Search in sources :

Example 76 with DateWritable

use of org.apache.hadoop.hive.serde2.io.DateWritable 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());
            }
    }
}
Also used : LongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.LongObjectInspector) DateObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DateObjectInspector) IntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector) BinaryObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BinaryObjectInspector) TimestampWritable(org.apache.hadoop.hive.serde2.io.TimestampWritable) StringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.StringObjectInspector) FloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.FloatObjectInspector) ByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ByteObjectInspector) TimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampObjectInspector) ShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ShortObjectInspector) MapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector) ListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector) HiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveDecimalObjectInspector) HiveCharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveCharObjectInspector) UnionObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector) UnionObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.UnionObjectInspector) HiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveDecimalObjectInspector) BooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector) ShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ShortObjectInspector) HiveIntervalYearMonthObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveIntervalYearMonthObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) MapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector) FloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.FloatObjectInspector) StringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.StringObjectInspector) HiveIntervalDayTimeObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveIntervalDayTimeObjectInspector) DateObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DateObjectInspector) ListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector) TimestampLocalTZObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampLocalTZObjectInspector) HiveVarcharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveVarcharObjectInspector) HiveCharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveCharObjectInspector) IntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector) PrimitiveObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector) LongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.LongObjectInspector) BinaryObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BinaryObjectInspector) ByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ByteObjectInspector) DoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DoubleObjectInspector) TimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampObjectInspector) DateWritable(org.apache.hadoop.hive.serde2.io.DateWritable) HiveDecimalWritable(org.apache.hadoop.hive.serde2.io.HiveDecimalWritable) HiveIntervalDayTimeObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveIntervalDayTimeObjectInspector) TimestampLocalTZObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampLocalTZObjectInspector) Text(org.apache.hadoop.io.Text) BytesWritable(org.apache.hadoop.io.BytesWritable) HiveIntervalDayTimeWritable(org.apache.hadoop.hive.serde2.io.HiveIntervalDayTimeWritable) HiveIntervalYearMonthWritable(org.apache.hadoop.hive.serde2.io.HiveIntervalYearMonthWritable) DoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DoubleObjectInspector) HiveVarcharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveVarcharObjectInspector) HiveIntervalYearMonthObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveIntervalYearMonthObjectInspector) TimestampLocalTZWritable(org.apache.hadoop.hive.serde2.io.TimestampLocalTZWritable) PrimitiveObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector) BooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector) Map(java.util.Map) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)

Example 77 with DateWritable

use of org.apache.hadoop.hive.serde2.io.DateWritable in project hive by apache.

the class VerifyLazy method lazyCompare.

public static boolean lazyCompare(TypeInfo typeInfo, Object lazyObject, Object expectedObject) {
    if (expectedObject == null) {
        if (lazyObject != null) {
            throw new RuntimeException("Expected object is null but object is not null " + lazyObject.toString() + " typeInfo " + typeInfo.toString());
        }
        return true;
    } else if (lazyObject == null) {
        throw new RuntimeException("Expected object is not null \"" + expectedObject.toString() + "\" typeInfo " + typeInfo.toString() + " but object is null");
    }
    if (lazyObject instanceof Writable) {
        if (!lazyObject.equals(expectedObject)) {
            throw new RuntimeException("Expected object " + expectedObject.toString() + " and actual object " + lazyObject.toString() + " is not equal typeInfo " + typeInfo.toString());
        }
        return true;
    }
    if (lazyObject instanceof LazyPrimitive) {
        Object primitiveObject = ((LazyPrimitive) lazyObject).getObject();
        PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) typeInfo;
        switch(primitiveTypeInfo.getPrimitiveCategory()) {
            case BOOLEAN:
                {
                    if (!(primitiveObject instanceof LazyBoolean)) {
                        throw new RuntimeException("Expected LazyBoolean");
                    }
                    boolean value = ((LazyBoolean) primitiveObject).getWritableObject().get();
                    boolean expected = ((BooleanWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Boolean field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case BYTE:
                {
                    if (!(primitiveObject instanceof LazyByte)) {
                        throw new RuntimeException("Expected LazyByte");
                    }
                    byte value = ((LazyByte) primitiveObject).getWritableObject().get();
                    byte expected = ((ByteWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Byte field mismatch (expected " + (int) expected + " found " + (int) value + ")");
                    }
                }
                break;
            case SHORT:
                {
                    if (!(primitiveObject instanceof LazyShort)) {
                        throw new RuntimeException("Expected LazyShort");
                    }
                    short value = ((LazyShort) primitiveObject).getWritableObject().get();
                    short expected = ((ShortWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Short field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case INT:
                {
                    if (!(primitiveObject instanceof LazyInteger)) {
                        throw new RuntimeException("Expected LazyInteger");
                    }
                    int value = ((LazyInteger) primitiveObject).getWritableObject().get();
                    int expected = ((IntWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Int field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case LONG:
                {
                    if (!(primitiveObject instanceof LazyLong)) {
                        throw new RuntimeException("Expected LazyLong");
                    }
                    long value = ((LazyLong) primitiveObject).getWritableObject().get();
                    long expected = ((LongWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Long field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case FLOAT:
                {
                    if (!(primitiveObject instanceof LazyFloat)) {
                        throw new RuntimeException("Expected LazyFloat");
                    }
                    float value = ((LazyFloat) primitiveObject).getWritableObject().get();
                    float expected = ((FloatWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Float field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case DOUBLE:
                {
                    if (!(primitiveObject instanceof LazyDouble)) {
                        throw new RuntimeException("Expected LazyDouble");
                    }
                    double value = ((LazyDouble) primitiveObject).getWritableObject().get();
                    double expected = ((DoubleWritable) expectedObject).get();
                    if (value != expected) {
                        throw new RuntimeException("Double field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case STRING:
                {
                    if (!(primitiveObject instanceof LazyString)) {
                        throw new RuntimeException("Text expected writable not Text");
                    }
                    Text value = ((LazyString) primitiveObject).getWritableObject();
                    Text expected = ((Text) expectedObject);
                    if (!value.equals(expected)) {
                        throw new RuntimeException("String field mismatch (expected '" + expected + "' found '" + value + "')");
                    }
                }
                break;
            case CHAR:
                {
                    if (!(primitiveObject instanceof LazyHiveChar)) {
                        throw new RuntimeException("Expected LazyHiveChar");
                    }
                    HiveChar value = ((LazyHiveChar) primitiveObject).getWritableObject().getHiveChar();
                    HiveChar expected = ((HiveCharWritable) expectedObject).getHiveChar();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveChar field mismatch (expected '" + expected + "' found '" + value + "')");
                    }
                }
                break;
            case VARCHAR:
                {
                    if (!(primitiveObject instanceof LazyHiveVarchar)) {
                        throw new RuntimeException("Expected LazyHiveVarchar");
                    }
                    HiveVarchar value = ((LazyHiveVarchar) primitiveObject).getWritableObject().getHiveVarchar();
                    HiveVarchar expected = ((HiveVarcharWritable) expectedObject).getHiveVarchar();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveVarchar field mismatch (expected '" + expected + "' found '" + value + "')");
                    }
                }
                break;
            case DECIMAL:
                {
                    if (!(primitiveObject instanceof LazyHiveDecimal)) {
                        throw new RuntimeException("Expected LazyDecimal");
                    }
                    HiveDecimal value = ((LazyHiveDecimal) primitiveObject).getWritableObject().getHiveDecimal();
                    HiveDecimal expected = ((HiveDecimalWritable) expectedObject).getHiveDecimal();
                    if (!value.equals(expected)) {
                        DecimalTypeInfo decimalTypeInfo = (DecimalTypeInfo) primitiveTypeInfo;
                        int precision = decimalTypeInfo.getPrecision();
                        int scale = decimalTypeInfo.getScale();
                        throw new RuntimeException("Decimal field mismatch (expected " + expected.toString() + " found " + value.toString() + ") precision " + precision + ", scale " + scale);
                    }
                }
                break;
            case DATE:
                {
                    if (!(primitiveObject instanceof LazyDate)) {
                        throw new RuntimeException("Expected LazyDate");
                    }
                    Date value = ((LazyDate) primitiveObject).getWritableObject().get();
                    Date expected = ((DateWritable) expectedObject).get();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("Date field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case TIMESTAMP:
                {
                    if (!(primitiveObject instanceof LazyTimestamp)) {
                        throw new RuntimeException("TimestampWritable expected writable not TimestampWritable");
                    }
                    Timestamp value = ((LazyTimestamp) primitiveObject).getWritableObject().getTimestamp();
                    Timestamp expected = ((TimestampWritable) expectedObject).getTimestamp();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("Timestamp field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case INTERVAL_YEAR_MONTH:
                {
                    if (!(primitiveObject instanceof LazyHiveIntervalYearMonth)) {
                        throw new RuntimeException("Expected LazyHiveIntervalYearMonth");
                    }
                    HiveIntervalYearMonth value = ((LazyHiveIntervalYearMonth) primitiveObject).getWritableObject().getHiveIntervalYearMonth();
                    HiveIntervalYearMonth expected = ((HiveIntervalYearMonthWritable) expectedObject).getHiveIntervalYearMonth();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveIntervalYearMonth field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case INTERVAL_DAY_TIME:
                {
                    if (!(primitiveObject instanceof LazyHiveIntervalDayTime)) {
                        throw new RuntimeException("Expected writable LazyHiveIntervalDayTime");
                    }
                    HiveIntervalDayTime value = ((LazyHiveIntervalDayTime) primitiveObject).getWritableObject().getHiveIntervalDayTime();
                    HiveIntervalDayTime expected = ((HiveIntervalDayTimeWritable) expectedObject).getHiveIntervalDayTime();
                    if (!value.equals(expected)) {
                        throw new RuntimeException("HiveIntervalDayTime field mismatch (expected " + expected + " found " + value + ")");
                    }
                }
                break;
            case BINARY:
                {
                    if (!(primitiveObject instanceof LazyBinary)) {
                        throw new RuntimeException("Expected LazyBinary");
                    }
                    BytesWritable bytesWritable = ((LazyBinary) primitiveObject).getWritableObject();
                    byte[] value = Arrays.copyOfRange(bytesWritable.getBytes(), 0, bytesWritable.getLength());
                    BytesWritable bytesWritableExpected = (BytesWritable) expectedObject;
                    byte[] expected = Arrays.copyOfRange(bytesWritableExpected.getBytes(), 0, bytesWritableExpected.getLength());
                    if (value.length != expected.length) {
                        throw new RuntimeException("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(value) + ")");
                    }
                    for (int b = 0; b < value.length; b++) {
                        if (value[b] != expected[b]) {
                            throw new RuntimeException("Byte Array field mismatch (expected " + Arrays.toString(expected) + " found " + Arrays.toString(value) + ")");
                        }
                    }
                }
                break;
            default:
                throw new Error("Unknown primitive category " + primitiveTypeInfo.getPrimitiveCategory());
        }
    } else if (lazyObject instanceof LazyArray) {
        LazyArray lazyArray = (LazyArray) lazyObject;
        List<Object> list = lazyArray.getList();
        List<Object> expectedList = (List<Object>) expectedObject;
        ListTypeInfo listTypeInfo = (ListTypeInfo) typeInfo;
        if (list.size() != expectedList.size()) {
            throw new RuntimeException("SerDe deserialized list length does not match (list " + list.toString() + " list.size() " + list.size() + " expectedList " + expectedList.toString() + " expectedList.size() " + expectedList.size() + ")" + " elementTypeInfo " + listTypeInfo.getListElementTypeInfo().toString());
        }
        return lazyCompareList((ListTypeInfo) typeInfo, list, expectedList);
    } else if (typeInfo instanceof ListTypeInfo) {
        List<Object> list;
        if (lazyObject instanceof LazyBinaryArray) {
            list = ((LazyBinaryArray) lazyObject).getList();
        } else {
            list = (List<Object>) lazyObject;
        }
        List<Object> expectedList = (List<Object>) expectedObject;
        if (list.size() != expectedList.size()) {
            throw new RuntimeException("SerDe deserialized list length does not match (list " + list.toString() + " list.size() " + list.size() + " expectedList " + expectedList.toString() + " expectedList.size() " + expectedList.size() + ")");
        }
        return lazyCompareList((ListTypeInfo) typeInfo, list, expectedList);
    } else if (lazyObject instanceof LazyMap) {
        LazyMap lazyMap = (LazyMap) lazyObject;
        Map<Object, Object> map = lazyMap.getMap();
        Map<Object, Object> expectedMap = (Map<Object, Object>) expectedObject;
        return lazyCompareMap((MapTypeInfo) typeInfo, map, expectedMap);
    } else if (typeInfo instanceof MapTypeInfo) {
        Map<Object, Object> map;
        Map<Object, Object> expectedMap = (Map<Object, Object>) expectedObject;
        if (lazyObject instanceof LazyBinaryMap) {
            map = ((LazyBinaryMap) lazyObject).getMap();
        } else {
            map = (Map<Object, Object>) lazyObject;
        }
        return lazyCompareMap((MapTypeInfo) typeInfo, map, expectedMap);
    } else if (lazyObject instanceof LazyStruct) {
        LazyStruct lazyStruct = (LazyStruct) lazyObject;
        List<Object> fields = lazyStruct.getFieldsAsList();
        List<Object> expectedFields = (List<Object>) expectedObject;
        StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
        return lazyCompareStruct(structTypeInfo, fields, expectedFields);
    } else if (typeInfo instanceof StructTypeInfo) {
        ArrayList<Object> fields;
        if (lazyObject instanceof LazyBinaryStruct) {
            fields = ((LazyBinaryStruct) lazyObject).getFieldsAsList();
        } else {
            fields = (ArrayList<Object>) lazyObject;
        }
        List<Object> expectedFields = (List<Object>) expectedObject;
        StructTypeInfo structTypeInfo = (StructTypeInfo) typeInfo;
        return lazyCompareStruct(structTypeInfo, fields, expectedFields);
    } else if (lazyObject instanceof LazyUnion) {
        LazyUnion union = (LazyUnion) lazyObject;
        StandardUnionObjectInspector.StandardUnion expectedUnion = (StandardUnionObjectInspector.StandardUnion) expectedObject;
        UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
        return lazyCompareUnion(unionTypeInfo, union, expectedUnion);
    } else if (typeInfo instanceof UnionTypeInfo) {
        StandardUnionObjectInspector.StandardUnion expectedUnion = (StandardUnionObjectInspector.StandardUnion) expectedObject;
        UnionTypeInfo unionTypeInfo = (UnionTypeInfo) typeInfo;
        if (lazyObject instanceof LazyBinaryUnion) {
            return lazyCompareUnion(unionTypeInfo, (LazyBinaryUnion) lazyObject, expectedUnion);
        } else {
            return lazyCompareUnion(unionTypeInfo, (UnionObject) lazyObject, expectedUnion);
        }
    } else {
        System.err.println("Not implemented " + typeInfo.getClass().getName());
    }
    return true;
}
Also used : StandardUnionObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StandardUnionObjectInspector) ByteWritable(org.apache.hadoop.hive.serde2.io.ByteWritable) DateWritable(org.apache.hadoop.hive.serde2.io.DateWritable) Writable(org.apache.hadoop.io.Writable) LongWritable(org.apache.hadoop.io.LongWritable) HiveCharWritable(org.apache.hadoop.hive.serde2.io.HiveCharWritable) HiveIntervalYearMonthWritable(org.apache.hadoop.hive.serde2.io.HiveIntervalYearMonthWritable) HiveIntervalDayTimeWritable(org.apache.hadoop.hive.serde2.io.HiveIntervalDayTimeWritable) BytesWritable(org.apache.hadoop.io.BytesWritable) TimestampWritable(org.apache.hadoop.hive.serde2.io.TimestampWritable) DoubleWritable(org.apache.hadoop.hive.serde2.io.DoubleWritable) ShortWritable(org.apache.hadoop.hive.serde2.io.ShortWritable) IntWritable(org.apache.hadoop.io.IntWritable) HiveVarcharWritable(org.apache.hadoop.hive.serde2.io.HiveVarcharWritable) BooleanWritable(org.apache.hadoop.io.BooleanWritable) HiveDecimalWritable(org.apache.hadoop.hive.serde2.io.HiveDecimalWritable) FloatWritable(org.apache.hadoop.io.FloatWritable) LazyBinaryArray(org.apache.hadoop.hive.serde2.lazybinary.LazyBinaryArray) PrimitiveTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo) LazyBinaryStruct(org.apache.hadoop.hive.serde2.lazybinary.LazyBinaryStruct) ArrayList(java.util.ArrayList) List(java.util.List) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) UnionObject(org.apache.hadoop.hive.serde2.objectinspector.UnionObject) LazyBinaryMap(org.apache.hadoop.hive.serde2.lazybinary.LazyBinaryMap) LazyBinaryMap(org.apache.hadoop.hive.serde2.lazybinary.LazyBinaryMap) Map(java.util.Map) UnionTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.UnionTypeInfo) HiveChar(org.apache.hadoop.hive.common.type.HiveChar) StructTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.StructTypeInfo) Timestamp(java.sql.Timestamp) HiveDecimal(org.apache.hadoop.hive.common.type.HiveDecimal) LazyBinaryUnion(org.apache.hadoop.hive.serde2.lazybinary.LazyBinaryUnion) HiveIntervalDayTime(org.apache.hadoop.hive.common.type.HiveIntervalDayTime) Text(org.apache.hadoop.io.Text) BytesWritable(org.apache.hadoop.io.BytesWritable) HiveVarchar(org.apache.hadoop.hive.common.type.HiveVarchar) Date(java.sql.Date) HiveIntervalYearMonth(org.apache.hadoop.hive.common.type.HiveIntervalYearMonth) ListTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.ListTypeInfo) MapTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.MapTypeInfo)

Example 78 with DateWritable

use of org.apache.hadoop.hive.serde2.io.DateWritable in project hive by apache.

the class LazySimpleSerializeWrite method writeDate.

/*
   * DATE.
   */
@Override
public void writeDate(Date date) throws IOException {
    beginPrimitive();
    if (dateWritable == null) {
        dateWritable = new DateWritable();
    }
    dateWritable.set(date);
    LazyDate.writeUTF8(output, dateWritable);
    finishPrimitive();
}
Also used : DateWritable(org.apache.hadoop.hive.serde2.io.DateWritable)

Example 79 with DateWritable

use of org.apache.hadoop.hive.serde2.io.DateWritable in project presto by prestodb.

the class TestObjectEncoders method testPrimitiveObjectEncoders.

@Test
public void testPrimitiveObjectEncoders() {
    ObjectInspector inspector;
    ObjectEncoder encoder;
    inspector = writableLongObjectInspector;
    encoder = createEncoder(BIGINT, inspector);
    assertTrue(encoder.encode(new LongWritable(123456L)) instanceof Long);
    inspector = writableIntObjectInspector;
    encoder = createEncoder(INTEGER, inspector);
    assertTrue(encoder.encode(new IntWritable(12345)) instanceof Long);
    inspector = writableShortObjectInspector;
    encoder = createEncoder(SMALLINT, inspector);
    assertTrue(encoder.encode(new ShortWritable((short) 1234)) instanceof Long);
    inspector = writableByteObjectInspector;
    encoder = createEncoder(TINYINT, inspector);
    assertTrue(encoder.encode(new ByteWritable((byte) 123)) instanceof Long);
    inspector = writableBooleanObjectInspector;
    encoder = createEncoder(BOOLEAN, inspector);
    assertTrue(encoder.encode(new BooleanWritable(true)) instanceof Boolean);
    inspector = writableDoubleObjectInspector;
    encoder = createEncoder(DOUBLE, inspector);
    assertTrue(encoder.encode(new DoubleWritable(0.1)) instanceof Double);
    inspector = writableDateObjectInspector;
    encoder = createEncoder(DATE, inspector);
    assertTrue(encoder.encode(new DateWritable(DateTimeUtils.createDate(18380L))) instanceof Long);
    inspector = writableHiveDecimalObjectInspector;
    encoder = createEncoder(createDecimalType(11, 10), inspector);
    assertTrue(encoder.encode(new HiveDecimalWritable("1.2345678910")) instanceof Long);
    encoder = createEncoder(createDecimalType(34, 33), inspector);
    assertTrue(encoder.encode(new HiveDecimalWritable("1.281734081274028174012432412423134")) instanceof Slice);
}
Also used : PrimitiveObjectInspectorFactory.writableByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableByteObjectInspector) PrimitiveObjectInspectorFactory.writableDateObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableDateObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) PrimitiveObjectInspectorFactory.writableShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableShortObjectInspector) PrimitiveObjectInspectorFactory.writableStringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableStringObjectInspector) PrimitiveObjectInspectorFactory.writableBooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableBooleanObjectInspector) PrimitiveObjectInspectorFactory.writableHiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableHiveDecimalObjectInspector) PrimitiveObjectInspectorFactory.writableBinaryObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableBinaryObjectInspector) PrimitiveObjectInspectorFactory.writableIntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableIntObjectInspector) PrimitiveObjectInspectorFactory.writableDoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableDoubleObjectInspector) PrimitiveObjectInspectorFactory.writableLongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.writableLongObjectInspector) DateWritable(org.apache.hadoop.hive.serde2.io.DateWritable) HiveDecimalWritable(org.apache.hadoop.hive.serde2.io.HiveDecimalWritable) DoubleWritable(org.apache.hadoop.io.DoubleWritable) ShortWritable(org.apache.hadoop.hive.serde2.io.ShortWritable) BooleanWritable(org.apache.hadoop.io.BooleanWritable) Slice(io.airlift.slice.Slice) LongWritable(org.apache.hadoop.io.LongWritable) ByteWritable(org.apache.hadoop.io.ByteWritable) IntWritable(org.apache.hadoop.io.IntWritable) Test(org.testng.annotations.Test)

Example 80 with DateWritable

use of org.apache.hadoop.hive.serde2.io.DateWritable in project presto by prestodb.

the class TestHiveBucketing method testHashingCompare.

@Test
public void testHashingCompare() throws Exception {
    assertBucketEquals("boolean", null);
    assertBucketEquals("boolean", true);
    assertBucketEquals("boolean", false);
    assertBucketEquals("tinyint", null);
    assertBucketEquals("tinyint", (byte) 5);
    assertBucketEquals("tinyint", Byte.MIN_VALUE);
    assertBucketEquals("tinyint", Byte.MAX_VALUE);
    assertBucketEquals("smallint", null);
    assertBucketEquals("smallint", (short) 300);
    assertBucketEquals("smallint", Short.MIN_VALUE);
    assertBucketEquals("smallint", Short.MAX_VALUE);
    assertBucketEquals("int", null);
    assertBucketEquals("int", 300_000);
    assertBucketEquals("int", Integer.MIN_VALUE);
    assertBucketEquals("int", Integer.MAX_VALUE);
    assertBucketEquals("bigint", null);
    assertBucketEquals("bigint", 300_000_000_000L);
    assertBucketEquals("bigint", Long.MIN_VALUE);
    assertBucketEquals("bigint", Long.MAX_VALUE);
    assertBucketEquals("float", null);
    assertBucketEquals("float", 12.34F);
    assertBucketEquals("float", -Float.MAX_VALUE);
    assertBucketEquals("float", Float.MIN_VALUE);
    assertBucketEquals("float", Float.POSITIVE_INFINITY);
    assertBucketEquals("float", Float.NEGATIVE_INFINITY);
    assertBucketEquals("double", null);
    assertBucketEquals("double", 12.34);
    assertBucketEquals("double", -Double.MAX_VALUE);
    assertBucketEquals("double", Double.MIN_VALUE);
    assertBucketEquals("double", Double.POSITIVE_INFINITY);
    assertBucketEquals("double", Double.NEGATIVE_INFINITY);
    assertBucketEquals("varchar(15)", null);
    assertBucketEquals("varchar(15)", "");
    assertBucketEquals("varchar(15)", "test string");
    // 3-byte UTF-8 sequences (in Basic Plane, i.e. Plane 0)
    assertBucketEquals("varchar(15)", "\u5f3a\u5927\u7684Presto\u5f15\u64ce");
    // 4 code points: 20FFC - 20FFF. 4-byte UTF-8 sequences in Supplementary Plane 2
    assertBucketEquals("varchar(15)", "\uD843\uDFFC\uD843\uDFFD\uD843\uDFFE\uD843\uDFFF");
    assertBucketEquals("string", null);
    assertBucketEquals("string", "");
    assertBucketEquals("string", "test string");
    // 3-byte UTF-8 sequences (in Basic Plane, i.e. Plane 0)
    assertBucketEquals("string", "\u5f3a\u5927\u7684Presto\u5f15\u64ce");
    // 4 code points: 20FFC - 20FFF. 4-byte UTF-8 sequences in Supplementary Plane 2
    assertBucketEquals("string", "\uD843\uDFFC\uD843\uDFFD\uD843\uDFFE\uD843\uDFFF");
    assertBucketEquals("date", null);
    assertBucketEquals("date", new DateWritable(toIntExact(LocalDate.of(1970, 1, 1).toEpochDay())).get());
    assertBucketEquals("date", new DateWritable(toIntExact(LocalDate.of(2015, 11, 19).toEpochDay())).get());
    assertBucketEquals("date", new DateWritable(toIntExact(LocalDate.of(1950, 11, 19).toEpochDay())).get());
    assertBucketEquals("timestamp", null);
    assertBucketEquals("timestamp", new Timestamp(1000 * LocalDateTime.of(1970, 1, 1, 0, 0, 0, 0).toEpochSecond(ZoneOffset.UTC)));
    assertBucketEquals("timestamp", new Timestamp(1000 * LocalDateTime.of(1969, 12, 31, 23, 59, 59, 999_000_000).toEpochSecond(ZoneOffset.UTC)));
    assertBucketEquals("timestamp", new Timestamp(1000 * LocalDateTime.of(1950, 11, 19, 12, 34, 56, 789_000_000).toEpochSecond(ZoneOffset.UTC)));
    assertBucketEquals("timestamp", new Timestamp(1000 * LocalDateTime.of(2015, 11, 19, 7, 6, 5, 432_000_000).toEpochSecond(ZoneOffset.UTC)));
    assertBucketEquals("array<double>", null);
    assertBucketEquals("array<boolean>", ImmutableList.of());
    assertBucketEquals("array<smallint>", ImmutableList.of((short) 5, (short) 8, (short) 13));
    assertBucketEquals("array<string>", ImmutableList.of("test1", "test2", "test3", "test4"));
    assertBucketEquals("map<float,date>", null);
    assertBucketEquals("map<double,timestamp>", ImmutableMap.of());
    assertBucketEquals("map<string,bigint>", ImmutableMap.of("key", 123L, "key2", 123456789L, "key3", -123456L));
    assertBucketEquals("array<array<bigint>>", ImmutableList.of(ImmutableList.of(10L, 20L), ImmutableList.of(-10L, -20L), asList((Object) null)));
    assertBucketEquals("map<array<double>,map<int,timestamp>>", ImmutableMap.of(ImmutableList.of(12.3, 45.7), ImmutableMap.of(123, new Timestamp(1_234_567_890_000L))));
    // multiple bucketing columns
    assertBucketEquals(ImmutableList.of("float", "array<smallint>", "map<string,bigint>"), ImmutableList.of(12.34F, ImmutableList.of((short) 5, (short) 8, (short) 13), ImmutableMap.of("key", 123L)));
    assertBucketEquals(ImmutableList.of("double", "array<smallint>", "boolean", "map<string,bigint>", "tinyint"), asList(null, ImmutableList.of((short) 5, (short) 8, (short) 13), null, ImmutableMap.of("key", 123L), null));
}
Also used : DateWritable(org.apache.hadoop.hive.serde2.io.DateWritable) Timestamp(java.sql.Timestamp) Test(org.testng.annotations.Test)

Aggregations

DateWritable (org.apache.hadoop.hive.serde2.io.DateWritable)77 ObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector)36 Text (org.apache.hadoop.io.Text)36 TimestampWritable (org.apache.hadoop.hive.serde2.io.TimestampWritable)35 Date (java.sql.Date)31 DeferredJavaObject (org.apache.hadoop.hive.ql.udf.generic.GenericUDF.DeferredJavaObject)29 DeferredObject (org.apache.hadoop.hive.ql.udf.generic.GenericUDF.DeferredObject)29 HiveDecimalWritable (org.apache.hadoop.hive.serde2.io.HiveDecimalWritable)25 IntWritable (org.apache.hadoop.io.IntWritable)25 Timestamp (java.sql.Timestamp)24 BytesWritable (org.apache.hadoop.io.BytesWritable)24 ShortWritable (org.apache.hadoop.hive.serde2.io.ShortWritable)22 LongWritable (org.apache.hadoop.io.LongWritable)22 ByteWritable (org.apache.hadoop.hive.serde2.io.ByteWritable)20 BooleanWritable (org.apache.hadoop.io.BooleanWritable)20 DoubleWritable (org.apache.hadoop.hive.serde2.io.DoubleWritable)19 HiveIntervalDayTimeWritable (org.apache.hadoop.hive.serde2.io.HiveIntervalDayTimeWritable)19 FloatWritable (org.apache.hadoop.io.FloatWritable)19 HiveChar (org.apache.hadoop.hive.common.type.HiveChar)18 HiveDecimal (org.apache.hadoop.hive.common.type.HiveDecimal)18