Question

我正在尝试将Reader monad和State monad组合在一起。这是我的代码：

object StatePoc extends App {
  def startsWithState : Reader[String, State[Seq[String], Unit]] = Reader { s => State { ss => (ss.filter(_.startsWith(s)), Unit)}}
  def endsWithState : Reader[String, State[Seq[String], Unit]] = Reader { s => State { ss => (ss.filter(_.endsWith(s)), Unit)}}
  def process: Kleisli[Id.Id, String, State[Seq[String], Unit]] = {
    for {
      s <- startsWithState
      e <- endsWithState
    } yield e
  }

  val all = Seq("AB", "BA", "ABA", "ABBA")

  println(process("A")(all))
}

不幸的是，此代码未返回预期结果：List(BA, ABA, ABBA)而不是List(ABA, ABBA)。

Answer 1

在你的函数process中，你从两个读者中读取状态对象，但是你只返回最后一个状态对象，而不是两者的组合。这样的事情会同时结合s和e：

  def process: Kleisli[Id.Id, String, State[Seq[String], Unit]] = {
    for {
      s <- startsWithState
      e <- endsWithState
    } yield {
      State { seq => e.run(s.run(seq)._1)}
    }
  }

Answer 2

另一个救世主可能是monad变压器，这需要一些样板，特别是没有种类的投影仪：

type StringReader[a] = Reader[String, a]
type StateChoose[a, b] = StateT[StringReader, a, b]
type StringsChoose[a] = StateChoose[Seq[String], a]
type StringsTrans[m[_], b] = StateT[m, Seq[String], b]

def ST = MonadState[StateChoose, Seq[String]]
def R = MonadReader[Reader, String]
def T = MonadTrans[StringsTrans]

def transform(action: String => Seq[String] => Seq[String]): StringsChoose[Unit] = for {
  s <- T.liftMU(R.ask)
  _ <- ST.modify(action(s))
} yield ()
def startsWithState = transform( s => _.filter(_.startsWith(s)))
def endsWithState = transform(s => _.filter(_.endsWith(s)))

def process: StringsChoose[Unit] = {
  for {
    s <- startsWithState
    e <- endsWithState
  } yield e
}
val all = Seq("AB", "BA", "ABA", "ABBA")

println(process.exec(all).apply("A"))

如何结合Reader和State monad

2 个答案: