use of org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory in project asterixdb by apache.
the class InMemoryStableSortPOperator method contributeRuntimeOperator.
@Override
public void contributeRuntimeOperator(IHyracksJobBuilder builder, JobGenContext context, ILogicalOperator op, IOperatorSchema opSchema, IOperatorSchema[] inputSchemas, IOperatorSchema outerPlanSchema) throws AlgebricksException {
RecordDescriptor recDescriptor = JobGenHelper.mkRecordDescriptor(context.getTypeEnvironment(op), opSchema, context);
int n = sortColumns.length;
int[] sortFields = new int[n];
IBinaryComparatorFactory[] comps = new IBinaryComparatorFactory[n];
int i = 0;
INormalizedKeyComputerFactoryProvider nkcfProvider = context.getNormalizedKeyComputerFactoryProvider();
INormalizedKeyComputerFactory nkcf = null;
IVariableTypeEnvironment env = context.getTypeEnvironment(op);
for (OrderColumn oc : sortColumns) {
LogicalVariable var = oc.getColumn();
sortFields[i] = opSchema.findVariable(var);
Object type = env.getVarType(var);
OrderKind order = oc.getOrder();
if (i == 0 && nkcfProvider != null && type != null) {
nkcf = nkcfProvider.getNormalizedKeyComputerFactory(type, order == OrderKind.ASC);
}
IBinaryComparatorFactoryProvider bcfp = context.getBinaryComparatorFactoryProvider();
comps[i] = bcfp.getBinaryComparatorFactory(type, oc.getOrder() == OrderKind.ASC);
i++;
}
IPushRuntimeFactory runtime = new InMemorySortRuntimeFactory(sortFields, nkcf, comps, null);
builder.contributeMicroOperator(op, runtime, recDescriptor);
ILogicalOperator src = op.getInputs().get(0).getValue();
builder.contributeGraphEdge(src, 0, op, 0);
}
use of org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory in project asterixdb by apache.
the class Dataset method getResourceFactory.
/**
* Create the index dataflow helper factory for a particular index on the dataset
*
* @param mdProvider
* metadata provider to get metadata information, components, and runtimes
* @param index
* the index to get the dataflow helper factory for
* @param recordType
* the record type for the dataset
* @param metaType
* the meta type for the dataset
* @param mergePolicyFactory
* the merge policy factory of the dataset
* @param mergePolicyProperties
* the merge policy properties for the dataset
* @return indexDataflowHelperFactory
* an instance of {@link org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory}
* @throws AlgebricksException
* if dataflow helper factory could not be created
*/
public IResourceFactory getResourceFactory(MetadataProvider mdProvider, Index index, ARecordType recordType, ARecordType metaType, ILSMMergePolicyFactory mergePolicyFactory, Map<String, String> mergePolicyProperties) throws AlgebricksException {
ITypeTraits[] filterTypeTraits = DatasetUtil.computeFilterTypeTraits(this, recordType);
IBinaryComparatorFactory[] filterCmpFactories = DatasetUtil.computeFilterBinaryComparatorFactories(this, recordType, mdProvider.getStorageComponentProvider().getComparatorFactoryProvider());
IResourceFactory resourceFactory;
switch(index.getIndexType()) {
case BTREE:
resourceFactory = bTreeResourceFactoryProvider.getResourceFactory(mdProvider, this, index, recordType, metaType, mergePolicyFactory, mergePolicyProperties, filterTypeTraits, filterCmpFactories);
break;
case RTREE:
resourceFactory = rTreeResourceFactoryProvider.getResourceFactory(mdProvider, this, index, recordType, metaType, mergePolicyFactory, mergePolicyProperties, filterTypeTraits, filterCmpFactories);
break;
case LENGTH_PARTITIONED_NGRAM_INVIX:
case LENGTH_PARTITIONED_WORD_INVIX:
case SINGLE_PARTITION_NGRAM_INVIX:
case SINGLE_PARTITION_WORD_INVIX:
resourceFactory = invertedIndexResourceFactoryProvider.getResourceFactory(mdProvider, this, index, recordType, metaType, mergePolicyFactory, mergePolicyProperties, filterTypeTraits, filterCmpFactories);
break;
default:
throw new CompilationException(ErrorCode.COMPILATION_UNKNOWN_INDEX_TYPE, index.getIndexType().toString());
}
return new DatasetLocalResourceFactory(datasetId, resourceFactory);
}
use of org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory in project asterixdb by apache.
the class VSizeFrameSortMergeTest method sortTask.
public void sortTask(int frameSize, int frameLimit) throws Exception {
JobSpecification spec = new JobSpecification();
FileScanOperatorDescriptor ordScanner = new FileScanOperatorDescriptor(spec, ordersSplitProvider, new DelimitedDataTupleParserFactory(new IValueParserFactory[] { UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE }, '|'), ordersDesc);
PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, ordScanner, NC1_ID, NC2_ID);
spec.setFrameSize(frameSize);
ExternalSortOperatorDescriptor sorter = new ExternalSortOperatorDescriptor(spec, frameLimit, new int[] { 1, 0 }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY), PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, ordersDesc);
PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, sorter, NC1_ID, NC2_ID);
String path = getClass().getSimpleName() + aInteger.getAndIncrement() + ".tmp";
IFileSplitProvider outputSplitProvider = new ConstantFileSplitProvider(new FileSplit[] { new ManagedFileSplit(NC1_ID, path) });
IOperatorDescriptor printer = new PlainFileWriterOperatorDescriptor(spec, outputSplitProvider, "|");
PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, printer, NC1_ID);
spec.connect(new OneToOneConnectorDescriptor(spec), ordScanner, 0, sorter, 0);
spec.connect(new MToNPartitioningMergingConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 1, 0 }, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY), PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }), new int[] { 1, 0 }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY), PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, new UTF8StringNormalizedKeyComputerFactory()), sorter, 0, printer, 0);
spec.addRoot(printer);
runTest(spec);
System.out.println("Result write into :" + path);
}
use of org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory in project asterixdb by apache.
the class SortMergeTest method sortMergeTest01.
@Test
public void sortMergeTest01() throws Exception {
JobSpecification spec = new JobSpecification();
FileSplit[] ordersSplits = new FileSplit[] { new ManagedFileSplit(NC1_ID, "data" + File.separator + "tpch0.001" + File.separator + "orders-part1.tbl"), new ManagedFileSplit(NC2_ID, "data" + File.separator + "tpch0.001" + File.separator + "orders-part2.tbl") };
IFileSplitProvider ordersSplitProvider = new ConstantFileSplitProvider(ordersSplits);
RecordDescriptor ordersDesc = new RecordDescriptor(new ISerializerDeserializer[] { new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer(), new UTF8StringSerializerDeserializer() });
FileScanOperatorDescriptor ordScanner = new FileScanOperatorDescriptor(spec, ordersSplitProvider, new DelimitedDataTupleParserFactory(new IValueParserFactory[] { UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE, UTF8StringParserFactory.INSTANCE }, '|'), ordersDesc);
PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, ordScanner, NC1_ID, NC2_ID);
InMemorySortOperatorDescriptor sorter = new InMemorySortOperatorDescriptor(spec, new int[] { 1 }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, ordersDesc);
PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, sorter, NC1_ID, NC2_ID);
ResultSetId rsId = new ResultSetId(1);
spec.addResultSetId(rsId);
IOperatorDescriptor printer = new ResultWriterOperatorDescriptor(spec, rsId, true, false, ResultSerializerFactoryProvider.INSTANCE.getResultSerializerFactoryProvider());
PartitionConstraintHelper.addAbsoluteLocationConstraint(spec, printer, NC1_ID);
spec.connect(new OneToOneConnectorDescriptor(spec), ordScanner, 0, sorter, 0);
spec.connect(new MToNPartitioningMergingConnectorDescriptor(spec, new FieldHashPartitionComputerFactory(new int[] { 1 }, new IBinaryHashFunctionFactory[] { PointableBinaryHashFunctionFactory.of(UTF8StringPointable.FACTORY) }), new int[] { 1 }, new IBinaryComparatorFactory[] { PointableBinaryComparatorFactory.of(UTF8StringPointable.FACTORY) }, new UTF8StringNormalizedKeyComputerFactory()), sorter, 0, printer, 0);
runTest(spec);
}
use of org.apache.hyracks.api.dataflow.value.IBinaryComparatorFactory in project asterixdb by apache.
the class AbstractRTreeExamplesTest method bulkLoadExample.
/**
* Bulk load example. Load a tree with 10,000 tuples.
*/
@Test
public void bulkLoadExample() throws Exception {
if (LOGGER.isLoggable(Level.INFO)) {
LOGGER.info("Bulk load example");
}
// 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 RTree keys.
int rtreeKeyFieldCount = 4;
IBinaryComparatorFactory[] rtreeCmpFactories = new IBinaryComparatorFactory[rtreeKeyFieldCount];
rtreeCmpFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
rtreeCmpFactories[1] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
rtreeCmpFactories[2] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
rtreeCmpFactories[3] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
// Declare BTree keys, this will only be used for LSMRTree
int btreeKeyFieldCount;
IBinaryComparatorFactory[] btreeCmpFactories;
int[] btreeFields = null;
if (rTreeType == RTreeType.LSMRTREE) {
//Parameters look different for LSM RTREE from LSM RTREE WITH ANTI MATTER TUPLES
btreeKeyFieldCount = 1;
btreeCmpFactories = new IBinaryComparatorFactory[btreeKeyFieldCount];
btreeCmpFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
btreeFields = new int[btreeKeyFieldCount];
for (int i = 0; i < btreeKeyFieldCount; i++) {
btreeFields[i] = rtreeKeyFieldCount + i;
}
} else {
btreeKeyFieldCount = 5;
btreeCmpFactories = new IBinaryComparatorFactory[btreeKeyFieldCount];
btreeCmpFactories[0] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
btreeCmpFactories[1] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
btreeCmpFactories[2] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
btreeCmpFactories[3] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
btreeCmpFactories[4] = PointableBinaryComparatorFactory.of(IntegerPointable.FACTORY);
}
// create value providers
IPrimitiveValueProviderFactory[] valueProviderFactories = RTreeUtils.createPrimitiveValueProviderFactories(rtreeCmpFactories.length, IntegerPointable.FACTORY);
//6
ITreeIndex treeIndex = createTreeIndex(typeTraits, rtreeCmpFactories, btreeCmpFactories, valueProviderFactories, RTreePolicyType.RTREE, null, btreeFields, null, null, null);
treeIndex.create();
treeIndex.activate();
// Load records.
int numInserts = 10000;
if (LOGGER.isLoggable(Level.INFO)) {
LOGGER.info("Bulk loading " + numInserts + " tuples");
}
long start = System.currentTimeMillis();
IIndexBulkLoader bulkLoader = treeIndex.createBulkLoader(0.7f, false, numInserts, true);
ArrayTupleBuilder tb = new ArrayTupleBuilder(fieldCount);
ArrayTupleReference tuple = new ArrayTupleReference();
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 = 5;
TupleUtils.createIntegerTuple(tb, tuple, Math.min(p1x, p2x), Math.min(p1y, p2y), Math.max(p1x, p2x), Math.max(p1y, p2y), pk);
bulkLoader.add(tuple);
}
bulkLoader.end();
long end = System.currentTimeMillis();
if (LOGGER.isLoggable(Level.INFO)) {
LOGGER.info(numInserts + " tuples loaded in " + (end - start) + "ms");
}
IIndexAccessor indexAccessor = treeIndex.createAccessor(NoOpOperationCallback.INSTANCE, NoOpOperationCallback.INSTANCE);
// Build key.
ArrayTupleBuilder keyTb = new ArrayTupleBuilder(rtreeKeyFieldCount);
ArrayTupleReference key = new ArrayTupleReference();
TupleUtils.createIntegerTuple(keyTb, key, -1000, -1000, 1000, 1000);
rangeSearch(rtreeCmpFactories, indexAccessor, fieldSerdes, key, null, null);
treeIndex.deactivate();
treeIndex.destroy();
}
Aggregations