001 /**
002 * Licensed to the Apache Software Foundation (ASF) under one
003 * or more contributor license agreements. See the NOTICE file
004 * distributed with this work for additional information
005 * regarding copyright ownership. The ASF licenses this file
006 * to you under the Apache License, Version 2.0 (the
007 * "License"); you may not use this file except in compliance
008 * with the License. You may obtain a copy of the License at
009 *
010 * http://www.apache.org/licenses/LICENSE-2.0
011 *
012 * Unless required by applicable law or agreed to in writing, software
013 * distributed under the License is distributed on an "AS IS" BASIS,
014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
015 * See the License for the specific language governing permissions and
016 * limitations under the License.
017 */
018
019 package org.apache.hadoop.mapred.join;
020
021 import java.io.IOException;
022 import java.util.ArrayList;
023 import java.util.PriorityQueue;
024
025 import org.apache.hadoop.classification.InterfaceAudience;
026 import org.apache.hadoop.classification.InterfaceStability;
027 import org.apache.hadoop.io.Writable;
028 import org.apache.hadoop.io.WritableComparable;
029 import org.apache.hadoop.io.WritableComparator;
030 import org.apache.hadoop.mapred.JobConf;
031
032 /**
033 * Prefer the "rightmost" data source for this key.
034 * For example, <tt>override(S1,S2,S3)</tt> will prefer values
035 * from S3 over S2, and values from S2 over S1 for all keys
036 * emitted from all sources.
037 */
038 @InterfaceAudience.Public
039 @InterfaceStability.Stable
040 public class OverrideRecordReader<K extends WritableComparable,
041 V extends Writable>
042 extends MultiFilterRecordReader<K,V> {
043
044 OverrideRecordReader(int id, JobConf conf, int capacity,
045 Class<? extends WritableComparator> cmpcl) throws IOException {
046 super(id, conf, capacity, cmpcl);
047 }
048
049 /**
050 * Emit the value with the highest position in the tuple.
051 */
052 @SuppressWarnings("unchecked") // No static typeinfo on Tuples
053 protected V emit(TupleWritable dst) {
054 return (V) dst.iterator().next();
055 }
056
057 /**
058 * Instead of filling the JoinCollector with iterators from all
059 * data sources, fill only the rightmost for this key.
060 * This not only saves space by discarding the other sources, but
061 * it also emits the number of key-value pairs in the preferred
062 * RecordReader instead of repeating that stream n times, where
063 * n is the cardinality of the cross product of the discarded
064 * streams for the given key.
065 */
066 protected void fillJoinCollector(K iterkey) throws IOException {
067 final PriorityQueue<ComposableRecordReader<K,?>> q = getRecordReaderQueue();
068 if (!q.isEmpty()) {
069 int highpos = -1;
070 ArrayList<ComposableRecordReader<K,?>> list =
071 new ArrayList<ComposableRecordReader<K,?>>(kids.length);
072 q.peek().key(iterkey);
073 final WritableComparator cmp = getComparator();
074 while (0 == cmp.compare(q.peek().key(), iterkey)) {
075 ComposableRecordReader<K,?> t = q.poll();
076 if (-1 == highpos || list.get(highpos).id() < t.id()) {
077 highpos = list.size();
078 }
079 list.add(t);
080 if (q.isEmpty())
081 break;
082 }
083 ComposableRecordReader<K,?> t = list.remove(highpos);
084 t.accept(jc, iterkey);
085 for (ComposableRecordReader<K,?> rr : list) {
086 rr.skip(iterkey);
087 }
088 list.add(t);
089 for (ComposableRecordReader<K,?> rr : list) {
090 if (rr.hasNext()) {
091 q.add(rr);
092 }
093 }
094 }
095 }
096
097 }