Examples with IFileSplitProvider org.apache.hyracks.dataflow.std.file.IFileSplitProvider used on opensource projects

Example 76 with IFileSplitProvider

use of org.apache.hyracks.dataflow.std.file.IFileSplitProvider in project asterixdb by apache.

the class ExternalIndexingOperations method buildFilesIndexUpdateJobSpec.

public static JobSpecification buildFilesIndexUpdateJobSpec(Dataset dataset, List<ExternalFile> externalFilesSnapshot, MetadataProvider metadataProvider) throws AlgebricksException {
    IStorageComponentProvider storageComponentProvider = metadataProvider.getStorageComponentProvider();
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    Pair<IFileSplitProvider, AlgebricksPartitionConstraint> secondarySplitsAndConstraint = metadataProvider.getSplitProviderAndConstraints(dataset, IndexingConstants.getFilesIndexName(dataset.getDatasetName()));
    IFileSplitProvider secondaryFileSplitProvider = secondarySplitsAndConstraint.first;
    IIndexDataflowHelperFactory dataflowHelperFactory = new IndexDataflowHelperFactory(storageComponentProvider.getStorageManager(), secondaryFileSplitProvider);
    ExternalFilesIndexModificationOperatorDescriptor externalFilesOp = new ExternalFilesIndexModificationOperatorDescriptor(spec, dataflowHelperFactory, externalFilesSnapshot);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, externalFilesOp, secondarySplitsAndConstraint.second);
    spec.addRoot(externalFilesOp);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}

Example 77 with IFileSplitProvider

use of org.apache.hyracks.dataflow.std.file.IFileSplitProvider in project asterixdb by apache.

the class ExternalIndexingOperations method buildRecoverOp.

public static JobSpecification buildRecoverOp(Dataset ds, List<Index> indexes, MetadataProvider metadataProvider) throws AlgebricksException {
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    IStorageManager storageMgr = metadataProvider.getStorageComponentProvider().getStorageManager();
    ArrayList<IIndexDataflowHelperFactory> treeDataflowHelperFactories = new ArrayList<>();
    AlgebricksPartitionConstraint constraints = null;
    for (Index index : indexes) {
        IFileSplitProvider indexSplitProvider;
        if (isValidIndexName(index.getDatasetName(), index.getIndexName())) {
            Pair<IFileSplitProvider, AlgebricksPartitionConstraint> sAndConstraints = metadataProvider.getSplitProviderAndConstraints(ds, index.getIndexName());
            indexSplitProvider = sAndConstraints.first;
            constraints = sAndConstraints.second;
        } else {
            indexSplitProvider = metadataProvider.getSplitProviderAndConstraints(ds, IndexingConstants.getFilesIndexName(ds.getDatasetName())).first;
        }
        IIndexDataflowHelperFactory indexDataflowHelperFactory = new IndexDataflowHelperFactory(storageMgr, indexSplitProvider);
        treeDataflowHelperFactories.add(indexDataflowHelperFactory);
    }
    ExternalDatasetIndexesRecoverOperatorDescriptor op = new ExternalDatasetIndexesRecoverOperatorDescriptor(spec, treeDataflowHelperFactories);
    spec.addRoot(op);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, op, constraints);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}

Example 78 with IFileSplitProvider

use of org.apache.hyracks.dataflow.std.file.IFileSplitProvider in project asterixdb by apache.

the class ExternalIndexingOperations method buildCommitJob.

public static JobSpecification buildCommitJob(Dataset ds, List<Index> indexes, MetadataProvider metadataProvider) throws AlgebricksException {
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    IStorageManager storageMgr = metadataProvider.getStorageComponentProvider().getStorageManager();
    ArrayList<IIndexDataflowHelperFactory> treeDataflowHelperFactories = new ArrayList<>();
    AlgebricksPartitionConstraint constraints = null;
    for (Index index : indexes) {
        IFileSplitProvider indexSplitProvider;
        if (isValidIndexName(index.getDatasetName(), index.getIndexName())) {
            Pair<IFileSplitProvider, AlgebricksPartitionConstraint> sAndConstraints = metadataProvider.getSplitProviderAndConstraints(ds, index.getIndexName());
            indexSplitProvider = sAndConstraints.first;
            constraints = sAndConstraints.second;
        } else {
            indexSplitProvider = metadataProvider.getSplitProviderAndConstraints(ds, IndexingConstants.getFilesIndexName(ds.getDatasetName())).first;
        }
        IIndexDataflowHelperFactory indexDataflowHelperFactory = new IndexDataflowHelperFactory(storageMgr, indexSplitProvider);
        treeDataflowHelperFactories.add(indexDataflowHelperFactory);
    }
    ExternalDatasetIndexesCommitOperatorDescriptor op = new ExternalDatasetIndexesCommitOperatorDescriptor(spec, treeDataflowHelperFactories);
    spec.addRoot(op);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, op, constraints);
    spec.setConnectorPolicyAssignmentPolicy(new ConnectorPolicyAssignmentPolicy());
    return spec;
}

Example 79 with IFileSplitProvider

use of org.apache.hyracks.dataflow.std.file.IFileSplitProvider 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;
}

Example 80 with IFileSplitProvider

use of org.apache.hyracks.dataflow.std.file.IFileSplitProvider in project asterixdb by apache.

the class ExternalIndexingOperations method buildDropFilesIndexJobSpec.

public static JobSpecification buildDropFilesIndexJobSpec(MetadataProvider metadataProvider, Dataset dataset) throws AlgebricksException {
    String indexName = IndexingConstants.getFilesIndexName(dataset.getDatasetName());
    JobSpecification spec = RuntimeUtils.createJobSpecification(metadataProvider.getApplicationContext());
    Pair<IFileSplitProvider, AlgebricksPartitionConstraint> splitsAndConstraint = metadataProvider.getSplitProviderAndConstraints(dataset, indexName);
    IIndexDataflowHelperFactory dataflowHelperFactory = new IndexDataflowHelperFactory(metadataProvider.getStorageComponentProvider().getStorageManager(), splitsAndConstraint.first);
    IndexDropOperatorDescriptor btreeDrop = new IndexDropOperatorDescriptor(spec, dataflowHelperFactory);
    AlgebricksPartitionConstraintHelper.setPartitionConstraintInJobSpec(spec, btreeDrop, splitsAndConstraint.second);
    spec.addRoot(btreeDrop);
    return spec;
}