gpt4 book ai didi

haskell - 这个 Haskell kata 解决方案可以变得更惯用吗?

转载 作者:行者123 更新时间:2023-12-05 00:04:12 27 4
gpt4 key购买 nike

经过 10 年的中断,我正在重新学习 Haskell,部分是为了看看发生了什么变化,部分是为了消除在 C#、SQL 和 JavaScript 中度过的日子,部分是因为它突然变得很酷;-)

我决定将自己的汉诺塔设置为编码卡塔,足够简单的东西,但我已经觉得我的代码是非惯用的,并且很想听听任何 Haskell 老手可能有什么提示和技巧。

为了让套路更有趣,我将问题分成两部分,第一部分,函数 moves , 生成解决难题所需的移动序列。代码的其余部分旨在为塔建模并执行移动。

我绝对感到不满意的一个部分是 moveDisc功能,这将是繁琐的扩展到 4 个塔。

Hanoi.hs

module Hanoi 
where

import Data.Maybe

type Disc = Integer
type Towers = [[Disc]]
data Column = A | B | C deriving (Eq,Show)

getDisc :: Towers -> Column -> Maybe Disc
getDisc t A = listToMaybe $ t !! 0
getDisc t B = listToMaybe $ t !! 1
getDisc t C = listToMaybe $ t !! 2

validMove :: Towers -> Column -> Column -> Bool
validMove tower from to
| srcDisc == Nothing = False
| destDisc == Nothing = True
| otherwise = srcDisc < destDisc
where srcDisc = getDisc tower from
destDisc = getDisc tower to

moveDisc :: Towers -> Column -> Column -> Towers
moveDisc [a:as, b, c] A B = [as, a:b, c]
moveDisc [a:as, b, c] A C = [as, b, a:c]
moveDisc [a, b:bs, c] B A = [b:a, bs, c]
moveDisc [a, b:bs, c] B C = [a, bs, b:c]
moveDisc [a, b, c:cs] C A = [c:a, b, cs]
moveDisc [a, b, c:cs] C B = [a, c:b, cs]

moves :: Integer -> Column -> Column -> Column -> [(Column,Column)]
moves 1 a _ c = [(a,c)]
moves n a b c = moves (n-1) a c b ++ [(a,c)] ++ moves (n-1) b a c

solve :: Towers -> Towers
solve towers = foldl (\t (from,to) -> moveDisc t from to) towers (moves len A B C)
where len = height towers

height :: Towers -> Integer
height (t:_) = toInteger $ length t

newGame :: Integer -> Towers
newGame n = [[1..n],[],[]]

TestHanoi.hs
module TestHanoi
where

import Test.HUnit
import Hanoi

main = runTestTT $ "Hanoi Tests" ~: TestList [

getDisc [[1],[2],[2]] A ~?= Just 1 ,
getDisc [[1],[2],[3]] B ~?= Just 2 ,
getDisc [[1],[2],[3]] C ~?= Just 3 ,
getDisc [[],[2],[3]] A ~?= Nothing ,
getDisc [[1,2,3],[],[]] A ~?= Just 1 ,

validMove [[1,2,3],[],[]] A B ~?= True ,
validMove [[2,3],[1],[]] A B ~?= False ,
validMove [[3],[],[1,2]] A C ~?= False ,
validMove [[],[],[1,2,3]] A C ~?= False ,

moveDisc [[1],[],[]] A B ~?= [[],[1],[]] ,
moveDisc [[],[1],[]] B C ~?= [[],[],[1]] ,
moveDisc [[1,2],[],[]] A B ~?= [[2],[1],[]] ,
moveDisc [[],[2],[1]] C B ~?= [[],[1,2],[]] ,
moveDisc [[1,2],[],[]] A C ~?= [[2],[],[1]] ,
moveDisc [[3],[2],[1]] B A ~?= [[2,3],[],[1]] ,

moves 1 A B C ~?= [(A,C)] ,
moves 2 A B C ~?= [(A,B),(A,C),(B,C)] ,

"acceptance test" ~:
solve [[1,2,3,4,5,6], [], []] ~?= [[],[],[1,2,3,4,5,6]] ,

"is optimal" ~:
length (moves 3 A B C) ~?= 7
]

我期待听到任何改进意见或建议。

最佳答案

这是使用替代表示的实现。我没有存储三个挂钉尺寸列表,而是存储一个列列表,其中第一个元素对应于最小圆盘的位置,依此类推。这样做的好处是现在不可能表示非法状态,例如丢失的磁盘、较大的磁盘堆叠在较小的磁盘上等。它还使得许多功能难以实现。

Hanoi.hs

module Hanoi where

import Control.Applicative
import Control.Monad
import Data.List
import Data.Maybe

type Disc = Integer
type Towers = [Column]
data Column = A | B | C deriving (Eq, Show)

getDisc :: Column -> Towers -> Maybe Disc
getDisc c t = (+1) . toInteger <$> elemIndex c t

validMove :: Column -> Column -> Towers -> Bool
validMove from to = isJust . moveDisc from to

moveDisc :: Column -> Column -> Towers -> Maybe Towers
moveDisc from to = foldr check Nothing . tails
where check (c:cs)
| c == from = const . Just $ to : cs
| c == to = const Nothing
| otherwise = fmap (c:)

moves :: Integer -> Column -> Column -> Column -> [(Column,Column)]
moves 1 a _ c = [(a,c)]
moves n a b c = moves (n-1) a c b ++ [(a,c)] ++ moves (n-1) b a c

solve :: Towers -> Towers
solve towers = fromJust $ foldM (\t (from,to) -> moveDisc from to t) towers (moves len A B C)
where len = height towers

height :: Towers -> Integer
height = genericLength

newGame :: Integer -> Towers
newGame n = genericReplicate n A

HanoiTest.hs
module HanoiTest where

import Test.HUnit
import Hanoi

main = runTestTT $ "Hanoi Tests" ~: TestList [

getDisc A [A, B, C] ~?= Just 1 ,
getDisc B [A, B, C] ~?= Just 2 ,
getDisc C [A, B, C] ~?= Just 3 ,
getDisc A [B, B, C] ~?= Nothing ,
getDisc A [A, A, A] ~?= Just 1 ,

validMove A B [A, A, A] ~?= True ,
validMove A B [B, A, A] ~?= False ,
validMove A C [C, C, A] ~?= False ,
validMove A C [C, C, C] ~?= False ,

moveDisc A B [A] ~?= Just [B] ,
moveDisc B C [B] ~?= Just [C] ,
moveDisc A B [A, A] ~?= Just [B, A] ,
moveDisc C B [C, B] ~?= Just [B, B] ,
moveDisc A C [A, A] ~?= Just [C, A] ,
moveDisc B A [C, B, A] ~?= Just [C, A, A] ,

moves 1 A B C ~?= [(A,C)] ,
moves 2 A B C ~?= [(A,B),(A,C),(B,C)] ,

"acceptance test" ~:
solve [A, A, A, A, A, A] ~?= [C, C, C, C, C, C] ,

"is optimal" ~:
length (moves 3 A B C) ~?= 7
]

除了表示的变化,我还做了 moveDisc返回总数 Nothing在无效移动的情况下。这样我就可以轻松实现 validMove就其而言。我确实觉得有一种更优雅的方式来实现 moveDisc尽管。

请注意 solve仅当参数是初始位置时才有效。您的代码也是这种情况(由于 moveDisc 中的模式不完整而失败)。我回 Nothing在这种情况下。

编辑:增加了rampion的改进 moveDisc并更改了参数排序以将数据结构放在最后。

关于haskell - 这个 Haskell kata 解决方案可以变得更惯用吗?,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/5890782/

27 4 0
Copyright 2021 - 2024 cfsdn All Rights Reserved 蜀ICP备2022000587号
广告合作:1813099741@qq.com 6ren.com