Encoding character pairs

shrink.fsx

let allChars =
    Set.ofSeq
    <| seq {
        for c in (System.Char.MinValue) .. (System.Char.MaxValue) do
            if System.Char.IsControl(c) |> not then
                yield c
    }

let mostCommonPair (s: string) =
    s
    |> Seq.windowed 2
    |> Seq.countBy id
    |> Seq.sortBy snd
    |> Seq.last
    |> fst
    |> System.String.Concat

let getUnused (unused: Set<char>) (target: string) (s: string) =
    let used = Set.ofSeq s
    unused - used |> Set.minElement |> string

let shrink (s, swaps) =
    let pair = mostCommonPair s
    let t = getUnused allChars pair s
    s.Replace(pair, t), ((t, pair) :: swaps)

// - https://en.wikipedia.org/wiki/Byte_pair_encoding
// - https://en.wikipedia.org/wiki/Re-Pair