionColumns.get(0).getName();
    businessObjectDataServiceTestHelper.createDatabaseEntitiesForBusinessObjectDataDdlTesting(FileTypeEntity.TXT_FILE_TYPE, partitionKey, null, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, UNSORTED_PARTITION_VALUES, SUBPARTITION_VALUES, BACKSLASH, BACKSLASH, BACKSLASH, schemaColumnDaoTestHelper.getTestSchemaColumns(), partitionColumns, false, null, true, ALLOW_DUPLICATE_BUSINESS_OBJECT_DATA);
    // Retrieve business object data ddl.
    BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(UNSORTED_PARTITION_VALUES);
    BusinessObjectDataDdl resultDdl = businessObjectDataService.generateBusinessObjectDataDdl(request);
    // Validate the results - please note that we do not escape single backslash in null value.
    String expectedRowFormat = "ROW FORMAT DELIMITED FIELDS TERMINATED BY '\\\\' ESCAPED BY '\\\\' NULL DEFINED AS '\\'";
    String expectedDdl = businessObjectDataServiceTestHelper.getExpectedBusinessObjectDataDdl(PARTITION_COLUMNS.length, FIRST_COLUMN_NAME, FIRST_COLUMN_DATA_TYPE, expectedRowFormat, Hive13DdlGenerator.TEXT_HIVE_FILE_FORMAT, FileTypeEntity.TXT_FILE_TYPE, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, STORAGE_1_AVAILABLE_PARTITION_VALUES, SUBPARTITION_VALUES, false, true, true);
    businessObjectDataServiceTestHelper.validateBusinessObjectDataDdl(request, expectedDdl, resultDdl);
}

Example 89 with SchemaColumn

use of org.finra.herd.model.api.xml.SchemaColumn in project herd by FINRAOS.

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlMissingBusinessObjectDataDoNotAllowMissingData.

@Test
public void testGenerateBusinessObjectDataDdlMissingBusinessObjectDataDoNotAllowMissingData() {
    // Prepare test data.
    List<SchemaColumn> partitionColumns = schemaColumnDaoTestHelper.getTestPartitionColumns();
    businessObjectDataServiceTestHelper.createDatabaseEntitiesForBusinessObjectDataDdlTesting(FileTypeEntity.TXT_FILE_TYPE, FIRST_PARTITION_COLUMN_NAME, PARTITION_KEY_GROUP, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, UNSORTED_PARTITION_VALUES, SUBPARTITION_VALUES, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, SCHEMA_NULL_VALUE_BACKSLASH_N, schemaColumnDaoTestHelper.getTestSchemaColumns(), partitionColumns, false, CUSTOM_DDL_NAME, true, ALLOW_DUPLICATE_BUSINESS_OBJECT_DATA);
    // Try to retrieve business object data ddl for the non-existing business object data with "allow missing data" flag set to "false".
    BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(UNSORTED_PARTITION_VALUES, CUSTOM_DDL_NAME);
    request.getPartitionValueFilters().get(0).setPartitionValues(Arrays.asList("I_DO_NOT_EXIST"));
    request.setAllowMissingData(false);
    try {
        businessObjectDataService.generateBusinessObjectDataDdl(request);
        fail("Should throw an ObjectNotFoundException when business object data does not exist.");
    } catch (ObjectNotFoundException e) {
        assertEquals(String.format("Business object data {namespace: \"%s\", businessObjectDefinitionName: \"%s\", businessObjectFormatUsage: \"%s\", " + "businessObjectFormatFileType: \"%s\", businessObjectFormatVersion: %s, partitionValue: \"%s\", " + "subpartitionValues: \",,,\", businessObjectDataVersion: %d} is not available in \"%s\" storage(s).", request.getNamespace(), request.getBusinessObjectDefinitionName(), request.getBusinessObjectFormatUsage(), request.getBusinessObjectFormatFileType(), request.getBusinessObjectFormatVersion(), request.getPartitionValueFilters().get(0).getPartitionValues().get(request.getPartitionValueFilters().get(0).getPartitionValues().size() - 1), request.getBusinessObjectDataVersion(), request.getStorageName()), e.getMessage());
    }
}

Example 90 with SchemaColumn

use of org.finra.herd.model.api.xml.SchemaColumn in project herd by FINRAOS.

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlNoCustomDdlNotSupportedSchemaColumnDataType.

@Test
public void testGenerateBusinessObjectDataDdlNoCustomDdlNotSupportedSchemaColumnDataType() {
    // Prepare test data without custom ddl.
    List<SchemaColumn> schemaColumns = schemaColumnDaoTestHelper.getTestSchemaColumns();
    SchemaColumn schemaColumn = new SchemaColumn();
    schemaColumns.add(schemaColumn);
    schemaColumn.setName("COLUMN");
    schemaColumn.setType("UNKNOWN");
    String partitionKey = schemaColumns.get(0).getName();
    businessObjectDataServiceTestHelper.createDatabaseEntitiesForBusinessObjectDataDdlTesting(FileTypeEntity.TXT_FILE_TYPE, partitionKey, null, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, UNSORTED_PARTITION_VALUES, SUBPARTITION_VALUES, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, SCHEMA_NULL_VALUE_BACKSLASH_N, schemaColumns, schemaColumnDaoTestHelper.getTestPartitionColumns(), false, null, true, ALLOW_DUPLICATE_BUSINESS_OBJECT_DATA);
    // Try to retrieve business object data ddl for the format that uses unsupported schema column data type.
    BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(UNSORTED_PARTITION_VALUES);
    try {
        businessObjectDataService.generateBusinessObjectDataDdl(request);
        fail("Should throw an IllegalArgumentException when business object format has a column with an unsupported data type.");
    } catch (IllegalArgumentException e) {
        assertEquals(String.format("Column \"%s\" has an unsupported data type \"%s\" in the schema for business object format " + "{namespace: \"%s\", businessObjectDefinitionName: \"%s\", businessObjectFormatUsage: \"%s\", " + "businessObjectFormatFileType: \"%s\", businessObjectFormatVersion: %d}.", schemaColumn.getName(), schemaColumn.getType(), request.getNamespace(), request.getBusinessObjectDefinitionName(), request.getBusinessObjectFormatUsage(), request.getBusinessObjectFormatFileType(), request.getBusinessObjectFormatVersion()), e.getMessage());
    }
}