/*
* 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.db.aggregation;
import java.io.IOException;
import org.apache.cassandra.db.ClusteringComparator;
import org.apache.cassandra.db.TypeSizes;
import org.apache.cassandra.io.util.DataInputPlus;
import org.apache.cassandra.io.util.DataOutputPlus;
Defines how rows should be grouped for creating aggregates.
/**
* Defines how rows should be grouped for creating aggregates.
*/
public abstract class AggregationSpecification
{
public static final Serializer serializer = new Serializer();
AggregationSpecification
that group all the row together.
/**
* <code>AggregationSpecification</code> that group all the row together.
*/
public static final AggregationSpecification AGGREGATE_EVERYTHING = new AggregationSpecification(Kind.AGGREGATE_EVERYTHING)
{
@Override
public GroupMaker newGroupMaker(GroupingState state)
{
return GroupMaker.GROUP_EVERYTHING;
}
};
The AggregationSpecification
kind.
/**
* The <code>AggregationSpecification</code> kind.
*/
private final Kind kind;
The AggregationSpecification
kinds.
/**
* The <code>AggregationSpecification</code> kinds.
*/
public static enum Kind
{
AGGREGATE_EVERYTHING, AGGREGATE_BY_PK_PREFIX
}
Returns the AggregationSpecification
kind.
Returns: the AggregationSpecification
kind
/**
* Returns the <code>AggregationSpecification</code> kind.
* @return the <code>AggregationSpecification</code> kind
*/
public Kind kind()
{
return kind;
}
private AggregationSpecification(Kind kind)
{
this.kind = kind;
}
Creates a new GroupMaker
instance.
Returns: a new GroupMaker
instance
/**
* Creates a new <code>GroupMaker</code> instance.
*
* @return a new <code>GroupMaker</code> instance
*/
public final GroupMaker newGroupMaker()
{
return newGroupMaker(GroupingState.EMPTY_STATE);
}
Creates a new GroupMaker
instance.
Params: - state –
GroupMaker
state
Returns: a new GroupMaker
instance
/**
* Creates a new <code>GroupMaker</code> instance.
*
* @param state <code>GroupMaker</code> state
* @return a new <code>GroupMaker</code> instance
*/
public abstract GroupMaker newGroupMaker(GroupingState state);
Creates a new AggregationSpecification
instance that will build aggregates based on primary key
columns.
Params: - comparator – the comparator used to compare the clustering prefixes
- clusteringPrefixSize – the number of clustering columns used to create the aggregates
Returns: a new AggregationSpecification
instance that will build aggregates based on primary key
columns
/**
* Creates a new <code>AggregationSpecification</code> instance that will build aggregates based on primary key
* columns.
*
* @param comparator the comparator used to compare the clustering prefixes
* @param clusteringPrefixSize the number of clustering columns used to create the aggregates
* @return a new <code>AggregationSpecification</code> instance that will build aggregates based on primary key
* columns
*/
public static AggregationSpecification aggregatePkPrefix(ClusteringComparator comparator, int clusteringPrefixSize)
{
return new AggregateByPkPrefix(comparator, clusteringPrefixSize);
}
AggregationSpecification
that build aggregates based on primary key columns
/**
* <code>AggregationSpecification</code> that build aggregates based on primary key columns
*/
private static final class AggregateByPkPrefix extends AggregationSpecification
{
The number of clustering component to compare.
/**
* The number of clustering component to compare.
*/
private final int clusteringPrefixSize;
The comparator used to compare the clustering prefixes.
/**
* The comparator used to compare the clustering prefixes.
*/
private final ClusteringComparator comparator;
public AggregateByPkPrefix(ClusteringComparator comparator, int clusteringPrefixSize)
{
super(Kind.AGGREGATE_BY_PK_PREFIX);
this.comparator = comparator;
this.clusteringPrefixSize = clusteringPrefixSize;
}
@Override
public GroupMaker newGroupMaker(GroupingState state)
{
return GroupMaker.newInstance(comparator, clusteringPrefixSize, state);
}
}
public static class Serializer
{
public void serialize(AggregationSpecification aggregationSpec, DataOutputPlus out, int version) throws IOException
{
out.writeByte(aggregationSpec.kind().ordinal());
switch (aggregationSpec.kind())
{
case AGGREGATE_EVERYTHING:
break;
case AGGREGATE_BY_PK_PREFIX:
out.writeUnsignedVInt(((AggregateByPkPrefix) aggregationSpec).clusteringPrefixSize);
break;
default:
throw new AssertionError();
}
}
public AggregationSpecification deserialize(DataInputPlus in, int version, ClusteringComparator comparator) throws IOException
{
Kind kind = Kind.values()[in.readUnsignedByte()];
switch (kind)
{
case AGGREGATE_EVERYTHING:
return AggregationSpecification.AGGREGATE_EVERYTHING;
case AGGREGATE_BY_PK_PREFIX:
int clusteringPrefixSize = (int) in.readUnsignedVInt();
return AggregationSpecification.aggregatePkPrefix(comparator, clusteringPrefixSize);
default:
throw new AssertionError();
}
}
public long serializedSize(AggregationSpecification aggregationSpec, int version)
{
long size = TypeSizes.sizeof((byte) aggregationSpec.kind().ordinal());
switch (aggregationSpec.kind())
{
case AGGREGATE_EVERYTHING:
break;
case AGGREGATE_BY_PK_PREFIX:
size += TypeSizes.sizeofUnsignedVInt(((AggregateByPkPrefix) aggregationSpec).clusteringPrefixSize);
break;
default:
throw new AssertionError();
}
return size;
}
}
}