LinkRank.java 文件源码

/**
 * Runs the link analysis job. The link analysis job applies the link rank
 * formula to create a score per url and stores that score in the NodeDb.
 * 
 * Typically the link analysis job is run a number of times to allow the link
 * rank scores to converge.
 * 
 * @param nodeDb
 *          The node database from which we are getting previous link rank
 *          scores.
 * @param inverted
 *          The inverted inlinks
 * @param output
 *          The link analysis output.
 * @param iteration
 *          The current iteration number.
 * @param numIterations
 *          The total number of link analysis iterations
 * 
 * @throws IOException
 *           If an error occurs during link analysis.
 */
private void runAnalysis(Path nodeDb, Path inverted, Path output,
    int iteration, int numIterations, float rankOne) throws IOException {

  JobConf analyzer = new NutchJob(getConf());
  analyzer.set("link.analyze.iteration", String.valueOf(iteration + 1));
  analyzer.setJobName("LinkAnalysis Analyzer, iteration " + (iteration + 1)
      + " of " + numIterations);
  FileInputFormat.addInputPath(analyzer, nodeDb);
  FileInputFormat.addInputPath(analyzer, inverted);
  FileOutputFormat.setOutputPath(analyzer, output);
  analyzer.set("link.analyze.rank.one", String.valueOf(rankOne));
  analyzer.setMapOutputKeyClass(Text.class);
  analyzer.setMapOutputValueClass(ObjectWritable.class);
  analyzer.setInputFormat(SequenceFileInputFormat.class);
  analyzer.setMapperClass(Analyzer.class);
  analyzer.setReducerClass(Analyzer.class);
  analyzer.setOutputKeyClass(Text.class);
  analyzer.setOutputValueClass(Node.class);
  analyzer.setOutputFormat(MapFileOutputFormat.class);
  analyzer.setBoolean("mapreduce.fileoutputcommitter.marksuccessfuljobs",
      false);

  LOG.info("Starting analysis job");
  try {
    JobClient.runJob(analyzer);
  } catch (IOException e) {
    LOG.error(StringUtils.stringifyException(e));
    throw e;
  }
  LOG.info("Finished analysis job.");
}