Examples with StructTypeInfo org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo used on opensource projects

Example 6 with StructTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo in project hive by apache.

the class VectorRandomRowSource method getObjectInspector.

private ObjectInspector getObjectInspector(TypeInfo typeInfo) {
    final ObjectInspector objectInspector;
    switch(typeInfo.getCategory()) {
        case PRIMITIVE:
            {
                final PrimitiveTypeInfo primitiveType = (PrimitiveTypeInfo) typeInfo;
                objectInspector = PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(primitiveType);
            }
            break;
        case MAP:
            {
                final MapTypeInfo mapType = (MapTypeInfo) typeInfo;
                final MapObjectInspector mapInspector = ObjectInspectorFactory.getStandardMapObjectInspector(getObjectInspector(mapType.getMapKeyTypeInfo()), getObjectInspector(mapType.getMapValueTypeInfo()));
                objectInspector = mapInspector;
            }
            break;
        case LIST:
            {
                final ListTypeInfo listType = (ListTypeInfo) typeInfo;
                final ListObjectInspector listInspector = ObjectInspectorFactory.getStandardListObjectInspector(getObjectInspector(listType.getListElementTypeInfo()));
                objectInspector = listInspector;
            }
            break;
        case STRUCT:
            {
                final StructTypeInfo structType = (StructTypeInfo) typeInfo;
                final List<TypeInfo> fieldTypes = structType.getAllStructFieldTypeInfos();
                final List<ObjectInspector> fieldInspectors = new ArrayList<ObjectInspector>();
                for (TypeInfo fieldType : fieldTypes) {
                    fieldInspectors.add(getObjectInspector(fieldType));
                }
                final StructObjectInspector structInspector = ObjectInspectorFactory.getStandardStructObjectInspector(structType.getAllStructFieldNames(), fieldInspectors);
                objectInspector = structInspector;
            }
            break;
        case UNION:
            {
                final UnionTypeInfo unionType = (UnionTypeInfo) typeInfo;
                final List<TypeInfo> fieldTypes = unionType.getAllUnionObjectTypeInfos();
                final List<ObjectInspector> fieldInspectors = new ArrayList<ObjectInspector>();
                for (TypeInfo fieldType : fieldTypes) {
                    fieldInspectors.add(getObjectInspector(fieldType));
                }
                final UnionObjectInspector unionInspector = ObjectInspectorFactory.getStandardUnionObjectInspector(fieldInspectors);
                objectInspector = unionInspector;
            }
            break;
        default:
            throw new RuntimeException("Unexpected category " + typeInfo.getCategory());
    }
    Preconditions.checkState(objectInspector != null);
    return objectInspector;
}

Example 7 with StructTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo in project carbondata by apache.

the class CarbonHiveSerDe method initialize.

@Override
public void initialize(@Nullable Configuration configuration, Properties tbl) throws SerDeException {
    final TypeInfo rowTypeInfo;
    final List<String> columnNames;
    final List<TypeInfo> columnTypes;
    // Get column names and sort order
    assert configuration != null;
    final String columnNameProperty = tbl.getProperty(serdeConstants.LIST_COLUMNS);
    final String columnTypeProperty = tbl.getProperty(serdeConstants.LIST_COLUMN_TYPES);
    if (columnNameProperty.length() == 0) {
        columnNames = new ArrayList<String>();
    } else {
        columnNames = Arrays.asList(columnNameProperty.split(","));
    }
    if (columnTypeProperty.length() == 0) {
        columnTypes = new ArrayList<TypeInfo>();
    } else {
        columnTypes = TypeInfoUtils.getTypeInfosFromTypeString(columnTypeProperty);
    }
    // Create row related objects
    rowTypeInfo = TypeInfoFactory.getStructTypeInfo(columnNames, columnTypes);
    this.objInspector = new CarbonObjectInspector((StructTypeInfo) rowTypeInfo);
    // Stats part
    serializedSize = 0;
    deserializedSize = 0;
    status = LAST_OPERATION.UNKNOWN;
}

Example 8 with StructTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo in project incubator-gobblin by apache.

the class HiveAvroORCQueryGenerator method escapeHiveType.

/**
 * Escape the Hive nested field names.
 * @param type Primitive or nested Hive type.
 * @return Escaped Hive nested field.
 */
public static String escapeHiveType(String type) {
    TypeInfo typeInfo = TypeInfoUtils.getTypeInfoFromTypeString(type);
    // Primitve
    if (ObjectInspector.Category.PRIMITIVE.equals(typeInfo.getCategory())) {
        return type;
    } else // List
    if (ObjectInspector.Category.LIST.equals(typeInfo.getCategory())) {
        ListTypeInfo listTypeInfo = (ListTypeInfo) typeInfo;
        return org.apache.hadoop.hive.serde.serdeConstants.LIST_TYPE_NAME + "<" + escapeHiveType(listTypeInfo.getListElementTypeInfo().getTypeName()) + ">";
    } else // Map
    if (ObjectInspector.Category.MAP.equals(typeInfo.getCategory())) {
        MapTypeInfo mapTypeInfo = (MapTypeInfo) typeInfo;
        return org.apache.hadoop.hive.serde.serdeConstants.MAP_TYPE_NAME + "<" + escapeHiveType(mapTypeInfo.getMapKeyTypeInfo().getTypeName()) + "," + escapeHiveType(mapTypeInfo.getMapValueTypeInfo().getTypeName()) + ">";
    } else // Struct
    if (ObjectInspector.Category.STRUCT.equals(typeInfo.getCategory())) {
        StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
        List<String> allStructFieldNames = structTypeInfo.getAllStructFieldNames();
        List<TypeInfo> allStructFieldTypeInfos = structTypeInfo.getAllStructFieldTypeInfos();
        StringBuilder sb = new StringBuilder();
        sb.append(serdeConstants.STRUCT_TYPE_NAME + "<");
        for (int i = 0; i < allStructFieldNames.size(); i++) {
            if (i > 0) {
                sb.append(",");
            }
            sb.append("`");
            sb.append(allStructFieldNames.get(i));
            sb.append("`");
            sb.append(":");
            sb.append(escapeHiveType(allStructFieldTypeInfos.get(i).getTypeName()));
        }
        sb.append(">");
        return sb.toString();
    } else // Union
    if (ObjectInspector.Category.UNION.equals(typeInfo.getCategory())) {
        UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
        List<TypeInfo> allUnionObjectTypeInfos = unionTypeInfo.getAllUnionObjectTypeInfos();
        StringBuilder sb = new StringBuilder();
        sb.append(serdeConstants.UNION_TYPE_NAME + "<");
        for (int i = 0; i < allUnionObjectTypeInfos.size(); i++) {
            if (i > 0) {
                sb.append(",");
            }
            sb.append(escapeHiveType(allUnionObjectTypeInfos.get(i).getTypeName()));
        }
        sb.append(">");
        return sb.toString();
    } else {
        throw new RuntimeException("Unknown type encountered: " + type);
    }
}

Example 9 with StructTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo in project drill by axbaretto.

the class HiveRecordReader method init.

private void init() throws ExecutionSetupException {
    final JobConf job = new JobConf(hiveConf);
    // Get the configured default val
    defaultPartitionValue = hiveConf.get(ConfVars.DEFAULTPARTITIONNAME.varname);
    Properties tableProperties;
    try {
        tableProperties = MetaStoreUtils.getTableMetadata(table);
        final Properties partitionProperties = (partition == null) ? tableProperties : HiveUtilities.getPartitionMetadata(partition, table);
        HiveUtilities.addConfToJob(job, partitionProperties);
        final SerDe tableSerDe = createSerDe(job, table.getSd().getSerdeInfo().getSerializationLib(), tableProperties);
        final StructObjectInspector tableOI = getStructOI(tableSerDe);
        if (partition != null) {
            partitionSerDe = createSerDe(job, partition.getSd().getSerdeInfo().getSerializationLib(), partitionProperties);
            partitionOI = getStructOI(partitionSerDe);
            finalOI = (StructObjectInspector) ObjectInspectorConverters.getConvertedOI(partitionOI, tableOI);
            partTblObjectInspectorConverter = ObjectInspectorConverters.getConverter(partitionOI, finalOI);
            job.setInputFormat(HiveUtilities.getInputFormatClass(job, partition.getSd(), table));
        } else {
            // For non-partitioned tables, there is no need to create converter as there are no schema changes expected.
            partitionSerDe = tableSerDe;
            partitionOI = tableOI;
            partTblObjectInspectorConverter = null;
            finalOI = tableOI;
            job.setInputFormat(HiveUtilities.getInputFormatClass(job, table.getSd(), table));
        }
        // Get list of partition column names
        final List<String> partitionNames = Lists.newArrayList();
        for (FieldSchema field : table.getPartitionKeys()) {
            partitionNames.add(field.getName());
        }
        // We should always get the columns names from ObjectInspector. For some of the tables (ex. avro) metastore
        // may not contain the schema, instead it is derived from other sources such as table properties or external file.
        // SerDe object knows how to get the schema with all the config and table properties passed in initialization.
        // ObjectInspector created from the SerDe object has the schema.
        final StructTypeInfo sTypeInfo = (StructTypeInfo) TypeInfoUtils.getTypeInfoFromObjectInspector(finalOI);
        final List<String> tableColumnNames = sTypeInfo.getAllStructFieldNames();
        // Select list of columns for project pushdown into Hive SerDe readers.
        final List<Integer> columnIds = Lists.newArrayList();
        if (isStarQuery()) {
            selectedColumnNames = tableColumnNames;
            for (int i = 0; i < selectedColumnNames.size(); i++) {
                columnIds.add(i);
            }
            selectedPartitionNames = partitionNames;
        } else {
            selectedColumnNames = Lists.newArrayList();
            for (SchemaPath field : getColumns()) {
                String columnName = field.getRootSegment().getPath();
                if (partitionNames.contains(columnName)) {
                    selectedPartitionNames.add(columnName);
                } else {
                    columnIds.add(tableColumnNames.indexOf(columnName));
                    selectedColumnNames.add(columnName);
                }
            }
        }
        ColumnProjectionUtils.appendReadColumns(job, columnIds, selectedColumnNames);
        for (String columnName : selectedColumnNames) {
            ObjectInspector fieldOI = finalOI.getStructFieldRef(columnName).getFieldObjectInspector();
            TypeInfo typeInfo = TypeInfoUtils.getTypeInfoFromTypeString(fieldOI.getTypeName());
            selectedColumnObjInspectors.add(fieldOI);
            selectedColumnTypes.add(typeInfo);
            selectedColumnFieldConverters.add(HiveFieldConverter.create(typeInfo, fragmentContext));
        }
        for (int i = 0; i < table.getPartitionKeys().size(); i++) {
            FieldSchema field = table.getPartitionKeys().get(i);
            if (selectedPartitionNames.contains(field.getName())) {
                TypeInfo pType = TypeInfoUtils.getTypeInfoFromTypeString(field.getType());
                selectedPartitionTypes.add(pType);
                if (partition != null) {
                    selectedPartitionValues.add(HiveUtilities.convertPartitionType(pType, partition.getValues().get(i), defaultPartitionValue));
                }
            }
        }
    } catch (Exception e) {
        throw new ExecutionSetupException("Failure while initializing HiveRecordReader: " + e.getMessage(), e);
    }
    if (!empty) {
        try {
            reader = (org.apache.hadoop.mapred.RecordReader<Object, Object>) job.getInputFormat().getRecordReader(inputSplit, job, Reporter.NULL);
        } catch (Exception e) {
            throw new ExecutionSetupException("Failed to get o.a.hadoop.mapred.RecordReader from Hive InputFormat", e);
        }
        key = reader.createKey();
        skipRecordsInspector = new SkipRecordsInspector(tableProperties, reader);
    }
}

Example 10 with StructTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo in project drill by axbaretto.

the class HiveAbstractReader method init.

private void init() throws ExecutionSetupException {
    final JobConf job = new JobConf(hiveConf);
    // Get the configured default val
    defaultPartitionValue = hiveConf.get(ConfVars.DEFAULTPARTITIONNAME.varname);
    Properties tableProperties;
    try {
        tableProperties = HiveUtilities.getTableMetadata(table);
        final Properties partitionProperties = (partition == null) ? tableProperties : HiveUtilities.getPartitionMetadata(partition, table);
        HiveUtilities.addConfToJob(job, partitionProperties);
        final SerDe tableSerDe = createSerDe(job, table.getSd().getSerdeInfo().getSerializationLib(), tableProperties);
        final StructObjectInspector tableOI = getStructOI(tableSerDe);
        if (partition != null) {
            partitionSerDe = createSerDe(job, partition.getSd().getSerdeInfo().getSerializationLib(), partitionProperties);
            partitionOI = getStructOI(partitionSerDe);
            finalOI = (StructObjectInspector) ObjectInspectorConverters.getConvertedOI(partitionOI, tableOI);
            partTblObjectInspectorConverter = ObjectInspectorConverters.getConverter(partitionOI, finalOI);
            job.setInputFormat(HiveUtilities.getInputFormatClass(job, partition.getSd(), table));
        } else {
            // For non-partitioned tables, there is no need to create converter as there are no schema changes expected.
            partitionSerDe = tableSerDe;
            partitionOI = tableOI;
            partTblObjectInspectorConverter = null;
            finalOI = tableOI;
            job.setInputFormat(HiveUtilities.getInputFormatClass(job, table.getSd(), table));
        }
        if (logger.isTraceEnabled()) {
            for (StructField field : finalOI.getAllStructFieldRefs()) {
                logger.trace("field in finalOI: {}", field.getClass().getName());
            }
            logger.trace("partitionSerDe class is {} {}", partitionSerDe.getClass().getName());
        }
        // Get list of partition column names
        final List<String> partitionNames = Lists.newArrayList();
        for (FieldSchema field : table.getPartitionKeys()) {
            partitionNames.add(field.getName());
        }
        // We should always get the columns names from ObjectInspector. For some of the tables (ex. avro) metastore
        // may not contain the schema, instead it is derived from other sources such as table properties or external file.
        // SerDe object knows how to get the schema with all the config and table properties passed in initialization.
        // ObjectInspector created from the SerDe object has the schema.
        final StructTypeInfo sTypeInfo = (StructTypeInfo) TypeInfoUtils.getTypeInfoFromObjectInspector(finalOI);
        final List<String> tableColumnNames = sTypeInfo.getAllStructFieldNames();
        // Select list of columns for project pushdown into Hive SerDe readers.
        final List<Integer> columnIds = Lists.newArrayList();
        if (isStarQuery()) {
            selectedColumnNames = tableColumnNames;
            for (int i = 0; i < selectedColumnNames.size(); i++) {
                columnIds.add(i);
            }
            selectedPartitionNames = partitionNames;
        } else {
            selectedColumnNames = Lists.newArrayList();
            for (SchemaPath field : getColumns()) {
                String columnName = field.getRootSegment().getPath();
                if (partitionNames.contains(columnName)) {
                    selectedPartitionNames.add(columnName);
                } else {
                    columnIds.add(tableColumnNames.indexOf(columnName));
                    selectedColumnNames.add(columnName);
                }
            }
        }
        ColumnProjectionUtils.appendReadColumns(job, columnIds, selectedColumnNames);
        for (String columnName : selectedColumnNames) {
            StructField fieldRef = finalOI.getStructFieldRef(columnName);
            selectedStructFieldRefs.add(fieldRef);
            ObjectInspector fieldOI = fieldRef.getFieldObjectInspector();
            TypeInfo typeInfo = TypeInfoUtils.getTypeInfoFromTypeString(fieldOI.getTypeName());
            selectedColumnObjInspectors.add(fieldOI);
            selectedColumnTypes.add(typeInfo);
            selectedColumnFieldConverters.add(HiveFieldConverter.create(typeInfo, fragmentContext));
        }
        for (int i = 0; i < selectedColumnNames.size(); ++i) {
            logger.trace("inspector:typeName={}, className={}, TypeInfo: {}, converter:{}", selectedColumnObjInspectors.get(i).getTypeName(), selectedColumnObjInspectors.get(i).getClass().getName(), selectedColumnTypes.get(i).toString(), selectedColumnFieldConverters.get(i).getClass().getName());
        }
        for (int i = 0; i < table.getPartitionKeys().size(); i++) {
            FieldSchema field = table.getPartitionKeys().get(i);
            if (selectedPartitionNames.contains(field.getName())) {
                TypeInfo pType = TypeInfoUtils.getTypeInfoFromTypeString(field.getType());
                selectedPartitionTypes.add(pType);
                if (partition != null) {
                    selectedPartitionValues.add(HiveUtilities.convertPartitionType(pType, partition.getValues().get(i), defaultPartitionValue));
                }
            }
        }
    } catch (Exception e) {
        throw new ExecutionSetupException("Failure while initializing Hive Reader " + this.getClass().getName(), e);
    }
    if (!empty) {
        try {
            reader = (org.apache.hadoop.mapred.RecordReader<Object, Object>) job.getInputFormat().getRecordReader(inputSplit, job, Reporter.NULL);
            logger.trace("hive reader created: {} for inputSplit {}", reader.getClass().getName(), inputSplit.toString());
        } catch (Exception e) {
            throw new ExecutionSetupException("Failed to get o.a.hadoop.mapred.RecordReader from Hive InputFormat", e);
        }
        internalInit(tableProperties, reader);
    }
}