Search in sources :

Example 71 with Tuple

use of edu.iu.dsc.tws.api.comms.structs.Tuple in project twister2 by DSC-SPIDAL.

the class SortJoinUtilsTest method fullOuterJoinComparision.

/**
 * This test compares the results of in memory and disk based full outer joins.
 * Purpose is to verify the accuracy of disk based full outer join
 */
@Test
public void fullOuterJoinComparision() {
    List<Tuple> left = new ArrayList<>();
    List<Tuple> right = new ArrayList<>();
    Random random = new Random();
    for (int i = 0; i < 100; i++) {
        left.add(Tuple.of(random.nextInt(10), random.nextInt()));
        right.add(Tuple.of(random.nextInt(10), random.nextInt()));
    }
    FSKeyedSortedMerger2 fsk1 = new FSKeyedSortedMerger2(10, 100, "/tmp", "op-1-" + UUID.randomUUID().toString(), MessageTypes.INTEGER, MessageTypes.INTEGER, (Comparator<Integer>) Integer::compare, 0, false, 1);
    for (Tuple tuple : left) {
        byte[] data = MessageTypes.INTEGER.getDataPacker().packToByteArray((Integer) tuple.getValue());
        fsk1.add(tuple.getKey(), data, data.length);
        fsk1.run();
    }
    FSKeyedSortedMerger2 fsk2 = new FSKeyedSortedMerger2(10, 100, "/tmp", "op-2-" + UUID.randomUUID().toString(), MessageTypes.INTEGER, MessageTypes.INTEGER, (Comparator<Integer>) Integer::compare, 0, false, 1);
    for (Tuple tuple : right) {
        byte[] data = MessageTypes.INTEGER.getDataPacker().packToByteArray((Integer) tuple.getValue());
        fsk2.add(tuple.getKey(), data, data.length);
        fsk2.run();
    }
    CommonThreadPool.init(Config.newBuilder().build());
    fsk1.switchToReading();
    fsk2.switchToReading();
    Iterator iterator = SortJoinUtils.fullOuterJoin((RestorableIterator) fsk1.readIterator(), (RestorableIterator) fsk2.readIterator(), new KeyComparatorWrapper((Comparator<Integer>) Integer::compare));
    List<Object> objects = SortJoinUtils.fullOuterJoin(left, right, new KeyComparatorWrapper(Comparator.naturalOrder()));
    objects.sort(Comparator.comparingInt(o -> (Integer) ((JoinedTuple) o).getKey()));
    int i = 0;
    while (iterator.hasNext()) {
        JoinedTuple nextFromIt = (JoinedTuple) iterator.next();
        JoinedTuple nextFromList = (JoinedTuple) objects.get(i++);
        Assert.assertEquals(nextFromIt.getKey(), nextFromList.getKey());
    }
    Assert.assertEquals(i, objects.size());
}
Also used : CommunicationContext(edu.iu.dsc.tws.api.comms.CommunicationContext) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) FSKeyedSortedMerger2(edu.iu.dsc.tws.comms.shuffle.FSKeyedSortedMerger2) RestorableIterator(edu.iu.dsc.tws.comms.shuffle.RestorableIterator) Iterator(java.util.Iterator) CommonThreadPool(edu.iu.dsc.tws.api.util.CommonThreadPool) Random(java.util.Random) Test(org.junit.Test) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Config(edu.iu.dsc.tws.api.config.Config) UUID(java.util.UUID) MessageTypes(edu.iu.dsc.tws.api.comms.messaging.types.MessageTypes) Logger(java.util.logging.Logger) ArrayList(java.util.ArrayList) List(java.util.List) Comparator(java.util.Comparator) Assert(org.junit.Assert) FSKeyedSortedMerger2(edu.iu.dsc.tws.comms.shuffle.FSKeyedSortedMerger2) ArrayList(java.util.ArrayList) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Comparator(java.util.Comparator) Random(java.util.Random) RestorableIterator(edu.iu.dsc.tws.comms.shuffle.RestorableIterator) Iterator(java.util.Iterator) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Test(org.junit.Test)

Example 72 with Tuple

use of edu.iu.dsc.tws.api.comms.structs.Tuple in project twister2 by DSC-SPIDAL.

the class SortJoinUtilsTest method rightOuterJoinComparision.

/**
 * This test compares the results of in memory and disk based right outer joins.
 * Purpose is to verify the accuracy of disk based right outer join
 */
@Test
public void rightOuterJoinComparision() {
    List<Tuple> left = new ArrayList<>();
    List<Tuple> right = new ArrayList<>();
    Random random = new Random();
    for (int i = 0; i < 100; i++) {
        left.add(Tuple.of(random.nextInt(10), random.nextInt()));
        right.add(Tuple.of(random.nextInt(10), random.nextInt()));
    }
    FSKeyedSortedMerger2 fsk1 = new FSKeyedSortedMerger2(10, 100, "/tmp", "op-1-" + UUID.randomUUID().toString(), MessageTypes.INTEGER, MessageTypes.INTEGER, (Comparator<Integer>) Integer::compare, 0, false, 1);
    for (Tuple tuple : left) {
        byte[] data = MessageTypes.INTEGER.getDataPacker().packToByteArray((Integer) tuple.getValue());
        fsk1.add(tuple.getKey(), data, data.length);
        fsk1.run();
    }
    FSKeyedSortedMerger2 fsk2 = new FSKeyedSortedMerger2(10, 100, "/tmp", "op-2-" + UUID.randomUUID().toString(), MessageTypes.INTEGER, MessageTypes.INTEGER, (Comparator<Integer>) Integer::compare, 0, false, 1);
    for (Tuple tuple : right) {
        byte[] data = MessageTypes.INTEGER.getDataPacker().packToByteArray((Integer) tuple.getValue());
        fsk2.add(tuple.getKey(), data, data.length);
        fsk2.run();
    }
    CommonThreadPool.init(Config.newBuilder().build());
    fsk1.switchToReading();
    fsk2.switchToReading();
    Iterator iterator = SortJoinUtils.rightOuterJoin((RestorableIterator) fsk1.readIterator(), (RestorableIterator) fsk2.readIterator(), new KeyComparatorWrapper((Comparator<Integer>) Integer::compare));
    List<Object> objects = SortJoinUtils.rightOuterJoin(left, right, new KeyComparatorWrapper(Comparator.naturalOrder()));
    objects.sort(Comparator.comparingInt(o -> (Integer) ((JoinedTuple) o).getKey()));
    int i = 0;
    while (iterator.hasNext()) {
        JoinedTuple nextFromIt = (JoinedTuple) iterator.next();
        JoinedTuple nextFromList = (JoinedTuple) objects.get(i++);
        Assert.assertEquals(nextFromIt.getKey(), nextFromList.getKey());
    }
    Assert.assertEquals(i, objects.size());
}
Also used : CommunicationContext(edu.iu.dsc.tws.api.comms.CommunicationContext) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) FSKeyedSortedMerger2(edu.iu.dsc.tws.comms.shuffle.FSKeyedSortedMerger2) RestorableIterator(edu.iu.dsc.tws.comms.shuffle.RestorableIterator) Iterator(java.util.Iterator) CommonThreadPool(edu.iu.dsc.tws.api.util.CommonThreadPool) Random(java.util.Random) Test(org.junit.Test) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Config(edu.iu.dsc.tws.api.config.Config) UUID(java.util.UUID) MessageTypes(edu.iu.dsc.tws.api.comms.messaging.types.MessageTypes) Logger(java.util.logging.Logger) ArrayList(java.util.ArrayList) List(java.util.List) Comparator(java.util.Comparator) Assert(org.junit.Assert) FSKeyedSortedMerger2(edu.iu.dsc.tws.comms.shuffle.FSKeyedSortedMerger2) ArrayList(java.util.ArrayList) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Comparator(java.util.Comparator) Random(java.util.Random) RestorableIterator(edu.iu.dsc.tws.comms.shuffle.RestorableIterator) Iterator(java.util.Iterator) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Test(org.junit.Test)

Example 73 with Tuple

use of edu.iu.dsc.tws.api.comms.structs.Tuple in project twister2 by DSC-SPIDAL.

the class SortJoinUtilsTest method innerJoinComparision.

/**
 * This test compares the results of in memory and disk based inner joins.
 * Purpose is to verify the accuracy of disk based inner join
 */
@Test
public void innerJoinComparision() {
    List<Tuple> left = new ArrayList<>();
    List<Tuple> right = new ArrayList<>();
    Random random = new Random();
    for (int i = 0; i < 100; i++) {
        left.add(Tuple.of(random.nextInt(10), random.nextInt()));
        right.add(Tuple.of(random.nextInt(10), random.nextInt()));
    }
    FSKeyedSortedMerger2 fsk1 = new FSKeyedSortedMerger2(10, 100, "/tmp", "op-1-" + UUID.randomUUID().toString(), MessageTypes.INTEGER, MessageTypes.INTEGER, (Comparator<Integer>) Integer::compare, 0, false, 1);
    for (Tuple tuple : left) {
        byte[] data = MessageTypes.INTEGER.getDataPacker().packToByteArray((Integer) tuple.getValue());
        fsk1.add(tuple.getKey(), data, data.length);
        fsk1.run();
    }
    FSKeyedSortedMerger2 fsk2 = new FSKeyedSortedMerger2(10, 100, "/tmp", "op-2-" + UUID.randomUUID().toString(), MessageTypes.INTEGER, MessageTypes.INTEGER, (Comparator<Integer>) Integer::compare, 0, false, 1);
    for (Tuple tuple : right) {
        byte[] data = MessageTypes.INTEGER.getDataPacker().packToByteArray((Integer) tuple.getValue());
        fsk2.add(tuple.getKey(), data, data.length);
        fsk2.run();
    }
    CommonThreadPool.init(Config.newBuilder().build());
    fsk1.switchToReading();
    fsk2.switchToReading();
    Iterator iterator = SortJoinUtils.innerJoin((RestorableIterator) fsk1.readIterator(), (RestorableIterator) fsk2.readIterator(), new KeyComparatorWrapper((Comparator<Integer>) Integer::compare));
    List<Object> objects = SortJoinUtils.innerJoin(left, right, new KeyComparatorWrapper(Comparator.naturalOrder()));
    objects.sort(Comparator.comparingInt(o -> (Integer) ((JoinedTuple) o).getKey()));
    int i = 0;
    while (iterator.hasNext()) {
        JoinedTuple nextFromIt = (JoinedTuple) iterator.next();
        JoinedTuple nextFromList = (JoinedTuple) objects.get(i++);
        Assert.assertEquals(nextFromIt.getKey(), nextFromList.getKey());
    }
    Assert.assertEquals(i, objects.size());
}
Also used : CommunicationContext(edu.iu.dsc.tws.api.comms.CommunicationContext) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) FSKeyedSortedMerger2(edu.iu.dsc.tws.comms.shuffle.FSKeyedSortedMerger2) RestorableIterator(edu.iu.dsc.tws.comms.shuffle.RestorableIterator) Iterator(java.util.Iterator) CommonThreadPool(edu.iu.dsc.tws.api.util.CommonThreadPool) Random(java.util.Random) Test(org.junit.Test) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Config(edu.iu.dsc.tws.api.config.Config) UUID(java.util.UUID) MessageTypes(edu.iu.dsc.tws.api.comms.messaging.types.MessageTypes) Logger(java.util.logging.Logger) ArrayList(java.util.ArrayList) List(java.util.List) Comparator(java.util.Comparator) Assert(org.junit.Assert) FSKeyedSortedMerger2(edu.iu.dsc.tws.comms.shuffle.FSKeyedSortedMerger2) ArrayList(java.util.ArrayList) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Comparator(java.util.Comparator) Random(java.util.Random) RestorableIterator(edu.iu.dsc.tws.comms.shuffle.RestorableIterator) Iterator(java.util.Iterator) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) JoinedTuple(edu.iu.dsc.tws.api.comms.structs.JoinedTuple) Test(org.junit.Test)

Example 74 with Tuple

use of edu.iu.dsc.tws.api.comms.structs.Tuple in project twister2 by DSC-SPIDAL.

the class FSKeyedMergerTest method testStart.

@Test
public void testStart() throws Exception {
    ByteBuffer buffer = ByteBuffer.allocate(4);
    for (int i = 0; i < 1000; i++) {
        buffer.clear();
        buffer.putInt(i);
        byte[] serialize = serializer.serialize(i);
        fsMerger.add(i, serialize, serialize.length);
        fsMerger.run();
    }
    fsMerger.switchToReading();
    Iterator<Object> it = fsMerger.readIterator();
    int count = 0;
    Set<Integer> set = new HashSet<>();
    while (it.hasNext()) {
        LOG.info("Reading value: " + count);
        Tuple val = (Tuple) it.next();
        int k = (int) val.getKey();
        if (set.contains(k)) {
            Assert.fail("Duplicate value");
        }
        set.add(k);
        count++;
    }
    if (count != 1000) {
        Assert.fail("Count = " + count);
    }
}
Also used : ByteBuffer(java.nio.ByteBuffer) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) HashSet(java.util.HashSet) Test(org.junit.Test)

Example 75 with Tuple

use of edu.iu.dsc.tws.api.comms.structs.Tuple in project twister2 by DSC-SPIDAL.

the class KeyedSerializerTest method testBuildLargeByteMessage.

@Test
public void testBuildLargeByteMessage() {
    int numBuffers = 10;
    int size = 1000;
    MessageType type = MessageTypes.BYTE_ARRAY;
    Object data = createKeyedData(800, type, MessageTypes.BYTE_ARRAY);
    InMessage inMessage = keyedSingleValueCase(numBuffers, size, type, MessageTypes.BYTE_ARRAY, data);
    Tuple deserializedData = (Tuple) inMessage.getDeserializedData();
    Assert.assertArrayEquals((byte[]) deserializedData.getKey(), (byte[]) ((Tuple) data).getKey());
    Assert.assertArrayEquals((byte[]) deserializedData.getValue(), (byte[]) ((Tuple) data).getValue());
}
Also used : InMessage(edu.iu.dsc.tws.comms.dfw.InMessage) MessageType(edu.iu.dsc.tws.api.comms.messaging.types.MessageType) Tuple(edu.iu.dsc.tws.api.comms.structs.Tuple) Test(org.junit.Test)

Aggregations

Tuple (edu.iu.dsc.tws.api.comms.structs.Tuple)98 Iterator (java.util.Iterator)38 List (java.util.List)35 Logger (java.util.logging.Logger)34 ArrayList (java.util.ArrayList)29 Config (edu.iu.dsc.tws.api.config.Config)27 WorkerEnvironment (edu.iu.dsc.tws.api.resource.WorkerEnvironment)24 Test (org.junit.Test)24 BatchEnvironment (edu.iu.dsc.tws.tset.env.BatchEnvironment)18 InMessage (edu.iu.dsc.tws.comms.dfw.InMessage)17 HashMap (java.util.HashMap)16 TSetEnvironment (edu.iu.dsc.tws.tset.env.TSetEnvironment)15 JobConfig (edu.iu.dsc.tws.api.JobConfig)14 MessageTypes (edu.iu.dsc.tws.api.comms.messaging.types.MessageTypes)14 JoinedTuple (edu.iu.dsc.tws.api.comms.structs.JoinedTuple)14 ResourceAllocator (edu.iu.dsc.tws.rsched.core.ResourceAllocator)14 SourceTSet (edu.iu.dsc.tws.tset.sets.batch.SourceTSet)13 CommunicationContext (edu.iu.dsc.tws.api.comms.CommunicationContext)11 MessageType (edu.iu.dsc.tws.api.comms.messaging.types.MessageType)11 Comparator (java.util.Comparator)11