在这种情况下，如何按类型正确进行模式匹配？

Question

I have a use-case where I'd like to serialize a Saddle Frame to a Scala-Play Json. I had it working for simple element types eg Double ie Frame[DateTime, String, Double] but now it becomes a bit more hairy trying to serialize Frame types whose elements are complex types eg Seq[(DateTime, Double)]

So I have the basic frameWrites implementation that works for primitive element types:

import play.api.libs.json._
import org.saddle._
import org.joda.time._

object JsonUtils {
implicit def frameWrites[RX, CX, E] = new Writes[Frame[RX, CX, E]] {
  override def writes(frame: Frame[RX, CX, E]): JsValue = {
    val json: JsArray = Json.arr(
      (0 until frame.numRows).map { i =>
        Json.obj(
          frame.rowIx.at(i).toString ->
            (0 until frame.numCols).map { j =>
              Json.obj(
                frame.colIx.at(j).toString -> cellWrites(frame.at(i, j))
              )
            }
        )
      })
    json
  }
}}

So I created a cellWrites to be able to handle complex element types like so:

object JsonUtils {
implicit def cellWrites[E](elem: E)(implicit tag: ClassTag[E]): JsValue = {
  elem match {
    case seq: Seq[(DateTime, Double)] => { 
    // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ this should but doesn't happen
      Json.arr(
        seq.map { case (dt: DateTime, value: Double) =>
          Json.obj(dt.toString -> JsNumber(value))
        }
      )
    }
    case doubleValue: Double => JsNumber(doubleValue)
    case _ => JsString(elem.toString)
  }
}}

The problem is that my Frame or type Frame[String, String, Seq[(DateTime, Double)]] is never matching the appropriate first case in cellWrites . I also tried using the implicit tag type information but could not find the way to use it for Pattern matching. This is a test-case sample Frame I use:

val today = DateTime.parse("2017-11-14T13:00:00.000+01:00")
val colIx = Index("x", "y")
val data = Seq(
  Series(
    "a" -> Seq(
        today.minusHours(2) -> 2.0,
        today.minusHours(1) -> 1.0,
        today.minusHours(0) -> 0.0
    ),
    "b" -> Seq(
      today.minusHours(1) -> 1.0,
      today.minusHours(0) -> 0.0
    )
  ),
  Series(
    "a" -> Seq(
      today.minusHours(5) -> 5.0,
      today.minusHours(4) -> 4.0,
      today.minusHours(3) -> 3.0
    ),
    "b" -> Seq(
      today.minusHours(4) -> 4.0,
      today.minusHours(3) -> 3.0
    )
  )
)
val frame: Frame[String, String, Seq[(DateTime, Double)]] = Frame(data, colIx)
println(frame)

Note that all the code above compiles fine. However, the Pattern matching by cell element type is not working.

Answer 1

OK found it, elem was actually of type org.saddle.scalar.Scalar[E] therefore this works:

implicit def cellWrites[E](scalar: Scalar[E]): JsValue = {
  val elem = scalar.get
  elem match {
    case seq: Seq[(DateTime, Double)] => {
      Json.arr(
        seq.map { case (dt: DateTime, value: Double) =>
          Json.obj(dt.toString -> JsNumber(value))
        }
      )
    }
    case doubleValue: Double => JsNumber(doubleValue)
    case _ => JsString(elem.toString)
  }
}