Examples with MapRDBStatistics org.apache.drill.exec.planner.index.MapRDBStatistics used on opensource projects

Example 1 with MapRDBStatistics

use of org.apache.drill.exec.planner.index.MapRDBStatistics in project drill by apache.

the class ConvertHiveMapRDBJsonScanToDrillMapRDBJsonScan method createNativeScanRel.

/**
 * Helper method which creates a DrillScanRel with native Drill HiveScan.
 */
private DrillScanRel createNativeScanRel(DrillScanRel hiveScanRel, PlannerSettings settings) throws IOException {
    RelDataTypeFactory typeFactory = hiveScanRel.getCluster().getTypeFactory();
    HiveScan hiveScan = (HiveScan) hiveScanRel.getGroupScan();
    HiveReadEntry hiveReadEntry = hiveScan.getHiveReadEntry();
    Map<String, String> parameters = hiveReadEntry.getHiveTableWrapper().getParameters();
    JsonScanSpec scanSpec = new JsonScanSpec(parameters.get(MAPRDB_TABLE_NAME), null, null);
    List<SchemaPath> hiveScanCols = hiveScanRel.getColumns().stream().map(colNameSchemaPath -> replaceOverriddenSchemaPath(parameters, colNameSchemaPath)).collect(Collectors.toList());
    // creates TupleMetadata based on Hive's schema (with optional data modes) to be used in the reader
    // for the case when column type wasn't discovered
    HiveToRelDataTypeConverter dataTypeConverter = new HiveToRelDataTypeConverter(typeFactory);
    TupleMetadata schema = new TupleSchema();
    hiveReadEntry.getTable().getColumnListsCache().getTableSchemaColumns().forEach(column -> schema.addColumn(HiveUtilities.getColumnMetadata(replaceOverriddenColumnId(parameters, column.getName()), dataTypeConverter.convertToNullableRelDataType(column))));
    MapRDBFormatPluginConfig formatConfig = new MapRDBFormatPluginConfig();
    formatConfig.readTimestampWithZoneOffset = settings.getOptions().getBoolean(ExecConstants.HIVE_READ_MAPRDB_JSON_TIMESTAMP_WITH_TIMEZONE_OFFSET);
    formatConfig.allTextMode = settings.getOptions().getBoolean(ExecConstants.HIVE_MAPRDB_JSON_ALL_TEXT_MODE);
    JsonTableGroupScan nativeMapRDBScan = new JsonTableGroupScan(hiveScan.getUserName(), hiveScan.getStoragePlugin(), // TODO: We should use Hive format plugins here, once it will be implemented. DRILL-6621
    (MapRDBFormatPlugin) hiveScan.getStoragePlugin().getFormatPlugin(formatConfig), scanSpec, hiveScanCols, new MapRDBStatistics(), FileSystemMetadataProviderManager.getMetadataProviderForSchema(schema));
    List<String> nativeScanColNames = hiveScanRel.getRowType().getFieldList().stream().map(field -> replaceOverriddenColumnId(parameters, field.getName())).collect(Collectors.toList());
    List<RelDataType> nativeScanColTypes = hiveScanRel.getRowType().getFieldList().stream().map(RelDataTypeField::getType).collect(Collectors.toList());
    RelDataType nativeScanRowType = typeFactory.createStructType(nativeScanColTypes, nativeScanColNames);
    return new DrillScanRel(hiveScanRel.getCluster(), hiveScanRel.getTraitSet(), hiveScanRel.getTable(), nativeMapRDBScan, nativeScanRowType, hiveScanCols);
}