Examples with PrimitiveTypeInfo org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo used on opensource projects

Example 11 with PrimitiveTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo 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 12 with PrimitiveTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo 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 13 with PrimitiveTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo in project hive by apache.

the class GenericUDFInBloomFilter method evaluate.

@Override
public Object evaluate(DeferredObject[] arguments) throws HiveException {
    // Return if either of the arguments is null
    if (arguments[0].get() == null || arguments[1].get() == null) {
        return null;
    }
    if (!initializedBloomFilter) {
        // Setup the bloom filter once
        try {
            BytesWritable bw = (BytesWritable) arguments[1].get();
            byte[] bytes = new byte[bw.getLength()];
            System.arraycopy(bw.getBytes(), 0, bytes, 0, bw.getLength());
            bloomFilter = BloomFilter.deserialize(new ByteArrayInputStream(bytes));
        } catch (IOException e) {
            throw new HiveException(e);
        }
        initializedBloomFilter = true;
    }
    // Check if the value is in bloom filter
    switch(((PrimitiveObjectInspector) valObjectInspector).getTypeInfo().getPrimitiveCategory()) {
        case BOOLEAN:
            boolean vBoolean = ((BooleanObjectInspector) valObjectInspector).get(arguments[0].get());
            return bloomFilter.testLong(vBoolean ? 1 : 0);
        case BYTE:
            byte vByte = ((ByteObjectInspector) valObjectInspector).get(arguments[0].get());
            return bloomFilter.testLong(vByte);
        case SHORT:
            short vShort = ((ShortObjectInspector) valObjectInspector).get(arguments[0].get());
            return bloomFilter.testLong(vShort);
        case INT:
            int vInt = ((IntObjectInspector) valObjectInspector).get(arguments[0].get());
            return bloomFilter.testLong(vInt);
        case LONG:
            long vLong = ((LongObjectInspector) valObjectInspector).get(arguments[0].get());
            return bloomFilter.testLong(vLong);
        case FLOAT:
            float vFloat = ((FloatObjectInspector) valObjectInspector).get(arguments[0].get());
            return bloomFilter.testDouble(vFloat);
        case DOUBLE:
            double vDouble = ((DoubleObjectInspector) valObjectInspector).get(arguments[0].get());
            return bloomFilter.testDouble(vDouble);
        case DECIMAL:
            HiveDecimalWritable vDecimal = ((HiveDecimalObjectInspector) valObjectInspector).getPrimitiveWritableObject(arguments[0].get());
            int startIdx = vDecimal.toBytes(scratchBuffer);
            return bloomFilter.testBytes(scratchBuffer, startIdx, scratchBuffer.length - startIdx);
        case DATE:
            DateWritable vDate = ((DateObjectInspector) valObjectInspector).getPrimitiveWritableObject(arguments[0].get());
            return bloomFilter.testLong(vDate.getDays());
        case TIMESTAMP:
            Timestamp vTimeStamp = ((TimestampObjectInspector) valObjectInspector).getPrimitiveJavaObject(arguments[0].get());
            return bloomFilter.testLong(vTimeStamp.getTime());
        case CHAR:
            Text vChar = ((HiveCharObjectInspector) valObjectInspector).getPrimitiveWritableObject(arguments[0].get()).getStrippedValue();
            return bloomFilter.testBytes(vChar.getBytes(), 0, vChar.getLength());
        case VARCHAR:
            Text vVarchar = ((HiveVarcharObjectInspector) valObjectInspector).getPrimitiveWritableObject(arguments[0].get()).getTextValue();
            return bloomFilter.testBytes(vVarchar.getBytes(), 0, vVarchar.getLength());
        case STRING:
            Text vString = ((StringObjectInspector) valObjectInspector).getPrimitiveWritableObject(arguments[0].get());
            return bloomFilter.testBytes(vString.getBytes(), 0, vString.getLength());
        case BINARY:
            BytesWritable vBytes = ((BinaryObjectInspector) valObjectInspector).getPrimitiveWritableObject(arguments[0].get());
            return bloomFilter.testBytes(vBytes.getBytes(), 0, vBytes.getLength());
        default:
            throw new UDFArgumentTypeException(0, "Bad primitive category " + ((PrimitiveTypeInfo) valObjectInspector).getPrimitiveCategory());
    }
}

Example 14 with PrimitiveTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo in project hive by apache.

the class TestVectorMapJoinFastRowHashMap method addAndVerifyRows.

private void addAndVerifyRows(VectorRandomRowSource valueSource, Object[][] rows, VectorMapJoinFastHashTable map, HashTableKeyType hashTableKeyType, VerifyFastRowHashMap verifyTable, String[] keyTypeNames, boolean doClipping, boolean useExactBytes) throws HiveException, IOException, SerDeException {
    final int keyCount = keyTypeNames.length;
    PrimitiveTypeInfo[] keyPrimitiveTypeInfos = new PrimitiveTypeInfo[keyCount];
    PrimitiveCategory[] keyPrimitiveCategories = new PrimitiveCategory[keyCount];
    ArrayList<ObjectInspector> keyPrimitiveObjectInspectorList = new ArrayList<ObjectInspector>(keyCount);
    for (int i = 0; i < keyCount; i++) {
        PrimitiveTypeInfo primitiveTypeInfo = (PrimitiveTypeInfo) TypeInfoUtils.getTypeInfoFromTypeString(keyTypeNames[i]);
        keyPrimitiveTypeInfos[i] = primitiveTypeInfo;
        PrimitiveCategory primitiveCategory = primitiveTypeInfo.getPrimitiveCategory();
        keyPrimitiveCategories[i] = primitiveCategory;
        keyPrimitiveObjectInspectorList.add(PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(primitiveTypeInfo));
    }
    boolean[] keyColumnSortOrderIsDesc = new boolean[keyCount];
    Arrays.fill(keyColumnSortOrderIsDesc, false);
    byte[] keyColumnNullMarker = new byte[keyCount];
    Arrays.fill(keyColumnNullMarker, BinarySortableSerDe.ZERO);
    byte[] keyColumnNotNullMarker = new byte[keyCount];
    Arrays.fill(keyColumnNotNullMarker, BinarySortableSerDe.ONE);
    BinarySortableSerializeWrite keySerializeWrite = new BinarySortableSerializeWrite(keyColumnSortOrderIsDesc, keyColumnNullMarker, keyColumnNotNullMarker);
    PrimitiveTypeInfo[] valuePrimitiveTypeInfos = valueSource.primitiveTypeInfos();
    final int columnCount = valuePrimitiveTypeInfos.length;
    SerializeWrite valueSerializeWrite = new LazyBinarySerializeWrite(columnCount);
    final int count = rows.length;
    for (int i = 0; i < count; i++) {
        Object[] valueRow = rows[i];
        Output valueOutput = new Output();
        ((LazyBinarySerializeWrite) valueSerializeWrite).set(valueOutput);
        for (int index = 0; index < columnCount; index++) {
            Writable writable = (Writable) valueRow[index];
            VerifyFastRow.serializeWrite(valueSerializeWrite, valuePrimitiveTypeInfos[index], writable);
        }
        byte[] value = Arrays.copyOf(valueOutput.getData(), valueOutput.getLength());
        // Add a new key or add a value to an existing key?
        byte[] key;
        if (random.nextBoolean() || verifyTable.getCount() == 0) {
            Object[] keyRow = VectorRandomRowSource.randomRow(keyCount, random, keyPrimitiveObjectInspectorList, keyPrimitiveCategories, keyPrimitiveTypeInfos);
            Output keyOutput = new Output();
            keySerializeWrite.set(keyOutput);
            for (int index = 0; index < keyCount; index++) {
                Writable writable = (Writable) keyRow[index];
                VerifyFastRow.serializeWrite(keySerializeWrite, keyPrimitiveTypeInfos[index], writable);
            }
            key = Arrays.copyOf(keyOutput.getData(), keyOutput.getLength());
            verifyTable.add(key, keyRow, value, valueRow);
        } else {
            key = verifyTable.addRandomExisting(value, valueRow, random);
        }
        // Serialize keyRow into key bytes.
        BytesWritable keyWritable = new BytesWritable(key);
        BytesWritable valueWritable = new BytesWritable(value);
        map.putRow(keyWritable, valueWritable);
    // verifyTable.verify(map);
    }
    verifyTable.verify(map, hashTableKeyType, valuePrimitiveTypeInfos, doClipping, useExactBytes, random);
}

Example 15 with PrimitiveTypeInfo

use of org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo in project hive by apache.

the class CheckFastRowHashMap method verifyHashMapRows.

public static void verifyHashMapRows(List<Object[]> rows, int[] actualToValueMap, VectorMapJoinHashMapResult hashMapResult, TypeInfo[] typeInfos) throws IOException {
    final int count = rows.size();
    final int columnCount = typeInfos.length;
    WriteBuffers.ByteSegmentRef ref = hashMapResult.first();
    for (int a = 0; a < count; a++) {
        int valueIndex = actualToValueMap[a];
        Object[] row = rows.get(valueIndex);
        byte[] bytes = ref.getBytes();
        int offset = (int) ref.getOffset();
        int length = ref.getLength();
        LazyBinaryDeserializeRead lazyBinaryDeserializeRead = new LazyBinaryDeserializeRead(typeInfos, /* useExternalBuffer */
        false);
        lazyBinaryDeserializeRead.set(bytes, offset, length);
        for (int index = 0; index < columnCount; index++) {
            Writable writable = (Writable) row[index];
            VerifyFastRow.verifyDeserializeRead(lazyBinaryDeserializeRead, (PrimitiveTypeInfo) typeInfos[index], writable);
        }
        TestCase.assertTrue(lazyBinaryDeserializeRead.isEndOfInputReached());
        ref = hashMapResult.next();
        if (a == count - 1) {
            TestCase.assertTrue(ref == null);
        } else {
            TestCase.assertTrue(ref != null);
        }
    }
}