hadoop1.0.4 map数设置

/** 
 * A base class for file-based {@link InputFormat}s.
 * 
 * <p><code>FileInputFormat</code> is the base class for all file-based 
 * <code>InputFormat</code>s. This provides a generic implementation of
 * {@link #getSplits(JobContext)}.
 * Subclasses of <code>FileInputFormat</code> can also override the 
 * {@link #isSplitable(JobContext, Path)} method to ensure input-files are
 * not split-up and are processed as a whole by {@link Mapper}s.
 */
public abstract class FileInputFormat<K, V> extends InputFormat<K, V> {


  public static enum Counter { 
    BYTES_READ
  }
  
  private static final Log LOG = LogFactory.getLog(FileInputFormat.class);


  private static final double SPLIT_SLOP = 1.1;   // 10% slop


  private static final PathFilter hiddenFileFilter = new PathFilter(){
      public boolean accept(Path p){
        String name = p.getName(); 
        return !name.startsWith("_") && !name.startsWith("."); 
      }
    }; 


  static final String NUM_INPUT_FILES = "mapreduce.input.num.files";


  /**
   * Proxy PathFilter that accepts a path only if all filters given in the
   * constructor do. Used by the listPaths() to apply the built-in
   * hiddenFileFilter together with a user provided one (if any).
   */
  private static class MultiPathFilter implements PathFilter {
    private List<PathFilter> filters;


    public MultiPathFilter(List<PathFilter> filters) {
      this.filters = filters;
    }


    public boolean accept(Path path) {
      for (PathFilter filter : filters) {
        if (!filter.accept(path)) {
          return false;
        }
      }
      return true;
    }
  }


  /**
   * Get the lower bound on split size imposed by the format.
   * @return the number of bytes of the minimal split for this format
   */
  protected long getFormatMinSplitSize() {
    return 1;
  }


  /**
   * Is the given filename splitable? Usually, true, but if the file is
   * stream compressed, it will not be.
   * 
   * <code>FileInputFormat</code> implementations can override this and return
   * <code>false</code> to ensure that individual input files are never split-up
   * so that {@link Mapper}s process entire files.
   * 
   * @param context the job context
   * @param filename the file name to check
   * @return is this file splitable?
   */
  protected boolean isSplitable(JobContext context, Path filename) {
    return true;
  }


  /**
   * Set a PathFilter to be applied to the input paths for the map-reduce job.
   * @param job the job to modify
   * @param filter the PathFilter class use for filtering the input paths.
   */
  public static void setInputPathFilter(Job job,
                                        Class<? extends PathFilter> filter) {
    job.getConfiguration().setClass("mapred.input.pathFilter.class", filter, 
                                    PathFilter.class);
  }


  /**
   * Set the minimum input split size
   * @param job the job to modify
   * @param size the minimum size
   */
  public static void setMinInputSplitSize(Job job,
                                          long size) {
    job.getConfiguration().setLong("mapred.min.split.size", size);
  }


  /**
   * Get the minimum split size
   * @param job the job
   * @return the minimum number of bytes that can be in a split
   */
  public static long getMinSplitSize(JobContext job) {
    return job.getConfiguration().getLong("mapred.min.split.size", 1L);
  }


  /**
   * Set the maximum split size
   * @param job the job to modify
   * @param size the maximum split size
   */
  public static void setMaxInputSplitSize(Job job,
                                          long size) {
    job.getConfiguration().setLong("mapred.max.split.size", size);
  }


  /**
   * Get the maximum split size.
   * @param context the job to look at.
   * @return the maximum number of bytes a split can include
   */
  public static long getMaxSplitSize(JobContext context) {
    return context.getConfiguration().getLong("mapred.max.split.size", 
                                              Long.MAX_VALUE);
  }


  /**
   * Get a PathFilter instance of the filter set for the input paths.
   *
   * @return the PathFilter instance set for the job, NULL if none has been set.
   */
  public static PathFilter getInputPathFilter(JobContext context) {
    Configuration conf = context.getConfiguration();
    Class<?> filterClass = conf.getClass("mapred.input.pathFilter.class", null,
        PathFilter.class);
    return (filterClass != null) ?
        (PathFilter) ReflectionUtils.newInstance(filterClass, conf) : null;
  }


  /** List input directories.
   * Subclasses may override to, e.g., select only files matching a regular
   * expression. 
   * 
   * @param job the job to list input paths for
   * @return array of FileStatus objects
   * @throws IOException if zero items.
   */
  protected List<FileStatus> listStatus(JobContext job
                                        ) throws IOException {
    List<FileStatus> result = new ArrayList<FileStatus>();
    Path[] dirs = getInputPaths(job);
    if (dirs.length == 0) {
      throw new IOException("No input paths specified in job");
    }
    
    // get tokens for all the required FileSystems..
    TokenCache.obtainTokensForNamenodes(job.getCredentials(), dirs, 
                                        job.getConfiguration());


    List<IOException> errors = new ArrayList<IOException>();
    
    // creates a MultiPathFilter with the hiddenFileFilter and the
    // user provided one (if any).
    List<PathFilter> filters = new ArrayList<PathFilter>();
    filters.add(hiddenFileFilter);
    PathFilter jobFilter = getInputPathFilter(job);
    if (jobFilter != null) {
      filters.add(jobFilter);
    }
    PathFilter inputFilter = new MultiPathFilter(filters);
    
    for (int i=0; i < dirs.length; ++i) {
      Path p = dirs[i];
      FileSystem fs = p.getFileSystem(job.getConfiguration()); 
      FileStatus[] matches = fs.globStatus(p, inputFilter);
      if (matches == null) {
        errors.add(new IOException("Input path does not exist: " + p));
      } else if (matches.length == 0) {
        errors.add(new IOException("Input Pattern " + p + " matches 0 files"));
      } else {
        for (FileStatus globStat: matches) {
          if (globStat.isDir()) {
            for(FileStatus stat: fs.listStatus(globStat.getPath(),
                inputFilter)) {
              result.add(stat);
            }          
          } else {
            result.add(globStat);
          }
        }
      }
    }


    if (!errors.isEmpty()) {
      throw new InvalidInputException(errors);
    }
    LOG.info("Total input paths to process : " + result.size()); 
    return result;
  }
  


  /** 
   * Generate the list of files and make them into FileSplits.
   */ 
  public List<InputSplit> getSplits(JobContext job
                                    ) throws IOException {
    long minSize = Math.max(getFormatMinSplitSize(), getMinSplitSize(job));
    long maxSize = getMaxSplitSize(job);


    // generate splits
    List<InputSplit> splits = new ArrayList<InputSplit>();
    List<FileStatus>files = listStatus(job);
    for (FileStatus file: files) {
      Path path = file.getPath();
      FileSystem fs = path.getFileSystem(job.getConfiguration());
      long length = file.getLen();
      BlockLocation[] blkLocations = fs.getFileBlockLocations(file, 0, length);
      if ((length != 0) && isSplitable(job, path)) { 
        long blockSize = file.getBlockSize();
        long splitSize = computeSplitSize(blockSize, minSize, maxSize);


        long bytesRemaining = length;
        while (((double) bytesRemaining)/splitSize > SPLIT_SLOP) {
          int blkIndex = getBlockIndex(blkLocations, length-bytesRemaining);
          splits.add(new FileSplit(path, length-bytesRemaining, splitSize, 
                                   blkLocations[blkIndex].getHosts()));
          bytesRemaining -= splitSize;
        }
        
        if (bytesRemaining != 0) {
          splits.add(new FileSplit(path, length-bytesRemaining, bytesRemaining, 
                     blkLocations[blkLocations.length-1].getHosts()));
        }
      } else if (length != 0) {
        splits.add(new FileSplit(path, 0, length, blkLocations[0].getHosts()));
      } else { 
        //Create empty hosts array for zero length files
        splits.add(new FileSplit(path, 0, length, new String[0]));
      }
    }
    
    // Save the number of input files in the job-conf
    job.getConfiguration().setLong(NUM_INPUT_FILES, files.size());


    LOG.debug("Total # of splits: " + splits.size());
    return splits;
  }


  protected long computeSplitSize(long blockSize, long minSize,
                                  long maxSize) {
    return Math.max(minSize, Math.min(maxSize, blockSize));
  }



  protected int getBlockIndex(BlockLocation[] blkLocations, 
                              long offset) {
    for (int i = 0 ; i < blkLocations.length; i++) {
      // is the offset inside this block?
      if ((blkLocations[i].getOffset() <= offset) &&
          (offset < blkLocations[i].getOffset() + blkLocations[i].getLength())){
        return i;
      }
    }
    BlockLocation last = blkLocations[blkLocations.length -1];
    long fileLength = last.getOffset() + last.getLength() -1;
    throw new IllegalArgumentException("Offset " + offset + 
                                       " is outside of file (0.." +
                                       fileLength + ")");
  }


  /**
   * Sets the given comma separated paths as the list of inputs 
   * for the map-reduce job.
   * 
   * @param job the job
   * @param commaSeparatedPaths Comma separated paths to be set as 
   *        the list of inputs for the map-reduce job.
   */
  public static void setInputPaths(Job job, 
                                   String commaSeparatedPaths
                                   ) throws IOException {
    setInputPaths(job, StringUtils.stringToPath(
                        getPathStrings(commaSeparatedPaths)));
  }


  /**
   * Add the given comma separated paths to the list of inputs for
   *  the map-reduce job.
   * 
   * @param job The job to modify
   * @param commaSeparatedPaths Comma separated paths to be added to
   *        the list of inputs for the map-reduce job.
   */
  public static void addInputPaths(Job job, 
                                   String commaSeparatedPaths
                                   ) throws IOException {
    for (String str : getPathStrings(commaSeparatedPaths)) {
      addInputPath(job, new Path(str));
    }
  }


  /**
   * Set the array of {@link Path}s as the list of inputs
   * for the map-reduce job.
   * 
   * @param job The job to modify 
   * @param inputPaths the {@link Path}s of the input directories/files 
   * for the map-reduce job.
   */ 
  public static void setInputPaths(Job job, 
                                   Path... inputPaths) throws IOException {
    Configuration conf = job.getConfiguration();
    Path path = inputPaths[0].getFileSystem(conf).makeQualified(inputPaths[0]);
    StringBuffer str = new StringBuffer(StringUtils.escapeString(path.toString()));
    for(int i = 1; i < inputPaths.length;i++) {
      str.append(StringUtils.COMMA_STR);
      path = inputPaths[i].getFileSystem(conf).makeQualified(inputPaths[i]);
      str.append(StringUtils.escapeString(path.toString()));
    }
    conf.set("mapred.input.dir", str.toString());
  }


  /**
   * Add a {@link Path} to the list of inputs for the map-reduce job.
   * 
   * @param job The {@link Job} to modify
   * @param path {@link Path} to be added to the list of inputs for 
   *            the map-reduce job.
   */
  public static void addInputPath(Job job, 
                                  Path path) throws IOException {
    Configuration conf = job.getConfiguration();
    path = path.getFileSystem(conf).makeQualified(path);
    String dirStr = StringUtils.escapeString(path.toString());
    String dirs = conf.get("mapred.input.dir");
    conf.set("mapred.input.dir", dirs == null ? dirStr : dirs + "," + dirStr);
  }
  
  // This method escapes commas in the glob pattern of the given paths.
  private static String[] getPathStrings(String commaSeparatedPaths) {
    int length = commaSeparatedPaths.length();
    int curlyOpen = 0;
    int pathStart = 0;
    boolean globPattern = false;
    List<String> pathStrings = new ArrayList<String>();
    
    for (int i=0; i<length; i++) {
      char ch = commaSeparatedPaths.charAt(i);
      switch(ch) {
        case '{' : {
          curlyOpen++;
          if (!globPattern) {
            globPattern = true;
          }
          break;
        }
        case '}' : {
          curlyOpen--;
          if (curlyOpen == 0 && globPattern) {
            globPattern = false;
          }
          break;
        }
        case ',' : {
          if (!globPattern) {
            pathStrings.add(commaSeparatedPaths.substring(pathStart, i));
            pathStart = i + 1 ;
          }
          break;
        }
      }
    }
    pathStrings.add(commaSeparatedPaths.substring(pathStart, length));
    
    return pathStrings.toArray(new String[0]);
  }
  
  /**
   * Get the list of input {@link Path}s for the map-reduce job.
   * 
   * @param context The job
   * @return the list of input {@link Path}s for the map-reduce job.
   */
  public static Path[] getInputPaths(JobContext context) {
    String dirs = context.getConfiguration().get("mapred.input.dir", "");
    String [] list = StringUtils.split(dirs);
    Path[] result = new Path[list.length];
    for (int i = 0; i < list.length; i++) {
      result[i] = new Path(StringUtils.unEscapeString(list[i]));
    }
    return result;
  }


}

你可能感兴趣的:(hadoop,map)