Examples with LazySerDeParameters org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters used on opensource projects

Example 1 with LazySerDeParameters

use of org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters in project hive by apache.

the class TestVectorSerDeRow method testVectorSerializeRow.

void testVectorSerializeRow(Random r, SerializationType serializationType) throws HiveException, IOException, SerDeException {
    String[] emptyScratchTypeNames = new String[0];
    VectorRandomRowSource source = new VectorRandomRowSource();
    source.init(r);
    VectorizedRowBatchCtx batchContext = new VectorizedRowBatchCtx();
    batchContext.init(source.rowStructObjectInspector(), emptyScratchTypeNames);
    VectorizedRowBatch batch = batchContext.createVectorizedRowBatch();
    VectorAssignRow vectorAssignRow = new VectorAssignRow();
    vectorAssignRow.init(source.typeNames());
    int fieldCount = source.typeNames().size();
    DeserializeRead deserializeRead;
    SerializeWrite serializeWrite;
    switch(serializationType) {
        case BINARY_SORTABLE:
            deserializeRead = new BinarySortableDeserializeRead(source.primitiveTypeInfos(), /* useExternalBuffer */
            false);
            serializeWrite = new BinarySortableSerializeWrite(fieldCount);
            break;
        case LAZY_BINARY:
            deserializeRead = new LazyBinaryDeserializeRead(source.primitiveTypeInfos(), /* useExternalBuffer */
            false);
            serializeWrite = new LazyBinarySerializeWrite(fieldCount);
            break;
        case LAZY_SIMPLE:
            {
                StructObjectInspector rowObjectInspector = source.rowStructObjectInspector();
                LazySerDeParameters lazySerDeParams = getSerDeParams(rowObjectInspector);
                byte separator = (byte) '\t';
                deserializeRead = new LazySimpleDeserializeRead(source.primitiveTypeInfos(), /* useExternalBuffer */
                false, separator, lazySerDeParams);
                serializeWrite = new LazySimpleSerializeWrite(fieldCount, separator, lazySerDeParams);
            }
            break;
        default:
            throw new Error("Unknown serialization type " + serializationType);
    }
    VectorSerializeRow vectorSerializeRow = new VectorSerializeRow(serializeWrite);
    vectorSerializeRow.init(source.typeNames());
    Object[][] randomRows = source.randomRows(100000);
    int firstRandomRowIndex = 0;
    for (int i = 0; i < randomRows.length; i++) {
        Object[] row = randomRows[i];
        vectorAssignRow.assignRow(batch, batch.size, row);
        batch.size++;
        if (batch.size == batch.DEFAULT_SIZE) {
            serializeBatch(batch, vectorSerializeRow, deserializeRead, source, randomRows, firstRandomRowIndex);
            firstRandomRowIndex = i + 1;
            batch.reset();
        }
    }
    if (batch.size > 0) {
        serializeBatch(batch, vectorSerializeRow, deserializeRead, source, randomRows, firstRandomRowIndex);
    }
}

Example 2 with LazySerDeParameters

use of org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters in project hive by apache.

the class TestLazySimpleFast method testLazySimpleFast.

private void testLazySimpleFast(SerdeRandomRowSource source, Object[][] rows, LazySimpleSerDe serde, StructObjectInspector rowOI, LazySimpleSerDe serde_fewer, StructObjectInspector writeRowOI, byte separator, LazySerDeParameters serdeParams, LazySerDeParameters serdeParams_fewer, PrimitiveTypeInfo[] primitiveTypeInfos, boolean useIncludeColumns, boolean doWriteFewerColumns, Random r) throws Throwable {
    int rowCount = rows.length;
    int columnCount = primitiveTypeInfos.length;
    boolean[] columnsToInclude = null;
    if (useIncludeColumns) {
        columnsToInclude = new boolean[columnCount];
        for (int i = 0; i < columnCount; i++) {
            columnsToInclude[i] = r.nextBoolean();
        }
    }
    int writeColumnCount = columnCount;
    PrimitiveTypeInfo[] writePrimitiveTypeInfos = primitiveTypeInfos;
    if (doWriteFewerColumns) {
        writeColumnCount = writeRowOI.getAllStructFieldRefs().size();
        writePrimitiveTypeInfos = Arrays.copyOf(primitiveTypeInfos, writeColumnCount);
    }
    // Try to serialize
    BytesWritable[] serializeWriteBytes = new BytesWritable[rowCount];
    for (int i = 0; i < rowCount; i++) {
        Object[] row = rows[i];
        Output output = new Output();
        LazySimpleSerializeWrite lazySimpleSerializeWrite = new LazySimpleSerializeWrite(columnCount, separator, serdeParams);
        lazySimpleSerializeWrite.set(output);
        for (int index = 0; index < columnCount; index++) {
            Writable writable = (Writable) row[index];
            VerifyFast.serializeWrite(lazySimpleSerializeWrite, primitiveTypeInfos[index], writable);
        }
        BytesWritable bytesWritable = new BytesWritable();
        bytesWritable.set(output.getData(), 0, output.getLength());
        serializeWriteBytes[i] = bytesWritable;
    }
    // Try to deserialize
    for (int i = 0; i < rowCount; i++) {
        Object[] row = rows[i];
        LazySimpleDeserializeRead lazySimpleDeserializeRead = new LazySimpleDeserializeRead(writePrimitiveTypeInfos, /* useExternalBuffer */
        false, separator, serdeParams);
        BytesWritable bytesWritable = serializeWriteBytes[i];
        byte[] bytes = bytesWritable.getBytes();
        int length = bytesWritable.getLength();
        lazySimpleDeserializeRead.set(bytes, 0, length);
        char[] chars = new char[length];
        for (int c = 0; c < chars.length; c++) {
            chars[c] = (char) (bytes[c] & 0xFF);
        }
        for (int index = 0; index < columnCount; index++) {
            if (useIncludeColumns && !columnsToInclude[index]) {
                lazySimpleDeserializeRead.skipNextField();
            } else if (index >= writeColumnCount) {
                // Should come back a null.
                VerifyFast.verifyDeserializeRead(lazySimpleDeserializeRead, primitiveTypeInfos[index], null);
            } else {
                Writable writable = (Writable) row[index];
                VerifyFast.verifyDeserializeRead(lazySimpleDeserializeRead, primitiveTypeInfos[index], writable);
            }
        }
        if (writeColumnCount == columnCount) {
            TestCase.assertTrue(lazySimpleDeserializeRead.isEndOfInputReached());
        }
    }
    // Try to deserialize using SerDe class our Writable row objects created by SerializeWrite.
    for (int i = 0; i < rowCount; i++) {
        BytesWritable bytesWritable = serializeWriteBytes[i];
        LazyStruct lazySimpleStruct = (LazyStruct) serde.deserialize(bytesWritable);
        Object[] row = rows[i];
        for (int index = 0; index < columnCount; index++) {
            PrimitiveTypeInfo primitiveTypeInfo = primitiveTypeInfos[index];
            Writable writable = (Writable) row[index];
            LazyPrimitive lazyPrimitive = (LazyPrimitive) lazySimpleStruct.getField(index);
            Object object;
            if (lazyPrimitive != null) {
                object = lazyPrimitive.getWritableObject();
            } else {
                object = null;
            }
            if (writable == null || object == null) {
                if (writable != null || object != null) {
                    fail("SerDe deserialized NULL column mismatch");
                }
            } else {
                if (!object.equals(writable)) {
                    fail("SerDe deserialized value does not match");
                }
            }
        }
    }
    // One Writable per row.
    byte[][] serdeBytes = new byte[rowCount][];
    // Serialize using the SerDe, then below deserialize using DeserializeRead.
    Object[] serdeRow = new Object[columnCount];
    for (int i = 0; i < rowCount; i++) {
        Object[] row = rows[i];
        // LazySimple seems to work better with an row object array instead of a Java object...
        for (int index = 0; index < columnCount; index++) {
            serdeRow[index] = row[index];
        }
        Text serialized = (Text) serde.serialize(serdeRow, rowOI);
        byte[] bytes1 = Arrays.copyOfRange(serialized.getBytes(), 0, serialized.getLength());
        byte[] bytes2 = Arrays.copyOfRange(serializeWriteBytes[i].getBytes(), 0, serializeWriteBytes[i].getLength());
        if (!Arrays.equals(bytes1, bytes2)) {
            fail("SerializeWrite and SerDe serialization does not match");
        }
        serdeBytes[i] = copyBytes(serialized);
    }
    // Try to deserialize using DeserializeRead our Writable row objects created by SerDe.
    for (int i = 0; i < rowCount; i++) {
        Object[] row = rows[i];
        LazySimpleDeserializeRead lazySimpleDeserializeRead = new LazySimpleDeserializeRead(writePrimitiveTypeInfos, /* useExternalBuffer */
        false, separator, serdeParams);
        byte[] bytes = serdeBytes[i];
        lazySimpleDeserializeRead.set(bytes, 0, bytes.length);
        for (int index = 0; index < columnCount; index++) {
            if (useIncludeColumns && !columnsToInclude[index]) {
                lazySimpleDeserializeRead.skipNextField();
            } else if (index >= writeColumnCount) {
                // Should come back a null.
                VerifyFast.verifyDeserializeRead(lazySimpleDeserializeRead, primitiveTypeInfos[index], null);
            } else {
                Writable writable = (Writable) row[index];
                VerifyFast.verifyDeserializeRead(lazySimpleDeserializeRead, primitiveTypeInfos[index], writable);
            }
        }
        if (writeColumnCount == columnCount) {
            TestCase.assertTrue(lazySimpleDeserializeRead.isEndOfInputReached());
        }
    }
}

Example 3 with LazySerDeParameters

use of org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters in project hive by apache.

the class TestLazySimpleFast method testLazySimpleFastCase.

public void testLazySimpleFastCase(int caseNum, boolean doNonRandomFill, Random r) throws Throwable {
    SerdeRandomRowSource source = new SerdeRandomRowSource();
    source.init(r);
    int rowCount = 1000;
    Object[][] rows = source.randomRows(rowCount);
    if (doNonRandomFill) {
        MyTestClass.nonRandomRowFill(rows, source.primitiveCategories());
    }
    StructObjectInspector rowStructObjectInspector = source.rowStructObjectInspector();
    PrimitiveTypeInfo[] primitiveTypeInfos = source.primitiveTypeInfos();
    int columnCount = primitiveTypeInfos.length;
    int writeColumnCount = columnCount;
    StructObjectInspector writeRowStructObjectInspector = rowStructObjectInspector;
    boolean doWriteFewerColumns = r.nextBoolean();
    if (doWriteFewerColumns) {
        writeColumnCount = 1 + r.nextInt(columnCount);
        if (writeColumnCount == columnCount) {
            doWriteFewerColumns = false;
        } else {
            writeRowStructObjectInspector = source.partialRowStructObjectInspector(writeColumnCount);
        }
    }
    String fieldNames = ObjectInspectorUtils.getFieldNames(rowStructObjectInspector);
    String fieldTypes = ObjectInspectorUtils.getFieldTypes(rowStructObjectInspector);
    LazySimpleSerDe serde = getSerDe(fieldNames, fieldTypes);
    LazySerDeParameters serdeParams = getSerDeParams(fieldNames, fieldTypes);
    LazySimpleSerDe serde_fewer = null;
    LazySerDeParameters serdeParams_fewer = null;
    if (doWriteFewerColumns) {
        String partialFieldNames = ObjectInspectorUtils.getFieldNames(writeRowStructObjectInspector);
        String partialFieldTypes = ObjectInspectorUtils.getFieldTypes(writeRowStructObjectInspector);
        serde_fewer = getSerDe(fieldNames, fieldTypes);
        serdeParams_fewer = getSerDeParams(partialFieldNames, partialFieldTypes);
    }
    byte separator = (byte) '\t';
    testLazySimpleFast(source, rows, serde, rowStructObjectInspector, serde_fewer, writeRowStructObjectInspector, separator, serdeParams, serdeParams_fewer, primitiveTypeInfos, /* useIncludeColumns */
    false, /* doWriteFewerColumns */
    false, r);
    testLazySimpleFast(source, rows, serde, rowStructObjectInspector, serde_fewer, writeRowStructObjectInspector, separator, serdeParams, serdeParams_fewer, primitiveTypeInfos, /* useIncludeColumns */
    true, /* doWriteFewerColumns */
    false, r);
    if (doWriteFewerColumns) {
        testLazySimpleFast(source, rows, serde, rowStructObjectInspector, serde_fewer, writeRowStructObjectInspector, separator, serdeParams, serdeParams_fewer, primitiveTypeInfos, /* useIncludeColumns */
        false, /* doWriteFewerColumns */
        true, r);
        testLazySimpleFast(source, rows, serde, rowStructObjectInspector, serde_fewer, writeRowStructObjectInspector, separator, serdeParams, serdeParams_fewer, primitiveTypeInfos, /* useIncludeColumns */
        true, /* doWriteFewerColumns */
        true, r);
    }
}

Example 4 with LazySerDeParameters

use of org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters in project phoenix by apache.

the class PhoenixSerDe method initialize.

@Override
public void initialize(Configuration conf, Properties tbl) throws SerDeException {
    tableProperties = tbl;
    if (LOG.isDebugEnabled()) {
        LOG.debug("SerDe initialize : " + tbl.getProperty("name"));
    }
    serdeParams = new LazySerDeParameters(conf, tbl, getClass().getName());
    objectInspector = createLazyPhoenixInspector(conf, tbl);
    String inOutWork = tbl.getProperty(PhoenixStorageHandlerConstants.IN_OUT_WORK);
    if (inOutWork == null) {
        return;
    }
    serializer = new PhoenixSerializer(conf, tbl);
    row = new PhoenixRow(serdeParams.getColumnNames());
}

Example 5 with LazySerDeParameters

use of org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters in project cdap by caskdata.

the class SimpleMapEqualComparerTest method serializeAndDeserialize.

Object serializeAndDeserialize(StringTextMapHolder o1, StructObjectInspector oi1, LazySimpleSerDe serde, LazySerDeParameters serdeParams) throws IOException, SerDeException {
    ByteStream.Output serializeStream = new ByteStream.Output();
    LazySimpleSerDe.serialize(serializeStream, o1, oi1, serdeParams.getSeparators(), 0, serdeParams.getNullSequence(), serdeParams.isEscaped(), serdeParams.getEscapeChar(), serdeParams.getNeedsEscape());
    Text t = new Text(serializeStream.toByteArray());
    return serde.deserialize(t);
}