org.apache.cassandra.index.sai.plan.QueryViewBuilder Maven / Gradle / Ivy
Show all versions of cassandra-all Show documentation
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.cassandra.index.sai.plan;
import java.util.ArrayList;
import java.util.Collection;
import java.util.Collections;
import java.util.HashSet;
import java.util.List;
import java.util.NavigableSet;
import java.util.Set;
import java.util.TreeSet;
import java.util.stream.Collectors;
import org.apache.cassandra.db.PartitionPosition;
import org.apache.cassandra.dht.AbstractBounds;
import org.apache.cassandra.index.sai.disk.SSTableIndex;
import org.apache.cassandra.index.sai.view.View;
import org.apache.cassandra.io.sstable.format.SSTableReader;
import org.apache.cassandra.utils.Pair;
/**
* Build a query specific view of the on-disk indexes for a query. This will return a
* {@link Collection} of {@link Expression} and {@link SSTableIndex}s that represent
* the on-disk data for a query.
*
* The query view will include all the indexed expressions even if they don't have any
* on-disk data. This in necessary because the query view is used to query in-memory
* data as well as the attached on-disk indexes.
*/
public class QueryViewBuilder
{
private final Collection expressions;
private final AbstractBounds range;
QueryViewBuilder(Collection expressions, AbstractBounds range)
{
this.expressions = expressions;
this.range = range;
}
public static class QueryView
{
public final Collection>> view;
public final Set referencedIndexes;
public QueryView(Collection>> view, Set referencedIndexes)
{
this.view = view;
this.referencedIndexes = referencedIndexes;
}
}
protected QueryView build()
{
Set referencedIndexes = new HashSet<>();
while (true)
{
referencedIndexes.clear();
boolean failed = false;
Collection>> view = getQueryView(expressions);
for (SSTableIndex index : view.stream().map(pair -> pair.right).flatMap(Collection::stream).collect(Collectors.toList()))
{
if (index.reference())
referencedIndexes.add(index);
else
failed = true;
}
if (failed)
referencedIndexes.forEach(SSTableIndex::release);
else
return new QueryView(view, referencedIndexes);
}
}
private Collection>> getQueryView(Collection expressions)
{
// first let's determine the most selective expression
Pair> mostSelective = calculateMostSelective(expressions);
List>> queryView = new ArrayList<>();
for (Expression expression : expressions)
{
// Non-index column query should only act as FILTER BY for satisfiedBy(Row) method
// because otherwise it likely to go through the whole index.
if (expression.isNotIndexed())
continue;
// If we didn't get a most selective expression then none of the
// expressions select anything so, add an empty entry for the
// expression. We need the empty entry because we may have in-memory
// data for the expression
if (mostSelective == null)
{
queryView.add(Pair.create(expression, Collections.emptyList()));
continue;
}
// If this expression is the most selective then just add it to the
// query view
if (expression.equals(mostSelective.left))
{
queryView.add(mostSelective);
continue;
}
// Finally, we select all the sstable indexes for this expression that
// have overlapping keys with the sstable indexes of the most selective
// and have a term range that is satisfied by the expression.
View view = expression.getIndex().view();
Set indexes = new TreeSet<>(SSTableIndex.COMPARATOR);
indexes.addAll(view.match(expression)
.stream()
.filter(index -> sstableIndexOverlaps(index, mostSelective.right))
.collect(Collectors.toList()));
queryView.add(Pair.create(expression, indexes));
}
return queryView;
}
private boolean sstableIndexOverlaps(SSTableIndex sstableIndex, Collection sstableIndexes)
{
return sstableIndexes.stream().anyMatch(index -> index.bounds().contains(sstableIndex.bounds().left) ||
index.bounds().contains(sstableIndex.bounds().right));
}
// The purpose of this method is to calculate the most selective expression. This is the
// expression with the most sstable indexes that match the expression by term and lie
// within the key range being queried.
//
// The result can be null. This indicates that none of the expressions select any
// sstable indexes.
private Pair> calculateMostSelective(Collection expressions)
{
Expression mostSelectiveExpression = null;
NavigableSet mostSelectiveIndexes = null;
for (Expression expression : expressions)
{
if (expression.isNotIndexed())
continue;
View view = expression.getIndex().view();
NavigableSet indexes = new TreeSet<>(SSTableIndex.COMPARATOR);
indexes.addAll(selectIndexesInRange(view.match(expression)));
if (indexes.isEmpty())
continue;
if (mostSelectiveExpression == null || mostSelectiveIndexes.size() > indexes.size())
{
mostSelectiveIndexes = indexes;
mostSelectiveExpression = expression;
}
}
return mostSelectiveExpression == null ? null : Pair.create(mostSelectiveExpression, mostSelectiveIndexes);
}
private List selectIndexesInRange(Collection indexes)
{
return indexes.stream().filter(this::indexInRange).collect(Collectors.toList());
}
private boolean indexInRange(SSTableIndex index)
{
SSTableReader sstable = index.getSSTable();
return range.left.compareTo(sstable.getLast()) <= 0 && (range.right.isMinimum() || sstable.getFirst().compareTo(range.right) <= 0);
}
}