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

Example 61 with IBinaryComparatorFactory

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

the class Groupby method createJob.

private static JobSpecification createJob(FileSplit[] inSplits, FileSplit[] outSplits, int htSize, long fileSize, int frameLimit, int frameSize, String alg, boolean outPlain) {
    JobSpecification spec = new JobSpecification(frameSize);
    IFileSplitProvider splitsProvider = new ConstantFileSplitProvider(inSplits);
    FileScanOperatorDescriptor fileScanner = new FileScanOperatorDescriptor(spec, splitsProvider, new DelimitedDataTupleParserFactory(lineitemParserFactories, '|'), lineitemDesc);
    createPartitionConstraint(spec, fileScanner, inSplits);
    // Output: each unique string with an integer count
    RecordDescriptor outDesc = new RecordDescriptor(new ISerializerDeserializer[] { IntegerSerializerDeserializer.INSTANCE, // IntegerSerializerDeserializer.INSTANCE,
    IntegerSerializerDeserializer.INSTANCE });
    // Specify the grouping key, which will be the string extracted during
    // the scan.
    int[] keys = new int[] { 0 };
    AbstractOperatorDescriptor grouper;
    if (alg.equalsIgnoreCase("hash")) {
        // external hash graph
        grouper = new ExternalGroupOperatorDescriptor(spec, htSize, fileSize, keys, frameLimit, new IBinaryComparatorFactory[] { // PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY),
        PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY) }, new IntegerNormalizedKeyComputerFactory(), new MultiFieldsAggregatorFactory(new IFieldAggregateDescriptorFactory[] { new CountFieldAggregatorFactory(false) }), new MultiFieldsAggregatorFactory(new IFieldAggregateDescriptorFactory[] { new IntSumFieldAggregatorFactory(keys.length, false) }), outDesc, outDesc, new HashSpillableTableFactory(new IBinaryHashFunctionFamily[] { MurmurHash3BinaryHashFunctionFamily.INSTANCE }));
        createPartitionConstraint(spec, grouper, outSplits);
    } else if (alg.equalsIgnoreCase("sort")) {
        grouper = new SortGroupByOperatorDescriptor(spec, frameLimit, keys, keys, new IntegerNormalizedKeyComputerFactory(), new IBinaryComparatorFactory[] { // PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY),
        PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY) }, new MultiFieldsAggregatorFactory(new IFieldAggregateDescriptorFactory[] { new CountFieldAggregatorFactory(true) }), new MultiFieldsAggregatorFactory(new IFieldAggregateDescriptorFactory[] { new IntSumFieldAggregatorFactory(keys.length, true) }), outDesc, outDesc, false);
        createPartitionConstraint(spec, grouper, outSplits);
    } else {
        System.err.println("unknow groupby alg:" + alg);
        return null;
    }
    // Connect scanner with the grouper
    IConnectorDescriptor scanGroupConnDef2 = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(keys, new IBinaryHashFunctionFactory[] { // PointableBinaryHashFunctionFactory.of(IntegerPointable.FACTORY),
    PointableBinaryHashFunctionFactory.of(IntegerPointable.FACTORY) }));
    spec.connect(scanGroupConnDef2, fileScanner, 0, grouper, 0);
    IFileSplitProvider outSplitProvider = new ConstantFileSplitProvider(outSplits);
    AbstractSingleActivityOperatorDescriptor writer = outPlain ? new PlainFileWriterOperatorDescriptor(spec, outSplitProvider, "|") : new FrameFileWriterOperatorDescriptor(spec, outSplitProvider);
    createPartitionConstraint(spec, writer, outSplits);
    IConnectorDescriptor groupOutConn = new OneToOneConnectorDescriptor(spec);
    spec.connect(groupOutConn, grouper, 0, writer, 0);
    spec.addRoot(writer);
    return spec;
}

Example 62 with IBinaryComparatorFactory

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

the class LSMRTreeTestContext method create.

public static LSMRTreeTestContext create(IIOManager ioManager, List<IVirtualBufferCache> virtualBufferCaches, FileReference file, IBufferCache diskBufferCache, IFileMapProvider diskFileMapProvider, ISerializerDeserializer[] fieldSerdes, IPrimitiveValueProviderFactory[] valueProviderFactories, int numKeyFields, RTreePolicyType rtreePolicyType, double bloomFilterFalsePositiveRate, ILSMMergePolicy mergePolicy, ILSMOperationTracker opTracker, ILSMIOOperationScheduler ioScheduler, ILSMIOOperationCallback ioOpCallback, IMetadataPageManagerFactory metadataPageManagerFactory) throws Exception {
    ITypeTraits[] typeTraits = SerdeUtils.serdesToTypeTraits(fieldSerdes);
    IBinaryComparatorFactory[] rtreeCmpFactories = SerdeUtils.serdesToComparatorFactories(fieldSerdes, numKeyFields);
    int numBtreeFields = fieldSerdes.length - numKeyFields;
    ISerializerDeserializer[] btreeFieldSerdes = new ISerializerDeserializer[numBtreeFields];
    int[] btreeFields = new int[numBtreeFields];
    for (int i = 0; i < numBtreeFields; i++) {
        btreeFields[i] = numKeyFields + i;
        btreeFieldSerdes[i] = fieldSerdes[numKeyFields + i];
    }
    IBinaryComparatorFactory[] btreeCmpFactories = SerdeUtils.serdesToComparatorFactories(btreeFieldSerdes, numBtreeFields);
    LSMRTree lsmTree = LSMRTreeUtils.createLSMTree(ioManager, virtualBufferCaches, file, diskBufferCache, diskFileMapProvider, typeTraits, rtreeCmpFactories, btreeCmpFactories, valueProviderFactories, rtreePolicyType, bloomFilterFalsePositiveRate, mergePolicy, opTracker, ioScheduler, ioOpCallback, LSMRTreeUtils.proposeBestLinearizer(typeTraits, rtreeCmpFactories.length), null, btreeFields, null, null, null, true, false, metadataPageManagerFactory);
    LSMRTreeTestContext testCtx = new LSMRTreeTestContext(fieldSerdes, lsmTree);
    return testCtx;
}

Example 63 with IBinaryComparatorFactory

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

the class LSMRTreeWithAntiMatterTuplesTestContext method create.

public static LSMRTreeWithAntiMatterTuplesTestContext create(IIOManager ioManager, List<IVirtualBufferCache> virtualBufferCaches, FileReference file, IBufferCache diskBufferCache, IFileMapProvider diskFileMapProvider, ISerializerDeserializer[] fieldSerdes, IPrimitiveValueProviderFactory[] valueProviderFactories, int numKeyFields, RTreePolicyType rtreePolicyType, ILSMMergePolicy mergePolicy, ILSMOperationTracker opTracker, ILSMIOOperationScheduler ioScheduler, ILSMIOOperationCallback ioOpCallback, IMetadataPageManagerFactory metadataPageManagerFactory) throws Exception {
    ITypeTraits[] typeTraits = SerdeUtils.serdesToTypeTraits(fieldSerdes);
    IBinaryComparatorFactory[] rtreeCmpFactories = SerdeUtils.serdesToComparatorFactories(fieldSerdes, numKeyFields);
    IBinaryComparatorFactory[] btreeCmpFactories = SerdeUtils.serdesToComparatorFactories(fieldSerdes, fieldSerdes.length);
    LSMRTreeWithAntiMatterTuples lsmTree = LSMRTreeUtils.createLSMTreeWithAntiMatterTuples(ioManager, virtualBufferCaches, file, diskBufferCache, diskFileMapProvider, typeTraits, rtreeCmpFactories, btreeCmpFactories, valueProviderFactories, rtreePolicyType, mergePolicy, opTracker, ioScheduler, ioOpCallback, LSMRTreeUtils.proposeBestLinearizer(typeTraits, rtreeCmpFactories.length), null, null, null, null, true, false, metadataPageManagerFactory);
    LSMRTreeWithAntiMatterTuplesTestContext testCtx = new LSMRTreeWithAntiMatterTuplesTestContext(fieldSerdes, lsmTree);
    return testCtx;
}

Example 64 with IBinaryComparatorFactory

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

the class RTreeSearchCursorTest method rangeSearchTest.

@SuppressWarnings({ "unchecked", "rawtypes" })
@Test
public void rangeSearchTest() throws Exception {
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("TESTING RANGE SEARCH CURSOR FOR RTREE");
    }
    IBufferCache bufferCache = harness.getBufferCache();
    // Declare fields.
    int fieldCount = 5;
    ITypeTraits[] typeTraits = new ITypeTraits[fieldCount];
    typeTraits[0] = IntegerPointable.TYPE_TRAITS;
    typeTraits[1] = IntegerPointable.TYPE_TRAITS;
    typeTraits[2] = IntegerPointable.TYPE_TRAITS;
    typeTraits[3] = IntegerPointable.TYPE_TRAITS;
    typeTraits[4] = IntegerPointable.TYPE_TRAITS;
    // Declare field serdes.
    ISerializerDeserializer[] fieldSerdes = { IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE };
    // Declare keys.
    int keyFieldCount = 4;
    IBinaryComparatorFactory[] cmpFactories = new IBinaryComparatorFactory[keyFieldCount];
    cmpFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    cmpFactories[1] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    cmpFactories[2] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    cmpFactories[3] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    // create value providers
    IPrimitiveValueProviderFactory[] valueProviderFactories = RTreeUtils.createPrimitiveValueProviderFactories(cmpFactories.length, IntegerPointable.FACTORY);
    RTreeTypeAwareTupleWriterFactory tupleWriterFactory = new RTreeTypeAwareTupleWriterFactory(typeTraits);
    ITreeIndexMetadataFrameFactory metaFrameFactory = new LIFOMetaDataFrameFactory();
    ITreeIndexFrameFactory interiorFrameFactory = new RTreeNSMInteriorFrameFactory(tupleWriterFactory, valueProviderFactories, RTreePolicyType.RTREE, false);
    ITreeIndexFrameFactory leafFrameFactory = new RTreeNSMLeafFrameFactory(tupleWriterFactory, valueProviderFactories, RTreePolicyType.RTREE, false);
    IRTreeInteriorFrame interiorFrame = (IRTreeInteriorFrame) interiorFrameFactory.createFrame();
    IRTreeLeafFrame leafFrame = (IRTreeLeafFrame) leafFrameFactory.createFrame();
    IMetadataPageManager freePageManager = new LinkedMetaDataPageManager(bufferCache, metaFrameFactory);
    RTree rtree = new RTree(bufferCache, harness.getFileMapProvider(), freePageManager, interiorFrameFactory, leafFrameFactory, cmpFactories, fieldCount, harness.getFileReference(), false);
    rtree.create();
    rtree.activate();
    ArrayTupleBuilder tb = new ArrayTupleBuilder(fieldCount);
    ArrayTupleReference tuple = new ArrayTupleReference();
    ITreeIndexAccessor indexAccessor = rtree.createAccessor(NoOpOperationCallback.INSTANCE, NoOpOperationCallback.INSTANCE);
    int numInserts = 10000;
    ArrayList<RTreeCheckTuple> checkTuples = new ArrayList<>();
    for (int i = 0; i < numInserts; i++) {
        int p1x = rnd.nextInt();
        int p1y = rnd.nextInt();
        int p2x = rnd.nextInt();
        int p2y = rnd.nextInt();
        int pk = rnd.nextInt();
        ;
        TupleUtils.createIntegerTuple(tb, tuple, Math.min(p1x, p2x), Math.min(p1y, p2y), Math.max(p1x, p2x), Math.max(p1y, p2y), pk);
        try {
            indexAccessor.insert(tuple);
        } catch (HyracksDataException e) {
            if (e.getErrorCode() != ErrorCode.DUPLICATE_KEY) {
                throw e;
            }
        }
        RTreeCheckTuple checkTuple = new RTreeCheckTuple(fieldCount, keyFieldCount);
        checkTuple.appendField(Math.min(p1x, p2x));
        checkTuple.appendField(Math.min(p1y, p2y));
        checkTuple.appendField(Math.max(p1x, p2x));
        checkTuple.appendField(Math.max(p1y, p2y));
        checkTuple.appendField(pk);
        checkTuples.add(checkTuple);
    }
    // Build key.
    ArrayTupleBuilder keyTb = new ArrayTupleBuilder(keyFieldCount);
    ArrayTupleReference key = new ArrayTupleReference();
    TupleUtils.createIntegerTuple(keyTb, key, -1000, -1000, 1000, 1000);
    MultiComparator cmp = MultiComparator.create(cmpFactories);
    ITreeIndexCursor searchCursor = new RTreeSearchCursor(interiorFrame, leafFrame);
    SearchPredicate searchPredicate = new SearchPredicate(key, cmp);
    RTreeCheckTuple keyCheck = (RTreeCheckTuple) rTreeTestUtils.createCheckTupleFromTuple(key, fieldSerdes, keyFieldCount);
    HashMultiSet<RTreeCheckTuple> expectedResult = rTreeTestUtils.getRangeSearchExpectedResults(checkTuples, keyCheck);
    rTreeTestUtils.getRangeSearchExpectedResults(checkTuples, keyCheck);
    indexAccessor.search(searchCursor, searchPredicate);
    rTreeTestUtils.checkExpectedResults(searchCursor, expectedResult, fieldSerdes, keyFieldCount, null);
    rtree.deactivate();
    rtree.destroy();
}

Example 65 with IBinaryComparatorFactory

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

the class BTreeStatsTest method test01.

@Test
public void test01() throws Exception {
    TestStorageManagerComponentHolder.init(PAGE_SIZE, NUM_PAGES, MAX_OPEN_FILES);
    IBufferCache bufferCache = harness.getBufferCache();
    IFileMapProvider fmp = harness.getFileMapProvider();
    // declare fields
    int fieldCount = 2;
    ITypeTraits[] typeTraits = new ITypeTraits[fieldCount];
    typeTraits[0] = IntegerPointable.TYPE_TRAITS;
    typeTraits[1] = IntegerPointable.TYPE_TRAITS;
    // declare keys
    int keyFieldCount = 1;
    IBinaryComparatorFactory[] cmpFactories = new IBinaryComparatorFactory[keyFieldCount];
    cmpFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    TypeAwareTupleWriterFactory tupleWriterFactory = new TypeAwareTupleWriterFactory(typeTraits);
    ITreeIndexFrameFactory leafFrameFactory = new BTreeNSMLeafFrameFactory(tupleWriterFactory);
    ITreeIndexFrameFactory interiorFrameFactory = new BTreeNSMInteriorFrameFactory(tupleWriterFactory);
    ITreeIndexMetadataFrameFactory metaFrameFactory = new LIFOMetaDataFrameFactory();
    IBTreeLeafFrame leafFrame = (IBTreeLeafFrame) leafFrameFactory.createFrame();
    IBTreeInteriorFrame interiorFrame = (IBTreeInteriorFrame) interiorFrameFactory.createFrame();
    ITreeIndexMetadataFrame metaFrame = metaFrameFactory.createFrame();
    IMetadataPageManager freePageManager = new LinkedMetaDataPageManager(bufferCache, metaFrameFactory);
    BTree btree = new BTree(bufferCache, fmp, freePageManager, interiorFrameFactory, leafFrameFactory, cmpFactories, fieldCount, harness.getFileReference());
    btree.create();
    btree.activate();
    Random rnd = new Random();
    rnd.setSeed(50);
    long start = System.currentTimeMillis();
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("INSERTING INTO TREE");
    }
    IFrame frame = new VSizeFrame(ctx);
    FrameTupleAppender appender = new FrameTupleAppender();
    ArrayTupleBuilder tb = new ArrayTupleBuilder(fieldCount);
    DataOutput dos = tb.getDataOutput();
    ISerializerDeserializer[] recDescSers = { IntegerSerializerDeserializer.INSTANCE, IntegerSerializerDeserializer.INSTANCE };
    RecordDescriptor recDesc = new RecordDescriptor(recDescSers);
    IFrameTupleAccessor accessor = new FrameTupleAccessor(recDesc);
    accessor.reset(frame.getBuffer());
    FrameTupleReference tuple = new FrameTupleReference();
    ITreeIndexAccessor indexAccessor = btree.createAccessor(TestOperationCallback.INSTANCE, TestOperationCallback.INSTANCE);
    // 10000
    for (int i = 0; i < 100000; i++) {
        int f0 = rnd.nextInt() % 100000;
        int f1 = 5;
        tb.reset();
        IntegerSerializerDeserializer.INSTANCE.serialize(f0, dos);
        tb.addFieldEndOffset();
        IntegerSerializerDeserializer.INSTANCE.serialize(f1, dos);
        tb.addFieldEndOffset();
        appender.reset(frame, true);
        appender.append(tb.getFieldEndOffsets(), tb.getByteArray(), 0, tb.getSize());
        tuple.reset(accessor, 0);
        if (LOGGER.isLoggable(Level.INFO)) {
            if (i % 10000 == 0) {
                long end = System.currentTimeMillis();
                LOGGER.info("INSERTING " + i + " : " + f0 + " " + f1 + " " + (end - start));
            }
        }
        try {
            indexAccessor.insert(tuple);
        } catch (HyracksDataException e) {
            if (e.getErrorCode() != ErrorCode.DUPLICATE_KEY) {
                e.printStackTrace();
                throw e;
            }
        }
    }
    int fileId = fmp.lookupFileId(harness.getFileReference());
    TreeIndexStatsGatherer statsGatherer = new TreeIndexStatsGatherer(bufferCache, freePageManager, fileId, btree.getRootPageId());
    TreeIndexStats stats = statsGatherer.gatherStats(leafFrame, interiorFrame, metaFrame);
    if (LOGGER.isLoggable(Level.INFO)) {
        LOGGER.info("\n" + stats.toString());
    }
    TreeIndexBufferCacheWarmup bufferCacheWarmup = new TreeIndexBufferCacheWarmup(bufferCache, freePageManager, fileId);
    bufferCacheWarmup.warmup(leafFrame, metaFrame, new int[] { 1, 2 }, new int[] { 2, 5 });
    btree.deactivate();
    btree.destroy();
    bufferCache.close();
}