我已经问过一个有关了解单子状态的问题,但是现在我觉得这应该是另外一个了。
鉴于我实现了状态monad,我希望我的state
是类似于Environment
的数据结构:
环境
data Env=Env{
envName::String,
fileNames::[String]
}
instance Show Env where
show Env{envName=x,fileNames=xs} = "{ envName:"++x++
" , files: ["++foldr (\t y-> t++","++y) "" xs ++"] }"
initEnv::IO Env
initEnv=do
name<- getLine
names<- getCurrentDirectory>>=listDirectory
return Env{envName=name,fileNames=names}
我不知道如何将这种数据结构作为State
集成到State monad中,以便能够更改环境的name
,打印或使用它。似乎太宽泛了,但是如果没有完整的例子,我将无法理解:
州实施monad
newtype State s a = State {run::s->(a,s)}
instance Functor (State s) where
fmap=Control.Monad.liftM
instance Applicative (State s) where
pure=return
(<*>)=Control.Monad.ap
instance Monad (State s) where
return a= State $ \k->(a,k)
(>>=) m f=State $ \s -> let (a,s')=run m s in
run (f a) s'
我要实现的目标
readEnv::State a Env->Env
readEnv m =
changeEnvName::State a Env->State a Env
changeEnvName m = --given the environment packed in a state ,
-- i want to change the name
getEnvFileLengths::State a Env->[Int]
getEnvFileLengths s a= s>>= getLengths
getLengths::[String]->[Int]
getLengths xs=map length xs
PS 我知道我应该使用Reader
或Writer
monad,但我希望使用all in one
方法来理解所有事物如何组合在一起。
有什么想法吗?
答案 0 :(得分:1)
如果正确获得类型签名,则可能会更容易取得进展:
readEnv::State Env Env
changeEnvName::String -> State Env ()
getEnvFileLengths::State Env [Int]
如果对您来说,这些类型看起来像是奇怪的类型选择,则可能值得尝试扩展newtype
,然后查看它们是否显得更明智:
-- give me an initial environment from your store, I'll give you the new environment
-- to store and another copy of the environment as the result of the computation
readEnv :: Env -> (Env, Env)
-- give me a new name and the old environment, I'll give you a new environment and
-- a trivial acknowledgement that I'm done
changeEnvName :: String -> Env -> ((), Env)
-- give me an initial environment that you're storing, I'll give you the new
-- environment to store (actually same as the old one) and the result of the
-- length computations
getEnvFileLengths :: Env -> ([Int], Env)