Markov Chain N-gram

scala

val file = scala.io.Source.fromFile("quijote.txt")
val txt = file.mkString
val splited = txt.replace("\n", " ").replace("[ *]", " ").split(" ")
val wgram = scala.collection.mutable.HashMap[String, List[String]]()

(0 until splited.size-1).foreach{ i =>
  val word = splited(i)
  wgram(word) = splited(i + 1) :: wgram.getOrElse(word, Nil)
}

val r = scala.util.Random

def getFirstWord: String = {
  val res = wgram.keys.toList(r.nextInt(wgram.size-2))
  if(res.charAt(0).isLetter && res.charAt(0).isUpper && res.last.isLetter)
    res
  else
    getFirstWord
}

def getText(start: String = getFirstWord, words: Int, endPunctuation: Boolean = false): String = {
  val lst = wgram(start)

  words match {
    case n if n<=0 && !endPunctuation =>
      start
    case n if n<=0 && endPunctuation && lst.exists(s=> s.endsWith(".") || lst.endsWith("!") || lst.endsWith("?")) =>
      lst.collectFirst {
        case s if s.endsWith(".") || lst.endsWith("!") || lst.endsWith("?") =>  s
      }.get
    case n =>
      val word = lst(r.nextInt(lst.size))
      s"$start ${getText(word, words-1, endPunctuation).toLowerCase}"
  }
}

(1 to 10).foreach { _ =>
  println(getText(words = 10, endPunctuation = true))
}