F# Markov chain first attempt

gistfile1.fs

open System
open System.IO
open System.Text.RegularExpressions

let inputText = File.ReadAllText(__SOURCE_DIRECTORY__ + "/frankenstein.txt")

type Prefix = string list 

type Suffix = string

type Chain = Map<Prefix, Suffix list>

let shift words word = if List.length words = 2 then List.tail words @ [word] else words @ [word]

let toWords s = 
    Regex.Split(inputText, @"\b") 
    |> Seq.filter (fun w -> String.IsNullOrWhiteSpace(w) = false)
    |> Seq.toList

let rec addWords chain prefix (words: string list) : Chain =
    match words with
    | [] -> chain
    | _ -> 
        let word = List.head words
        let suffixes = defaultArg (Map.tryFind prefix chain) []
        let nextSuffix = word :: suffixes
        let nextChain = Map.add prefix nextSuffix chain
        let nextPrefix = shift prefix word
        addWords nextChain nextPrefix (List.tail words)

let buildChain words = addWords Map.empty [] words

let rand = new Random()

let rec pickWords chain prefix n words =
    if n = 0 
    then words 
    else
        match Map.tryFind prefix chain with
        | None -> []
        | Some c -> 
            let choice = List.nth c (rand.Next(0, List.length c))
            let words' = words @ [choice]
            let prefix' = shift prefix choice
            pickWords chain prefix' (n - 1) words'

let generate chain n = pickWords chain [] n []


[<EntryPoint>]
let main argv = 
    let input = toWords inputText
    let chain = buildChain input
    printf "%A" <| generate chain 100
    0 // return an integer exit code