Add continuously growing collection Generators

[mdedetrich]

This PR adds generators which allow you to generate a collection/list that continuously grows until a condition/predicate is fullfilled (with the input of the condition being the current state of the collection being generated).

I am creating this PR because I currently have a usecase where I need to create a collection that is of specific size (in bytes, not number of elements) due to testing multipart uploads and having to make sure that generated test data is of least a specific chunk size but I suppose the problem is generic enough that it can apply to other issues.

Due to Gen being a sealed trait and certain methods inside Gen being private its not that easy to extend Gen with your own generator methods.

The naming of the methods can probably be done better, also not sure if I need to add tests for these and if so what I should be testing for/what they should look like. The failureHint's approach is also an open question due to the fact that its obviously not possible to figure out the size of the resulting collection.

[mdedetrich]

@armanbilge Would it be possible to look into this?

[satorg]

@mdedetrich but is it really necessary to extend Gen for that? There's a public method infiniteLazyList (or infitineStream) in Gen that (I think) can be used to achieve pretty much the same functionality, e.g.:

  def buildableOfCondGen[C, T](fillCondition: C => Boolean, g: Gen[T])(
      implicit
      evb: Buildable[T, C],
      evt: C => Iterable[T]
  ): Gen[C] =
    Gen.infiniteLazyList(g).map { ll =>
      val it = ll.iterator
      val bldr = evb.builder
      while (!fillCondition(bldr.result()))
        bldr += it.next()
      bldr.result() // sub-optimal: is called twice for the same result, can be improved!
    }

  val vgen: Gen[Vector[Int]] = buildableOfCondGen[Vector[Int], Int](_.size == 123, Gen.chooseNum(456, 789))

  Prop.forAll(vgen) { v =>
    v.size == 123
  }.check()

And perhaps you even don't have to worry about failureHint with it.

[mdedetrich]

@satorg

Perfect thanks, I didn't realize there was a flexible infiniteLazyList that can theoretically be used to do the same. Will look into this and report back!

[satorg]

Hi @mdedetrich , sorry for the long ping. Just to clarify with you: would you like to continue on this PR or do you think it can be closed? Thank you for your effort in any case!