hadoop-gremlin/src/main/java/org/apache/tinkerpop/gremlin/hadoop/process/computer/util/MapReduceHelper.java

/*
 * 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.tinkerpop.gremlin.hadoop.process.computer.util;

import org.apache.commons.configuration.BaseConfiguration;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.hadoop.mapreduce.Job;
import org.apache.hadoop.mapreduce.Mapper;
import org.apache.hadoop.mapreduce.OutputFormat;
import org.apache.hadoop.mapreduce.Reducer;
import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;
import org.apache.hadoop.mapreduce.lib.input.SequenceFileInputFormat;
import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;
import org.apache.hadoop.mapreduce.lib.output.SequenceFileOutputFormat;
import org.apache.tinkerpop.gremlin.hadoop.Constants;
import org.apache.tinkerpop.gremlin.hadoop.process.computer.GraphFilterInputFormat;
import org.apache.tinkerpop.gremlin.hadoop.process.computer.HadoopCombine;
import org.apache.tinkerpop.gremlin.hadoop.process.computer.HadoopMap;
import org.apache.tinkerpop.gremlin.hadoop.process.computer.HadoopReduce;
import org.apache.tinkerpop.gremlin.hadoop.structure.HadoopGraph;
import org.apache.tinkerpop.gremlin.hadoop.structure.io.InputOutputHelper;
import org.apache.tinkerpop.gremlin.hadoop.structure.io.ObjectWritable;
import org.apache.tinkerpop.gremlin.hadoop.structure.io.ObjectWritableComparator;
import org.apache.tinkerpop.gremlin.hadoop.structure.io.ObjectWritableIterator;
import org.apache.tinkerpop.gremlin.hadoop.structure.util.ConfUtil;
import org.apache.tinkerpop.gremlin.process.computer.MapReduce;
import org.apache.tinkerpop.gremlin.process.computer.Memory;
import org.apache.tinkerpop.gremlin.process.computer.VertexProgram;

import java.io.IOException;
import java.util.Comparator;
import java.util.Optional;

/**
 * @author Marko A. Rodriguez (http://markorodriguez.com)
 */
public final class MapReduceHelper {

    private MapReduceHelper() {
    }

    public static void executeMapReduceJob(final MapReduce mapReduce, final Memory.Admin memory, final Configuration configuration) throws IOException, ClassNotFoundException, InterruptedException {
        final Configuration newConfiguration = new Configuration(configuration);
        final boolean vertexProgramExists = newConfiguration.get(VertexProgram.VERTEX_PROGRAM, null) != null;
        if (vertexProgramExists) {
            newConfiguration.set(Constants.GREMLIN_HADOOP_GRAPH_READER, InputOutputHelper.getInputFormat((Class) newConfiguration.getClass(Constants.GREMLIN_HADOOP_GRAPH_WRITER, OutputFormat.class)).getCanonicalName());
            newConfiguration.unset(Constants.GREMLIN_HADOOP_GRAPH_FILTER);
        }
        final BaseConfiguration apacheConfiguration = new BaseConfiguration();
        apacheConfiguration.setDelimiterParsingDisabled(true);
        mapReduce.storeState(apacheConfiguration);
        ConfUtil.mergeApacheIntoHadoopConfiguration(apacheConfiguration, newConfiguration);

        final Optional<Comparator<?>> mapSort = mapReduce.getMapKeySort();
        final Optional<Comparator<?>> reduceSort = mapReduce.getReduceKeySort();
        newConfiguration.setClass(Constants.GREMLIN_HADOOP_MAP_REDUCE_CLASS, mapReduce.getClass(), MapReduce.class);
        final Job job = Job.getInstance(newConfiguration, mapReduce.toString());
        HadoopGraph.LOGGER.info(Constants.GREMLIN_HADOOP_JOB_PREFIX + mapReduce.toString());
        job.setJarByClass(HadoopGraph.class);
        if (mapSort.isPresent())
            job.setSortComparatorClass(ObjectWritableComparator.ObjectWritableMapComparator.class);
        job.setMapperClass(HadoopMap.class);
        if (mapReduce.doStage(MapReduce.Stage.REDUCE)) {
            if (mapReduce.doStage(MapReduce.Stage.COMBINE))
                job.setCombinerClass(HadoopCombine.class);
            job.setReducerClass(HadoopReduce.class);
        } else {
            if (mapSort.isPresent()) {
                job.setReducerClass(Reducer.class);
                job.setNumReduceTasks(1); // todo: is this necessary to ensure sorted order?
            } else {
                job.setNumReduceTasks(0);
            }
        }
        job.setMapOutputKeyClass(ObjectWritable.class);
        job.setMapOutputValueClass(ObjectWritable.class);
        job.setOutputKeyClass(ObjectWritable.class);
        job.setOutputValueClass(ObjectWritable.class);
        job.setInputFormatClass(GraphFilterInputFormat.class);
        job.setOutputFormatClass(SequenceFileOutputFormat.class);
        // if there is no vertex program, then grab the graph from the input location
        final Path graphPath;
        if (vertexProgramExists) {
            graphPath = new Path(Constants.getGraphLocation(newConfiguration.get(Constants.GREMLIN_HADOOP_OUTPUT_LOCATION)));
        } else {
            graphPath = new Path(newConfiguration.get(Constants.GREMLIN_HADOOP_INPUT_LOCATION));
        }

        Path memoryPath = new Path(Constants.getMemoryLocation(newConfiguration.get(Constants.GREMLIN_HADOOP_OUTPUT_LOCATION), (reduceSort.isPresent() ? mapReduce.getMemoryKey() + "-temp" : mapReduce.getMemoryKey())));
        if (FileSystem.get(newConfiguration).exists(memoryPath)) {
            FileSystem.get(newConfiguration).delete(memoryPath, true);
        }
        FileInputFormat.setInputPaths(job, graphPath);
        FileOutputFormat.setOutputPath(job, memoryPath);
        job.waitForCompletion(true);

        // if there is a reduce sort, we need to run another identity MapReduce job
        if (reduceSort.isPresent()) {
            final Job reduceSortJob = Job.getInstance(newConfiguration, "ReduceKeySort");
            reduceSortJob.setSortComparatorClass(ObjectWritableComparator.ObjectWritableReduceComparator.class);
            reduceSortJob.setMapperClass(Mapper.class);
            reduceSortJob.setReducerClass(Reducer.class);
            reduceSortJob.setMapOutputKeyClass(ObjectWritable.class);
            reduceSortJob.setMapOutputValueClass(ObjectWritable.class);
            reduceSortJob.setOutputKeyClass(ObjectWritable.class);
            reduceSortJob.setOutputValueClass(ObjectWritable.class);
            reduceSortJob.setInputFormatClass(SequenceFileInputFormat.class);
            reduceSortJob.setOutputFormatClass(SequenceFileOutputFormat.class);
            reduceSortJob.setNumReduceTasks(1); // todo: is this necessary to ensure sorted order?
            FileInputFormat.setInputPaths(reduceSortJob, memoryPath);
            final Path sortedMemoryPath = new Path(Constants.getMemoryLocation(newConfiguration.get(Constants.GREMLIN_HADOOP_OUTPUT_LOCATION), mapReduce.getMemoryKey()));
            FileOutputFormat.setOutputPath(reduceSortJob, sortedMemoryPath);
            reduceSortJob.waitForCompletion(true);
            FileSystem.get(newConfiguration).delete(memoryPath, true); // delete the temporary memory path
            memoryPath = sortedMemoryPath;
        }
        mapReduce.addResultToMemory(memory, new ObjectWritableIterator(newConfiguration, memoryPath));
    }
}