Passing a method and parameters to a Scala case class? - scala

Question

I am parsing an XML document and store its data in various other structured document formats. In this XML document, the elements reference other elements, such as:
<myCar id="12" name="Porsche XYZ" ...>
<connected refId="3" />
</myCar>
...
<myCar id="3" name="Audi XYZ" ...>
...
</myCar>
Here, refId maps to id. When creating the myCar instance with the id 12, I cannot reference to myCar with id 3, because it has not yet been parsed.
Obviously, the easy solution would be to parse the document twice (instantiate references in the second run, after all elements have been parsed and created). However, for performance reasons I only want to parse the document once. Thus, I thought I could just store the relevant reference data in a case class and build a list of instances that is passed from one method to another, in order to process it after having parsed the entire document.
However, my problem is that the logic for creating the references varies to a great extent. So, I cannot use something like this:
case class Ref (a: String, b: String)
val refs: List[Ref] = List.empty
// 1. fill the list with references during parsing
// 2. after parsing the document, process all references in the list
I think what I need is to move all my reference creation logic to separate methods, and then when parsing the document maintain a list with "pointers" to these methods including the appropriate parameters. In this way, I could just iterate through the list after parsing the entire document, and call every method with the correct parameters.
How can this be achieved?

Answer 1

I'm not 100% sure what you're asking but I think you are asking for a way to link the connected cars to the base case class such that after parsing the XML you have a list of cars and for any given car you can access the refId attribute (materialized as a full car object) from the connected tag.
Here is a simple approach:
given:
val xml = <root>
<myCar id="12" name="Porsche XYZ">
<connected refId="3" />
</myCar>
<myCar id="3" name="Audi XYZ">
</myCar>
</root>
First we'll make a case class to model a myCar:
case class Car(
id: String,
name: String,
connectedId: Option[String]
)
Then we parse the XML into Car instances. I'm going to parse it into a Map[String, Car] where the key is Car.id:
val result = (xml \ "myCar").foldLeft(Map.empty[String, Car]) {
case (acc, next) =>
val id: String = (next \# "id")
val name = (next \# "name")
val connectionStr = (next \ "connected" \# "refId")
val connection = Option.unless(connectionStr.isEmpty)(connectionStr)
val car = Car(
id,
name,
connection
)
acc + (id -> car)
}
Next we need a way to turn connectedId into an actual car. I did this by adding a method to Car changing the case class to:
case class Car(
id: String,
name: String,
connectedId: Option[String]
) {
def getConnected(cars: Map[String, Car]): Option[Car] = {
connectedId.flatMap { id =>
cars.get(id)
}
}
}
This method (getConnected) takes the Map produced in the previous step.
Get the list of cars with:
result.values // Iterable(Car(12,Porsche XYZ,Some(3)), Car(3,Audi XYZ,None))
Get the connected car for the first car in the list:
result.values.head.getConnected(result) // Some(Car(3,Audi XYZ,None))
If you want to "fill in" the connected cars add a field to hold the connected car (pass None in the initial foldLeft above):
case class Car(
id: String,
name: String,
connectedId: Option[String],
connected: Option[Car],
) {
def getConnected(cars: Map[String, Car]): Option[Car] = {
connectedId.flatMap { id =>
cars.get(id)
}
}
}
Then just map over the list, adding the connections:
result.values
.map { car =>
val connectedCar = car.connectedId.flatMap { id =>
car.getConnected(result)
}
car.copy(connected = connectedCar)
}
This produces:
List(Car(12,Porsche XYZ,Some(3),Some(Car(3,Audi XYZ,None,None))), Car(3,Audi XYZ,None,None))
This does not recursively fill in the connected cars. You'd have to switch to either make this recursive somehow or use a var in Car to track the connected car and modify references instead of using .copy to accomplish that. I haven't thought about this too much though.
Full working code here: https://scastie.scala-lang.org/YuhNdszQROKTaNExMchaCg