Examples with ArrayTupleReference org.apache.hyracks.dataflow.common.comm.io.ArrayTupleReference used on opensource projects

Example 6 with ArrayTupleReference

use of org.apache.hyracks.dataflow.common.comm.io.ArrayTupleReference in project asterixdb by apache.

the class LSMBTreeExamplesTest method additionalFilteringingExample.

/**
     * Test the LSM component filters.
     */
@Test
public void additionalFilteringingExample() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("Testing LSMBTree component filters.");
    }
    // Declare fields.
    int fieldCount = 2;
    ITypeTraits[] typeTraits = new ITypeTraits[fieldCount];
    typeTraits[0] = IntegerPointable.TYPE_TRAITS;
    typeTraits[1] = IntegerPointable.TYPE_TRAITS;
    // Declare field serdes.
    ISerializerDeserializer[] fieldSerdes = { IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE };
    // Declare keys.
    int keyFieldCount = 1;
    IBinaryComparatorFactory[] cmpFactories = new IBinaryComparatorFactory[keyFieldCount];
    cmpFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    // This is only used for the LSM-BTree.
    int[] bloomFilterKeyFields = new int[keyFieldCount];
    bloomFilterKeyFields[0] = 0;
    ITypeTraits[] filterTypeTraits = { IntegerPointable.TYPE_TRAITS };
    IBinaryComparatorFactory[] filterCmpFactories = { PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY) };
    int[] filterFields = { 1 };
    int[] btreeFields = { 1 };
    ITreeIndex treeIndex = createTreeIndex(typeTraits, cmpFactories, bloomFilterKeyFields, filterTypeTraits, filterCmpFactories, btreeFields, filterFields);
    treeIndex.create();
    treeIndex.activate();
    long start = System.currentTimeMillis();
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("Inserting into tree...");
    }
    ArrayTupleBuilder tb = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    IIndexAccessor indexAccessor = treeIndex.createAccessor(TestOperationCallback.INSTANCE, TestOperationCallback.INSTANCE);
    int numInserts = 10000;
    for (int i = 0; i < numInserts; i++) {
        int f0 = rnd.nextInt() % numInserts;
        int f1 = i;
        TupleUtils.createIntegerTuple(tb, tuple, f0, f1);
        if (LOGGER.isLoggable(Level.INFO)) {
            if (i % 1000 == 0) {
                LOGGER.info("Inserting " + i + " : " + f0 + " " + f1);
            }
        }
        indexAccessor.insert(tuple);
    }
    long end = System.currentTimeMillis();
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info(numInserts + " inserts in " + (end - start) + "ms");
    }
    orderedScan(indexAccessor, fieldSerdes);
    diskOrderScan(indexAccessor, fieldSerdes);
    // Build low key.
    ArrayTupleBuilder lowKeyTb = new ArrayTupleBuilder(keyFieldCount);
    ArrayTupleReference lowKey = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(lowKeyTb, lowKey, -1000);
    // Build high key.
    ArrayTupleBuilder highKeyTb = new ArrayTupleBuilder(keyFieldCount);
    ArrayTupleReference highKey = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(highKeyTb, highKey, 1000);
    // Build min filter key.
    ArrayTupleBuilder minFilterTb = new ArrayTupleBuilder(filterFields.length);
    ArrayTupleReference minTuple = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(minFilterTb, minTuple, 400);
    // Build max filter key.
    ArrayTupleBuilder maxFilterTb = new ArrayTupleBuilder(filterFields.length);
    ArrayTupleReference maxTuple = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(maxFilterTb, maxTuple, 500);
    rangeSearch(cmpFactories, indexAccessor, fieldSerdes, lowKey, highKey, minTuple, maxTuple);
    treeIndex.validate();
    treeIndex.deactivate();
    treeIndex.destroy();
}

Example 7 with ArrayTupleReference

use of org.apache.hyracks.dataflow.common.comm.io.ArrayTupleReference in project asterixdb by apache.

the class BloomFilterTest method singleFieldTest.

@Test
public void singleFieldTest() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("TESTING BLOOM FILTER");
    }
    IBufferCache bufferCache = harness.getBufferCache();
    int numElements = 100;
    int[] keyFields = { 0 };
    BloomFilter bf = new BloomFilter(bufferCache, harness.getFileMapProvider(), harness.getFileReference(), keyFields);
    double acceptanleFalsePositiveRate = 0.1;
    int maxBucketsPerElement = BloomCalculations.maxBucketsPerElement(numElements);
    BloomFilterSpecification bloomFilterSpec = BloomCalculations.computeBloomSpec(maxBucketsPerElement, acceptanleFalsePositiveRate);
    bf.create();
    bf.activate();
    IIndexBulkLoader builder = bf.createBuilder(numElements, bloomFilterSpec.getNumHashes(), bloomFilterSpec.getNumBucketsPerElements());
    int fieldCount = 2;
    ArrayTupleBuilder tupleBuilder = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    // generate keys
    int maxKey = 1000;
    TreeSet<Integer> uniqueKeys = new TreeSet<>();
    ArrayList<Integer> keys = new ArrayList<>();
    while (uniqueKeys.size() < numElements) {
        int key = rnd.nextInt() % maxKey;
        uniqueKeys.add(key);
    }
    for (Integer i : uniqueKeys) {
        keys.add(i);
    }
    // Insert tuples in the bloom filter
    for (int i = 0; i < keys.size(); ++i) {
        TupleUtils.createIntegerTuple(tupleBuilder, tuple, keys.get(i), i);
        builder.add(tuple);
    }
    builder.end();
    // Check all the inserted tuples can be found.
    long[] hashes = new long[2];
    for (int i = 0; i < keys.size(); ++i) {
        TupleUtils.createIntegerTuple(tupleBuilder, tuple, keys.get(i), i);
        Assert.assertTrue(bf.contains(tuple, hashes));
    }
    bf.deactivate();
    bf.destroy();
}

Example 8 with ArrayTupleReference

use of org.apache.hyracks.dataflow.common.comm.io.ArrayTupleReference in project asterixdb by apache.

the class BloomFilterTest method multiFieldTest.

@Test
public void multiFieldTest() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("TESTING BLOOM FILTER");
    }
    IBufferCache bufferCache = harness.getBufferCache();
    int numElements = 10000;
    int[] keyFields = { 2, 4, 1 };
    BloomFilter bf = new BloomFilter(bufferCache, harness.getFileMapProvider(), harness.getFileReference(), keyFields);
    double acceptanleFalsePositiveRate = 0.1;
    int maxBucketsPerElement = BloomCalculations.maxBucketsPerElement(numElements);
    BloomFilterSpecification bloomFilterSpec = BloomCalculations.computeBloomSpec(maxBucketsPerElement, acceptanleFalsePositiveRate);
    bf.create();
    bf.activate();
    IIndexBulkLoader builder = bf.createBuilder(numElements, bloomFilterSpec.getNumHashes(), bloomFilterSpec.getNumBucketsPerElements());
    int fieldCount = 5;
    ISerializerDeserializer[] fieldSerdes = { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), IntegerSerializerDeserializer.INSTANCE, new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() };
    ArrayTupleBuilder tupleBuilder = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    int maxLength = 20;
    ArrayList<String> s1 = new ArrayList<>();
    ArrayList<String> s2 = new ArrayList<>();
    ArrayList<String> s3 = new ArrayList<>();
    ArrayList<String> s4 = new ArrayList<>();
    for (int i = 0; i < numElements; ++i) {
        s1.add(randomString(rnd.nextInt() % maxLength, rnd));
        s2.add(randomString(rnd.nextInt() % maxLength, rnd));
        s3.add(randomString(rnd.nextInt() % maxLength, rnd));
        s4.add(randomString(rnd.nextInt() % maxLength, rnd));
    }
    for (int i = 0; i < numElements; ++i) {
        TupleUtils.createTuple(tupleBuilder, tuple, fieldSerdes, s1.get(i), s2.get(i), i, s3.get(i), s4.get(i));
        builder.add(tuple);
    }
    builder.end();
    long[] hashes = new long[2];
    for (int i = 0; i < numElements; ++i) {
        TupleUtils.createTuple(tupleBuilder, tuple, fieldSerdes, s1.get(i), s2.get(i), i, s3.get(i), s4.get(i));
        Assert.assertTrue(bf.contains(tuple, hashes));
    }
    bf.deactivate();
    bf.destroy();
}

Example 9 with ArrayTupleReference

use of org.apache.hyracks.dataflow.common.comm.io.ArrayTupleReference in project asterixdb by apache.

the class MurmurHashForITupleReferenceTest method murmurhashTwoIntegerFieldsTest.

@Test
public void murmurhashTwoIntegerFieldsTest() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("TESTING MURMUR HASH TWO INTEGER FIELDS");
    }
    int fieldCount = 2;
    ArrayTupleBuilder tupleBuilder = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(tupleBuilder, tuple, rnd.nextInt(), rnd.nextInt());
    tuple.reset(tupleBuilder.getFieldEndOffsets(), tupleBuilder.getByteArray());
    int[] keyFields = { 0, 1 };
    int length = getTupleSize(tuple, keyFields);
    long[] actuals = new long[NUM_LONG_VARS_FOR_128_BIT_HASH];
    MurmurHash128Bit.hash3_x64_128(tuple, keyFields, 0L, actuals);
    ByteBuffer buffer;
    byte[] array = new byte[length];
    fillArrayWithData(array, keyFields, tuple, length);
    buffer = ByteBuffer.wrap(array);
    long[] expecteds = hash3_x64_128(buffer, 0, length, 0L);
    Assert.assertArrayEquals(expecteds, actuals);
}

Example 10 with ArrayTupleReference

use of org.apache.hyracks.dataflow.common.comm.io.ArrayTupleReference in project asterixdb by apache.

the class MurmurHashForITupleReferenceTest method murmurhashONEIntegerFieldTest.

@Test
public void murmurhashONEIntegerFieldTest() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("TESTING MURMUR HASH ONE INTEGER FIELD");
    }
    int fieldCount = 2;
    ArrayTupleBuilder tupleBuilder = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(tupleBuilder, tuple, rnd.nextInt());
    tuple.reset(tupleBuilder.getFieldEndOffsets(), tupleBuilder.getByteArray());
    int[] keyFields = { 0 };
    int length = getTupleSize(tuple, keyFields);
    long[] actuals = new long[NUM_LONG_VARS_FOR_128_BIT_HASH];
    MurmurHash128Bit.hash3_x64_128(tuple, keyFields, 0L, actuals);
    ByteBuffer buffer;
    byte[] array = new byte[length];
    fillArrayWithData(array, keyFields, tuple, length);
    buffer = ByteBuffer.wrap(array);
    long[] expecteds = hash3_x64_128(buffer, 0, length, 0L);
    Assert.assertArrayEquals(expecteds, actuals);
}