Examples with ISerializerDeserializer org.apache.hyracks.api.dataflow.value.ISerializerDeserializer used on opensource projects

Example 56 with ISerializerDeserializer

use of org.apache.hyracks.api.dataflow.value.ISerializerDeserializer in project asterixdb by apache.

the class LSMBTreePrimaryIndexSearchOperatorTest method shouldWriteFilterValueIfAppendFilterIsTrue.

@Test
public void shouldWriteFilterValueIfAppendFilterIsTrue() throws Exception {
    JobSpecification spec = new JobSpecification();
    // build tuple containing low and high search key
    // high key and low key
    ArrayTupleBuilder tb = new ArrayTupleBuilder(DataSetConstants.primaryKeyFieldCount * 2);
    DataOutput dos = tb.getDataOutput();
    tb.reset();
    // low key
    new UTF8StringSerializerDeserializer().serialize("100", dos);
    tb.addFieldEndOffset();
    // high key
    new UTF8StringSerializerDeserializer().serialize("200", dos);
    tb.addFieldEndOffset();
    ISerializerDeserializer[] keyRecDescSers = { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() };
    RecordDescriptor keyRecDesc = new RecordDescriptor(keyRecDescSers);
    ConstantTupleSourceOperatorDescriptor keyProviderOp = new ConstantTupleSourceOperatorDescriptor(spec, keyRecDesc, tb.getFieldEndOffsets(), tb.getByteArray(), tb.getSize());
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, keyProviderOp, NC1_ID);
    int[] lowKeyFields = { 0 };
    int[] highKeyFields = { 1 };
    BTreeSearchOperatorDescriptor primaryBtreeSearchOp = new BTreeSearchOperatorDescriptor(spec, DataSetConstants.primaryAndFilterRecDesc, lowKeyFields, highKeyFields, true, true, primaryHelperFactory, false, false, NoopMissingWriterFactory.INSTANCE, NoOpOperationCallbackFactory.INSTANCE, null, null, true);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, primaryBtreeSearchOp, NC1_ID);
    IFileSplitProvider outSplits = new ConstantFileSplitProvider(new FileSplit[] { createFile(nc1) });
    IOperatorDescriptor printer = new PlainFileWriterOperatorDescriptor(spec, outSplits, ",");
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, printer, NC1_ID);
    spec.connect(new OneToOneConnectorDescriptor(spec), keyProviderOp, 0, primaryBtreeSearchOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), primaryBtreeSearchOp, 0, printer, 0);
    spec.addRoot(printer);
    runTest(spec);
}

Example 57 with ISerializerDeserializer

use of org.apache.hyracks.api.dataflow.value.ISerializerDeserializer in project asterixdb by apache.

the class LSMBTreeSecondaryIndexSearchOperatorTest method shouldCarryFilterValueToPrimaryIndexSearch.

@Test
public void shouldCarryFilterValueToPrimaryIndexSearch() throws Exception {
    JobSpecification spec = new JobSpecification();
    // build tuple containing search keys (only use the first key as search
    // key)
    ArrayTupleBuilder tb = new ArrayTupleBuilder(DataSetConstants.secondaryKeyFieldCount);
    DataOutput dos = tb.getDataOutput();
    tb.reset();
    // low key
    new UTF8StringSerializerDeserializer().serialize("1998-07-21", dos);
    tb.addFieldEndOffset();
    // high key
    new UTF8StringSerializerDeserializer().serialize("2000-10-18", dos);
    tb.addFieldEndOffset();
    ISerializerDeserializer[] keyRecDescSers = { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() };
    RecordDescriptor keyRecDesc = new RecordDescriptor(keyRecDescSers);
    ConstantTupleSourceOperatorDescriptor keyProviderOp = new ConstantTupleSourceOperatorDescriptor(spec, keyRecDesc, tb.getFieldEndOffsets(), tb.getByteArray(), tb.getSize());
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, keyProviderOp, NC1_ID);
    int[] secondaryLowKeyFields = { 0 };
    int[] secondaryHighKeyFields = { 1 };
    // search secondary index
    BTreeSearchOperatorDescriptor secondaryBtreeSearchOp = new BTreeSearchOperatorDescriptor(spec, DataSetConstants.secondaryWithFilterRecDesc, secondaryLowKeyFields, secondaryHighKeyFields, true, true, secondaryHelperFactory, false, false, NoopMissingWriterFactory.INSTANCE, NoOpOperationCallbackFactory.INSTANCE, null, null, true);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, secondaryBtreeSearchOp, NC1_ID);
    // second field from the tuples
    int[] primaryLowKeyFields = { 1 };
    // second field from the tuples
    int[] primaryHighKeyFields = { 1 };
    int[] minFilterFields = { 2 };
    int[] maxFilterFields = { 3 };
    // search primary index
    BTreeSearchOperatorDescriptor primaryBtreeSearchOp = new BTreeSearchOperatorDescriptor(spec, DataSetConstants.primaryRecDesc, primaryLowKeyFields, primaryHighKeyFields, true, true, primaryHelperFactory, false, false, NoopMissingWriterFactory.INSTANCE, NoOpOperationCallbackFactory.INSTANCE, minFilterFields, maxFilterFields, false);
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, primaryBtreeSearchOp, NC1_ID);
    IFileSplitProvider outSplits = new ConstantFileSplitProvider(new FileSplit[] { createFile(nc1) });
    IOperatorDescriptor printer = new PlainFileWriterOperatorDescriptor(spec, outSplits, ",");
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, printer, NC1_ID);
    spec.connect(new OneToOneConnectorDescriptor(spec), keyProviderOp, 0, secondaryBtreeSearchOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), secondaryBtreeSearchOp, 0, primaryBtreeSearchOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), primaryBtreeSearchOp, 0, printer, 0);
    spec.addRoot(printer);
    runTest(spec);
}

Example 58 with ISerializerDeserializer

use of org.apache.hyracks.api.dataflow.value.ISerializerDeserializer in project asterixdb by apache.

the class OrderedIndexExamplesTest method bulkLoadExample.

/**
     * Bulk load example. Load a tree with 100,000 tuples. BTree has a composite
     * key to "simulate" non-unique index creation.
     */
@Test
public void bulkLoadExample() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("Bulk load example");
    }
    // Declare fields.
    int fieldCount = 3;
    ITypeTraits[] typeTraits = new ITypeTraits[fieldCount];
    typeTraits[0] = IntegerPointable.TYPE_TRAITS;
    typeTraits[1] = IntegerPointable.TYPE_TRAITS;
    typeTraits[2] = IntegerPointable.TYPE_TRAITS;
    // Declare field serdes.
    ISerializerDeserializer[] fieldSerdes = { IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE };
    // declare keys
    int keyFieldCount = 2;
    IBinaryComparatorFactory[] cmpFactories = new IBinaryComparatorFactory[keyFieldCount];
    cmpFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    cmpFactories[1] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    // This is only used for the LSM-BTree.
    int[] bloomFilterKeyFields = new int[keyFieldCount];
    bloomFilterKeyFields[0] = 0;
    bloomFilterKeyFields[1] = 1;
    ITreeIndex treeIndex = createTreeIndex(typeTraits, cmpFactories, bloomFilterKeyFields, null, null, null, null);
    treeIndex.create();
    treeIndex.activate();
    // Load sorted records.
    int ins = 100000;
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("Bulk loading " + ins + " tuples");
    }
    long start = System.currentTimeMillis();
    IIndexBulkLoader bulkLoader = treeIndex.createBulkLoader(0.7f, false, ins, true);
    ArrayTupleBuilder tb = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    for (int i = 0; i < ins; i++) {
        TupleUtils.createIntegerTuple(tb, tuple, i, i, 5);
        bulkLoader.add(tuple);
    }
    bulkLoader.end();
    long end = System.currentTimeMillis();
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info(ins + " tuples loaded in " + (end - start) + "ms");
    }
    IIndexAccessor indexAccessor = treeIndex.createAccessor(TestOperationCallback.INSTANCE, TestOperationCallback.INSTANCE);
    // Build low key.
    ArrayTupleBuilder lowKeyTb = new ArrayTupleBuilder(1);
    ArrayTupleReference lowKey = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(lowKeyTb, lowKey, 44444);
    // Build high key.
    ArrayTupleBuilder highKeyTb = new ArrayTupleBuilder(1);
    ArrayTupleReference highKey = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(highKeyTb, highKey, 44500);
    // Prefix-Range search in [44444, 44500]
    rangeSearch(cmpFactories, indexAccessor, fieldSerdes, lowKey, highKey, null, null);
    treeIndex.validate();
    treeIndex.deactivate();
    treeIndex.destroy();
}

Example 59 with ISerializerDeserializer

use of org.apache.hyracks.api.dataflow.value.ISerializerDeserializer in project asterixdb by apache.

the class OrderedIndexExamplesTest method deleteExample.

/**
     * Deletion Example. Create a BTree with one variable-length key field and
     * one variable-length value field. Fill B-tree with random values using
     * insertions, then delete entries one-by-one. Repeat procedure a few times
     * on same BTree.
     */
@Test
public void deleteExample() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("Deletion Example");
    }
    // Declare fields.
    int fieldCount = 2;
    ITypeTraits[] typeTraits = new ITypeTraits[fieldCount];
    typeTraits[0] = UTF8StringPointable.TYPE_TRAITS;
    typeTraits[1] = UTF8StringPointable.TYPE_TRAITS;
    // Declare field serdes.
    ISerializerDeserializer[] fieldSerdes = { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() };
    // Declare keys.
    int keyFieldCount = 1;
    IBinaryComparatorFactory[] cmpFactories = new IBinaryComparatorFactory[keyFieldCount];
    cmpFactories[0] = PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY);
    // This is only used for the LSM-BTree.
    int[] bloomFilterKeyFields = new int[keyFieldCount];
    bloomFilterKeyFields[0] = 0;
    ITreeIndex treeIndex = createTreeIndex(typeTraits, cmpFactories, bloomFilterKeyFields, null, null, null, null);
    treeIndex.create();
    treeIndex.activate();
    ArrayTupleBuilder tb = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    IIndexAccessor indexAccessor = treeIndex.createAccessor(TestOperationCallback.INSTANCE, TestOperationCallback.INSTANCE);
    // Max string length to be generated.
    int runs = 3;
    for (int run = 0; run < runs; run++) {
        if (LOGGER.isLoggable(Level.INFO)) {
            LOGGER.info("Deletion example run: " + (run + 1) + "/" + runs);
            LOGGER.info("Inserting into tree...");
        }
        int maxLength = 10;
        int ins = 10000;
        String[] f0s = new String[ins];
        String[] f1s = new String[ins];
        int insDone = 0;
        int[] insDoneCmp = new int[ins];
        for (int i = 0; i < ins; i++) {
            String f0 = randomString(Math.abs(rnd.nextInt()) % maxLength + 1, rnd);
            String f1 = randomString(Math.abs(rnd.nextInt()) % maxLength + 1, rnd);
            TupleUtils.createTuple(tb, tuple, fieldSerdes, f0, f1);
            f0s[i] = f0;
            f1s[i] = f1;
            if (LOGGER.isLoggable(Level.INFO)) {
                if (i % 1000 == 0) {
                    LOGGER.info("Inserting " + i);
                }
            }
            try {
                indexAccessor.insert(tuple);
                insDone++;
            } catch (HyracksDataException e) {
                if (e.getErrorCode() != ErrorCode.DUPLICATE_KEY) {
                    throw e;
                }
            }
            insDoneCmp[i] = insDone;
        }
        if (LOGGER.isLoggable(Level.INFO)) {
            LOGGER.info("Deleting from tree...");
        }
        int delDone = 0;
        for (int i = 0; i < ins; i++) {
            TupleUtils.createTuple(tb, tuple, fieldSerdes, f0s[i], f1s[i]);
            if (LOGGER.isLoggable(Level.INFO)) {
                if (i % 1000 == 0) {
                    LOGGER.info("Deleting " + i);
                }
            }
            try {
                indexAccessor.delete(tuple);
                delDone++;
            } catch (HyracksDataException e) {
                if (e.getErrorCode() != ErrorCode.UPDATE_OR_DELETE_NON_EXISTENT_KEY) {
                    throw e;
                }
            }
            if (insDoneCmp[i] != delDone) {
                if (LOGGER.isLoggable(Level.INFO)) {
                    LOGGER.info("INCONSISTENT STATE, ERROR IN DELETION EXAMPLE.");
                    LOGGER.info("INSDONECMP: " + insDoneCmp[i] + " " + delDone);
                }
                break;
            }
        }
        if (insDone != delDone) {
            if (LOGGER.isLoggable(Level.INFO)) {
                LOGGER.info("ERROR! INSDONE: " + insDone + " DELDONE: " + delDone);
            }
            break;
        }
    }
    treeIndex.validate();
    treeIndex.deactivate();
    treeIndex.destroy();
}

Example 60 with ISerializerDeserializer

use of org.apache.hyracks.api.dataflow.value.ISerializerDeserializer in project asterixdb by apache.

the class OrderedIndexExamplesTest method fixedLengthKeyValueExample.

/**
     * Fixed-Length Key,Value Example. Create a tree index with one fixed-length
     * key field and one fixed-length value field. Fill index with random values
     * using insertions (not bulk load). Perform scans and range search.
     */
@Test
public void fixedLengthKeyValueExample() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("Fixed-Length Key,Value Example.");
    }
    // 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;
    ITreeIndex treeIndex = createTreeIndex(typeTraits, cmpFactories, bloomFilterKeyFields, null, null, null, null);
    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 = 5;
        TupleUtils.createIntegerTuple(tb, tuple, f0, f1);
        if (LOGGER.isLoggable(Level.INFO)) {
            if (i % 1000 == 0) {
                LOGGER.info("Inserting " + i + " : " + f0 + " " + f1);
            }
        }
        try {
            indexAccessor.insert(tuple);
        } catch (HyracksDataException e) {
            if (e.getErrorCode() != ErrorCode.DUPLICATE_KEY) {
                throw e;
            }
        }
    }
    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);
    rangeSearch(cmpFactories, indexAccessor, fieldSerdes, lowKey, highKey, null, null);
    treeIndex.validate();
    treeIndex.deactivate();
    treeIndex.destroy();
}