org.apache.hadoop.mapreduce.JobSubmissionFiles#getJobSplitFile ( )源码实例Demo

/**
 * Generate new-api mapreduce InputFormat splits
 * @param jobContext JobContext required by InputFormat
 * @param inputSplitDir Directory in which to generate splits information
 *
 * @return InputSplitInfo containing the split files' information and the
 * location hints for each split generated to be used to determining parallelism of
 * the map stage.
 *
 * @throws IOException
 * @throws InterruptedException
 * @throws ClassNotFoundException
 */
private static InputSplitInfoDisk writeNewSplits(JobContext jobContext,
    Path inputSplitDir) throws IOException, InterruptedException,
    ClassNotFoundException {
  
  org.apache.hadoop.mapreduce.InputSplit[] splits = 
      generateNewSplits(jobContext, null, 0);
  
  Configuration conf = jobContext.getConfiguration();

  JobSplitWriter.createSplitFiles(inputSplitDir, conf,
      inputSplitDir.getFileSystem(conf), splits);

  List<TaskLocationHint> locationHints =
      new ArrayList<TaskLocationHint>(splits.length);
  for (int i = 0; i < splits.length; ++i) {
    locationHints.add(
        new TaskLocationHint(new HashSet<String>(
            Arrays.asList(splits[i].getLocations())), null));
  }

  return new InputSplitInfoDisk(
      JobSubmissionFiles.getJobSplitFile(inputSplitDir),
      JobSubmissionFiles.getJobSplitMetaFile(inputSplitDir),
      splits.length, locationHints, jobContext.getCredentials());
}
 

/**
 * Generate old-api mapred InputFormat splits
 * @param jobConf JobConf required by InputFormat class
 * @param inputSplitDir Directory in which to generate splits information
 *
 * @return InputSplitInfo containing the split files' information and the
 * number of splits generated to be used to determining parallelism of
 * the map stage.
 *
 * @throws IOException
 */
private static InputSplitInfoDisk writeOldSplits(JobConf jobConf,
    Path inputSplitDir) throws IOException {
  
  org.apache.hadoop.mapred.InputSplit[] splits = 
      generateOldSplits(jobConf, null, 0);
  
  JobSplitWriter.createSplitFiles(inputSplitDir, jobConf,
      inputSplitDir.getFileSystem(jobConf), splits);

  List<TaskLocationHint> locationHints =
      new ArrayList<TaskLocationHint>(splits.length);
  for (int i = 0; i < splits.length; ++i) {
    locationHints.add(
        new TaskLocationHint(new HashSet<String>(
            Arrays.asList(splits[i].getLocations())), null));
  }

  return new InputSplitInfoDisk(
      JobSubmissionFiles.getJobSplitFile(inputSplitDir),
      JobSubmissionFiles.getJobSplitMetaFile(inputSplitDir),
      splits.length, locationHints, jobConf.getCredentials());
}
 

/**
 * Generate old-api mapred InputFormat splits
 * @param jobConf JobConf required by InputFormat class
 * @param inputSplitDir Directory in which to generate splits information
 *
 * @return InputSplitInfo containing the split files' information and the
 * number of splits generated to be used to determining parallelism of
 * the map stage.
 *
 * @throws IOException
 */
private static InputSplitInfoDisk writeOldSplits(JobConf jobConf,
                                                 Path inputSplitDir) throws IOException {

  org.apache.hadoop.mapred.InputSplit[] splits =
      generateOldSplits(jobConf, false, true, 0);

  JobSplitWriter.createSplitFiles(inputSplitDir, jobConf,
      inputSplitDir.getFileSystem(jobConf), splits);

  List<TaskLocationHint> locationHints =
      new ArrayList<TaskLocationHint>(splits.length);
  for (int i = 0; i < splits.length; ++i) {
    locationHints.add(
        TaskLocationHint.createTaskLocationHint(new HashSet<String>(
            Arrays.asList(splits[i].getLocations())), null)
    );
  }

  return new InputSplitInfoDisk(
      JobSubmissionFiles.getJobSplitFile(inputSplitDir),
      JobSubmissionFiles.getJobSplitMetaFile(inputSplitDir),
      splits.length, locationHints, jobConf.getCredentials());
}
 

/**
 * Generate new-api mapreduce InputFormat splits
 * @param jobContext JobContext required by InputFormat
 * @param inputSplitDir Directory in which to generate splits information
 *
 * @return InputSplitInfo containing the split files' information and the
 * location hints for each split generated to be used to determining parallelism of
 * the map stage.
 *
 * @throws IOException
 * @throws InterruptedException
 * @throws ClassNotFoundException
 */
private static InputSplitInfoDisk writeNewSplits(JobContext jobContext,
                                                 Path inputSplitDir) throws IOException, InterruptedException,
    ClassNotFoundException {

  org.apache.hadoop.mapreduce.InputSplit[] splits =
      generateNewSplits(jobContext, false, true, 0);

  Configuration conf = jobContext.getConfiguration();

  JobSplitWriter.createSplitFiles(inputSplitDir, conf,
      inputSplitDir.getFileSystem(conf), splits);

  List<TaskLocationHint> locationHints =
      new ArrayList<TaskLocationHint>(splits.length);
  for (int i = 0; i < splits.length; ++i) {
    locationHints.add(
        TaskLocationHint.createTaskLocationHint(new HashSet<String>(
            Arrays.asList(splits[i].getLocations())), null)
    );
  }

  return new InputSplitInfoDisk(
      JobSubmissionFiles.getJobSplitFile(inputSplitDir),
      JobSubmissionFiles.getJobSplitMetaFile(inputSplitDir),
      splits.length, locationHints, jobContext.getCredentials());
}