给出元素列表:
xs = [a, b, c, d, ... z]
其中a, b, c
等是任意值的占位符。
我想实现一个产生
adjacents :: [a] -> [(a, a)]
adjacentValues = [(a, b), (b, c), (c, d), ... (y, z)]
在Haskell中,递归定义相当简洁:
adjacents :: [a] -> [(a, a)]
adjacents (x:xs) = (x, head xs) : adjacents xs
adjacents [] = []
Purescript更加冗长:
adjacents :: forall a. List a -> List (Tuple a a)
adjacents list = case uncons list of
Nothing -> []
Just {head: x, tail: xs} -> case head xs of
Just next -> Tuple x next : adjacents xs
Nothing -> []
有没有办法在没有显式递归的情况下表达adjacents
(使用折叠)?
免责声明:此问题同时包含Purescript和Haskell标签,因为我想向更广泛的受众群体开放。我认为答案不依赖于haskells懒惰评估语义,因此在两种语言中都有效。
答案 0 :(得分:7)
在Haskell中,没有显式递归,你可以用它的尾部压缩列表。
let a = [1,2,3,4,5,6,7,8,9,0]
a `zip` tail a
=> [(1,2),(2,3),(3,4),(4,5),(5,6),(6,7),(7,8),(8,9),(9,0)]
答案 1 :(得分:2)
完整性的Purescript解决方案:
adjacent :: forall n. List n -> List (Tuple n n)
adjacent list = zip list $ fromMaybe empty $ tail list
可以更优雅地表达为:
adjacent :: forall n. List n -> List (Tuple n n)
adjacent list = zip list $ drop 1 list
答案 2 :(得分:0)
为了便于说明(基于zip
的解决方案肯定更好),这里是您明确递归的Haskell解决方案。我没有特别的原因把它打成一个单线。
{-# LANGUAGE LambdaCase #-}
import Data.List (unfoldr)
adjacent :: [a] -> [(a, a)]
adjacent = unfoldr (\case { x:y:ys -> Just ((x, y), ys); _ -> Nothing })
(请注意,此处的模式匹配处理列表,其中包含奇数个元素而不会崩溃。)
答案 3 :(得分:0)
自从我们看到zip
和unfoldr
后,我们应该使用foldr
:
adjacent :: [a] -> [(a,a)]
adjacent xs = foldr go (const []) xs Nothing
where
go a r Nothing = r (Just a)
go a r (Just prev) = (prev, a) : r (Just a)
现在,因为每个玩具问题都需要一个过度设计的解决方案,这就是你可以用来获得双面列表融合的方法:
import GHC.Exts (build)
adjacent :: [a] -> [(a,a)]
adjacent xs = build $ \c nil ->
let
go a r Nothing = r (Just a)
go a r (Just prev) = (prev, a) `c` r (Just a)
in foldr go (const nil) xs Nothing
{-# INLINE adjacent #-}
答案 4 :(得分:0)
折叠状态,其中状态是最后配对的项目:
在Haskell:
import Data.List (mapAccumL)
adjacents :: [a] -> [(a, a)]
adjacents [] = []
adjacents (x:xs) = snd $ mapAccumL op x xs
where
op x y = (y, (x,y))