Examples with LazyBinarySerDe org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe used on opensource projects

Example 11 with LazyBinarySerDe

use of org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe in project hive by apache.

the class LlapRowRecordReader method initSerDe.

protected AbstractSerDe initSerDe(Configuration conf) throws SerDeException {
    Properties props = new Properties();
    StringBuilder columnsBuffer = new StringBuilder();
    StringBuilder typesBuffer = new StringBuilder();
    boolean isFirst = true;
    for (FieldDesc colDesc : schema.getColumns()) {
        if (!isFirst) {
            columnsBuffer.append(',');
            typesBuffer.append(',');
        }
        columnsBuffer.append(colDesc.getName());
        typesBuffer.append(colDesc.getTypeInfo().toString());
        isFirst = false;
    }
    String columns = columnsBuffer.toString();
    String types = typesBuffer.toString();
    props.put(serdeConstants.LIST_COLUMNS, columns);
    props.put(serdeConstants.LIST_COLUMN_TYPES, types);
    props.put(serdeConstants.ESCAPE_CHAR, "\\");
    AbstractSerDe serde = new LazyBinarySerDe();
    serde.initialize(conf, props);
    return serde;
}

Example 12 with LazyBinarySerDe

use of org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe in project hive by apache.

the class TestStatsSerde method testLazyBinarySerDe.

/**
 * Test LazyBinarySerDe
 */
public void testLazyBinarySerDe() throws Throwable {
    try {
        System.out.println("test: testLazyBinarySerDe");
        int num = 1000;
        Random r = new Random(1234);
        MyTestClass[] rows = new MyTestClass[num];
        for (int i = 0; i < num; i++) {
            MyTestClass t = new MyTestClass();
            ExtraTypeInfo extraTypeInfo = new ExtraTypeInfo();
            t.randomFill(r, extraTypeInfo);
            rows[i] = t;
        }
        StructObjectInspector rowOI = (StructObjectInspector) ObjectInspectorFactory.getReflectionObjectInspector(MyTestClass.class, ObjectInspectorOptions.JAVA);
        String fieldNames = ObjectInspectorUtils.getFieldNames(rowOI);
        String fieldTypes = ObjectInspectorUtils.getFieldTypes(rowOI);
        Properties schema = new Properties();
        schema.setProperty(serdeConstants.LIST_COLUMNS, fieldNames);
        schema.setProperty(serdeConstants.LIST_COLUMN_TYPES, fieldTypes);
        LazyBinarySerDe serDe = new LazyBinarySerDe();
        SerDeUtils.initializeSerDe(serDe, new Configuration(), schema, null);
        deserializeAndSerializeLazyBinary(serDe, rows, rowOI);
        System.out.println("test: testLazyBinarySerDe - OK");
    } catch (Throwable e) {
        e.printStackTrace();
        throw e;
    }
}

Example 13 with LazyBinarySerDe

use of org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe in project hive by apache.

the class LazyBinarySerDe method serialize.

/**
 * A recursive function that serialize an object to a byte buffer based on its
 * object inspector.
 *
 * @param byteStream
 *          the byte stream storing the serialization data
 * @param obj
 *          the object to serialize
 * @param objInspector
 *          the object inspector
 * @param skipLengthPrefix a boolean indicating whether length prefix is
 *          needed for list/map/struct
 * @param warnedOnceNullMapKey a boolean indicating whether a warning
 *          has been issued once already when encountering null map keys
 */
public static void serialize(RandomAccessOutput byteStream, Object obj, ObjectInspector objInspector, boolean skipLengthPrefix, BooleanRef warnedOnceNullMapKey) throws SerDeException {
    // do nothing for null object
    if (null == obj) {
        return;
    }
    switch(objInspector.getCategory()) {
        case PRIMITIVE:
            {
                PrimitiveObjectInspector poi = (PrimitiveObjectInspector) objInspector;
                switch(poi.getPrimitiveCategory()) {
                    case VOID:
                        {
                            return;
                        }
                    case BOOLEAN:
                        {
                            boolean v = ((BooleanObjectInspector) poi).get(obj);
                            byteStream.write((byte) (v ? 1 : 0));
                            return;
                        }
                    case BYTE:
                        {
                            ByteObjectInspector boi = (ByteObjectInspector) poi;
                            byte v = boi.get(obj);
                            byteStream.write(v);
                            return;
                        }
                    case SHORT:
                        {
                            ShortObjectInspector spoi = (ShortObjectInspector) poi;
                            short v = spoi.get(obj);
                            byteStream.write((byte) (v >> 8));
                            byteStream.write((byte) (v));
                            return;
                        }
                    case INT:
                        {
                            IntObjectInspector ioi = (IntObjectInspector) poi;
                            int v = ioi.get(obj);
                            LazyBinaryUtils.writeVInt(byteStream, v);
                            return;
                        }
                    case LONG:
                        {
                            LongObjectInspector loi = (LongObjectInspector) poi;
                            long v = loi.get(obj);
                            LazyBinaryUtils.writeVLong(byteStream, v);
                            return;
                        }
                    case FLOAT:
                        {
                            FloatObjectInspector foi = (FloatObjectInspector) poi;
                            int v = Float.floatToIntBits(foi.get(obj));
                            byteStream.write((byte) (v >> 24));
                            byteStream.write((byte) (v >> 16));
                            byteStream.write((byte) (v >> 8));
                            byteStream.write((byte) (v));
                            return;
                        }
                    case DOUBLE:
                        {
                            DoubleObjectInspector doi = (DoubleObjectInspector) poi;
                            LazyBinaryUtils.writeDouble(byteStream, doi.get(obj));
                            return;
                        }
                    case STRING:
                        {
                            StringObjectInspector soi = (StringObjectInspector) poi;
                            Text t = soi.getPrimitiveWritableObject(obj);
                            serializeText(byteStream, t, skipLengthPrefix);
                            return;
                        }
                    case CHAR:
                        {
                            HiveCharObjectInspector hcoi = (HiveCharObjectInspector) poi;
                            Text t = hcoi.getPrimitiveWritableObject(obj).getTextValue();
                            serializeText(byteStream, t, skipLengthPrefix);
                            return;
                        }
                    case VARCHAR:
                        {
                            HiveVarcharObjectInspector hcoi = (HiveVarcharObjectInspector) poi;
                            Text t = hcoi.getPrimitiveWritableObject(obj).getTextValue();
                            serializeText(byteStream, t, skipLengthPrefix);
                            return;
                        }
                    case BINARY:
                        {
                            BinaryObjectInspector baoi = (BinaryObjectInspector) poi;
                            BytesWritable bw = baoi.getPrimitiveWritableObject(obj);
                            int length = bw.getLength();
                            if (!skipLengthPrefix) {
                                LazyBinaryUtils.writeVInt(byteStream, length);
                            } else {
                                if (length == 0) {
                                    throw new RuntimeException("LazyBinaryColumnarSerde cannot serialize a non-null zero " + "length binary field. Consider using either LazyBinarySerde or ColumnarSerde.");
                                }
                            }
                            byteStream.write(bw.getBytes(), 0, length);
                            return;
                        }
                    case DATE:
                        {
                            DateWritable d = ((DateObjectInspector) poi).getPrimitiveWritableObject(obj);
                            writeDateToByteStream(byteStream, d);
                            return;
                        }
                    case TIMESTAMP:
                        {
                            TimestampObjectInspector toi = (TimestampObjectInspector) poi;
                            TimestampWritable t = toi.getPrimitiveWritableObject(obj);
                            t.writeToByteStream(byteStream);
                            return;
                        }
                    case TIMESTAMPLOCALTZ:
                        {
                            TimestampLocalTZWritable t = ((TimestampLocalTZObjectInspector) poi).getPrimitiveWritableObject(obj);
                            t.writeToByteStream(byteStream);
                            return;
                        }
                    case INTERVAL_YEAR_MONTH:
                        {
                            HiveIntervalYearMonthWritable intervalYearMonth = ((HiveIntervalYearMonthObjectInspector) poi).getPrimitiveWritableObject(obj);
                            intervalYearMonth.writeToByteStream(byteStream);
                            return;
                        }
                    case INTERVAL_DAY_TIME:
                        {
                            HiveIntervalDayTimeWritable intervalDayTime = ((HiveIntervalDayTimeObjectInspector) poi).getPrimitiveWritableObject(obj);
                            intervalDayTime.writeToByteStream(byteStream);
                            return;
                        }
                    case DECIMAL:
                        {
                            HiveDecimalObjectInspector bdoi = (HiveDecimalObjectInspector) poi;
                            HiveDecimalWritable t = bdoi.getPrimitiveWritableObject(obj);
                            if (t == null) {
                                return;
                            }
                            writeToByteStream(byteStream, t);
                            return;
                        }
                    default:
                        {
                            throw new RuntimeException("Unrecognized type: " + poi.getPrimitiveCategory());
                        }
                }
            }
        case LIST:
            {
                ListObjectInspector loi = (ListObjectInspector) objInspector;
                ObjectInspector eoi = loi.getListElementObjectInspector();
                int byteSizeStart = 0;
                int listStart = 0;
                if (!skipLengthPrefix) {
                    // 1/ reserve spaces for the byte size of the list
                    // which is a integer and takes four bytes
                    byteSizeStart = byteStream.getLength();
                    byteStream.reserve(4);
                    listStart = byteStream.getLength();
                }
                // 2/ write the size of the list as a VInt
                int size = loi.getListLength(obj);
                LazyBinaryUtils.writeVInt(byteStream, size);
                // 3/ write the null bytes
                byte nullByte = 0;
                for (int eid = 0; eid < size; eid++) {
                    // set the bit to 1 if an element is not null
                    if (null != loi.getListElement(obj, eid)) {
                        nullByte |= 1 << (eid % 8);
                    }
                    // if this is the last element
                    if (7 == eid % 8 || eid == size - 1) {
                        byteStream.write(nullByte);
                        nullByte = 0;
                    }
                }
                // 4/ write element by element from the list
                for (int eid = 0; eid < size; eid++) {
                    serialize(byteStream, loi.getListElement(obj, eid), eoi, false, warnedOnceNullMapKey);
                }
                if (!skipLengthPrefix) {
                    // 5/ update the list byte size
                    int listEnd = byteStream.getLength();
                    int listSize = listEnd - listStart;
                    writeSizeAtOffset(byteStream, byteSizeStart, listSize);
                }
                return;
            }
        case MAP:
            {
                MapObjectInspector moi = (MapObjectInspector) objInspector;
                ObjectInspector koi = moi.getMapKeyObjectInspector();
                ObjectInspector voi = moi.getMapValueObjectInspector();
                Map<?, ?> map = moi.getMap(obj);
                int byteSizeStart = 0;
                int mapStart = 0;
                if (!skipLengthPrefix) {
                    // 1/ reserve spaces for the byte size of the map
                    // which is a integer and takes four bytes
                    byteSizeStart = byteStream.getLength();
                    byteStream.reserve(4);
                    mapStart = byteStream.getLength();
                }
                // 2/ write the size of the map which is a VInt
                int size = map.size();
                LazyBinaryUtils.writeVInt(byteStream, size);
                // 3/ write the null bytes
                int b = 0;
                byte nullByte = 0;
                for (Map.Entry<?, ?> entry : map.entrySet()) {
                    // set the bit to 1 if a key is not null
                    if (null != entry.getKey()) {
                        nullByte |= 1 << (b % 8);
                    } else if (warnedOnceNullMapKey != null) {
                        if (!warnedOnceNullMapKey.value) {
                            LOG.warn("Null map key encountered! Ignoring similar problems.");
                        }
                        warnedOnceNullMapKey.value = true;
                    }
                    b++;
                    // set the bit to 1 if a value is not null
                    if (null != entry.getValue()) {
                        nullByte |= 1 << (b % 8);
                    }
                    b++;
                    // or if this is the last key-value pair
                    if (0 == b % 8 || b == size * 2) {
                        byteStream.write(nullByte);
                        nullByte = 0;
                    }
                }
                // 4/ write key-value pairs one by one
                for (Map.Entry<?, ?> entry : map.entrySet()) {
                    serialize(byteStream, entry.getKey(), koi, false, warnedOnceNullMapKey);
                    serialize(byteStream, entry.getValue(), voi, false, warnedOnceNullMapKey);
                }
                if (!skipLengthPrefix) {
                    // 5/ update the byte size of the map
                    int mapEnd = byteStream.getLength();
                    int mapSize = mapEnd - mapStart;
                    writeSizeAtOffset(byteStream, byteSizeStart, mapSize);
                }
                return;
            }
        case STRUCT:
        case UNION:
            {
                int byteSizeStart = 0;
                int typeStart = 0;
                if (!skipLengthPrefix) {
                    // 1/ reserve spaces for the byte size of the struct
                    // which is a integer and takes four bytes
                    byteSizeStart = byteStream.getLength();
                    byteStream.reserve(4);
                    typeStart = byteStream.getLength();
                }
                if (ObjectInspector.Category.STRUCT.equals(objInspector.getCategory())) {
                    // 2/ serialize the struct
                    serializeStruct(byteStream, obj, (StructObjectInspector) objInspector, warnedOnceNullMapKey);
                } else {
                    // 2/ serialize the union
                    serializeUnion(byteStream, obj, (UnionObjectInspector) objInspector, warnedOnceNullMapKey);
                }
                if (!skipLengthPrefix) {
                    // 3/ update the byte size of the struct
                    int typeEnd = byteStream.getLength();
                    int typeSize = typeEnd - typeStart;
                    writeSizeAtOffset(byteStream, byteSizeStart, typeSize);
                }
                return;
            }
        default:
            {
                throw new RuntimeException("Unrecognized type: " + objInspector.getCategory());
            }
    }
}

Example 14 with LazyBinarySerDe

use of org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe in project hive by apache.

the class LazyBinarySerDe method initialize.

/**
 * Initialize the SerDe with configuration and table information.
 */
@Override
public void initialize(Configuration conf, Properties tbl) throws SerDeException {
    // Get column names and types
    String columnNameProperty = tbl.getProperty(serdeConstants.LIST_COLUMNS);
    String columnNameDelimiter = tbl.containsKey(serdeConstants.COLUMN_NAME_DELIMITER) ? tbl.getProperty(serdeConstants.COLUMN_NAME_DELIMITER) : String.valueOf(SerDeUtils.COMMA);
    String columnTypeProperty = tbl.getProperty(serdeConstants.LIST_COLUMN_TYPES);
    if (columnNameProperty.length() == 0) {
        columnNames = new ArrayList<String>();
    } else {
        columnNames = Arrays.asList(columnNameProperty.split(columnNameDelimiter));
    }
    if (columnTypeProperty.length() == 0) {
        columnTypes = new ArrayList<TypeInfo>();
    } else {
        columnTypes = TypeInfoUtils.getTypeInfosFromTypeString(columnTypeProperty);
    }
    assert (columnNames.size() == columnTypes.size());
    // Create row related objects
    rowTypeInfo = TypeInfoFactory.getStructTypeInfo(columnNames, columnTypes);
    // Create the object inspector and the lazy binary struct object
    cachedObjectInspector = LazyBinaryUtils.getLazyBinaryObjectInspectorFromTypeInfo(rowTypeInfo);
    cachedLazyBinaryStruct = (LazyBinaryStruct) LazyBinaryFactory.createLazyBinaryObject(cachedObjectInspector);
    // output debug info
    LOG.debug("LazyBinarySerDe initialized with: columnNames=" + columnNames + " columnTypes=" + columnTypes);
    serializedSize = 0;
    stats = new SerDeStats();
    lastOperationSerialize = false;
    lastOperationDeserialize = false;
}

Example 15 with LazyBinarySerDe

use of org.apache.hadoop.hive.serde2.lazybinary.LazyBinarySerDe in project hive by apache.

the class TestLazyBinarySerDe method testLazyBinarySerDe.

/**
 * The test entrance function.
 *
 * @throws Throwable
 */
public void testLazyBinarySerDe() throws Throwable {
    try {
        System.out.println("Beginning Test TestLazyBinarySerDe:");
        // generate the data
        int num = 1000;
        Random r = new Random(1234);
        MyTestClass[] rows = new MyTestClass[num];
        for (int i = 0; i < num; i++) {
            MyTestClass t = new MyTestClass();
            ExtraTypeInfo extraTypeInfo = new ExtraTypeInfo();
            t.randomFill(r, extraTypeInfo);
            rows[i] = t;
        }
        StructObjectInspector rowOI = (StructObjectInspector) ObjectInspectorFactory.getReflectionObjectInspector(MyTestClass.class, ObjectInspectorOptions.JAVA);
        String fieldNames = ObjectInspectorUtils.getFieldNames(rowOI);
        String fieldTypes = ObjectInspectorUtils.getFieldTypes(rowOI);
        // call the tests
        // 1/ test LazyBinarySerDe
        testLazyBinarySerDe(rows, rowOI, getSerDe(fieldNames, fieldTypes));
        // 2/ test LazyBinaryMap
        testLazyBinaryMap(r);
        // 3/ test serialization and deserialization with different schemas
        testShorterSchemaDeserialization(r);
        // 4/ test serialization and deserialization with different schemas
        testLongerSchemaDeserialization(r);
        // 5/ test serialization and deserialization with different schemas
        testShorterSchemaDeserialization1(r);
        // 6/ test serialization and deserialization with different schemas
        testLongerSchemaDeserialization1(r);
        System.out.println("Test TestLazyBinarySerDe passed!");
    } catch (Throwable e) {
        e.printStackTrace();
        throw e;
    }
}