Examples with LazyBinaryColumnarSerDe org.apache.hadoop.hive.serde2.columnar.LazyBinaryColumnarSerDe used on opensource projects

Example 6 with LazyBinaryColumnarSerDe

use of org.apache.hadoop.hive.serde2.columnar.LazyBinaryColumnarSerDe in project hive by apache.

the class TestLazyBinaryColumnarSerDe method testSerDe.

public void testSerDe() throws SerDeException {
    StructObjectInspector oi = (StructObjectInspector) ObjectInspectorFactory.getReflectionObjectInspector(OuterStruct.class, ObjectInspectorOptions.JAVA);
    String cols = ObjectInspectorUtils.getFieldNames(oi);
    Properties props = new Properties();
    props.setProperty(serdeConstants.LIST_COLUMNS, cols);
    props.setProperty(serdeConstants.LIST_COLUMN_TYPES, ObjectInspectorUtils.getFieldTypes(oi));
    LazyBinaryColumnarSerDe serde = new LazyBinaryColumnarSerDe();
    SerDeUtils.initializeSerDe(serde, new Configuration(), props, null);
    OuterStruct outerStruct = new OuterStruct();
    outerStruct.mByte = 1;
    outerStruct.mShort = 2;
    outerStruct.mInt = 3;
    outerStruct.mLong = 4l;
    outerStruct.mFloat = 5.01f;
    outerStruct.mDouble = 6.001d;
    outerStruct.mString = "seven";
    outerStruct.mBA = new byte[] { '2' };
    InnerStruct is1 = new InnerStruct(8, 9l);
    InnerStruct is2 = new InnerStruct(10, 11l);
    outerStruct.mArray = new ArrayList<InnerStruct>(2);
    outerStruct.mArray.add(is1);
    outerStruct.mArray.add(is2);
    outerStruct.mMap = new TreeMap<String, InnerStruct>();
    outerStruct.mMap.put(new String("twelve"), new InnerStruct(13, 14l));
    outerStruct.mMap.put(new String("fifteen"), new InnerStruct(16, 17l));
    outerStruct.mStruct = new InnerStruct(18, 19l);
    BytesRefArrayWritable braw = (BytesRefArrayWritable) serde.serialize(outerStruct, oi);
    ObjectInspector out_oi = serde.getObjectInspector();
    Object out_o = serde.deserialize(braw);
    if (0 != ObjectInspectorUtils.compare(outerStruct, oi, out_o, out_oi, new CrossMapEqualComparer())) {
        System.out.println("expected = " + SerDeUtils.getJSONString(outerStruct, oi));
        System.out.println("actual = " + SerDeUtils.getJSONString(out_o, out_oi));
        fail("Deserialized object does not compare");
    }
}

Example 7 with LazyBinaryColumnarSerDe

use of org.apache.hadoop.hive.serde2.columnar.LazyBinaryColumnarSerDe in project hive by apache.

the class TestLazyBinaryColumnarSerDe method testHandlingAlteredSchemas.

/**
 * HIVE-5788
 * <p>
 * Background: in cases of "add column", table metadata changes but data does not.  Columns
 * missing from the data but which are required by metadata are interpreted as null.
 * <p>
 * This tests the use-case of altering columns of a table with already some data, then adding more data
 * in the new schema, and seeing if this serde can to read both types of data from the resultant table.
 * @throws SerDeException
 */
public void testHandlingAlteredSchemas() throws SerDeException {
    StructObjectInspector oi = (StructObjectInspector) ObjectInspectorFactory.getReflectionObjectInspector(BeforeStruct.class, ObjectInspectorOptions.JAVA);
    String cols = ObjectInspectorUtils.getFieldNames(oi);
    Properties props = new Properties();
    props.setProperty(serdeConstants.LIST_COLUMNS, cols);
    props.setProperty(serdeConstants.LIST_COLUMN_TYPES, ObjectInspectorUtils.getFieldTypes(oi));
    // serialize some data in the schema before it is altered.
    LazyBinaryColumnarSerDe serde = new LazyBinaryColumnarSerDe();
    SerDeUtils.initializeSerDe(serde, new Configuration(), props, null);
    BeforeStruct bs1 = new BeforeStruct();
    bs1.l1 = 1L;
    bs1.l2 = 2L;
    BytesRefArrayWritable braw1 = (BytesRefArrayWritable) serde.serialize(bs1, oi);
    // alter table add column: change the metadata
    oi = (StructObjectInspector) ObjectInspectorFactory.getReflectionObjectInspector(AfterStruct.class, ObjectInspectorOptions.JAVA);
    cols = ObjectInspectorUtils.getFieldNames(oi);
    props = new Properties();
    props.setProperty(serdeConstants.LIST_COLUMNS, cols);
    props.setProperty(serdeConstants.LIST_COLUMN_TYPES, ObjectInspectorUtils.getFieldTypes(oi));
    serde = new LazyBinaryColumnarSerDe();
    SerDeUtils.initializeSerDe(serde, new Configuration(), props, null);
    // serialize some data in the schema after it is altered.
    AfterStruct as = new AfterStruct();
    as.l1 = 11L;
    as.l2 = 12L;
    as.l3 = 13L;
    BytesRefArrayWritable braw2 = (BytesRefArrayWritable) serde.serialize(as, oi);
    // fetch operator
    serde = new LazyBinaryColumnarSerDe();
    SerDeUtils.initializeSerDe(serde, new Configuration(), props, null);
    // fetch the row inserted before schema is altered and verify
    LazyBinaryColumnarStruct struct1 = (LazyBinaryColumnarStruct) serde.deserialize(braw1);
    oi = (StructObjectInspector) serde.getObjectInspector();
    List<Object> objs1 = oi.getStructFieldsDataAsList(struct1);
    Assert.assertEquals(((LongWritable) objs1.get(0)).get(), 1L);
    Assert.assertEquals(((LongWritable) objs1.get(1)).get(), 2L);
    Assert.assertNull(objs1.get(2));
    // fetch the row inserted after schema is altered and verify
    LazyBinaryColumnarStruct struct2 = (LazyBinaryColumnarStruct) serde.deserialize(braw2);
    List<Object> objs2 = struct2.getFieldsAsList();
    Assert.assertEquals(((LongWritable) objs2.get(0)).get(), 11L);
    Assert.assertEquals(((LongWritable) objs2.get(1)).get(), 12L);
    Assert.assertEquals(((LongWritable) objs2.get(2)).get(), 13L);
}

Example 8 with LazyBinaryColumnarSerDe

use of org.apache.hadoop.hive.serde2.columnar.LazyBinaryColumnarSerDe 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());
            }
    }
}