Search in sources :

Example 1 with ExternalFilesIndexCreateOperatorDescriptor

use of org.apache.asterix.external.operators.ExternalFilesIndexCreateOperatorDescriptor in project asterixdb by apache.

the class ExternalIndexingOperations method buildFilesIndexCreateJobSpec.

public static JobSpecification buildFilesIndexCreateJobSpec(Dataset dataset, List<ExternalFile> externalFilesSnapshot, MetadataProvider metadataProvider) throws AlgebricksException {
    IStorageComponentProvider storageComponentProvider = metadataProvider.getStorageComponentProvider();
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    Pair<ILSMMergePolicyFactory, Map<String, String>> compactionInfo = DatasetUtil.getMergePolicyFactory(dataset, metadataProvider.getMetadataTxnContext());
    ILSMMergePolicyFactory mergePolicyFactory = compactionInfo.first;
    Map<String, String> mergePolicyProperties = compactionInfo.second;
    Pair<IFileSplitProvider, AlgebricksPartitionConstraint> secondarySplitsAndConstraint = metadataProvider.getSplitProviderAndConstraints(dataset, IndexingConstants.getFilesIndexName(dataset.getDatasetName()));
    IFileSplitProvider secondaryFileSplitProvider = secondarySplitsAndConstraint.first;
    String fileIndexName = IndexingConstants.getFilesIndexName(dataset.getDatasetName());
    Index fileIndex = MetadataManager.INSTANCE.getIndex(metadataProvider.getMetadataTxnContext(), dataset.getDataverseName(), dataset.getDatasetName(), fileIndexName);
    ARecordType recordType = (ARecordType) metadataProvider.findType(dataset.getItemTypeDataverseName(), dataset.getItemTypeName());
    IResourceFactory resourceFactory = dataset.getResourceFactory(metadataProvider, fileIndex, recordType, null, mergePolicyFactory, mergePolicyProperties);
    IIndexBuilderFactory indexBuilderFactory = new IndexBuilderFactory(storageComponentProvider.getStorageManager(), secondaryFileSplitProvider, resourceFactory, !dataset.isTemp());
    IIndexDataflowHelperFactory dataflowHelperFactory = new IndexDataflowHelperFactory(storageComponentProvider.getStorageManager(), secondaryFileSplitProvider);
    ExternalFilesIndexCreateOperatorDescriptor externalFilesOp = new ExternalFilesIndexCreateOperatorDescriptor(spec, indexBuilderFactory, dataflowHelperFactory, externalFilesSnapshot);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, externalFilesOp, secondarySplitsAndConstraint.second);
    spec.addRoot(externalFilesOp);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}
Also used : IStorageComponentProvider(org.apache.asterix.common.context.IStorageComponentProvider) ConnectorPolicyAssignmentPolicy(org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy) IFileSplitProvider(org.apache.hyracks.dataflow.std.file.IFileSplitProvider) IIndexBuilderFactory(org.apache.hyracks.storage.am.common.api.IIndexBuilderFactory) IndexBuilderFactory(org.apache.hyracks.storage.am.common.build.IndexBuilderFactory) Index(org.apache.asterix.metadata.entities.Index) ExternalFilesIndexCreateOperatorDescriptor(org.apache.asterix.external.operators.ExternalFilesIndexCreateOperatorDescriptor) IIndexBuilderFactory(org.apache.hyracks.storage.am.common.api.IIndexBuilderFactory) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) AlgebricksPartitionConstraint(org.apache.hyracks.algebricks.common.constraints.AlgebricksPartitionConstraint) JobSpecification(org.apache.hyracks.api.job.JobSpecification) Map(java.util.Map) ARecordType(org.apache.asterix.om.types.ARecordType) IResourceFactory(org.apache.hyracks.storage.common.IResourceFactory) IIndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory) IndexDataflowHelperFactory(org.apache.hyracks.storage.am.common.dataflow.IndexDataflowHelperFactory) ILSMMergePolicyFactory(org.apache.hyracks.storage.am.lsm.common.api.ILSMMergePolicyFactory)

Aggregations

Map (java.util.Map)1 IStorageComponentProvider (org.apache.asterix.common.context.IStorageComponentProvider)1 ExternalFilesIndexCreateOperatorDescriptor (org.apache.asterix.external.operators.ExternalFilesIndexCreateOperatorDescriptor)1 Index (org.apache.asterix.metadata.entities.Index)1 ARecordType (org.apache.asterix.om.types.ARecordType)1 AlgebricksPartitionConstraint (org.apache.hyracks.algebricks.common.constraints.AlgebricksPartitionConstraint)1 ConnectorPolicyAssignmentPolicy (org.apache.hyracks.algebricks.core.jobgen.impl.ConnectorPolicyAssignmentPolicy)1 JobSpecification (org.apache.hyracks.api.job.JobSpecification)1 IFileSplitProvider (org.apache.hyracks.dataflow.std.file.IFileSplitProvider)1 IIndexBuilderFactory (org.apache.hyracks.storage.am.common.api.IIndexBuilderFactory)1 IndexBuilderFactory (org.apache.hyracks.storage.am.common.build.IndexBuilderFactory)1 IIndexDataflowHelperFactory (org.apache.hyracks.storage.am.common.dataflow.IIndexDataflowHelperFactory)1 IndexDataflowHelperFactory (org.apache.hyracks.storage.am.common.dataflow.IndexDataflowHelperFactory)1 ILSMMergePolicyFactory (org.apache.hyracks.storage.am.lsm.common.api.ILSMMergePolicyFactory)1 IResourceFactory (org.apache.hyracks.storage.common.IResourceFactory)1