Examples with CommandLine org.apache.commons.cli.CommandLine used on opensource projects

Example 36 with CommandLine

use of org.apache.commons.cli.CommandLine in project Cloud9 by lintool.

the class DocumentForwardIndexHttpServer method run.

@SuppressWarnings("static-access")
public int run(String[] args) throws Exception {
    Options options = new Options();
    options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("(required) forward index path").create(INDEX_OPTION));
    options.addOption(OptionBuilder.withArgName("path").hasArg().withDescription("(required) DocnoMapping data path").create(MAPPING_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(INDEX_OPTION) || !cmdline.hasOption(MAPPING_OPTION)) {
        HelpFormatter formatter = new HelpFormatter();
        formatter.printHelp(this.getClass().getName(), options);
        ToolRunner.printGenericCommandUsage(System.out);
        return -1;
    }
    String indexFile = cmdline.getOptionValue(INDEX_OPTION);
    String mappingFile = cmdline.getOptionValue(MAPPING_OPTION);
    LOG.info("Launching DocumentForwardIndexHttpServer");
    LOG.info(" - index file: " + indexFile);
    LOG.info(" - docno mapping data file: " + mappingFile);
    Configuration conf = getConf();
    FileSystem fs = FileSystem.get(conf);
    Random rand = new Random();
    int r = rand.nextInt();
    // This tmp file as a rendezvous point.
    Path tmpPath = new Path("/tmp/" + r);
    if (fs.exists(tmpPath)) {
        fs.delete(tmpPath, true);
    }
    Job job = new Job(conf, DocumentForwardIndexHttpServer.class.getSimpleName());
    job.setJarByClass(DocumentForwardIndexHttpServer.class);
    job.getConfiguration().set("mapred.child.java.opts", "-Xmx1024m");
    job.getConfiguration().set(INDEX_KEY, indexFile);
    job.getConfiguration().set(DOCNO_MAPPING_KEY, mappingFile);
    job.getConfiguration().set(TMP_KEY, tmpPath.toString());
    job.setNumReduceTasks(0);
    job.setInputFormatClass(NullInputFormat.class);
    job.setOutputFormatClass(NullOutputFormat.class);
    job.setMapperClass(MyMapper.class);
    job.submit();
    LOG.info("Waiting for server to start up...");
    while (!fs.exists(tmpPath)) {
        Thread.sleep(50000);
        LOG.info("...");
    }
    FSDataInputStream in = fs.open(tmpPath);
    String host = in.readUTF();
    in.close();
    LOG.info("host: " + host);
    LOG.info("port: 8888");
    return 0;
}

Example 37 with CommandLine

use of org.apache.commons.cli.CommandLine in project Cloud9 by lintool.

the class HadoopAlign method main.

@SuppressWarnings("static-access")
public static void main(String[] args) throws IOException {
    options = new Options();
    options.addOption(OptionBuilder.withDescription("path to XML-formatted parallel corpus").withArgName("path").hasArg().isRequired().create(INPUT_OPTION));
    options.addOption(OptionBuilder.withDescription("path to work/output directory on HDFS").withArgName("path").hasArg().isRequired().create(WORK_OPTION));
    options.addOption(OptionBuilder.withDescription("two-letter collection language code").withArgName("en|de|fr|zh|es|ar|tr").hasArg().isRequired().create(FLANG_OPTION));
    options.addOption(OptionBuilder.withDescription("two-letter collection language code").withArgName("en|de|fr|zh|es|ar|tr").hasArg().isRequired().create(ELANG_OPTION));
    options.addOption(OptionBuilder.withDescription("number of IBM Model 1 iterations").withArgName("positive integer").hasArg().create(MODEL1_OPTION));
    options.addOption(OptionBuilder.withDescription("number of HMM iterations").withArgName("positive integer").hasArg().create(HMM_OPTION));
    options.addOption(OptionBuilder.withDescription("truncate/stem text or not").create(TRUNCATE_OPTION));
    options.addOption(OptionBuilder.withDescription("number of reducers").withArgName("positive integer").hasArg().create(REDUCE_OPTION));
    options.addOption(OptionBuilder.withDescription("Hadoop option to load external jars").withArgName("jar packages").hasArg().create(LIBJARS_OPTION));
    CommandLine cmdline;
    CommandLineParser parser = new GnuParser();
    try {
        cmdline = parser.parse(options, args);
    } catch (ParseException exp) {
        printUsage();
        System.err.println("Error parsing command line: " + exp.getMessage());
        return;
    }
    String bitextPath = cmdline.getOptionValue(INPUT_OPTION);
    String workDir = cmdline.getOptionValue(WORK_OPTION);
    String srcLang = cmdline.getOptionValue(FLANG_OPTION);
    String trgLang = cmdline.getOptionValue(ELANG_OPTION);
    int model1Iters = cmdline.hasOption(MODEL1_OPTION) ? Integer.parseInt(cmdline.getOptionValue(MODEL1_OPTION)) : 0;
    int hmmIters = cmdline.hasOption(HMM_OPTION) ? Integer.parseInt(cmdline.getOptionValue(HMM_OPTION)) : 0;
    if (model1Iters + hmmIters == 0) {
        System.err.println("Please enter a positive number of iterations for either Model 1 or HMM");
        printUsage();
        return;
    }
    boolean isTruncate = cmdline.hasOption(TRUNCATE_OPTION) ? true : false;
    int numReducers = cmdline.hasOption(REDUCE_OPTION) ? Integer.parseInt(cmdline.getOptionValue(REDUCE_OPTION)) : 50;
    HadoopAlignConfig hac = new HadoopAlignConfig(workDir, trgLang, srcLang, bitextPath, model1Iters, hmmIters, // use null word
    true, // use variational bayes
    false, // use word truncation
    isTruncate, // alpha
    0.00f);
    hac.setHMMHomogeneous(false);
    hac.set("mapreduce.map.memory.mb", "2048");
    hac.set("mapreduce.map.java.opts", "-Xmx2048m");
    hac.set("mapreduce.reduce.memory.mb", "2048");
    hac.set("mapreduce.reduce.java.opts", "-Xmx2048m");
    hac.setHMMp0(0.2);
    hac.setMaxSentLen(15);
    doAlignment(50, numReducers, hac);
}

Example 38 with CommandLine

use of org.apache.commons.cli.CommandLine in project Cloud9 by lintool.

the class FileMerger method run.

@Override
public /**
   * TODO: add in hadoop configuration
   */
int run(String[] args) throws IOException {
    Options options = new Options();
    options.addOption(HELP_OPTION, false, "print the help message");
    options.addOption(OptionBuilder.withArgName(PATH_INDICATOR).hasArg().withDescription("input file or directory").create(INPUT_OPTION));
    options.addOption(OptionBuilder.withArgName(PATH_INDICATOR).hasArg().withDescription("output file").create(OUTPUT_OPTION));
    options.addOption(OptionBuilder.withArgName(INTEGER_INDICATOR).hasArg().withDescription("number of mappers (default to 0 and hence local merge mode, set to positive value to enable cluster merge mode)").create(MAPPER_OPTION));
    options.addOption(OptionBuilder.withArgName("property=value").hasArgs(2).withValueSeparator().withDescription("assign value for given property").create("D"));
    options.addOption(TEXT_FILE_INPUT_FORMAT, false, "input file in sequence format");
    options.addOption(DELETE_SOURCE_OPTION, false, "delete sources after merging");
    int mapperTasks = 0;
    boolean deleteSource = DELETE_SOURCE;
    boolean textFileFormat = TEXT_FILE_INPUT;
    String inputPath = "";
    String outputPath = "";
    GenericOptionsParser genericOptionsParser = new GenericOptionsParser(args);
    Configuration configuration = genericOptionsParser.getConfiguration();
    CommandLineParser parser = new GnuParser();
    HelpFormatter formatter = new HelpFormatter();
    try {
        CommandLine line = parser.parse(options, args);
        if (line.hasOption(HELP_OPTION)) {
            formatter.printHelp(FileMerger.class.getName(), options);
            System.exit(0);
        }
        if (line.hasOption(INPUT_OPTION)) {
            inputPath = line.getOptionValue(INPUT_OPTION);
        } else {
            throw new ParseException("Parsing failed due to " + INPUT_OPTION + " not initialized...");
        }
        if (line.hasOption(OUTPUT_OPTION)) {
            outputPath = line.getOptionValue(OUTPUT_OPTION);
        } else {
            throw new ParseException("Parsing failed due to " + OUTPUT_OPTION + " not initialized...");
        }
        if (line.hasOption(MAPPER_OPTION)) {
            mapperTasks = Integer.parseInt(line.getOptionValue(MAPPER_OPTION));
            if (mapperTasks <= 0) {
                sLogger.info("Warning: " + MAPPER_OPTION + " is not positive, merge in local model...");
                mapperTasks = 0;
            }
        }
        if (line.hasOption(DELETE_SOURCE_OPTION)) {
            deleteSource = true;
        }
        if (line.hasOption(TEXT_FILE_INPUT_FORMAT)) {
            textFileFormat = true;
        }
    } catch (ParseException pe) {
        System.err.println(pe.getMessage());
        formatter.printHelp(FileMerger.class.getName(), options);
        System.exit(0);
    } catch (NumberFormatException nfe) {
        System.err.println(nfe.getMessage());
        System.exit(0);
    }
    try {
        merge(configuration, inputPath, outputPath, mapperTasks, textFileFormat, deleteSource);
    } catch (InstantiationException ie) {
        ie.printStackTrace();
    } catch (IllegalAccessException iae) {
        iae.printStackTrace();
    }
    return 0;
}

Example 39 with CommandLine

use of org.apache.commons.cli.CommandLine in project databus by linkedin.

the class BootstrapConfigBase method loadConfigProperties.

@SuppressWarnings("static-access")
public static Properties loadConfigProperties(String[] args) throws IOException {
    CommandLineParser cliParser = new GnuParser();
    Option dbOption = OptionBuilder.withLongOpt(BOOTSTRAP_DB_PROPS_OPT_LONG_NAME).withDescription("Bootstrap producer properties to use").hasArg().withArgName("property_file").create(BOOTSTRAP_DB_PROP_OPT_CHAR);
    Options options = new Options();
    options.addOption(dbOption);
    CommandLine cmd = null;
    try {
        cmd = cliParser.parse(options, args);
    } catch (ParseException pe) {
        throw new RuntimeException("BootstrapConfig: failed to parse command-line options.", pe);
    }
    Properties props = null;
    if (cmd.hasOption(BOOTSTRAP_DB_PROP_OPT_CHAR)) {
        String propFile = cmd.getOptionValue(BOOTSTRAP_DB_PROP_OPT_CHAR);
        LOG.info("Loading bootstrap DB config from properties file " + propFile);
        props = new Properties();
        FileInputStream f = new FileInputStream(propFile);
        try {
            props.load(f);
        } finally {
            if (null != f)
                f.close();
        }
    } else {
        LOG.info("Using system properties for bootstrap DB config");
    }
    return props;
}

Example 40 with CommandLine

use of org.apache.commons.cli.CommandLine in project databus by linkedin.

the class BootstrapAvroFileSeederMain method parseArgs.

@SuppressWarnings("static-access")
public static void parseArgs(String[] args) throws IOException {
    CommandLineParser cliParser = new GnuParser();
    Option helpOption = OptionBuilder.withLongOpt(HELP_OPT_LONG_NAME).withDescription("Help screen").create(HELP_OPT_CHAR);
    Option sourcesOption = OptionBuilder.withLongOpt(PHYSICAL_CONFIG_OPT_LONG_NAME).withDescription("Bootstrap producer properties to use").hasArg().withArgName("property_file").create(PHYSICAL_CONFIG_OPT_CHAR);
    Option dbOption = OptionBuilder.withLongOpt(BOOTSTRAP_DB_PROPS_OPT_LONG_NAME).withDescription("Bootstrap producer properties to use").hasArg().withArgName("property_file").create(BOOTSTRAP_DB_PROP_OPT_CHAR);
    Option log4jPropsOption = OptionBuilder.withLongOpt(LOG4J_PROPS_OPT_LONG_NAME).withDescription("Log4j properties to use").hasArg().withArgName("property_file").create(LOG4J_PROPS_OPT_CHAR);
    Options options = new Options();
    options.addOption(helpOption);
    options.addOption(sourcesOption);
    options.addOption(dbOption);
    options.addOption(log4jPropsOption);
    CommandLine cmd = null;
    try {
        cmd = cliParser.parse(options, args);
    } catch (ParseException pe) {
        LOG.fatal("Bootstrap Physical Config: failed to parse command-line options.", pe);
        throw new RuntimeException("Bootstrap Physical Config: failed to parse command-line options.", pe);
    }
    if (cmd.hasOption(LOG4J_PROPS_OPT_CHAR)) {
        String log4jPropFile = cmd.getOptionValue(LOG4J_PROPS_OPT_CHAR);
        PropertyConfigurator.configure(log4jPropFile);
        LOG.info("Using custom logging settings from file " + log4jPropFile);
    } else {
        PatternLayout defaultLayout = new PatternLayout("%d{ISO8601} +%r [%t] (%p) {%c} %m%n");
        ConsoleAppender defaultAppender = new ConsoleAppender(defaultLayout);
        Logger.getRootLogger().removeAllAppenders();
        Logger.getRootLogger().addAppender(defaultAppender);
        //using info as the default log level
        Logger.getRootLogger().setLevel(Level.INFO);
        LOG.info("Using default logging settings. Log Level is :" + Logger.getRootLogger().getLevel());
    }
    if (cmd.hasOption(HELP_OPT_CHAR)) {
        printCliHelp(options);
        System.exit(0);
    }
    if (!cmd.hasOption(PHYSICAL_CONFIG_OPT_CHAR))
        throw new RuntimeException("Sources Config is not provided; use --help for usage");
    if (!cmd.hasOption(BOOTSTRAP_DB_PROP_OPT_CHAR))
        throw new RuntimeException("Bootstrap config is not provided; use --help for usage");
    _sSourcesConfigFile = cmd.getOptionValue(PHYSICAL_CONFIG_OPT_CHAR);
    String propFile = cmd.getOptionValue(BOOTSTRAP_DB_PROP_OPT_CHAR);
    LOG.info("Loading bootstrap DB config from properties file " + propFile);
    _sBootstrapConfigProps = new Properties();
    FileInputStream fis = new FileInputStream(propFile);
    try {
        _sBootstrapConfigProps.load(fis);
    } finally {
        fis.close();
    }
}