Method run

src/java/org/apache/nutch/parse/ParserJob.java:247–285

(Map<String, Object> args)

  }

  @Override
  public Map<String, Object> run(Map<String, Object> args) throws Exception {
    String batchId = (String) args.get(Nutch.ARG_BATCH);
    Boolean shouldResume = (Boolean) args.get(Nutch.ARG_RESUME);
    Boolean force = (Boolean) args.get(Nutch.ARG_FORCE);
    Boolean sitemap = (Boolean) args.get(Nutch.ARG_SITEMAP);

    if (batchId != null) {
      getConf().set(GeneratorJob.BATCH_ID, batchId);
    }
    if (shouldResume != null) {
      getConf().setBoolean(RESUME_KEY, shouldResume);
    }
    if (force != null) {
      getConf().setBoolean(FORCE_KEY, force);
    }
    if (sitemap != null) {
      getConf().setBoolean(SITEMAP_PARSE, sitemap);
    }
    LOG.info("ParserJob: resuming:\t{}", getConf().getBoolean(RESUME_KEY, false));
    LOG.info("ParserJob: forced reparse:\t {}", getConf().getBoolean(FORCE_KEY, false));
    if (batchId == null || batchId.equals(Nutch.ALL_BATCH_ID_STR)) {
      LOG.info("ParserJob: parsing all");
    } else {
      LOG.info("ParserJob: batchId:\t{}", batchId);
    }
    currentJob = NutchJob.getInstance(getConf(), "parse");

    Collection<WebPage.Field> fields = getFields(currentJob);
    MapFieldValueFilter<String, WebPage> batchIdFilter = getBatchIdFilter(batchId);
    StorageUtils.initMapperJob(currentJob, fields, String.class, WebPage.class,
        ParserMapper.class, batchIdFilter);
    StorageUtils.initReducerJob(currentJob, IdentityPageReducer.class);
    currentJob.setNumReduceTasks(0);

    currentJob.waitForCompletion(true);
    ToolUtil.recordJobStatus(null, currentJob, results);
    return results;
  }

  private MapFieldValueFilter<String, WebPage> getBatchIdFilter(String batchId) {
    if (batchId.equals(REPARSE.toString())