Efficient way to create abstract collection with cats

Question

I have some code that uses Monix Observable for stream processing of a file. To test this code, I'd like for the operations I do on the Observable to be type independent so I can also perform them on any other data structure like List. That why I have written the following code to abstract the underlying data structure:

def permutations[F[_] : Applicative : FunctorFilter : SemigroupK](chars: F[Char]): F[F[Char]] = {
  Range.inclusive('a', 'z').map(_.toChar)
    .map { c ⇒
      FunctorFilter[F].filter(chars)(Character.toLowerCase _ andThen (_ != c))
    }
    .map(Applicative[F].pure)
    .reduceLeft(SemigroupK[F].combineK)
}

The thing that bugs me, is that this code creates a lot of intermediary data-structures. Is there a type-class I could use that makes this process more efficient? Something that lifts one data-structure into another without too much overhead, like LiftIO but for collections of items?

Answer 1

It does not look like cats has anything to offer for this. And monix is not better, it only implements a handful of typeclasses from cats.

So, my best guess would be defining such typeclasses yourself:

import monix.execution.Scheduler.Implicits.global
import cats._
import cats.implicits._
import monix.reactive._

object Test {

  def main(args: Array[String]): Unit = {

    println(permutations(List('a', 'b', 'c')))

    permutations(Observable('a', 'b', 'c')).foreach{c =>
      print("Observable(")
      c.foreach(c1 => print(c1 + " "))
      print(") ")
    }
  }

  def permutations[F[_] : Applicative](chars: F[Char])(implicit seq: Sequence[F], fil: Filter[F]): F[F[Char]] = {

    val abc = seq.fromIterable(
      Range.inclusive('a', 'z').map(_.toChar)
    )

    abc.map(c => fil.filter(chars)(_ != c))
  }

  trait Sequence[F[_]] {

    def fromIterable[A](f: Iterable[A]): F[A]
  }

  implicit val listSequence: Sequence[List] = new Sequence[List] {

    def fromIterable[A](f: Iterable[A]): List[A] = f.toList
  }

  implicit val observableSequence: Sequence[Observable] = new Sequence[Observable] {

    def fromIterable[A](f: Iterable[A]): Observable[A] = Observable.fromIterable(f)
  }

  trait Filter[F[_]] {

    def filter[A](fa: F[A])(f: A => Boolean): F[A]
  }

  implicit val observableFilterFunctor: Filter[Observable] = new Filter[Observable] {

    def filter[A](fa: Observable[A])(f: A => Boolean): Observable[A] =
      fa.filter(f)
  }

  implicit val listFilterFunctor: Filter[List] = new Filter[List] {

    def filter[A](fa: List[A])(f: A => Boolean): List[A] =
      fa.filter(f)
  }

}

Result:

List(List(b, c), List(a, c), List(a, b), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c), List(a, b, c))
Observable(b c ) Observable(a c ) Observable(a b ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) Observable(a b c ) 

I sadly couldn't get this to work on scalafiddle or scastie, because both don't offer the correct cats (1.5.0) and monix (3.0.0-M3) versions.

I still hope this helps.