扩展 Scala 集合

Question

我想要一个在尝试覆盖现有键的值时抛出的映射。我尝试过：

trait Unoverwriteable[A, B] extends scala.collection.Map[A, B] {
    case class KeyAlreadyExistsException(e: String) extends Exception(e)

    abstract override def + [B1 >: B] (kv: (A, B1)): Unoverwriteable[A, B1] = {
        if (this contains(kv _1)) throw new KeyAlreadyExistsException(
            "key already exists in WritableOnce map: %s".format((kv _1) toString)
        )
        super.+(kv)
    }

    abstract override def get(key: A): Option[B] = super.get(key)
    abstract override def iterator: Iterator[(A, B)] = super.iterator
    abstract override def -(key: A): Unoverwriteable[A, B] = super.-(key)
}

并得到：

<console>:11: error: type mismatch;
 found   : scala.collection.Map[A,B1]
 required: Unoverwirteable[A,B1]
               super.+(kv)
                      ^
<console>:16: error: type mismatch;
 found   : scala.collection.Map[A,B]
 required: Unoverwirteable[A,B]
           abstract override def -(key: A): Unoverwirteable[A, B] = super.-(key)
                                                                           ^

我对 Scala 很陌生，无法找到克服这个问题的方法。有什么帮助吗？ :)

编辑：我正在使用 Scala 2.8.0.Beta1-prerelease （这给 scala.collection 带来了一些变化）

Answer 1

这修复了您的编译错误：

trait Unoverwriteable[A, B] extends scala.collection.Map[A, B] {
    case class KeyAlreadyExistsException(e: String) extends Exception(e)

    abstract override def + [B1 >: B] (kv: (A, B1)): scala.collection.Map[A, B1] = {
        if (this contains(kv _1)) throw new KeyAlreadyExistsException(
            "key already exists in WritableOnce map: %s".format((kv _1) toString)
        )
        super.+[B1](kv)
    }

    abstract override def get(key: A): Option[B] = super.get(key)
    abstract override def iterator: Iterator[(A, B)] = super.iterator
    abstract override def -(key: A): scala.collection.Map[A, B] = super.-(key)
}

但是，我认为您确实想装饰 collection.mutable.Map#+=，如下所示：

trait Unoverwriteable[A, B] extends collection.mutable.Map[A, B] {
  case class KeyAlreadyExistsException(e: String) extends Exception(e)

  abstract override def +=(kv: (A, B)): this.type = {
    if (this contains (kv _1))
      throw new KeyAlreadyExistsException("key already exists in WritableOnce map: %s".format((kv _1) toString))
    super.+=(kv)
  }
}

Answer 2

当您重写 Map 中的方法时，您无法将您的特征定义为返回类型。

最简单的解决方案是省略类型：

abstract override def + [B1 >: B] (kv: (A, B1)) = { /* ... */ }
// ...
abstract override def -(key: A) = super.-(key)

或者您可以显式地添加超级类型：

import scala.collection.Map
abstract override def +[B1 >: B] (kv: (A, B1)): Map[A, B1] = { /* ... */ }
// ...
abstract override def -(key: A) = super.-(key): Map[A, B]

我认为您只需要覆盖 + ，因为您的其他方法仅委托给 Map。

Answer 3

您可以使用带有一点隐式魔法的 scala.collection.immutable.Map 来完成此操作。也就是说，您在接口中定义一个附加方法和一个隐式转换。这是我在 2.7 中的做法，我确信在 2.8 中可以重写不同的方法，但您应该了解总体思路。

trait Unoverwriteable[A, B] extends scala.collection.immutable.Map[A, B] {
    import Unoverwriteable.unoverwriteableMap

    case class KeyAlreadyExistsException(e: String) extends Exception(e)

    def underlying: scala.collection.immutable.Map[A, B]

    def update [B1 >: B] (key: A, value: B1): Unoverwriteable[A, B1] = {
        if (this contains(key)) throw new KeyAlreadyExistsException(
            "key already exists in WritableOnce map: %s".format(key.toString)
        )
        underlying update (key, value)
    }

    def get(key: A): Option[B] = underlying get key 
    def elements: Iterator[(A, B)] = underlying.elements
    def -(key: A): Unoverwriteable[A,B] = underlying - key
    def empty[C]: Unoverwriteable[A,C] = underlying.empty[C]
    def size: Int = underlying.size
}

然后，您在伴随对象中定义隐式：

object Unoverwriteable {
   implicit def unoverwriteableMap[A, B](map0: scala.collection.immutable.Map[A, B]): Unoverwriteable[A, B] =
      new Unoverwriteable[A, B] { def underlying = map0 }

}

要使用它，请向您的映射添加 Unwriteable 类型注释。如果取消注释 main 方法中的最后 2 行，您将根据需要得到 KeyAlreadyExistsException。

object UOMain {
   def main(args: Array[String]): Unit = {
      val map0 = Map((1 -> 1), (2 -> 2)): Unoverwriteable[Int, Int]
      println("map0="+ map0)

      val map1 = map0 - 2
      println("map1="+ map1)

      //val map2 = map1 + (1 -> 1000)
      //println("map2" + map2)
   }
}