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

Example 91 with SchemaColumn

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

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlNoCustomDdlMissingSchemaNullValue.

@Test
public void testGenerateBusinessObjectDataDdlNoCustomDdlMissingSchemaNullValue() {
    // 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, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, null, 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.
    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 92 with SchemaColumn

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

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlFilterOnSubPartitionValues.

@Test
public void testGenerateBusinessObjectDataDdlFilterOnSubPartitionValues() {
    List<SchemaColumn> columns = schemaColumnDaoTestHelper.getTestSchemaColumns();
    List<SchemaColumn> partitionColumns = schemaColumnDaoTestHelper.getTestPartitionColumns();
    // Test generate business object data ddl using primary partition and each of the available subpartition columns.
    for (int i = 0; i < Math.min(BusinessObjectDataEntity.MAX_SUBPARTITIONS + 1, PARTITION_COLUMNS.length); i++) {
        // Prepare test data without custom ddl.
        businessObjectDataServiceTestHelper.createDatabaseEntitiesForBusinessObjectDataDdlTesting(FileTypeEntity.TXT_FILE_TYPE, FIRST_PARTITION_COLUMN_NAME, PARTITION_KEY_GROUP, i + 1, STORAGE_1_AVAILABLE_PARTITION_VALUES, SUBPARTITION_VALUES, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, SCHEMA_NULL_VALUE_BACKSLASH_N, columns, partitionColumns, false, null, true, ALLOW_DUPLICATE_BUSINESS_OBJECT_DATA);
        // Retrieve business object data ddl request without drop table statement.
        BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(STORAGE_1_AVAILABLE_PARTITION_VALUES);
        request.getPartitionValueFilters().get(0).setPartitionKey(partitionColumns.get(i).getName());
        request.setIncludeDropPartitions(INCLUDE_DROP_PARTITIONS);
        request.setAllowMissingData(NO_ALLOW_MISSING_DATA);
        BusinessObjectDataDdl resultDdl = businessObjectDataService.generateBusinessObjectDataDdl(request);
        // Validate the results.
        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, i + 1, STORAGE_1_AVAILABLE_PARTITION_VALUES, SUBPARTITION_VALUES, false, true, true, INCLUDE_DROP_PARTITIONS);
        businessObjectDataServiceTestHelper.validateBusinessObjectDataDdl(request, expectedDdl, resultDdl);
    }
}

Example 93 with SchemaColumn

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

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlNoCustomDdlAllKnownFileTypes.

@Test
public void testGenerateBusinessObjectDataDdlNoCustomDdlAllKnownFileTypes() {
    // Create an S3 storage entity with the relative attributes.
    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)));
    // Expected business object format file type to Hive file format mapping.
    HashMap<String, String> businessObjectFormatFileTypeMap = new HashMap<>();
    businessObjectFormatFileTypeMap.put(FileTypeEntity.BZ_FILE_TYPE, Hive13DdlGenerator.TEXT_HIVE_FILE_FORMAT);
    businessObjectFormatFileTypeMap.put(FileTypeEntity.GZ_FILE_TYPE, Hive13DdlGenerator.TEXT_HIVE_FILE_FORMAT);
    businessObjectFormatFileTypeMap.put(FileTypeEntity.ORC_FILE_TYPE, Hive13DdlGenerator.ORC_HIVE_FILE_FORMAT);
    businessObjectFormatFileTypeMap.put(FileTypeEntity.PARQUET_FILE_TYPE, Hive13DdlGenerator.PARQUET_HIVE_FILE_FORMAT);
    businessObjectFormatFileTypeMap.put(FileTypeEntity.TXT_FILE_TYPE, Hive13DdlGenerator.TEXT_HIVE_FILE_FORMAT);
    for (String businessObjectFormatFileType : businessObjectFormatFileTypeMap.keySet()) {
        // Prepare test data for the respective business object format file type.
        List<SchemaColumn> partitionColumns = schemaColumnDaoTestHelper.getTestPartitionColumns().subList(0, 1);
        String partitionKey = partitionColumns.get(0).getName();
        BusinessObjectFormatEntity businessObjectFormatEntity = businessObjectFormatDaoTestHelper.createBusinessObjectFormatEntity(NAMESPACE, BDEF_NAME, FORMAT_USAGE_CODE, businessObjectFormatFileType, FORMAT_VERSION, FORMAT_DESCRIPTION, LATEST_VERSION_FLAG_SET, partitionKey, NO_PARTITION_KEY_GROUP, NO_ATTRIBUTES, SCHEMA_DELIMITER_PIPE, SCHEMA_ESCAPE_CHARACTER_BACKSLASH, SCHEMA_NULL_VALUE_BACKSLASH_N, schemaColumnDaoTestHelper.getTestSchemaColumns(), partitionColumns);
        for (String partitionValue : STORAGE_1_AVAILABLE_PARTITION_VALUES) {
            BusinessObjectDataEntity businessObjectDataEntity = businessObjectDataDaoTestHelper.createBusinessObjectDataEntity(NAMESPACE, BDEF_NAME, FORMAT_USAGE_CODE, businessObjectFormatFileType, FORMAT_VERSION, partitionValue, NO_SUBPARTITION_VALUES, DATA_VERSION, true, BDATA_STATUS);
            String s3KeyPrefix = s3KeyPrefixHelper.buildS3KeyPrefix(S3_KEY_PREFIX_VELOCITY_TEMPLATE, businessObjectFormatEntity, businessObjectDataHelper.getBusinessObjectDataKey(businessObjectDataEntity), STORAGE_NAME);
            StorageUnitEntity storageUnitEntity = storageUnitDaoTestHelper.createStorageUnitEntity(storageEntity, businessObjectDataEntity, StorageUnitStatusEntity.ENABLED, NO_STORAGE_DIRECTORY_PATH);
            // Create two storage files.
            for (int i = 0; i < 2; i++) {
                storageFileDaoTestHelper.createStorageFileEntity(storageUnitEntity, String.format("%s/data%d.dat", s3KeyPrefix, i), FILE_SIZE_1_KB, ROW_COUNT_1000);
            }
            herdDao.saveAndRefresh(storageUnitEntity);
            herdDao.saveAndRefresh(businessObjectDataEntity);
        }
        // Retrieve business object data ddl.
        BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(UNSORTED_PARTITION_VALUES);
        request.setBusinessObjectFormatFileType(businessObjectFormatFileType);
        BusinessObjectDataDdl resultDdl = businessObjectDataService.generateBusinessObjectDataDdl(request);
        // Validate the results.
        String expectedHiveFileFormat = businessObjectFormatFileTypeMap.get(businessObjectFormatFileType);
        String expectedDdl = businessObjectDataServiceTestHelper.getExpectedBusinessObjectDataDdl(partitionColumns.size(), FIRST_COLUMN_NAME, FIRST_COLUMN_DATA_TYPE, ROW_FORMAT, expectedHiveFileFormat, businessObjectFormatFileType, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, STORAGE_1_AVAILABLE_PARTITION_VALUES, SUBPARTITION_VALUES, false, true, true);
        businessObjectDataServiceTestHelper.validateBusinessObjectDataDdl(request, expectedDdl, resultDdl);
    }
}

Example 94 with SchemaColumn

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

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlNoCustomDdlUnprintableCharactersInRowFormat.

@Test
public void testGenerateBusinessObjectDataDdlNoCustomDdlUnprintableCharactersInRowFormat() {
    // Prepare test data without custom ddl.
    List<SchemaColumn> partitionColumns = schemaColumnDaoTestHelper.getTestPartitionColumns();
    String partitionKey = partitionColumns.get(0).getName();
    // Set schemaDelimiterCharacter to char(1), schemaEscapeCharacter to char(10), and schemaNullValue to char(128).
    businessObjectDataServiceTestHelper.createDatabaseEntitiesForBusinessObjectDataDdlTesting(FileTypeEntity.TXT_FILE_TYPE, partitionKey, null, BusinessObjectDataEntity.FIRST_PARTITION_COLUMN_POSITION, UNSORTED_PARTITION_VALUES, SUBPARTITION_VALUES, String.valueOf((char) 1), String.valueOf((char) 10), String.valueOf((char) 128), schemaColumnDaoTestHelper.getTestSchemaColumns(), partitionColumns, false, null, true, ALLOW_DUPLICATE_BUSINESS_OBJECT_DATA);
    // Retrieve business object data ddl request without business object format and data versions.
    BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(UNSORTED_PARTITION_VALUES);
    BusinessObjectDataDdl resultDdl = businessObjectDataService.generateBusinessObjectDataDdl(request);
    // Validate the results - please note that 1 decimal = 1 octal, 10 decimal = 12 octal, and 128 decimal = 200 octal.
    String expectedRowFormat = "ROW FORMAT DELIMITED FIELDS TERMINATED BY '\\001' ESCAPED BY '\\012' NULL DEFINED AS '\\200'";
    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 95 with SchemaColumn

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

the class BusinessObjectDataServiceGenerateBusinessObjectDataDdlTest method testGenerateBusinessObjectDataDdlMissingBusinessObjectDataAllowMissingDataSomeDataNoExists.

@Test
public void testGenerateBusinessObjectDataDdlMissingBusinessObjectDataAllowMissingDataSomeDataNoExists() {
    // 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);
    // Retrieve business object data ddl when some of the business object data is not available and "allow missing data" flag is set to "true".
    BusinessObjectDataDdlRequest request = businessObjectDataServiceTestHelper.getTestBusinessObjectDataDdlRequest(UNSORTED_PARTITION_VALUES, CUSTOM_DDL_NAME);
    request.getPartitionValueFilters().get(0).getPartitionValues().add("I_DO_NOT_EXIST");
    assertTrue(request.isAllowMissingData());
    BusinessObjectDataDdl resultDdl = businessObjectDataService.generateBusinessObjectDataDdl(request);
    // Validate the results.
    businessObjectDataServiceTestHelper.validateBusinessObjectDataDdl(request, businessObjectDataServiceTestHelper.getExpectedBusinessObjectDataDdl(), resultDdl);
}