Allows an HFile to be used as the input to MapReduce.

HFileInputFormat.scala

import org.apache.hadoop.fs.Path
import org.apache.hadoop.hbase.io.hfile.{HFile,HFileScanner}
import org.apache.hadoop.hbase.io.hfile.HFile.Reader
import org.apache.hadoop.hbase.io.ImmutableBytesWritable
import org.apache.hadoop.hbase.KeyValue
import org.apache.hadoop.mapreduce.{JobContext,InputSplit,TaskAttemptContext,RecordReader}
import org.apache.hadoop.mapreduce.lib.input.{FileInputFormat,FileSplit}

class HFileInputFormat extends FileInputFormat[ImmutableBytesWritable, KeyValue] {
  override def isSplitable(context: JobContext, file: Path): Boolean = false

  def createRecordReader(split: InputSplit, context: TaskAttemptContext): RecordReader[ImmutableBytesWritable, KeyValue] = {
    new HFileRecordReader
  }

  private class HFileRecordReader extends RecordReader[ImmutableBytesWritable, KeyValue] {
    private var reader: HFile.Reader = _
    private var scanner: HFileScanner = _
    private var entryNumber: Int = 0

    def close {
      if (reader != null) {
        reader.close
      }
    }

    def getCurrentKey: ImmutableBytesWritable = new ImmutableBytesWritable(scanner.getKeyValue.getRow)

    def getCurrentValue: KeyValue = scanner.getKeyValue

    def getProgress: Float = entryNumber / 1.max(reader.getEntries).toFloat

    def initialize(split: InputSplit, context: TaskAttemptContext) {
      val path = split.asInstanceOf[FileSplit].getPath
      val fs = org.apache.hadoop.fs.FileSystem.get(context.getConfiguration)
      reader = new HFile.Reader(fs, path, null, false)
      scanner = reader.getScanner(false, false)
      reader.loadFileInfo // This is required or else seekTo throws a NPE
      scanner.seekTo // This is required or else scanner.next throws an error
    }

    def nextKeyValue: Boolean = {
      entryNumber += 1
      scanner.next
    }
  }
}

This will skip the first entry of an HFile. I've forked this gist and fixed the defect.

entryNumber will overflow when HFile is exterme large (because it is not splittalbe).