Package org.apache.cassandra.index.sai.disk.v1.segment

Interface SegmentOrdering

All Known Implementing Classes:

IndexSegmentSearcher, LiteralIndexSegmentSearcher, NumericIndexSegmentSearcher, Segment, VectorIndexSegmentSearcher

public interface SegmentOrdering

A SegmentOrdering orders an index and produces a stream of PrimaryKeyWithScores. The limit can be used to lazily order the PrimaryKeys. Due to the possiblity for shadowed or updated keys, a SegmentOrdering should be able to order the whole index until exhausted. When using SegmentOrdering there are several steps to build the list of Primary Keys to be ordered: 1. Find all primary keys that match each non-ordering query predicate. 2. Union and intersect the results of step 1 to build a single KeyRangeIterator ordered by PrimaryKey. 3. Fan the primary keys from step 2 out to each sstable segment to order the list of primary keys.

SegmentOrdering handles the third step.

Note: a segment ordering is only used when a query has both ordering and non-ordering predicates. Where a query has only ordering predicates, the ordering is handled by SSTableIndex.search(Expression, AbstractBounds, QueryContext).

Method Summary

Modifier and Type	Method	Description
default CloseableIterator<PrimaryKeyWithScore>	orderResultsBy(QueryContext queryContext, List<PrimaryKey> results, Expression orderer)	Reorder, limit, and put back into original order the results from a single sstable

Method Details

orderResultsBy

default CloseableIterator<PrimaryKeyWithScore> orderResultsBy(QueryContext queryContext,
 List<PrimaryKey> results,
 Expression orderer)
                                                       throws IOException

Reorder, limit, and put back into original order the results from a single sstable

Throws:

IOException