gistfile1.scala

    val conf = new Configuration
    conf.set("io.serializations", "org.apache.hadoop.io.serializer.JavaSerialization,"
      + "org.apache.hadoop.io.serializer.WritableSerialization")

    // the path, where the vectors will be stored to
    val vectorsPath = new Path("job/vectors")
    // enumeration of all users involved in a selected subset of hand history records
    val dictionaryPath = new Path("job/dictionary")
    // text file with the dictionary size
    val dictionarySizePath = new Path("job/dictionary-size")
    // indexed dictionary (every user ID in the dictionary is mapped to an index, from 0)
    val indexedDictionaryPath = new Path("job/indexed-dictionary")

    println("Building dictionary...")
    // extracts IDs of all the users, participating in selected subset of hand history records
    Tool.main(Array(classOf[Dictionary.Builder].getName, "--hdfs",
      "--hbasehost", "localhost", "--output", dictionaryPath.toString))
    // adds index to the dictionary
    Tool.main(Array(classOf[Dictionary.Indexer].getName, "--hdfs",
      "--input", dictionaryPath.toString, "--output", indexedDictionaryPath.toString))
    // calculates dictionary size, and stores it to the FS
    Tool.main(Array(classOf[Dictionary.Size].getName, "--hdfs",
      "--input", dictionaryPath.toString, "--output", dictionarySizePath.toString))

    // reads dictionary size
    val fs = FileSystem.get(dictionaryPath.toUri, conf)
    val dictionarySize = new BufferedReader(
      new InputStreamReader(
        fs.open(new Path(dictionarySizePath, "part-00000"))
      )).readLine().toInt

    println("Vectorizing...")
    // builds vectors (player -> other players in the game)
    // IDs of other players (in the vectors) are replaces with indices, taken from dictionary
    Tool.main(Array(classOf[VectorBuilder].getName, "--hdfs",
      "--dictionary", dictionaryPath.toString, "--hbasehost", "localhost",
      "--output", vectorsPath.toString, "--dictionarySize", dictionarySize.toString))