DkproMapper.java

/*******************************************************************************
 * Copyright 2012-13
 * TU Darmstadt, UKP Lab and FG Sprachtechnologie
 * 
 * Licensed 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.dkpro.bigdata.hadoop;

import java.io.IOException;
import java.util.Random;

import org.apache.hadoop.io.Text;
import org.apache.hadoop.mapred.JobConf;
import org.apache.hadoop.mapred.Mapper;
import org.apache.hadoop.mapred.OutputCollector;
import org.apache.hadoop.mapred.Reporter;
import org.apache.uima.analysis_engine.AnalysisEngineDescription;
import org.apache.uima.analysis_engine.AnalysisEngineProcessException;
import org.apache.uima.cas.CAS;
import org.apache.uima.resource.ResourceInitializationException;
import org.apache.uima.util.ProcessTrace;
import org.apache.uima.util.ProcessTraceEvent;
import org.dkpro.bigdata.io.hadoop.CASWritable;

/**
 * A mapper for building pipelines with M/R. The engine is _NOT_ supposed to be
 * a CasConsumer, the resulting cas will be written to HDFS and can be used as
 * input to a Mapper process again.
 * 
 * @author zorn
 */
public class DkproMapper extends UIMAMapReduceBase implements
		Mapper<Text, CASWritable, Text, CASWritable> {
	public enum INPUT_FORMAT {
		CAS, TEXT, WEBARCHIVE
	}

	private final Random random;
	private String docLanguage;

	public DkproMapper() {
		super();
		this.random = new Random();
	}

	@Override
	public void map(Text key, CASWritable value,
			OutputCollector<Text, CASWritable> output, Reporter reporter)
			throws IOException {
		final CAS aCAS = value.getCAS();
		/*
		 * SAMPLING: Process and emit only a sample of the corpus
		 */
		if (samplingPropability != 100) {
            if (random.nextInt(100) >= samplingPropability) {
				reporter.incrCounter("uima", "sampling: SKIPPED", 1);
				return;
			}
        }
		reporter.incrCounter("uima", "sampling: NOT SKIPPED", 1);

		try {
			if (docLanguage != null) {
				aCAS.setDocumentLanguage(docLanguage);
			}
			// let uima process the cas
			final ProcessTrace result = this.engine.process(aCAS);
			for (final ProcessTraceEvent event : result.getEvents()) {
				reporter.incrCounter("uima", "map event " + event.getType(), 1);
			}

			final Text outkey = getOutputKey(key, aCAS);
			// update counters
			if (aCAS.getDocumentText() != null) {
                reporter.incrCounter("uima", "overall doc size",
						aCAS.getDocumentText().length());
            }
			if (this.job.getBoolean("dkpro.output.writecas", true)) {
				outValue.setCAS(aCAS);
				output.collect(outkey, outValue);
			}
		} catch (final AnalysisEngineProcessException e) {
			reporter.incrCounter("uima", e.toString(), 1);
			if (failures++ > maxFailures) {
                throw new IOException(e);
            }

		}
	}

	/**
	 * Overwrite this method to generate keys for the map-outputs With the
	 * default implementation all cases will be passed through a single reducer,
	 * which disables parallelization but has the advantage to have one single
	 * output file for the whole collection.
	 */
	protected Text getOutputKey(Text key, CAS aCAS) {
		return key;
	}

	@Override
	AnalysisEngineDescription getEngineDescription(EngineFactory factory,
			JobConf job) throws ResourceInitializationException {
		return factory.buildMapperEngine(job);
	}

	@Override
	public void configure(JobConf job) {
		super.configure(job);
		try {
			// create an output writable of the appropriate type
			outValue = (CASWritable) job.getMapOutputValueClass().newInstance();
			docLanguage = job.get("dkpro.document.language");
		} catch (Exception e) {
			throw new RuntimeException(e);
		}
	}
}