漢数字変換 by scala parser combinator

build.sbt

libraryDependencies += "org.scala-tools.testing" %% "scalacheck" % "1.9"

parser.scala

import scala.util.parsing.combinator._

object PrsKsj {
  def apply(s:String) = {
    val p = new PrsKsj
    p.parseAll(p.root,s)
  }
}

class PrsKsj extends JavaTokenParsers {
  val k2d = Seq("壱","弐","参","四","五","六","七","八","九") zip (1 to 9) toMap
  def root = opt(千の位) ~ opt(百の位) ~ opt(十の位) ~ opt(一の位) ^^ {
    case s ~ h ~ j ~ i => Seq(s,h,j,i).foldLeft(0){(l,r) => l + r.getOrElse(0)}
  }
  def ksj = "壱" | "弐" | "参" | "四" | "五" | "六" | "七" | "八" | "九"
  def 一の位 = ksj ^^ k2d
  def 十の位 = unitPrs("拾",10)
  def 百の位 = unitPrs("百",100)
  def 千の位 = unitPrs("千",1000)
  def unitPrs(u:String,mul:Int) = opt(ksj) <~ u ^^ {
    case Some(d) => k2d(d) * mul
    case None => mul
  }
}

object Main extends App {
  Seq(("壱",1),("百",100),("千",1000),("千壱",1001),("千弐拾",1020),
      ("百五",105),("六千参百拾壱",6311),("unmatch data",33)
  ) foreach {t => PrsKsj(t._1) match {
      case x if(x.successful) => print("test:" + t);assert(x.get == t._2);println(":OK")
      case x => println(t,x)
  }}
}

randomTest.scala

import org.scalacheck._
import Prop._

object KsjSpecification extends Properties("漢数字テスト") {
  case class KsjMaker(ksj:Seq[(String,Int)]) {
    val ksjPair = ksj map {_._1} zip Seq("千","百","拾","")
    val 漢数字 = ksjPair filter(_._1 != "") map {x => x._1 + x._2} mkString("")
    val アラビア数字 = ksj map {_._2} zip Seq(1000,100,10,1) map {case (a,b) => a * b} sum
  }
  val seed = Seq("","壱","弐","参","四","五","六","七","八","九") zip (0 to 9)
  val gen = for{
    th <- Gen.oneOf(seed)
    h <- Gen.oneOf(seed)
    t <- Gen.oneOf(seed)
    o <- Gen.oneOf(seed)
  } yield KsjMaker(th :: h :: t :: o :: Nil)

  property("ランダム選択") = forAll(gen) { x =>
    collect("%s(%s)".format(x.漢数字,x.アラビア数字)) {
      PrsKsj(x.漢数字).get == x.アラビア数字
    }
  }
}

宿題: 万・億・兆の変換も対応する