A slow, super large heirarchial hash set thing using files.

RandomAccessFileHashSet.scala

import java.io.{File, RandomAccessFile}

class RandomAccessFileHashSet(file: File) {
  val raf = new RandomAccessFile(file, "rw")
  def filePosition = raf.getFilePointer
  def fileLength = raf.length
  var recordCount: Long = 0L

  raf.writeLong(recordCount)
  val hashStart = filePosition

  def seek(position: Long) = raf.seek(Math.abs(position))

  def incrementRecordCount() = {
    val originalPos = filePosition
    recordCount = recordCount + 1
    seek(0)
    raf.writeLong(recordCount)
    seek(originalPos)
    println(s"Count: $recordCount")
  }

  def add(value: Array[Byte]): Unit = {
    val originalPos = filePosition
    val isNew = filePosition == fileLength

    // Ensure record exists at byte offset
    if (isNew) {
      0.to(256).foreach((x) => {
        raf.writeLong(0L)
      })
      seek(originalPos)
      if (value.isEmpty) {
        incrementRecordCount()
      }
    }

    // Read record for next offset
    if (value.nonEmpty) {
      // Find the next offset
      val newPos: Long = value.head.toLong * java.lang.Long.BYTES + filePosition
      seek(newPos)
      val nextOffset = raf.readLong()
      if (nextOffset == 0L) {
        // We need to add another record.
        seek(newPos)
        if (value.tail.isEmpty){
          raf.writeLong(-1 *  fileLength)
        } else {
          raf.writeLong(fileLength)
        }

        seek(fileLength)
      } else {
        if (value.tail.isEmpty && nextOffset > 0L) {
          seek(newPos)
          raf.writeLong(-1 *  nextOffset)
          incrementRecordCount()
        }
        seek(nextOffset)
      }
      add(value.tail)
    } else {
      seek(hashStart)
    }

  }
}