Generic random generators to implement Arbitrary instances for QuickCheck

Automating the arbitrary boilerplate also ensures that when a type changes to have more or fewer constructors, then the generator either fixes itself to generate that new case (when using the uniform distribution) or causes a compilation error so you remember to fix it (when using an explicit distribution).

This package also offers a simple (optional) strategy to ensure termination for recursive types: make Test.QuickCheck.Gen's size parameter decrease at every recursive call; when it reaches zero, sample directly from a trivially terminating generator given explicitly (genericArbitraryRec and withBaseCase) or implicitly (genericArbitrary').

{-# LANGUAGE DeriveGeneric #-}

import GHC.Generics (Generic)
import Test.QuickCheck
import Generic.Random

data Tree a = Leaf | Node (Tree a) a (Tree a)
  deriving (Show, Generic)

instance Arbitrary a => Arbitrary (Tree a) where
  arbitrary = genericArbitraryRec uniform `withBaseCase` return Leaf

-- Equivalent to
-- > arbitrary =
-- >   sized $ \n ->
-- >     if n == 0 then
-- >       return Leaf
-- >     else
-- >       oneof
-- >         [ return Leaf
-- >         , resize (n `div` 3) $
-- >             Node <$> arbitrary <*> arbitrary <*> arbitrary
-- >         ]

main :: IO ()
main = sample (arbitrary :: Gen (Tree ()))

• The following two packages also derive random generators, but only with a uniform distribution of constructors:

  • quickcheck-arbitrary-template (TH)
  • generic-arbitrary (GHC Generics)

• testing-feat: derive enumerations for algebraic data types, which can be turned into random generators (TH).

• boltzmann-samplers: derive Boltzmann samplers (SYB).