Search in sources :

Example 1 with IterativeJoin

use of org.apache.rya.api.function.join.IterativeJoin in project incubator-rya by apache.

the class JoinBatchBindingSetUpdater method processBatch.

/**
 * Processes {@link JoinBatchInformation}. Updates the BindingSets
 * associated with the specified nodeId. The BindingSets are processed in
 * batch fashion, where the number of results is indicated by
 * {@link JoinBatchInformation#getBatchSize()}. BindingSets are either
 * Added, Deleted, or Updated according to
 * {@link JoinBatchInformation#getTask()}. In the event that the number of
 * entries that need to be updated exceeds the batch size, the row of the
 * first unprocessed BindingSets is used to create a new JoinBatch job to
 * process the remaining BindingSets.
 * @throws Exception
 */
@Override
public void processBatch(final TransactionBase tx, final Bytes row, final BatchInformation batch) throws Exception {
    super.processBatch(tx, row, batch);
    final String nodeId = BatchRowKeyUtil.getNodeId(row);
    Preconditions.checkArgument(batch instanceof JoinBatchInformation);
    final JoinBatchInformation joinBatch = (JoinBatchInformation) batch;
    final Task task = joinBatch.getTask();
    // Figure out which join algorithm we are going to use.
    final IterativeJoin joinAlgorithm;
    switch(joinBatch.getJoinType()) {
        case NATURAL_JOIN:
            joinAlgorithm = new NaturalJoin();
            break;
        case LEFT_OUTER_JOIN:
            joinAlgorithm = new LeftOuterJoin();
            break;
        default:
            throw new RuntimeException("Unsupported JoinType: " + joinBatch.getJoinType());
    }
    final Set<VisibilityBindingSet> bsSet = new HashSet<>();
    final Optional<RowColumn> rowCol = fillSiblingBatch(tx, joinBatch, bsSet);
    // Iterates over the resulting BindingSets from the join.
    final Iterator<VisibilityBindingSet> newJoinResults;
    final VisibilityBindingSet bs = joinBatch.getBs();
    if (joinBatch.getSide() == Side.LEFT) {
        newJoinResults = joinAlgorithm.newLeftResult(bs, bsSet.iterator());
    } else {
        newJoinResults = joinAlgorithm.newRightResult(bsSet.iterator(), bs);
    }
    // Read join metadata, create new join BindingSets and insert them into the Fluo table.
    final JoinMetadata joinMetadata = CACHE.readJoinMetadata(tx, nodeId);
    final VariableOrder joinVarOrder = joinMetadata.getVariableOrder();
    while (newJoinResults.hasNext()) {
        final VisibilityBindingSet newJoinResult = newJoinResults.next();
        // create BindingSet value
        final Bytes bsBytes = BS_SERDE.serialize(newJoinResult);
        // make rowId
        Bytes rowKey = BindingHashShardingFunction.addShard(nodeId, joinVarOrder, newJoinResult);
        final Column col = FluoQueryColumns.JOIN_BINDING_SET;
        processTask(tx, task, rowKey, col, bsBytes);
    }
    // update the span and register updated batch job
    if (rowCol.isPresent()) {
        final Span newSpan = getNewSpan(rowCol.get(), joinBatch.getSpan());
        joinBatch.setSpan(newSpan);
        BatchInformationDAO.addBatch(tx, nodeId, joinBatch);
    }
}
Also used : Task(org.apache.rya.indexing.pcj.fluo.app.batch.BatchInformation.Task) VisibilityBindingSet(org.apache.rya.api.model.VisibilityBindingSet) RowColumn(org.apache.fluo.api.data.RowColumn) VariableOrder(org.apache.rya.indexing.pcj.storage.accumulo.VariableOrder) IterativeJoin(org.apache.rya.api.function.join.IterativeJoin) JoinMetadata(org.apache.rya.indexing.pcj.fluo.app.query.JoinMetadata) Span(org.apache.fluo.api.data.Span) Bytes(org.apache.fluo.api.data.Bytes) RowColumn(org.apache.fluo.api.data.RowColumn) Column(org.apache.fluo.api.data.Column) NaturalJoin(org.apache.rya.api.function.join.NaturalJoin) LeftOuterJoin(org.apache.rya.api.function.join.LeftOuterJoin) HashSet(java.util.HashSet)

Example 2 with IterativeJoin

use of org.apache.rya.api.function.join.IterativeJoin in project incubator-rya by apache.

the class JoinResultUpdater method updateJoinResults.

/**
 * Updates the results of a Join node when one of its children has added a
 * new Binding Set to its results.
 *
 * @param tx - The transaction all Fluo queries will use. (not null)
 * @param childNodeId - The Node ID of the child whose results received a new Binding Set. (not null)
 * @param childBindingSet - The Binding Set that was just emitted by child node. (not null)
 * @param joinMetadata - The metadata for the Join that has been notified. (not null)
 * @throws Exception The update could not be successfully performed.
 */
public void updateJoinResults(final TransactionBase tx, final String childNodeId, final VisibilityBindingSet childBindingSet, final JoinMetadata joinMetadata) throws Exception {
    checkNotNull(tx);
    checkNotNull(childNodeId);
    checkNotNull(childBindingSet);
    checkNotNull(joinMetadata);
    log.trace("Transaction ID: " + tx.getStartTimestamp() + "\n" + "Join Node ID: " + joinMetadata.getNodeId() + "\n" + "Child Node ID: " + childNodeId + "\n" + "Child Binding Set:\n" + childBindingSet + "\n");
    // Figure out which join algorithm we are going to use.
    final IterativeJoin joinAlgorithm;
    switch(joinMetadata.getJoinType()) {
        case NATURAL_JOIN:
            joinAlgorithm = new NaturalJoin();
            break;
        case LEFT_OUTER_JOIN:
            joinAlgorithm = new LeftOuterJoin();
            break;
        default:
            throw new RuntimeException("Unsupported JoinType: " + joinMetadata.getJoinType());
    }
    // Figure out which side of the join the new binding set appeared on.
    final Side emittingSide;
    final String siblingId;
    if (childNodeId.equals(joinMetadata.getLeftChildNodeId())) {
        emittingSide = Side.LEFT;
        siblingId = joinMetadata.getRightChildNodeId();
    } else {
        emittingSide = Side.RIGHT;
        siblingId = joinMetadata.getLeftChildNodeId();
    }
    // Iterates over the sibling node's BindingSets that join with the new binding set.
    final Set<VisibilityBindingSet> siblingBindingSets = new HashSet<>();
    final Span siblingSpan = getSpan(tx, childNodeId, childBindingSet, siblingId);
    final Column siblingColumn = getScanColumnFamily(siblingId);
    final Optional<RowColumn> rowColumn = fillSiblingBatch(tx, siblingSpan, siblingColumn, siblingBindingSets, joinMetadata.getJoinBatchSize());
    // Iterates over the resulting BindingSets from the join.
    final Iterator<VisibilityBindingSet> newJoinResults;
    if (emittingSide == Side.LEFT) {
        newJoinResults = joinAlgorithm.newLeftResult(childBindingSet, siblingBindingSets.iterator());
    } else {
        newJoinResults = joinAlgorithm.newRightResult(siblingBindingSets.iterator(), childBindingSet);
    }
    // Insert the new join binding sets to the Fluo table.
    final VariableOrder joinVarOrder = joinMetadata.getVariableOrder();
    while (newJoinResults.hasNext()) {
        final VisibilityBindingSet newJoinResult = newJoinResults.next();
        // Create the Row Key for the emitted binding set. It does not contain visibilities.
        final Bytes resultRow = makeRowKey(joinMetadata.getNodeId(), joinVarOrder, newJoinResult);
        // Only insert the join Binding Set if it is new or BindingSet contains values not used in resultRow.
        if (tx.get(resultRow, FluoQueryColumns.JOIN_BINDING_SET) == null || joinVarOrder.getVariableOrders().size() < newJoinResult.size()) {
            // Create the Node Value. It does contain visibilities.
            final Bytes nodeValueBytes = BS_SERDE.serialize(newJoinResult);
            log.trace("Transaction ID: " + tx.getStartTimestamp() + "\n" + "New Join Result:\n" + newJoinResult + "\n");
            tx.set(resultRow, FluoQueryColumns.JOIN_BINDING_SET, nodeValueBytes);
        }
    }
    // update the span and register updated batch job
    if (rowColumn.isPresent()) {
        final Span newSpan = AbstractBatchBindingSetUpdater.getNewSpan(rowColumn.get(), siblingSpan);
        final JoinBatchInformation joinBatch = JoinBatchInformation.builder().setBatchSize(joinMetadata.getJoinBatchSize()).setBs(childBindingSet).setColumn(siblingColumn).setJoinType(joinMetadata.getJoinType()).setSide(emittingSide).setSpan(newSpan).setTask(Task.Add).build();
        BatchInformationDAO.addBatch(tx, joinMetadata.getNodeId(), joinBatch);
    }
}
Also used : VisibilityBindingSet(org.apache.rya.api.model.VisibilityBindingSet) RowColumn(org.apache.fluo.api.data.RowColumn) VariableOrder(org.apache.rya.indexing.pcj.storage.accumulo.VariableOrder) IterativeJoin(org.apache.rya.api.function.join.IterativeJoin) JoinBatchInformation(org.apache.rya.indexing.pcj.fluo.app.batch.JoinBatchInformation) Span(org.apache.fluo.api.data.Span) Side(org.apache.rya.api.function.join.LazyJoiningIterator.Side) Bytes(org.apache.fluo.api.data.Bytes) RowColumn(org.apache.fluo.api.data.RowColumn) Column(org.apache.fluo.api.data.Column) NaturalJoin(org.apache.rya.api.function.join.NaturalJoin) LeftOuterJoin(org.apache.rya.api.function.join.LeftOuterJoin) HashSet(java.util.HashSet)

Aggregations

HashSet (java.util.HashSet)2 Bytes (org.apache.fluo.api.data.Bytes)2 Column (org.apache.fluo.api.data.Column)2 RowColumn (org.apache.fluo.api.data.RowColumn)2 Span (org.apache.fluo.api.data.Span)2 IterativeJoin (org.apache.rya.api.function.join.IterativeJoin)2 LeftOuterJoin (org.apache.rya.api.function.join.LeftOuterJoin)2 NaturalJoin (org.apache.rya.api.function.join.NaturalJoin)2 VisibilityBindingSet (org.apache.rya.api.model.VisibilityBindingSet)2 VariableOrder (org.apache.rya.indexing.pcj.storage.accumulo.VariableOrder)2 Side (org.apache.rya.api.function.join.LazyJoiningIterator.Side)1 Task (org.apache.rya.indexing.pcj.fluo.app.batch.BatchInformation.Task)1 JoinBatchInformation (org.apache.rya.indexing.pcj.fluo.app.batch.JoinBatchInformation)1 JoinMetadata (org.apache.rya.indexing.pcj.fluo.app.query.JoinMetadata)1