Search in sources :

Example 1 with UTF8StringSerializerDeserializer

use of org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer in project asterixdb by apache.

the class PigletMetadataProvider method getScannerRuntime.

@Override
public Pair<IOperatorDescriptor, AlgebricksPartitionConstraint> getScannerRuntime(IDataSource<String> dataSource, List<LogicalVariable> scanVariables, List<LogicalVariable> projectVariables, boolean projectPushed, List<LogicalVariable> minFilterVars, List<LogicalVariable> maxFilterVars, IOperatorSchema opSchema, IVariableTypeEnvironment typeEnv, JobGenContext context, JobSpecification jobSpec, Object implConfig) throws AlgebricksException {
    PigletFileDataSource ds = (PigletFileDataSource) dataSource;
    FileSplit[] fileSplits = ds.getFileSplits();
    String[] locations = new String[fileSplits.length];
    for (int i = 0; i < fileSplits.length; ++i) {
        locations[i] = fileSplits[i].getNodeName();
    }
    IFileSplitProvider fsp = new ConstantFileSplitProvider(fileSplits);
    Object[] colTypes = ds.getSchemaTypes();
    IValueParserFactory[] vpfs = new IValueParserFactory[colTypes.length];
    ISerializerDeserializer[] serDesers = new ISerializerDeserializer[colTypes.length];
    for (int i = 0; i < colTypes.length; ++i) {
        Type colType = (Type) colTypes[i];
        IValueParserFactory vpf;
        ISerializerDeserializer serDeser;
        switch(colType.getTag()) {
            case INTEGER:
                vpf = IntegerParserFactory.INSTANCE;
                serDeser = IntegerSerializerDeserializer.INSTANCE;
                break;
            case CHAR_ARRAY:
                vpf = UTF8StringParserFactory.INSTANCE;
                serDeser = new UTF8StringSerializerDeserializer();
                break;
            case FLOAT:
                vpf = FloatParserFactory.INSTANCE;
                serDeser = FloatSerializerDeserializer.INSTANCE;
                break;
            default:
                throw new UnsupportedOperationException();
        }
        vpfs[i] = vpf;
        serDesers[i] = serDeser;
    }
    ITupleParserFactory tpf = new DelimitedDataTupleParserFactory(vpfs, ',');
    RecordDescriptor rDesc = new RecordDescriptor(serDesers);
    IOperatorDescriptor scanner = new FileScanOperatorDescriptor(jobSpec, fsp, tpf, rDesc);
    AlgebricksAbsolutePartitionConstraint constraint = new AlgebricksAbsolutePartitionConstraint(locations);
    return new Pair<>(scanner, constraint);
}
Also used : IValueParserFactory(org.apache.hyracks.dataflow.common.data.parsers.IValueParserFactory) IFileSplitProvider(org.apache.hyracks.dataflow.std.file.IFileSplitProvider) RecordDescriptor(org.apache.hyracks.api.dataflow.value.RecordDescriptor) ConstantFileSplitProvider(org.apache.hyracks.dataflow.std.file.ConstantFileSplitProvider) DelimitedDataTupleParserFactory(org.apache.hyracks.dataflow.std.file.DelimitedDataTupleParserFactory) FileSplit(org.apache.hyracks.api.io.FileSplit) UTF8StringSerializerDeserializer(org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer) AlgebricksPartitionConstraint(org.apache.hyracks.algebricks.common.constraints.AlgebricksPartitionConstraint) AlgebricksAbsolutePartitionConstraint(org.apache.hyracks.algebricks.common.constraints.AlgebricksAbsolutePartitionConstraint) ISerializerDeserializer(org.apache.hyracks.api.dataflow.value.ISerializerDeserializer) Type(org.apache.hyracks.algebricks.examples.piglet.types.Type) ITupleParserFactory(org.apache.hyracks.dataflow.std.file.ITupleParserFactory) AlgebricksAbsolutePartitionConstraint(org.apache.hyracks.algebricks.common.constraints.AlgebricksAbsolutePartitionConstraint) IOperatorDescriptor(org.apache.hyracks.api.dataflow.IOperatorDescriptor) FileScanOperatorDescriptor(org.apache.hyracks.dataflow.std.file.FileScanOperatorDescriptor) Pair(org.apache.hyracks.algebricks.common.utils.Pair)

Example 2 with UTF8StringSerializerDeserializer

use of org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer in project asterixdb by apache.

the class SecondaryIndexSearchExample method createJob.

private static JobSpecification createJob(Options options) throws HyracksDataException {
    JobSpecification spec = new JobSpecification(options.frameSize);
    String[] splitNCs = options.ncs.split(",");
    IStorageManager storageManager = BTreeHelperStorageManager.INSTANCE;
    // schema of tuples coming out of secondary index
    RecordDescriptor secondaryRecDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer(), IntegerSerializerDeserializer.INSTANCE });
    int secondaryFieldCount = 2;
    ITypeTraits[] secondaryTypeTraits = new ITypeTraits[secondaryFieldCount];
    secondaryTypeTraits[0] = UTF8StringPointable.TYPE_TRAITS;
    secondaryTypeTraits[1] = IntegerPointable.TYPE_TRAITS;
    // comparators for sort fields and BTree fields
    IBinaryComparatorFactory[] secondaryComparatorFactories = new IBinaryComparatorFactory[2];
    secondaryComparatorFactories[0] = PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY);
    secondaryComparatorFactories[1] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    // comparators for primary index
    IBinaryComparatorFactory[] primaryComparatorFactories = new IBinaryComparatorFactory[1];
    primaryComparatorFactories[1] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    // schema of tuples coming out of primary index
    RecordDescriptor primaryRecDesc = new RecordDescriptor(new ISerializerDeserializer[] { IntegerSerializerDeserializer.INSTANCE, new UTF8StringSerializerDeserializer(), IntegerSerializerDeserializer.INSTANCE, new UTF8StringSerializerDeserializer() });
    int primaryFieldCount = 4;
    ITypeTraits[] primaryTypeTraits = new ITypeTraits[primaryFieldCount];
    primaryTypeTraits[0] = IntegerPointable.TYPE_TRAITS;
    primaryTypeTraits[1] = UTF8StringPointable.TYPE_TRAITS;
    primaryTypeTraits[2] = IntegerPointable.TYPE_TRAITS;
    primaryTypeTraits[3] = UTF8StringPointable.TYPE_TRAITS;
    // comparators for btree, note that we only need a comparator for the
    // non-unique key
    // i.e. we will have a range condition on the first field only (implying
    // [-infinity, +infinity] for the second field)
    IBinaryComparatorFactory[] searchComparatorFactories = new IBinaryComparatorFactory[1];
    searchComparatorFactories[0] = PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY);
    // build tuple containing low and high search keys
    // low
    ArrayTupleBuilder tb = new ArrayTupleBuilder(searchComparatorFactories.length * 2);
    // and
    // high
    // key
    DataOutput dos = tb.getDataOutput();
    tb.reset();
    // low
    new UTF8StringSerializerDeserializer().serialize("0", dos);
    // key
    tb.addFieldEndOffset();
    // high
    new UTF8StringSerializerDeserializer().serialize("f", dos);
    // key
    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());
    JobHelper.createPartitionConstraint(spec, keyProviderOp, splitNCs);
    // low key is in field 0 of tuples
    int[] secondaryLowKeyFields = { 0 };
    // going into secondary index
    // search op
    // high key is in field 1 of
    int[] secondaryHighKeyFields = { 1 };
    // tuples going into secondary
    // index search op
    IFileSplitProvider secondarySplitProvider = JobHelper.createFileSplitProvider(splitNCs, options.secondaryBTreeName);
    IIndexDataflowHelperFactory secondaryHelperFactory = new IndexDataflowHelperFactory(storageManager, secondarySplitProvider);
    BTreeSearchOperatorDescriptor secondarySearchOp = new BTreeSearchOperatorDescriptor(spec, secondaryRecDesc, secondaryLowKeyFields, secondaryHighKeyFields, true, true, secondaryHelperFactory, false, false, null, NoOpOperationCallbackFactory.INSTANCE, null, null, false);
    JobHelper.createPartitionConstraint(spec, secondarySearchOp, splitNCs);
    // secondary index will output tuples with [UTF8String, Integer]
    // the Integer field refers to the key in the primary index of the
    // source data records
    // low key is in field 0 of tuples
    int[] primaryLowKeyFields = { 1 };
    // going into primary index search op
    // high key is in field 1 of tuples
    int[] primaryHighKeyFields = { 1 };
    // going into primary index search
    // op
    IFileSplitProvider primarySplitProvider = JobHelper.createFileSplitProvider(splitNCs, options.primaryBTreeName);
    IIndexDataflowHelperFactory primaryHelperFactory = new IndexDataflowHelperFactory(storageManager, primarySplitProvider);
    BTreeSearchOperatorDescriptor primarySearchOp = new BTreeSearchOperatorDescriptor(spec, primaryRecDesc, primaryLowKeyFields, primaryHighKeyFields, true, true, primaryHelperFactory, false, false, null, NoOpOperationCallbackFactory.INSTANCE, null, null, false);
    JobHelper.createPartitionConstraint(spec, primarySearchOp, splitNCs);
    // have each node print the results of its respective B-Tree
    PrinterOperatorDescriptor printer = new PrinterOperatorDescriptor(spec);
    JobHelper.createPartitionConstraint(spec, printer, splitNCs);
    spec.connect(new OneToOneConnectorDescriptor(spec), keyProviderOp, 0, secondarySearchOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), secondarySearchOp, 0, primarySearchOp, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), primarySearchOp, 0, printer, 0);
    spec.addRoot(printer);
    return spec;
}
Also used : DataOutput(java.io.DataOutput) ITypeTraits(org.apache.hyracks.api.dataflow.value.ITypeTraits) RecordDescriptor(org.apache.hyracks.api.dataflow.value.RecordDescriptor) IFileSplitProvider(org.apache.hyracks.dataflow.std.file.IFileSplitProvider) IBinaryComparatorFactory(org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory) BTreeSearchOperatorDescriptor(org.apache.hyracks.storage.am.btree.dataflow.BTreeSearchOperatorDescriptor) ArrayTupleBuilder(org.apache.hyracks.dataflow.common.comm.io.ArrayTupleBuilder) OneToOneConnectorDescriptor(org.apache.hyracks.dataflow.std.connectors.OneToOneConnectorDescriptor) UTF8StringSerializerDeserializer(org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer) ISerializerDeserializer(org.apache.hyracks.api.dataflow.value.ISerializerDeserializer) IStorageManager(org.apache.hyracks.storage.common.IStorageManager) ConstantTupleSourceOperatorDescriptor(org.apache.hyracks.dataflow.std.misc.ConstantTupleSourceOperatorDescriptor) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) PrinterOperatorDescriptor(org.apache.hyracks.dataflow.std.misc.PrinterOperatorDescriptor) JobSpecification(org.apache.hyracks.api.job.JobSpecification) IndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IndexDataflowHelperFactory) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory)

Example 3 with UTF8StringSerializerDeserializer

use of org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer in project asterixdb by apache.

the class DataGenOperatorDescriptor method createPushRuntime.

@Override
public IOperatorNodePushable createPushRuntime(IHyracksTaskContext ctx, IRecordDescriptorProvider recordDescProvider, int partition, int nPartitions) throws HyracksDataException {
    final FrameTupleAppender appender = new FrameTupleAppender(new VSizeFrame(ctx));
    final RecordDescriptor recDesc = outRecDescs[0];
    final ArrayTupleBuilder tb = new ArrayTupleBuilder(recDesc.getFields().length);
    final Random rnd = new Random(randomSeed);
    final int maxUniqueAttempts = 20;
    return new AbstractUnaryOutputSourceOperatorNodePushable() {

        // for quick & dirty exclusion of duplicates
        // WARNING: could contain numRecord entries and use a lot of memory
        HashSet<String> stringHs = new HashSet<String>();

        HashSet<Integer> intHs = new HashSet<Integer>();

        @Override
        public void initialize() throws HyracksDataException {
            try {
                writer.open();
                for (int i = 0; i < numRecords; i++) {
                    tb.reset();
                    for (int j = 0; j < recDesc.getFieldCount(); j++) {
                        genField(tb, j);
                    }
                    if (!appender.append(tb.getFieldEndOffsets(), tb.getByteArray(), 0, tb.getSize())) {
                        appender.write(writer, true);
                        if (!appender.append(tb.getFieldEndOffsets(), tb.getByteArray(), 0, tb.getSize())) {
                            throw new HyracksDataException("Record size (" + tb.getSize() + ") larger than frame size (" + appender.getBuffer().capacity() + ")");
                        }
                    }
                }
                appender.write(writer, true);
            } catch (Throwable th) {
                writer.fail();
                throw new HyracksDataException(th);
            } finally {
                writer.close();
            }
        }

        private void genField(ArrayTupleBuilder tb, int fieldIndex) throws HyracksDataException {
            DataOutput dos = tb.getDataOutput();
            if (recDesc.getFields()[fieldIndex] instanceof IntegerSerializerDeserializer) {
                int val = -1;
                if (fieldIndex == uniqueField) {
                    int attempt = 0;
                    while (attempt < maxUniqueAttempts) {
                        int tmp = Math.abs(rnd.nextInt()) % (intMaxVal - intMinVal) + intMinVal;
                        if (intHs.contains(tmp))
                            attempt++;
                        else {
                            val = tmp;
                            intHs.add(val);
                            break;
                        }
                    }
                    if (attempt == maxUniqueAttempts)
                        throw new HyracksDataException("MaxUnique attempts reached in datagen");
                } else {
                    val = Math.abs(rnd.nextInt()) % (intMaxVal - intMinVal) + intMinVal;
                }
                recDesc.getFields()[fieldIndex].serialize(val, dos);
                tb.addFieldEndOffset();
            } else if (recDesc.getFields()[fieldIndex] instanceof UTF8StringSerializerDeserializer) {
                String val = null;
                if (fieldIndex == uniqueField) {
                    int attempt = 0;
                    while (attempt < maxUniqueAttempts) {
                        String tmp = randomString(maxStrLen, rnd);
                        if (stringHs.contains(tmp))
                            attempt++;
                        else {
                            val = tmp;
                            stringHs.add(val);
                            break;
                        }
                    }
                    if (attempt == maxUniqueAttempts)
                        throw new HyracksDataException("MaxUnique attempts reached in datagen");
                } else {
                    val = randomString(maxStrLen, rnd);
                }
                recDesc.getFields()[fieldIndex].serialize(val, dos);
                tb.addFieldEndOffset();
            } else {
                throw new HyracksDataException("Type unsupported in data generator. Only integers and strings allowed");
            }
        }

        private String randomString(int length, Random random) {
            String s = Long.toHexString(Double.doubleToLongBits(random.nextDouble()));
            StringBuilder strBuilder = new StringBuilder();
            for (int i = 0; i < s.length() && i < length; i++) {
                strBuilder.append(s.charAt(Math.abs(random.nextInt()) % s.length()));
            }
            return strBuilder.toString();
        }
    };
}
Also used : DataOutput(java.io.DataOutput) AbstractUnaryOutputSourceOperatorNodePushable(org.apache.hyracks.dataflow.std.base.AbstractUnaryOutputSourceOperatorNodePushable) RecordDescriptor(org.apache.hyracks.api.dataflow.value.RecordDescriptor) ArrayTupleBuilder(org.apache.hyracks.dataflow.common.comm.io.ArrayTupleBuilder) UTF8StringSerializerDeserializer(org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer) VSizeFrame(org.apache.hyracks.api.comm.VSizeFrame) HyracksDataException(org.apache.hyracks.api.exceptions.HyracksDataException) IntegerSerializerDeserializer(org.apache.hyracks.dataflow.common.data.marshalling.IntegerSerializerDeserializer) Random(java.util.Random) FrameTupleAppender(org.apache.hyracks.dataflow.common.comm.io.FrameTupleAppender) HashSet(java.util.HashSet)

Example 4 with UTF8StringSerializerDeserializer

use of org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer in project asterixdb by apache.

the class InsertPipelineExample method createJob.

private static JobSpecification createJob(Options options) {
    JobSpecification spec = new JobSpecification(options.frameSize);
    String[] splitNCs = options.ncs.split(",");
    // schema of tuples to be generated: 4 fields with int, string, string,
    // string
    // we will use field 2 as primary key to fill a clustered index
    RecordDescriptor recDesc = new RecordDescriptor(new ISerializerDeserializer[] { // this field will not go into B-Tree
    new UTF8StringSerializerDeserializer(), // we will use this as payload
    new UTF8StringSerializerDeserializer(), // we will use this field as key
    IntegerSerializerDeserializer.INSTANCE, // we will use this as payload
    IntegerSerializerDeserializer.INSTANCE, // we will use this as payload
    new UTF8StringSerializerDeserializer() });
    // generate numRecords records with field 2 being unique, integer values
    // in [0, 100000], and strings with max length of 10 characters, and
    // random seed 100
    DataGenOperatorDescriptor dataGen = new DataGenOperatorDescriptor(spec, recDesc, options.numTuples, 2, 0, 100000, 10, 100);
    // run data generator on first nodecontroller given
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, dataGen, splitNCs[0]);
    IStorageManager storageManager = BTreeHelperStorageManager.INSTANCE;
    // prepare insertion into primary index
    // tuples to be put into B-Tree shall have 4 fields
    int primaryFieldCount = 4;
    ITypeTraits[] primaryTypeTraits = new ITypeTraits[primaryFieldCount];
    primaryTypeTraits[0] = IntegerPointable.TYPE_TRAITS;
    primaryTypeTraits[1] = UTF8StringPointable.TYPE_TRAITS;
    primaryTypeTraits[2] = IntegerPointable.TYPE_TRAITS;
    primaryTypeTraits[3] = UTF8StringPointable.TYPE_TRAITS;
    // comparator factories for primary index
    IBinaryComparatorFactory[] primaryComparatorFactories = new IBinaryComparatorFactory[1];
    primaryComparatorFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    // the B-Tree expects its keyfields to be at the front of its input
    // tuple
    // map field 2 of input
    int[] primaryFieldPermutation = { 2, 1, 3, 4 };
    // tuple to field 0 of
    // B-Tree tuple, etc.
    IFileSplitProvider primarySplitProvider = JobHelper.createFileSplitProvider(splitNCs, options.primaryBTreeName);
    IIndexDataflowHelperFactory primaryHelperFactory = new IndexDataflowHelperFactory(storageManager, primarySplitProvider);
    // create operator descriptor
    TreeIndexInsertUpdateDeleteOperatorDescriptor primaryInsert = new TreeIndexInsertUpdateDeleteOperatorDescriptor(spec, recDesc, primaryFieldPermutation, IndexOperation.INSERT, primaryHelperFactory, null, NoOpOperationCallbackFactory.INSTANCE);
    JobHelper.createPartitionConstraint(spec, primaryInsert, splitNCs);
    // prepare insertion into secondary index
    // tuples to be put into B-Tree shall have 2 fields
    int secondaryFieldCount = 2;
    ITypeTraits[] secondaryTypeTraits = new ITypeTraits[secondaryFieldCount];
    secondaryTypeTraits[0] = UTF8StringPointable.TYPE_TRAITS;
    secondaryTypeTraits[1] = IntegerPointable.TYPE_TRAITS;
    // comparator factories for secondary index
    IBinaryComparatorFactory[] secondaryComparatorFactories = new IBinaryComparatorFactory[2];
    secondaryComparatorFactories[0] = PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY);
    secondaryComparatorFactories[1] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    // the B-Tree expects its keyfields to be at the front of its input
    // tuple
    int[] secondaryFieldPermutation = { 1, 2 };
    IFileSplitProvider secondarySplitProvider = JobHelper.createFileSplitProvider(splitNCs, options.secondaryBTreeName);
    IIndexDataflowHelperFactory secondaryHelperFactory = new IndexDataflowHelperFactory(storageManager, secondarySplitProvider);
    // create operator descriptor
    TreeIndexInsertUpdateDeleteOperatorDescriptor secondaryInsert = new TreeIndexInsertUpdateDeleteOperatorDescriptor(spec, recDesc, secondaryFieldPermutation, IndexOperation.INSERT, secondaryHelperFactory, null, NoOpOperationCallbackFactory.INSTANCE);
    JobHelper.createPartitionConstraint(spec, secondaryInsert, splitNCs);
    // end the insert pipeline at this sink operator
    NullSinkOperatorDescriptor nullSink = new NullSinkOperatorDescriptor(spec);
    JobHelper.createPartitionConstraint(spec, nullSink, splitNCs);
    // distribute the records from the datagen via hashing to the bulk load
    // ops
    IBinaryHashFunctionFactory[] hashFactories = new IBinaryHashFunctionFactory[1];
    hashFactories[0] = PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY);
    IConnectorDescriptor hashConn = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 0 }, hashFactories));
    // connect the ops
    spec.connect(hashConn, dataGen, 0, primaryInsert, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), primaryInsert, 0, secondaryInsert, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), secondaryInsert, 0, nullSink, 0);
    spec.addRoot(nullSink);
    return spec;
}
Also used : NullSinkOperatorDescriptor(org.apache.hyracks.dataflow.std.misc.NullSinkOperatorDescriptor) IConnectorDescriptor(org.apache.hyracks.api.dataflow.IConnectorDescriptor) ITypeTraits(org.apache.hyracks.api.dataflow.value.ITypeTraits) RecordDescriptor(org.apache.hyracks.api.dataflow.value.RecordDescriptor) IFileSplitProvider(org.apache.hyracks.dataflow.std.file.IFileSplitProvider) IBinaryComparatorFactory(org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory) MToNPartitioningConnectorDescriptor(org.apache.hyracks.dataflow.std.connectors.MToNPartitioningConnectorDescriptor) OneToOneConnectorDescriptor(org.apache.hyracks.dataflow.std.connectors.OneToOneConnectorDescriptor) UTF8StringSerializerDeserializer(org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer) IBinaryHashFunctionFactory(org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory) FieldHashPartitionComputerFactory(org.apache.hyracks.dataflow.common.data.partition.FieldHashPartitionComputerFactory) IStorageManager(org.apache.hyracks.storage.common.IStorageManager) TreeIndexInsertUpdateDeleteOperatorDescriptor(org.apache.hyracks.storage.am.common.dataflow.TreeIndexInsertUpdateDeleteOperatorDescriptor) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) DataGenOperatorDescriptor(org.apache.hyracks.examples.btree.helper.DataGenOperatorDescriptor) JobSpecification(org.apache.hyracks.api.job.JobSpecification) IndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IndexDataflowHelperFactory) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory)

Example 5 with UTF8StringSerializerDeserializer

use of org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer in project asterixdb by apache.

the class PrimaryIndexBulkLoadExample method createJob.

private static JobSpecification createJob(Options options) {
    JobSpecification spec = new JobSpecification(options.frameSize);
    String[] splitNCs = options.ncs.split(",");
    // schema of tuples to be generated: 5 fields with string, string, int,
    // int, string
    // we will use field-index 2 as primary key to fill a clustered index
    RecordDescriptor recDesc = new RecordDescriptor(new ISerializerDeserializer[] { // this field will not go into B-Tree
    new UTF8StringSerializerDeserializer(), // we will use this as payload
    new UTF8StringSerializerDeserializer(), // we will use this field as key
    IntegerSerializerDeserializer.INSTANCE, // we will use this as payload
    IntegerSerializerDeserializer.INSTANCE, // we will use this as payload
    new UTF8StringSerializerDeserializer() });
    // generate numRecords records with field 2 being unique, integer values
    // in [0, 100000], and strings with max length of 10 characters, and
    // random seed 50
    DataGenOperatorDescriptor dataGen = new DataGenOperatorDescriptor(spec, recDesc, options.numTuples, 2, 0, 100000, 10, 50);
    // run data generator on first nodecontroller given
    PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, dataGen, splitNCs[0]);
    // sort the tuples as preparation for bulk load
    // fields to sort on
    int[] sortFields = { 2 };
    // comparators for sort fields
    IBinaryComparatorFactory[] comparatorFactories = new IBinaryComparatorFactory[1];
    comparatorFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
    ExternalSortOperatorDescriptor sorter = new ExternalSortOperatorDescriptor(spec, options.sbSize, sortFields, comparatorFactories, recDesc);
    JobHelper.createPartitionConstraint(spec, sorter, splitNCs);
    // tuples to be put into B-Tree shall have 4 fields
    int fieldCount = 4;
    ITypeTraits[] typeTraits = new ITypeTraits[fieldCount];
    typeTraits[0] = IntegerPointable.TYPE_TRAITS;
    typeTraits[1] = UTF8StringPointable.TYPE_TRAITS;
    typeTraits[2] = IntegerPointable.TYPE_TRAITS;
    typeTraits[3] = UTF8StringPointable.TYPE_TRAITS;
    // create providers for B-Tree
    IStorageManager storageManager = BTreeHelperStorageManager.INSTANCE;
    // the B-Tree expects its keyfields to be at the front of its input
    // tuple
    // map field 2 of input tuple
    int[] fieldPermutation = { 2, 1, 3, 4 };
    // to field 0 of B-Tree tuple,
    // etc.
    IFileSplitProvider btreeSplitProvider = JobHelper.createFileSplitProvider(splitNCs, options.btreeName);
    IIndexDataflowHelperFactory dataflowHelperFactory = new IndexDataflowHelperFactory(storageManager, btreeSplitProvider);
    TreeIndexBulkLoadOperatorDescriptor btreeBulkLoad = new TreeIndexBulkLoadOperatorDescriptor(spec, recDesc, fieldPermutation, 0.7f, false, 1000L, true, dataflowHelperFactory);
    JobHelper.createPartitionConstraint(spec, btreeBulkLoad, splitNCs);
    // distribute the records from the datagen via hashing to the bulk load
    // ops
    IBinaryHashFunctionFactory[] hashFactories = new IBinaryHashFunctionFactory[1];
    hashFactories[0] = PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY);
    IConnectorDescriptor hashConn = new MToNPartitioningConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 0 }, hashFactories));
    NullSinkOperatorDescriptor nsOpDesc = new NullSinkOperatorDescriptor(spec);
    JobHelper.createPartitionConstraint(spec, nsOpDesc, splitNCs);
    spec.connect(hashConn, dataGen, 0, sorter, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), sorter, 0, btreeBulkLoad, 0);
    spec.connect(new OneToOneConnectorDescriptor(spec), btreeBulkLoad, 0, nsOpDesc, 0);
    spec.addRoot(nsOpDesc);
    return spec;
}
Also used : RecordDescriptor(org.apache.hyracks.api.dataflow.value.RecordDescriptor) IFileSplitProvider(org.apache.hyracks.dataflow.std.file.IFileSplitProvider) OneToOneConnectorDescriptor(org.apache.hyracks.dataflow.std.connectors.OneToOneConnectorDescriptor) UTF8StringSerializerDeserializer(org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer) IBinaryHashFunctionFactory(org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory) DataGenOperatorDescriptor(org.apache.hyracks.examples.btree.helper.DataGenOperatorDescriptor) JobSpecification(org.apache.hyracks.api.job.JobSpecification) TreeIndexBulkLoadOperatorDescriptor(org.apache.hyracks.storage.am.common.dataflow.TreeIndexBulkLoadOperatorDescriptor) IndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IndexDataflowHelperFactory) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) IConnectorDescriptor(org.apache.hyracks.api.dataflow.IConnectorDescriptor) NullSinkOperatorDescriptor(org.apache.hyracks.dataflow.std.misc.NullSinkOperatorDescriptor) ITypeTraits(org.apache.hyracks.api.dataflow.value.ITypeTraits) IBinaryComparatorFactory(org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory) MToNPartitioningConnectorDescriptor(org.apache.hyracks.dataflow.std.connectors.MToNPartitioningConnectorDescriptor) FieldHashPartitionComputerFactory(org.apache.hyracks.dataflow.common.data.partition.FieldHashPartitionComputerFactory) IStorageManager(org.apache.hyracks.storage.common.IStorageManager) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) ExternalSortOperatorDescriptor(org.apache.hyracks.dataflow.std.sort.ExternalSortOperatorDescriptor)

Aggregations

UTF8StringSerializerDeserializer (org.apache.hyracks.dataflow.common.data.marshalling.UTF8StringSerializerDeserializer)94 RecordDescriptor (org.apache.hyracks.api.dataflow.value.RecordDescriptor)74 Test (org.junit.Test)69 JobSpecification (org.apache.hyracks.api.job.JobSpecification)67 OneToOneConnectorDescriptor (org.apache.hyracks.dataflow.std.connectors.OneToOneConnectorDescriptor)62 FileScanOperatorDescriptor (org.apache.hyracks.dataflow.std.file.FileScanOperatorDescriptor)51 IFileSplitProvider (org.apache.hyracks.dataflow.std.file.IFileSplitProvider)48 ConstantFileSplitProvider (org.apache.hyracks.dataflow.std.file.ConstantFileSplitProvider)45 IConnectorDescriptor (org.apache.hyracks.api.dataflow.IConnectorDescriptor)37 IOperatorDescriptor (org.apache.hyracks.api.dataflow.IOperatorDescriptor)37 DelimitedDataTupleParserFactory (org.apache.hyracks.dataflow.std.file.DelimitedDataTupleParserFactory)36 ManagedFileSplit (org.apache.hyracks.api.io.ManagedFileSplit)34 FieldHashPartitionComputerFactory (org.apache.hyracks.dataflow.common.data.partition.FieldHashPartitionComputerFactory)34 FileSplit (org.apache.hyracks.api.io.FileSplit)33 IValueParserFactory (org.apache.hyracks.dataflow.common.data.parsers.IValueParserFactory)33 ISerializerDeserializer (org.apache.hyracks.api.dataflow.value.ISerializerDeserializer)31 IBinaryHashFunctionFactory (org.apache.hyracks.api.dataflow.value.IBinaryHashFunctionFactory)28 MToNPartitioningConnectorDescriptor (org.apache.hyracks.dataflow.std.connectors.MToNPartitioningConnectorDescriptor)26 ResultSetId (org.apache.hyracks.api.dataset.ResultSetId)24 ArrayTupleBuilder (org.apache.hyracks.dataflow.common.comm.io.ArrayTupleBuilder)24