Example usage for org.apache.hadoop.mapred JobConf set

List of usage examples for org.apache.hadoop.mapred JobConf set

  1. HOME
  2. Java
  3. org.apache.hadoop
  4. org.apache.hadoop.mapred.*
  5. JobConf
  6. set

Introduction

In this page you can find the example usage for org.apache.hadoop.mapred JobConf set.

Prototype

public void set(String name, String value)

Source Link

Document

Set the value of the name property.

Usage

From source file:edu.ucsb.cs.lsh.projection.ProjectionsGenerator.java

License:Apache License

public static void main(JobConf job) throws IOException {
    int nBits/*D*/, nFeatures/*K*/, nReducers;
    job.setJobName(ProjectionsGenerator.class.getSimpleName());
    FileSystem fs = FileSystem.get(job);

    nBits = job.getInt(ProjectionLshDriver.LSH_NBITS_PROPERTY, ProjectionLshDriver.LSH_NBITS_VALUE);
    nFeatures = readCollectionFeatureCount(fs, job);
    setParameters(nBits, nFeatures);//w  w w. jav a  2  s.  c  om
    nReducers = job.getInt(ProjectionLshDriver.LSH_NREDUCER_PROPERTY, ProjectionLshDriver.LSH_NREDUCER_VALUE);
    Path inputPath = new Path(INPUT_DIR);
    Path outputPath = new Path(OUTPUT_DIR);
    if (fs.exists(outputPath))
        fs.delete(outputPath, true);
    if (fs.exists(inputPath))
        fs.delete(inputPath, true);

    SequenceFile.Writer writer = SequenceFile.createWriter(fs, job, new Path(inputPath.toString() + "/file"),
            IntWritable.class, IntWritable.class);
    for (int i = 0; i < nReducers; i++)
        writer.append(new IntWritable(i), new IntWritable(i));
    writer.close();

    job.setInputFormat(SequenceFileInputFormat.class);
    job.setOutputFormat(SequenceFileOutputFormat.class);
    SequenceFileInputFormat.setInputPaths(job, inputPath);
    FileOutputFormat.setOutputPath(job, outputPath);
    FileOutputFormat.setCompressOutput(job, false);

    job.set("mapred.child.java.opts", "-Xmx2048m");
    job.setInt("mapred.map.max.attempts", 10);
    job.setInt("mapred.reduce.max.attempts", 10);

    job.setNumMapTasks(1);
    job.setNumReduceTasks(nReducers);

    job.setMapperClass(IdentityMapper.class);
    job.setMapOutputKeyClass(IntWritable.class);
    job.setMapOutputValueClass(IntWritable.class);

    job.setReducerClass(ProjectionReducer.class);
    job.setOutputKeyClass(IntWritable.class);
    job.setOutputValueClass(RandomVector.class);

    JobSubmitter.run(job, "LSH", job.getFloat(Config.THRESHOLD_PROPERTY, Config.THRESHOLD_VALUE));
}

From source file:edu.ucsb.cs.lsh.projection.SignaturesGenerator.java

License:Apache License

public static void main(String[] args) throws Exception {
    JobConf job = new JobConf(SignaturesGenerator.class);
    new GenericOptionsParser(job, args);
    job.setJobName(SignaturesGenerator.class.getSimpleName());
    int nBits = job.getInt(ProjectionLshDriver.LSH_NBITS_PROPERTY, ProjectionLshDriver.LSH_NBITS_VALUE);
    setParameters();//w  w  w  .ja  v a  2  s . c  o  m
    FileSystem fs = FileSystem.get(job);
    prepareDistributedCache(job, fs, new Path(ProjectionsGenerator.OUTPUT_DIR));
    Path outputPath = new Path(OUTPUT_DIR);
    if (fs.exists(outputPath))
        fs.delete(outputPath);

    FileInputFormat.setInputPaths(job, INPUT_DIR);
    // Path(INPUT_DIR));
    FileOutputFormat.setOutputPath(job, outputPath);
    // FileOutputFormat.setCompressOutput(job, false);
    job.setInputFormat(SequenceFileInputFormat.class);
    job.setOutputFormat(SequenceFileOutputFormat.class);

    job.set("mapred.child.java.opts", "-Xmx2048m");
    job.setInt("mapred.map.max.attempts", 10);
    job.setInt("mapred.reduce.max.attempts", 10);
    job.setInt("mapred.task.timeout", 6000000);

    job.setMapperClass(SigMapper.class);
    job.setMapOutputKeyClass(LongWritable.class);
    job.setMapOutputValueClass(BitSignature.class);
    job.setNumReduceTasks(0);
    job.setOutputKeyClass(LongWritable.class);
    job.setOutputValueClass(BitSignature.class);

    JobSubmitter.run(job, "LSH", -1);
}

From source file:edu.ucsb.cs.partitioning.cosine.Partitioner.java

License:Apache License

/**
 * Uniformly partitions the sequence vectors given the number of partitions
 * input in the configuration file. It also prepares partitions information
 * about its partitions: maximum p-norms, weights or norm/weights/lengths in
 * a file to guide the core static partitioning next for skipping.
 * /*from   ww w.ja  v  a 2  s .  c  o  m*/
 * @param norm_weight_all
 * @return
 */
public static JobConf main(String[] args, int norm_weight_all) throws IOException {

    JobConf job = new JobConf();
    new GenericOptionsParser(job, args);
    job.setJarByClass(Partitioner.class);
    System.out.println(
            JobSubmitter.stars() + "\n Running partitioner to prepare uniform partitionins (Single JVM) ");

    String inputDir = SortDriver.OUTPUT_DIR, maxDir;

    if (norm_weight_all == 1)
        maxDir = "/maxpnorm";
    //      maxDir = inputDir + "/maxpnorm";
    else if (norm_weight_all == 2)
        maxDir = "/maxweight";
    //      maxDir = inputDir + "/maxweight";
    else
        maxDir = "/maxall";
    //      maxDir = inputDir + "/maxall";

    if (!(new Path(inputDir).getFileSystem(job)).exists(new Path(inputDir)))
        throw new UnsupportedOperationException("ERROR: " + inputDir + " directory not set.");

    job.set(MAX_DIR_PATH, maxDir);
    job.set(Config.NUM_PARTITIONS_PROPERTY, Integer.toString(produceStaticParitions(job, inputDir, OUTPUT_DIR,
            maxDir, job.getInt(Config.NUM_PARTITIONS_PROPERTY, Config.NUM_PARTITIONS_VALUE), norm_weight_all)));
    return job;
}

From source file:edu.umd.cloud9.collection.clue.RepackClueWarcRecords.java

License:Apache License

/**
 * Runs this tool.// w  w w .j ava 2 s  .com
 */
public int run(String[] args) throws Exception {
    if (args.length != 5) {
        printUsage();
        return -1;
    }

    String basePath = args[0];
    String outputPath = args[1];
    int segment = Integer.parseInt(args[2]);
    String data = args[3];
    String compressionType = args[4];

    if (!compressionType.equals("block") && !compressionType.equals("record")
            && !compressionType.equals("none")) {
        System.err.println("Error: \"" + compressionType + "\" unknown compression type!");
        System.exit(-1);
    }

    // Default block size.
    int blocksize = 1000000;

    JobConf conf = new JobConf(RepackClueWarcRecords.class);
    conf.setJobName("RepackClueWarcRecords:segment" + segment);

    conf.set("DocnoMappingDataFile", data);

    LOG.info("Tool name: RepackClueWarcRecords");
    LOG.info(" - base path: " + basePath);
    LOG.info(" - output path: " + outputPath);
    LOG.info(" - segment number: " + segment);
    LOG.info(" - docno mapping data file: " + data);
    LOG.info(" - compression type: " + compressionType);

    if (compressionType.equals("block")) {
        LOG.info(" - block size: " + blocksize);
    }

    int mapTasks = 10;

    conf.setNumMapTasks(mapTasks);
    conf.setNumReduceTasks(0);

    ClueCollectionPathConstants.addEnglishCollectionPart(conf, basePath, segment);

    SequenceFileOutputFormat.setOutputPath(conf, new Path(outputPath));

    if (compressionType.equals("none")) {
        SequenceFileOutputFormat.setCompressOutput(conf, false);
    } else {
        SequenceFileOutputFormat.setCompressOutput(conf, true);

        if (compressionType.equals("record")) {
            SequenceFileOutputFormat.setOutputCompressionType(conf, SequenceFile.CompressionType.RECORD);
        } else {
            SequenceFileOutputFormat.setOutputCompressionType(conf, SequenceFile.CompressionType.BLOCK);
            conf.setInt("io.seqfile.compress.blocksize", blocksize);
        }
    }

    conf.setInputFormat(ClueWarcInputFormat.class);
    conf.setOutputFormat(SequenceFileOutputFormat.class);
    conf.setOutputKeyClass(IntWritable.class);
    conf.setOutputValueClass(ClueWarcRecord.class);

    conf.setMapperClass(MyMapper.class);

    // Delete the output directory if it exists already.
    FileSystem.get(conf).delete(new Path(outputPath), true);

    JobClient.runJob(conf);

    return 0;
}

From source file:edu.umd.cloud9.collection.medline.DemoCountMedlineCitations.java

License:Apache License

/**
 * Runs this tool.//from w w  w .j a va2 s .  c o  m
 */
public int run(String[] args) throws Exception {
    if (args.length != 3) {
        printUsage();
        return -1;
    }

    String inputPath = args[0];
    String outputPath = args[1];
    String mappingFile = args[2];

    sLogger.info("input: " + inputPath);
    sLogger.info("output dir: " + outputPath);
    sLogger.info("docno mapping file: " + mappingFile);

    JobConf conf = new JobConf(DemoCountMedlineCitations.class);
    conf.setJobName("DemoCountMedlineCitations");

    conf.setNumReduceTasks(0);

    // pass in the class name as a String; this is makes the mapper general
    // in being able to load any collection of Indexable objects that has
    // docid/docno mapping specified by a DocnoMapping object
    conf.set("DocnoMappingClass", "edu.umd.cloud9.collection.medline.MedlineDocnoMapping");

    // put the mapping file in the distributed cache so each map worker will
    // have it
    DistributedCache.addCacheFile(new URI(mappingFile), conf);

    FileInputFormat.setInputPaths(conf, new Path(inputPath));
    FileOutputFormat.setOutputPath(conf, new Path(outputPath));
    FileOutputFormat.setCompressOutput(conf, false);

    conf.setInputFormat(MedlineCitationInputFormat.class);
    conf.setOutputKeyClass(Text.class);
    conf.setOutputValueClass(IntWritable.class);

    conf.setMapperClass(MyMapper.class);

    // delete the output directory if it exists already
    FileSystem.get(conf).delete(new Path(outputPath), true);

    JobClient.runJob(conf);

    return 0;
}

From source file:edu.umd.cloud9.collection.trec.BuildTrecForwardIndex.java

License:Apache License

/**
 * Runs this tool.//from  w w w .ja va 2  s.c  o  m
 */
public int run(String[] args) throws Exception {
    if (args.length != 4) {
        printUsage();
        return -1;
    }

    JobConf conf = new JobConf(getConf(), BuildTrecForwardIndex.class);
    FileSystem fs = FileSystem.get(getConf());

    String collectionPath = args[0];
    String outputPath = args[1];
    String indexFile = args[2];
    String mappingFile = args[3];

    LOG.info("Tool name: " + BuildTrecForwardIndex.class.getCanonicalName());
    LOG.info(" - collection path: " + collectionPath);
    LOG.info(" - output path: " + outputPath);
    LOG.info(" - index file: " + indexFile);
    LOG.info(" - mapping file: " + mappingFile);

    conf.setJobName(BuildTrecForwardIndex.class.getSimpleName());

    conf.set("mapred.child.java.opts", "-Xmx1024m");
    conf.setNumReduceTasks(1);

    if (conf.get("mapred.job.tracker").equals("local")) {
        conf.set(DOCNO_MAPPING_FILE_PROPERTY, mappingFile);
    } else {
        DistributedCache.addCacheFile(new URI(mappingFile), conf);
    }

    FileInputFormat.setInputPaths(conf, new Path(collectionPath));
    FileOutputFormat.setOutputPath(conf, new Path(outputPath));
    FileOutputFormat.setCompressOutput(conf, false);

    conf.setInputFormat(TrecDocumentInputFormat.class);
    conf.setOutputKeyClass(IntWritable.class);
    conf.setOutputValueClass(Text.class);

    conf.setMapperClass(MyMapper.class);
    conf.setReducerClass(IdentityReducer.class);

    // delete the output directory if it exists already
    FileSystem.get(conf).delete(new Path(outputPath), true);

    RunningJob job = JobClient.runJob(conf);

    Counters counters = job.getCounters();
    int numDocs = (int) counters.findCounter(Count.DOCS).getCounter();

    String inputFile = outputPath + "/" + "part-00000";

    LOG.info("Writing " + numDocs + " doc offseta to " + indexFile);
    FSLineReader reader = new FSLineReader(inputFile, fs);

    FSDataOutputStream writer = fs.create(new Path(indexFile), true);

    writer.writeUTF(edu.umd.cloud9.collection.trec.TrecForwardIndex.class.getCanonicalName());
    writer.writeUTF(collectionPath);
    writer.writeInt(numDocs);

    int cnt = 0;
    Text line = new Text();
    while (reader.readLine(line) > 0) {
        String[] arr = line.toString().split("\\t");
        long offset = Long.parseLong(arr[1]);
        int len = Integer.parseInt(arr[2]);

        writer.writeLong(offset);
        writer.writeInt(len);

        cnt++;
        if (cnt % 100000 == 0) {
            LOG.info(cnt + " docs");
        }
    }
    reader.close();
    writer.close();
    LOG.info(cnt + " docs total. Done!");

    if (numDocs != cnt) {
        throw new RuntimeException("Unexpected number of documents in building forward index!");
    }

    return 0;
}

From source file:edu.umd.cloud9.collection.trec.DemoCountTrecDocuments.java

License:Apache License

/**
 * Runs this tool./*from w ww.j  a va 2  s  .  c  om*/
 */
public int run(String[] args) throws Exception {
    if (args.length != 3) {
        printUsage();
        return -1;
    }

    String inputPath = args[0];
    String outputPath = args[1];
    String mappingFile = args[2];

    LOG.info("Tool: " + DemoCountTrecDocuments.class.getCanonicalName());
    LOG.info(" - input: " + inputPath);
    LOG.info(" - output dir: " + outputPath);
    LOG.info(" - docno mapping file: " + mappingFile);

    JobConf conf = new JobConf(getConf(), DemoCountTrecDocuments.class);
    conf.setJobName(DemoCountTrecDocuments.class.getSimpleName());

    conf.setNumReduceTasks(0);

    // Pass in the class name as a String; this is makes the mapper general in being able to load
    // any collection of Indexable objects that has docid/docno mapping specified by a DocnoMapping
    // object.
    conf.set("DocnoMappingClass", edu.umd.cloud9.collection.trec.TrecDocnoMapping.class.getCanonicalName());

    // Put the mapping file in the distributed cache so each map worker will have it.
    DistributedCache.addCacheFile(new URI(mappingFile), conf);

    FileInputFormat.setInputPaths(conf, new Path(inputPath));
    FileOutputFormat.setOutputPath(conf, new Path(outputPath));
    FileOutputFormat.setCompressOutput(conf, false);

    conf.setInputFormat(TrecDocumentInputFormat.class);
    conf.setOutputKeyClass(Text.class);
    conf.setOutputValueClass(IntWritable.class);

    conf.setMapperClass(MyMapper.class);

    // Delete the output directory if it exists already.
    FileSystem.get(conf).delete(new Path(outputPath), true);

    JobClient.runJob(conf);

    return 0;
}

From source file:edu.umd.cloud9.collection.wikipedia.BuildWikipediaDocnoMapping.java

License:Apache License

@SuppressWarnings("static-access")
@Override//from   w w  w  . j av a2 s  .c o  m
public int run(String[] args) throws Exception {
    Options options = new Options();
    options.addOption(
            OptionBuilder.withArgName("path").hasArg().withDescription("XML dump file").create(INPUT_OPTION));
    options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("tmp output directory")
            .create(OUTPUT_PATH_OPTION));
    options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("output file")
            .create(OUTPUT_FILE_OPTION));
    options.addOption(OptionBuilder.withArgName("en|sv|de|cs|es|zh|ar|tr").hasArg()
            .withDescription("two-letter language code").create(LANGUAGE_OPTION));
    options.addOption(KEEP_ALL_OPTION, false, "keep all pages");

    CommandLine cmdline;
    CommandLineParser parser = new GnuParser();
    try {
        cmdline = parser.parse(options, args);
    } catch (ParseException exp) {
        System.err.println("Error parsing command line: " + exp.getMessage());
        return -1;
    }

    if (!cmdline.hasOption(INPUT_OPTION) || !cmdline.hasOption(OUTPUT_PATH_OPTION)
            || !cmdline.hasOption(OUTPUT_FILE_OPTION)) {
        HelpFormatter formatter = new HelpFormatter();
        formatter.printHelp(this.getClass().getName(), options);
        ToolRunner.printGenericCommandUsage(System.out);
        return -1;
    }

    String language = null;
    if (cmdline.hasOption(LANGUAGE_OPTION)) {
        language = cmdline.getOptionValue(LANGUAGE_OPTION);
        if (language.length() != 2) {
            System.err.println("Error: \"" + language + "\" unknown language!");
            return -1;
        }
    }

    String inputPath = cmdline.getOptionValue(INPUT_OPTION);
    String outputPath = cmdline.getOptionValue(OUTPUT_PATH_OPTION);
    String outputFile = cmdline.getOptionValue(OUTPUT_FILE_OPTION);
    boolean keepAll = cmdline.hasOption(KEEP_ALL_OPTION);

    LOG.info("Tool name: " + this.getClass().getName());
    LOG.info(" - input: " + inputPath);
    LOG.info(" - output path: " + outputPath);
    LOG.info(" - output file: " + outputFile);
    LOG.info(" - keep all pages: " + keepAll);
    LOG.info(" - language: " + language);

    JobConf conf = new JobConf(getConf(), BuildWikipediaDocnoMapping.class);
    conf.setJobName(String.format("BuildWikipediaDocnoMapping[%s: %s, %s: %s, %s: %s]", INPUT_OPTION, inputPath,
            OUTPUT_FILE_OPTION, outputFile, LANGUAGE_OPTION, language));

    conf.setBoolean(KEEP_ALL_OPTION, keepAll);
    if (language != null) {
        conf.set("wiki.language", language);
    }
    conf.setNumReduceTasks(1);

    FileInputFormat.setInputPaths(conf, new Path(inputPath));
    FileOutputFormat.setOutputPath(conf, new Path(outputPath));
    FileOutputFormat.setCompressOutput(conf, false);

    conf.setInputFormat(WikipediaPageInputFormat.class);
    conf.setOutputKeyClass(IntWritable.class);
    conf.setOutputValueClass(IntWritable.class);
    conf.setOutputFormat(TextOutputFormat.class);

    conf.setMapperClass(MyMapper.class);
    conf.setReducerClass(MyReducer.class);

    // Delete the output directory if it exists already.
    FileSystem.get(conf).delete(new Path(outputPath), true);

    RunningJob job = JobClient.runJob(conf);
    Counters c = job.getCounters();
    long cnt = keepAll ? c.getCounter(PageTypes.TOTAL) : c.getCounter(PageTypes.ARTICLE);

    WikipediaDocnoMapping.writeDocnoMappingData(FileSystem.get(conf), outputPath + "/part-00000", (int) cnt,
            outputFile);

    return 0;
}

From source file:edu.umd.cloud9.collection.wikipedia.BuildWikipediaForwardIndex.java

License:Apache License

@SuppressWarnings("static-access")
@Override//w w w .  java 2 s . c o  m
public int run(String[] args) throws Exception {
    Options options = new Options();
    options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("input").create(INPUT_OPTION));
    options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("tmp output directory")
            .create(OUTPUT_OPTION));
    options.addOption(
            OptionBuilder.withArgName("path").hasArg().withDescription("index file").create(INDEX_FILE_OPTION));
    options.addOption(OptionBuilder.withArgName("en|sv|de|cs|es|zh|ar|tr").hasArg()
            .withDescription("two-letter language code").create(LANGUAGE_OPTION));

    CommandLine cmdline;
    CommandLineParser parser = new GnuParser();
    try {
        cmdline = parser.parse(options, args);
    } catch (ParseException exp) {
        System.err.println("Error parsing command line: " + exp.getMessage());
        return -1;
    }

    if (!cmdline.hasOption(INPUT_OPTION) || !cmdline.hasOption(OUTPUT_OPTION)
            || !cmdline.hasOption(INDEX_FILE_OPTION)) {
        HelpFormatter formatter = new HelpFormatter();
        formatter.printHelp(this.getClass().getName(), options);
        ToolRunner.printGenericCommandUsage(System.out);
        return -1;
    }

    Path inputPath = new Path(cmdline.getOptionValue(INPUT_OPTION));
    String outputPath = cmdline.getOptionValue(OUTPUT_OPTION);
    String indexFile = cmdline.getOptionValue(INDEX_FILE_OPTION);

    if (!inputPath.isAbsolute()) {
        System.err.println("Error: " + INPUT_OPTION + " must be an absolute path!");
        return -1;
    }

    String language = null;
    if (cmdline.hasOption(LANGUAGE_OPTION)) {
        language = cmdline.getOptionValue(LANGUAGE_OPTION);
        if (language.length() != 2) {
            System.err.println("Error: \"" + language + "\" unknown language!");
            return -1;
        }
    }

    JobConf conf = new JobConf(getConf(), BuildWikipediaForwardIndex.class);
    FileSystem fs = FileSystem.get(conf);

    LOG.info("Tool name: " + this.getClass().getName());
    LOG.info(" - input path: " + inputPath);
    LOG.info(" - output path: " + outputPath);
    LOG.info(" - index file: " + indexFile);
    LOG.info("Note: This tool only works on block-compressed SequenceFiles!");
    LOG.info(" - language: " + language);

    conf.setJobName(String.format("BuildWikipediaForwardIndex[%s: %s, %s: %s, %s: %s]", INPUT_OPTION, inputPath,
            INDEX_FILE_OPTION, indexFile, LANGUAGE_OPTION, language));

    conf.setNumReduceTasks(1);

    FileInputFormat.setInputPaths(conf, inputPath);
    FileOutputFormat.setOutputPath(conf, new Path(outputPath));
    FileOutputFormat.setCompressOutput(conf, false);

    if (language != null) {
        conf.set("wiki.language", language);
    }

    conf.setInputFormat(NoSplitSequenceFileInputFormat.class);
    conf.setOutputKeyClass(IntWritable.class);
    conf.setOutputValueClass(Text.class);

    conf.setMapRunnerClass(MyMapRunner.class);
    conf.setReducerClass(IdentityReducer.class);

    // delete the output directory if it exists already
    fs.delete(new Path(outputPath), true);

    RunningJob job = JobClient.runJob(conf);

    Counters counters = job.getCounters();
    int blocks = (int) counters.getCounter(Blocks.Total);

    LOG.info("number of blocks: " + blocks);

    LOG.info("Writing index file...");
    LineReader reader = new LineReader(fs.open(new Path(outputPath + "/part-00000")));
    FSDataOutputStream out = fs.create(new Path(indexFile), true);

    out.writeUTF("edu.umd.cloud9.collection.wikipedia.WikipediaForwardIndex");
    out.writeUTF(inputPath.toString());
    out.writeInt(blocks);

    int cnt = 0;
    Text line = new Text();
    while (reader.readLine(line) > 0) {
        String[] arr = line.toString().split("\\s+");

        int docno = Integer.parseInt(arr[0]);
        int offset = Integer.parseInt(arr[1]);
        short fileno = Short.parseShort(arr[2]);

        out.writeInt(docno);
        out.writeInt(offset);
        out.writeShort(fileno);

        cnt++;

        if (cnt % 100000 == 0) {
            LOG.info(cnt + " blocks written");
        }
    }

    reader.close();
    out.close();

    if (cnt != blocks) {
        throw new RuntimeException("Error: mismatch in block count!");
    }

    return 0;
}

From source file:edu.umd.cloud9.collection.wikipedia.WikipediaForwardIndexBuilder.java

License:Apache License

@SuppressWarnings("static-access")
@Override/*from   www .  j  a v  a 2s .c  om*/
public int run(String[] args) throws Exception {
    Options options = new Options();
    options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("input").create(INPUT_OPTION));
    options.addOption(
            OptionBuilder.withArgName("path").hasArg().withDescription("index file").create(INDEX_FILE_OPTION));
    options.addOption(OptionBuilder
            .withArgName("en|sv|nl|de|fr|ru|it|es|vi|pl|ja|pt|zh|uk|ca|fa|no|fi|id|ar|sr|ko|hi|zh_yue|cs|tr")
            .hasArg().withDescription("two-letter or six-letter language code").create(LANGUAGE_OPTION));

    CommandLine cmdline;
    CommandLineParser parser = new GnuParser();
    try {
        cmdline = parser.parse(options, args);
    } catch (ParseException exp) {
        System.err.println("Error parsing command line: " + exp.getMessage());
        return -1;
    }

    if (!cmdline.hasOption(INPUT_OPTION) || !cmdline.hasOption(INDEX_FILE_OPTION)) {
        HelpFormatter formatter = new HelpFormatter();
        formatter.printHelp(this.getClass().getName(), options);
        ToolRunner.printGenericCommandUsage(System.out);
        return -1;
    }

    Path inputPath = new Path(cmdline.getOptionValue(INPUT_OPTION));
    String indexFile = cmdline.getOptionValue(INDEX_FILE_OPTION);

    String tmpPath = "tmp-" + WikipediaForwardIndexBuilder.class.getSimpleName() + "-" + RANDOM.nextInt(10000);

    if (!inputPath.isAbsolute()) {
        System.err.println("Error: " + INPUT_OPTION + " must be an absolute path!");
        return -1;
    }

    String language = null;
    if (cmdline.hasOption(LANGUAGE_OPTION)) {
        language = cmdline.getOptionValue(LANGUAGE_OPTION);
        if (!(language.length() == 2 || language.length() == 6)) {
            System.err.println("Error: \"" + language + "\" unknown language!");
            return -1;
        }
    }

    JobConf conf = new JobConf(getConf(), WikipediaForwardIndexBuilder.class);
    FileSystem fs = FileSystem.get(conf);

    LOG.info("Tool name: " + this.getClass().getName());
    LOG.info(" - input path: " + inputPath);
    LOG.info(" - index file: " + indexFile);
    LOG.info(" - language: " + language);
    LOG.info("Note: This tool only works on block-compressed SequenceFiles!");

    conf.setJobName(String.format("BuildWikipediaForwardIndex[%s: %s, %s: %s, %s: %s]", INPUT_OPTION, inputPath,
            INDEX_FILE_OPTION, indexFile, LANGUAGE_OPTION, language));

    conf.setNumReduceTasks(1);

    FileInputFormat.setInputPaths(conf, inputPath);
    FileOutputFormat.setOutputPath(conf, new Path(tmpPath));
    FileOutputFormat.setCompressOutput(conf, false);

    if (language != null) {
        conf.set("wiki.language", language);
    }

    conf.setInputFormat(NoSplitSequenceFileInputFormat.class);
    conf.setOutputKeyClass(IntWritable.class);
    conf.setOutputValueClass(Text.class);

    conf.setMapRunnerClass(MyMapRunner.class);
    conf.setReducerClass(IdentityReducer.class);

    // Delete the output directory if it exists already.
    fs.delete(new Path(tmpPath), true);

    RunningJob job = JobClient.runJob(conf);

    Counters counters = job.getCounters();
    int blocks = (int) counters.getCounter(Blocks.Total);

    LOG.info("number of blocks: " + blocks);

    LOG.info("Writing index file...");
    LineReader reader = new LineReader(fs.open(new Path(tmpPath + "/part-00000")));
    FSDataOutputStream out = fs.create(new Path(indexFile), true);

    out.writeUTF(edu.umd.cloud9.collection.wikipedia.WikipediaForwardIndex.class.getCanonicalName());
    out.writeUTF(inputPath.toString());
    out.writeInt(blocks);

    int cnt = 0;
    Text line = new Text();
    while (reader.readLine(line) > 0) {
        String[] arr = line.toString().split("\\s+");

        int docno = Integer.parseInt(arr[0]);
        int offset = Integer.parseInt(arr[1]);
        short fileno = Short.parseShort(arr[2]);

        out.writeInt(docno);
        out.writeInt(offset);
        out.writeShort(fileno);

        cnt++;

        if (cnt % 100000 == 0) {
            LOG.info(cnt + " blocks written");
        }
    }

    reader.close();
    out.close();

    if (cnt != blocks) {
        throw new RuntimeException("Error: mismatch in block count!");
    }

    // Clean up.
    fs.delete(new Path(tmpPath), true);

    return 0;
}