Examples with MapObjectInspector org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector used on opensource projects

Example 11 with MapObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project presto by prestodb.

the class TestHiveFileFormats method hasType.

public static boolean hasType(ObjectInspector objectInspector, PrimitiveCategory... types) {
    if (objectInspector instanceof PrimitiveObjectInspector) {
        PrimitiveObjectInspector primitiveInspector = (PrimitiveObjectInspector) objectInspector;
        PrimitiveCategory primitiveCategory = primitiveInspector.getPrimitiveCategory();
        for (PrimitiveCategory type : types) {
            if (primitiveCategory == type) {
                return true;
            }
        }
        return false;
    }
    if (objectInspector instanceof ListObjectInspector) {
        ListObjectInspector listInspector = (ListObjectInspector) objectInspector;
        return hasType(listInspector.getListElementObjectInspector(), types);
    }
    if (objectInspector instanceof MapObjectInspector) {
        MapObjectInspector mapInspector = (MapObjectInspector) objectInspector;
        return hasType(mapInspector.getMapKeyObjectInspector(), types) || hasType(mapInspector.getMapValueObjectInspector(), types);
    }
    if (objectInspector instanceof StructObjectInspector) {
        for (StructField field : ((StructObjectInspector) objectInspector).getAllStructFieldRefs()) {
            if (hasType(field.getFieldObjectInspector(), types)) {
                return true;
            }
        }
        return false;
    }
    throw new IllegalArgumentException("Unknown object inspector type " + objectInspector);
}

Example 12 with MapObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project presto by prestodb.

the class BlockInputDecoders method createForMap.

private static BlockInputDecoder createForMap(MapObjectInspector inspector, MapType mapType) {
    Type keyType = mapType.getKeyType();
    Type valueType = mapType.getValueType();
    ObjectInspector keyInspector = inspector.getMapKeyObjectInspector();
    ObjectInspector valueInspector = inspector.getMapValueObjectInspector();
    BlockInputDecoder keyDecoder = createBlockInputDecoder(keyInspector, keyType);
    BlockInputDecoder valueDecoder = createBlockInputDecoder(valueInspector, valueType);
    return (b, i) -> {
        if (b.isNull(i)) {
            return null;
        }
        SingleMapBlock single = (SingleMapBlock) b.getBlock(i);
        int positions = single.getPositionCount();
        HashMap<Object, Object> result = new HashMap<>();
        for (int j = 0; j < positions; j += 2) {
            Object key = keyDecoder.decode(single, j);
            Object value = valueDecoder.decode(single, j + 1);
            if (key != null) {
                result.put(key, value);
            }
        }
        return result;
    };
}

Example 13 with MapObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project presto by prestodb.

the class SerDeUtils method serializeMap.

private static Block serializeMap(Type type, BlockBuilder builder, Object object, MapObjectInspector inspector, boolean filterNullMapKeys) {
    Map<?, ?> map = inspector.getMap(object);
    if (map == null) {
        requireNonNull(builder, "parent builder is null").appendNull();
        return null;
    }
    List<Type> typeParameters = type.getTypeParameters();
    checkArgument(typeParameters.size() == 2, "map must have exactly 2 type parameter");
    Type keyType = typeParameters.get(0);
    Type valueType = typeParameters.get(1);
    ObjectInspector keyInspector = inspector.getMapKeyObjectInspector();
    ObjectInspector valueInspector = inspector.getMapValueObjectInspector();
    BlockBuilder currentBuilder;
    boolean builderSynthesized = false;
    if (builder == null) {
        builderSynthesized = true;
        builder = type.createBlockBuilder(null, 1);
    }
    currentBuilder = builder.beginBlockEntry();
    for (Map.Entry<?, ?> entry : map.entrySet()) {
        // Hive skips map entries with null keys
        if (!filterNullMapKeys || entry.getKey() != null) {
            serializeObject(keyType, currentBuilder, entry.getKey(), keyInspector);
            serializeObject(valueType, currentBuilder, entry.getValue(), valueInspector);
        }
    }
    builder.closeEntry();
    if (builderSynthesized) {
        return (Block) type.getObject(builder, 0);
    } else {
        return null;
    }
}

Example 14 with MapObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project hive by apache.

the class AccumuloRowSerializer method writeWithLevel.

/**
 * Recursively serialize an Object using its {@link ObjectInspector}, respecting the
 * separators defined by the {@link LazySerDeParameters}.
 * @param oi ObjectInspector for the current object
 * @param value The current object
 * @param output A buffer output is written to
 * @param mapping The mapping for this Hive column
 * @param level The current level/offset for the SerDe separator
 * @throws IOException
 */
protected void writeWithLevel(ObjectInspector oi, Object value, ByteStream.Output output, ColumnMapping mapping, int level) throws IOException {
    switch(oi.getCategory()) {
        case PRIMITIVE:
            if (mapping.getEncoding() == ColumnEncoding.BINARY) {
                this.writeBinary(output, value, (PrimitiveObjectInspector) oi);
            } else {
                this.writeString(output, value, (PrimitiveObjectInspector) oi);
            }
            return;
        case LIST:
            char separator = (char) serDeParams.getSeparators()[level];
            ListObjectInspector loi = (ListObjectInspector) oi;
            List<?> list = loi.getList(value);
            ObjectInspector eoi = loi.getListElementObjectInspector();
            if (list == null) {
                log.debug("No objects found when serializing list");
                return;
            } else {
                for (int i = 0; i < list.size(); i++) {
                    if (i > 0) {
                        output.write(separator);
                    }
                    writeWithLevel(eoi, list.get(i), output, mapping, level + 1);
                }
            }
            return;
        case MAP:
            char sep = (char) serDeParams.getSeparators()[level];
            char keyValueSeparator = (char) serDeParams.getSeparators()[level + 1];
            MapObjectInspector moi = (MapObjectInspector) oi;
            ObjectInspector koi = moi.getMapKeyObjectInspector();
            ObjectInspector voi = moi.getMapValueObjectInspector();
            Map<?, ?> map = moi.getMap(value);
            if (map == null) {
                log.debug("No object found when serializing map");
                return;
            } else {
                boolean first = true;
                for (Map.Entry<?, ?> entry : map.entrySet()) {
                    if (first) {
                        first = false;
                    } else {
                        output.write(sep);
                    }
                    writeWithLevel(koi, entry.getKey(), output, mapping, level + 2);
                    output.write(keyValueSeparator);
                    writeWithLevel(voi, entry.getValue(), output, mapping, level + 2);
                }
            }
            return;
        case STRUCT:
            sep = (char) serDeParams.getSeparators()[level];
            StructObjectInspector soi = (StructObjectInspector) oi;
            List<? extends StructField> fields = soi.getAllStructFieldRefs();
            list = soi.getStructFieldsDataAsList(value);
            if (list == null) {
                log.debug("No object found when serializing struct");
                return;
            } else {
                for (int i = 0; i < list.size(); i++) {
                    if (i > 0) {
                        output.write(sep);
                    }
                    writeWithLevel(fields.get(i).getFieldObjectInspector(), list.get(i), output, mapping, level + 1);
                }
            }
            return;
        default:
            throw new RuntimeException("Unknown category type: " + oi.getCategory());
    }
}

Example 15 with MapObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector in project hive by apache.

the class VectorExtractRow method extractRowColumn.

public Object extractRowColumn(ColumnVector colVector, TypeInfo typeInfo, ObjectInspector objectInspector, int batchIndex) {
    if (colVector == null) {
        // may ask for them..
        return null;
    }
    final int adjustedIndex = (colVector.isRepeating ? 0 : batchIndex);
    if (!colVector.noNulls && colVector.isNull[adjustedIndex]) {
        return null;
    }
    final Category category = typeInfo.getCategory();
    switch(category) {
        case PRIMITIVE:
            {
                final PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) typeInfo;
                final PrimitiveCategory primitiveCategory = primitiveTypeInfo.getPrimitiveCategory();
                final Writable primitiveWritable = VectorizedBatchUtil.getPrimitiveWritable(primitiveCategory);
                switch(primitiveCategory) {
                    case VOID:
                        return null;
                    case BOOLEAN:
                        ((BooleanWritable) primitiveWritable).set(((LongColumnVector) colVector).vector[adjustedIndex] == 0 ? false : true);
                        return primitiveWritable;
                    case BYTE:
                        ((ByteWritable) primitiveWritable).set((byte) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case SHORT:
                        ((ShortWritable) primitiveWritable).set((short) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case INT:
                        ((IntWritable) primitiveWritable).set((int) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case LONG:
                        ((LongWritable) primitiveWritable).set(((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case TIMESTAMP:
                        // From java.sql.Timestamp used by vectorization to serializable org.apache.hadoop.hive.common.type.Timestamp
                        java.sql.Timestamp ts = ((TimestampColumnVector) colVector).asScratchTimestamp(adjustedIndex);
                        Timestamp serializableTS = Timestamp.ofEpochMilli(ts.getTime(), ts.getNanos());
                        ((TimestampWritableV2) primitiveWritable).set(serializableTS);
                        return primitiveWritable;
                    case DATE:
                        ((DateWritableV2) primitiveWritable).set((int) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case FLOAT:
                        ((FloatWritable) primitiveWritable).set((float) ((DoubleColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case DOUBLE:
                        ((DoubleWritable) primitiveWritable).set(((DoubleColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case BINARY:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            BytesWritable bytesWritable = (BytesWritable) primitiveWritable;
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                bytesWritable.set(EMPTY_BYTES, 0, 0);
                            } else {
                                bytesWritable.set(bytes, start, length);
                            }
                            return primitiveWritable;
                        }
                    case STRING:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                ((Text) primitiveWritable).set(EMPTY_BYTES, 0, 0);
                            } else {
                                // Use org.apache.hadoop.io.Text as our helper to go from byte[] to String.
                                ((Text) primitiveWritable).set(bytes, start, length);
                            }
                            return primitiveWritable;
                        }
                    case VARCHAR:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            final HiveVarcharWritable hiveVarcharWritable = (HiveVarcharWritable) primitiveWritable;
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                hiveVarcharWritable.set(EMPTY_STRING, -1);
                            } else {
                                final int adjustedLength = StringExpr.truncate(bytes, start, length, ((VarcharTypeInfo) primitiveTypeInfo).getLength());
                                if (adjustedLength == 0) {
                                    hiveVarcharWritable.set(EMPTY_STRING, -1);
                                } else {
                                    hiveVarcharWritable.set(new String(bytes, start, adjustedLength, Charsets.UTF_8), -1);
                                }
                            }
                            return primitiveWritable;
                        }
                    case CHAR:
                        {
                            final BytesColumnVector bytesColVector = ((BytesColumnVector) colVector);
                            final byte[] bytes = bytesColVector.vector[adjustedIndex];
                            final int start = bytesColVector.start[adjustedIndex];
                            final int length = bytesColVector.length[adjustedIndex];
                            final HiveCharWritable hiveCharWritable = (HiveCharWritable) primitiveWritable;
                            final int maxLength = ((CharTypeInfo) primitiveTypeInfo).getLength();
                            if (bytes == null || length == 0) {
                                if (length > 0) {
                                    nullBytesReadError(primitiveCategory, batchIndex);
                                }
                                hiveCharWritable.set(EMPTY_STRING, maxLength);
                            } else {
                                final int adjustedLength = StringExpr.rightTrimAndTruncate(bytes, start, length, ((CharTypeInfo) primitiveTypeInfo).getLength());
                                if (adjustedLength == 0) {
                                    hiveCharWritable.set(EMPTY_STRING, maxLength);
                                } else {
                                    hiveCharWritable.set(new String(bytes, start, adjustedLength, Charsets.UTF_8), maxLength);
                                }
                            }
                            return primitiveWritable;
                        }
                    case DECIMAL:
                        if (colVector instanceof Decimal64ColumnVector) {
                            Decimal64ColumnVector dec32ColVector = (Decimal64ColumnVector) colVector;
                            ((HiveDecimalWritable) primitiveWritable).deserialize64(dec32ColVector.vector[adjustedIndex], dec32ColVector.scale);
                        } else {
                            // The HiveDecimalWritable set method will quickly copy the deserialized decimal writable fields.
                            ((HiveDecimalWritable) primitiveWritable).set(((DecimalColumnVector) colVector).vector[adjustedIndex]);
                        }
                        return primitiveWritable;
                    case INTERVAL_YEAR_MONTH:
                        ((HiveIntervalYearMonthWritable) primitiveWritable).set((int) ((LongColumnVector) colVector).vector[adjustedIndex]);
                        return primitiveWritable;
                    case INTERVAL_DAY_TIME:
                        ((HiveIntervalDayTimeWritable) primitiveWritable).set(((IntervalDayTimeColumnVector) colVector).asScratchIntervalDayTime(adjustedIndex));
                        return primitiveWritable;
                    default:
                        throw new RuntimeException("Primitive category " + primitiveCategory.name() + " not supported");
                }
            }
        case LIST:
            {
                final ListColumnVector listColumnVector = (ListColumnVector) colVector;
                final ListTypeInfo listTypeInfo = (ListTypeInfo) typeInfo;
                final ListObjectInspector listObjectInspector = (ListObjectInspector) objectInspector;
                final int offset = (int) listColumnVector.offsets[adjustedIndex];
                final int size = (int) listColumnVector.lengths[adjustedIndex];
                final List list = new ArrayList();
                for (int i = 0; i < size; i++) {
                    list.add(extractRowColumn(listColumnVector.child, listTypeInfo.getListElementTypeInfo(), listObjectInspector.getListElementObjectInspector(), offset + i));
                }
                return list;
            }
        case MAP:
            {
                final MapColumnVector mapColumnVector = (MapColumnVector) colVector;
                final MapTypeInfo mapTypeInfo = (MapTypeInfo) typeInfo;
                final MapObjectInspector mapObjectInspector = (MapObjectInspector) objectInspector;
                final int offset = (int) mapColumnVector.offsets[adjustedIndex];
                final int size = (int) mapColumnVector.lengths[adjustedIndex];
                final Map<Object, Object> map = new LinkedHashMap<Object, Object>();
                for (int i = 0; i < size; i++) {
                    final Object key = extractRowColumn(mapColumnVector.keys, mapTypeInfo.getMapKeyTypeInfo(), mapObjectInspector.getMapKeyObjectInspector(), offset + i);
                    final Object value = extractRowColumn(mapColumnVector.values, mapTypeInfo.getMapValueTypeInfo(), mapObjectInspector.getMapValueObjectInspector(), offset + i);
                    map.put(key, value);
                }
                return map;
            }
        case STRUCT:
            {
                final StructColumnVector structColumnVector = (StructColumnVector) colVector;
                final StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
                final StandardStructObjectInspector structInspector = (StandardStructObjectInspector) objectInspector;
                final List<TypeInfo> fieldTypeInfos = structTypeInfo.getAllStructFieldTypeInfos();
                final int size = fieldTypeInfos.size();
                final List<? extends StructField> structFields = structInspector.getAllStructFieldRefs();
                final Object struct = structInspector.create();
                for (int i = 0; i < size; i++) {
                    final StructField structField = structFields.get(i);
                    final TypeInfo fieldTypeInfo = fieldTypeInfos.get(i);
                    final Object value = extractRowColumn(structColumnVector.fields[i], fieldTypeInfo, structField.getFieldObjectInspector(), adjustedIndex);
                    structInspector.setStructFieldData(struct, structField, value);
                }
                return struct;
            }
        case UNION:
            {
                final UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
                final List<TypeInfo> objectTypeInfos = unionTypeInfo.getAllUnionObjectTypeInfos();
                final UnionObjectInspector unionInspector = (UnionObjectInspector) objectInspector;
                final List<ObjectInspector> unionInspectors = unionInspector.getObjectInspectors();
                final UnionColumnVector unionColumnVector = (UnionColumnVector) colVector;
                final byte tag = (byte) unionColumnVector.tags[adjustedIndex];
                final Object object = extractRowColumn(unionColumnVector.fields[tag], objectTypeInfos.get(tag), unionInspectors.get(tag), adjustedIndex);
                final StandardUnion standardUnion = new StandardUnion();
                standardUnion.setTag(tag);
                standardUnion.setObject(object);
                return standardUnion;
            }
        default:
            throw new RuntimeException("Category " + category.name() + " not supported");
    }
}