Examples with UTF8StringBuilder org.apache.hyracks.data.std.util.UTF8StringBuilder used on opensource projects

Example 6 with UTF8StringBuilder

use of org.apache.hyracks.data.std.util.UTF8StringBuilder in project asterixdb by apache.

the class SubstringBeforeDescriptor method createEvaluatorFactory.

@Override
public IScalarEvaluatorFactory createEvaluatorFactory(final IScalarEvaluatorFactory[] args) {
    return new IScalarEvaluatorFactory() {

        private static final long serialVersionUID = 1L;

        @Override
        public IScalarEvaluator createScalarEvaluator(final IHyracksTaskContext ctx) throws HyracksDataException {
            return new IScalarEvaluator() {

                private ArrayBackedValueStorage resultStorage = new ArrayBackedValueStorage();

                private DataOutput out = resultStorage.getDataOutput();

                private IPointable array0 = new VoidPointable();

                private IPointable array1 = new VoidPointable();

                private IScalarEvaluator evalString = args[0].createScalarEvaluator(ctx);

                private IScalarEvaluator evalPattern = args[1].createScalarEvaluator(ctx);

                private final GrowableArray array = new GrowableArray();

                private final UTF8StringBuilder builder = new UTF8StringBuilder();

                private final UTF8StringPointable stringPtr = new UTF8StringPointable();

                private final UTF8StringPointable patternPtr = new UTF8StringPointable();

                @Override
                public void evaluate(IFrameTupleReference tuple, IPointable result) throws HyracksDataException {
                    resultStorage.reset();
                    evalString.evaluate(tuple, array0);
                    byte[] src = array0.getByteArray();
                    int srcOffset = array0.getStartOffset();
                    int srcLen = array0.getLength();
                    evalPattern.evaluate(tuple, array1);
                    byte[] pattern = array1.getByteArray();
                    int patternOffset = array1.getStartOffset();
                    int patternLen = array1.getLength();
                    if (src[srcOffset] != ATypeTag.SERIALIZED_STRING_TYPE_TAG) {
                        throw new TypeMismatchException(getIdentifier(), 0, src[srcOffset], ATypeTag.SERIALIZED_STRING_TYPE_TAG);
                    }
                    if (pattern[patternOffset] != ATypeTag.SERIALIZED_STRING_TYPE_TAG) {
                        throw new TypeMismatchException(getIdentifier(), 1, pattern[patternOffset], ATypeTag.SERIALIZED_STRING_TYPE_TAG);
                    }
                    try {
                        stringPtr.set(src, srcOffset + 1, srcLen - 1);
                        patternPtr.set(pattern, patternOffset + 1, patternLen - 1);
                        array.reset();
                        UTF8StringPointable.substrBefore(stringPtr, patternPtr, builder, array);
                        out.writeByte(ATypeTag.SERIALIZED_STRING_TYPE_TAG);
                        out.write(array.getByteArray(), 0, array.getLength());
                    } catch (IOException e) {
                        throw new HyracksDataException(e);
                    }
                    result.set(resultStorage);
                }
            };
        }
    };
}

Example 7 with UTF8StringBuilder

use of org.apache.hyracks.data.std.util.UTF8StringBuilder in project asterixdb by apache.

the class SubstringDescriptor method createEvaluatorFactory.

@Override
public IScalarEvaluatorFactory createEvaluatorFactory(final IScalarEvaluatorFactory[] args) {
    return new IScalarEvaluatorFactory() {

        private static final long serialVersionUID = 1L;

        @Override
        public IScalarEvaluator createScalarEvaluator(final IHyracksTaskContext ctx) throws HyracksDataException {
            return new IScalarEvaluator() {

                private final ArrayBackedValueStorage resultStorage = new ArrayBackedValueStorage();

                private final DataOutput out = resultStorage.getDataOutput();

                private IPointable argString = new VoidPointable();

                private IPointable argStart = new VoidPointable();

                private IPointable argLen = new VoidPointable();

                private final IScalarEvaluator evalString = args[0].createScalarEvaluator(ctx);

                private final IScalarEvaluator evalStart = args[1].createScalarEvaluator(ctx);

                private final IScalarEvaluator evalLen = args[2].createScalarEvaluator(ctx);

                private final GrowableArray array = new GrowableArray();

                private final UTF8StringBuilder builder = new UTF8StringBuilder();

                private final UTF8StringPointable string = new UTF8StringPointable();

                @Override
                public void evaluate(IFrameTupleReference tuple, IPointable result) throws HyracksDataException {
                    resultStorage.reset();
                    evalString.evaluate(tuple, argString);
                    evalStart.evaluate(tuple, argStart);
                    evalLen.evaluate(tuple, argLen);
                    byte[] bytes = argStart.getByteArray();
                    int offset = argStart.getStartOffset();
                    int start = ATypeHierarchy.getIntegerValue(getIdentifier().getName(), 0, bytes, offset) - 1;
                    bytes = argLen.getByteArray();
                    offset = argLen.getStartOffset();
                    int len = ATypeHierarchy.getIntegerValue(getIdentifier().getName(), 1, bytes, offset);
                    bytes = argString.getByteArray();
                    offset = argString.getStartOffset();
                    int length = argString.getLength();
                    if (bytes[offset] != ATypeTag.SERIALIZED_STRING_TYPE_TAG) {
                        throw new TypeMismatchException(getIdentifier(), 0, bytes[offset], ATypeTag.SERIALIZED_STRING_TYPE_TAG);
                    }
                    string.set(bytes, offset + 1, length - 1);
                    array.reset();
                    try {
                        UTF8StringPointable.substr(string, start, len, builder, array);
                    } catch (StringIndexOutOfBoundsException e) {
                        throw new RuntimeDataException(ErrorCode.OUT_OF_BOUND, getIdentifier(), 1, start + len - 1);
                    } catch (IOException e) {
                        throw new HyracksDataException(e);
                    }
                    try {
                        out.writeByte(ATypeTag.SERIALIZED_STRING_TYPE_TAG);
                        out.write(array.getByteArray(), 0, array.getLength());
                    } catch (IOException e) {
                        throw new HyracksDataException(e);
                    }
                    result.set(resultStorage);
                }
            };
        }
    };
}

Example 8 with UTF8StringBuilder

use of org.apache.hyracks.data.std.util.UTF8StringBuilder in project asterixdb by apache.

the class Substring2Descriptor method createEvaluatorFactory.

@Override
public IScalarEvaluatorFactory createEvaluatorFactory(final IScalarEvaluatorFactory[] args) {
    return new IScalarEvaluatorFactory() {

        private static final long serialVersionUID = 1L;

        @Override
        public IScalarEvaluator createScalarEvaluator(final IHyracksTaskContext ctx) throws HyracksDataException {
            return new IScalarEvaluator() {

                private ArrayBackedValueStorage resultStorage = new ArrayBackedValueStorage();

                private DataOutput out = resultStorage.getDataOutput();

                private IPointable argString = new VoidPointable();

                private IPointable argStart = new VoidPointable();

                private IScalarEvaluator evalString = args[0].createScalarEvaluator(ctx);

                private IScalarEvaluator evalStart = args[1].createScalarEvaluator(ctx);

                private final GrowableArray array = new GrowableArray();

                private final UTF8StringBuilder builder = new UTF8StringBuilder();

                private final UTF8StringPointable string = new UTF8StringPointable();

                @Override
                public void evaluate(IFrameTupleReference tuple, IPointable result) throws HyracksDataException {
                    resultStorage.reset();
                    evalString.evaluate(tuple, argString);
                    evalStart.evaluate(tuple, argStart);
                    byte[] bytes = argStart.getByteArray();
                    int offset = argStart.getStartOffset();
                    int start = ATypeHierarchy.getIntegerValue(getIdentifier().getName(), 1, bytes, offset) - 1;
                    bytes = argString.getByteArray();
                    offset = argString.getStartOffset();
                    int len = argString.getLength();
                    if (bytes[offset] != ATypeTag.SERIALIZED_STRING_TYPE_TAG) {
                        throw new TypeMismatchException(getIdentifier(), 0, bytes[offset], ATypeTag.SERIALIZED_STRING_TYPE_TAG);
                    }
                    string.set(bytes, offset + 1, len - 1);
                    array.reset();
                    try {
                        UTF8StringPointable.substr(string, start, Integer.MAX_VALUE, builder, array);
                    } catch (StringIndexOutOfBoundsException e) {
                        throw new RuntimeDataException(ErrorCode.OUT_OF_BOUND, getIdentifier(), 1, start);
                    } catch (IOException e) {
                        throw new HyracksDataException(e);
                    }
                    try {
                        out.writeByte(ATypeTag.SERIALIZED_STRING_TYPE_TAG);
                        out.write(array.getByteArray(), 0, array.getLength());
                    } catch (IOException e) {
                        throw new HyracksDataException(e);
                    }
                    result.set(resultStorage);
                }
            };
        }
    };
}

Example 9 with UTF8StringBuilder

use of org.apache.hyracks.data.std.util.UTF8StringBuilder in project asterixdb by apache.

the class UTF8StringPointableTest method testUppercase.

@Test
public void testUppercase() throws Exception {
    UTF8StringBuilder builder = new UTF8StringBuilder();
    GrowableArray storage = new GrowableArray();
    UTF8StringPointable result = new UTF8StringPointable();
    STRING_UTF8_MIX_LOWERCASE.uppercase(builder, storage);
    result.set(storage.getByteArray(), 0, storage.getLength());
    UTF8StringPointable expected = generateUTF8Pointable(UTF8StringSample.STRING_UTF8_MIX_LOWERCASE.toUpperCase());
    assertEquals(0, expected.compareTo(result));
}

Example 10 with UTF8StringBuilder

use of org.apache.hyracks.data.std.util.UTF8StringBuilder in project asterixdb by apache.

the class UTF8StringPointableTest method testSubstrBefore.

@Test
public void testSubstrBefore() throws Exception {
    UTF8StringBuilder builder = new UTF8StringBuilder();
    GrowableArray storage = new GrowableArray();
    STRING_LEN_128.substrBefore(STRING_LEN_127, builder, storage);
    UTF8StringPointable result = new UTF8StringPointable();
    result.set(storage.getByteArray(), 0, storage.getLength());
    assertEquals(0, STRING_EMPTY.compareTo(result));
    storage.reset();
    UTF8StringPointable testPtr = generateUTF8Pointable("Mix中文123");
    UTF8StringPointable pattern = generateUTF8Pointable("文");
    UTF8StringPointable expect = generateUTF8Pointable("Mix中");
    testPtr.substrBefore(pattern, builder, storage);
    result.set(storage.getByteArray(), 0, storage.getLength());
    assertEquals(0, expect.compareTo(result));
}