How to access and update a value in a mutable map of map of maps

I've a three-level data structure (indentation and line breaks for readability):

scala> import scala.collection.mutable.Map
import scala.collection.mutable.Map

scala> val m = Map("normal" -> Map("home" -> Map("wins" -> 0, "scores" -> 0),
                                   "away" -> Map("wins" -> 0, "scores" -> 0)))
m: scala.collection.mutable.Map[java.lang.String,
   scala.collection.mutable.Map[java.lang.String,
   scala.collection.mutable.Map[java.lang.String,Int]]] = 
Map((normal,Map(away -> Map(wins -> 0, scores -> 0),
     home -> Map(wins -> 0, scores -> 0))))

Accessing the innermost data (scores) requires a lot of typing:

import org.scalatest.{Assertions, FunSuite}

class MapExamplesSO extends FunSuite with Assertions {
  test("Update values in a mutable map of map of maps") {
    import scala.collection.mutable.Map
    // The m map is essentially an accumulator
    val m = Map("normal" -> 
                Map("home" -> Map("wins" -> 0, "scores" -> 0),
                    "away" -&g开发者_C百科t; Map("wins" -> 0, "scores" -> 0)
                  )
          )
    //
    // Is there a less verbose way to increment the scores ?
    //
    assert(m("normal").apply("home").apply("scores") === 0)

    val s1 = m("normal").apply("home").apply("scores") + 1
    m("normal").apply("home").update("scores", s1)

    assert(m("normal").apply("home").apply("scores") === 1)

    val s2 = m("normal").apply("home").apply("scores") + 2
    m("normal").apply("home").update("scores", s2)

    assert(m("normal").apply("home").apply("scores") === 3)
  }
}

Is there a less verbose way to modify the value of scores ?

I'm a Scala newbie, so all other observations of the code above are also welcome.

You don't have to use "apply" just do it normally with "()"

m("normal")("home")("scores") = 1

You can write

m("normal").apply("home").apply("scores")

as

m("normal")("home")("scores")

However I'm not sure if such a structure is a good idea. Maybe you should consider encapsulating this functionality in a specialized class.

Less verbose:

assert(m("normal")("home")("scores") === 0)

val s1 = m("normal")("home")("scores") + 1
m("normal")("home")("scores") = s1

assert(m("normal")("home")("scores") === 1)

val s2 = m("normal")("home")("scores") + 2
m("normal")("home")("scores") = s2

assert(m("normal")("home")("scores") === 3)

This takes advantage of the fact that both apply and update have syntactic sugars for them as seen above. Shorter still:

// On REPL, put both these definitions inside an object instead
// of entering them on different lines
def scores = m("normal")("home")("scores")
def scores_=(n: Int) = m("normal")("home")("scores") = n

assert(scores === 0)

val s1 = scores + 1
scores = s1

assert(scores === 1)

val s2 = scores + 2
scores = s2

// Just so you see these updates are being made to the map:
assert(m("normal")("home")("scores") === 3)

Which takes advantage of the syntactic sugar for getters and setters (the getter definition must exist for the setter definition to work).

Adding a local helper function is always good way to reduce code duplication:

class MapExamplesSO {
  def test {
    import scala.collection.mutable.Map
    // The m map is essentially an accumulator
    var m = Map("normal" -> 
                Map("home" -> Map("wins" -> 0, "scores" -> 0),
                    "away" -> Map("wins" -> 0, "scores" -> 0)))


    //Local Helper returns (Old, New)
    def updateScore(k1 : String,k2 : String,k3 : String)
                   (f : Int => Int) : (Int, Int) = {
      val old = m(k1)(k2)(k3)
      m(k1)(k2)(k3) = f(old)
      (old, m(k1)(k2)(k3))
    }

    assert(m("normal")(home")("scores") === 0)
    assert(updateScore("normal","home","scores")(_+1)._2 === 1)
    assert(updateScore("normal","home","scores")(_+2)._2 === 3)
  }
}

[Edit made code tighter]