在Haskell中解析化合物

时间:2015-04-21 16:50:58

标签: parsing haskell

我试图为自己做一个化合物解析器作为练习,但我卡住了。

以下是我尝试使用的数据类型:

data Compound = Monoatomic String Int | Poliatomic [Compound] Int

给出像" Ca(OH)2"这样的字符串,我希望得到类似的东西;

Poliatomic [Monoatomic "Ca" 1, Poliatomic [Monoatomic "O" 1, Monoatomic "H" 1] 2 ] 1

单原子的单原子类型构造函数,以及多个原子的Poliatomic构造函数。在该示例中,(OH)2表示内部Poliatomic结构,并且表示为Poliatomic [(Monoatomic O 1), (Monoatomic H 1 )] 2。数字2表示我们有两个poliosomic结构。

我做了这么多;

import Data.Char (isUpper)
data Compound = Monoatomic String Int | Poliatomic [Compound] Int

instance Functor Compound where
        fmap f (Monoatomic s i) = Monoatomic (f s) i
        fmap f (Poliatomic xs i) = Poliatomic (fmap f xs) i

-- Change number of a compound
changeNumber :: Compound -> Int -> Compound
changeNumber (Monoatomic xs _) n = Monoatomic xs n
changeNumber (Poliatomic xs _) n = Poliatomic xs n

-- Take a partial compound and next chracter return partial compound
parseCompound :: Compound -> Char -> Compound
parseCompound (Poliatomic x:xs n) c
        | isUpper c = Poliatomic ((Monoatomic [c] 1):x:xs) n -- add new atom to compound
        | isLower c = Poliatomic 

-- I want to do foldl parseCompound (Poliatomic [] 1) inputstring

但是我继续这太复杂了。

看起来它应该是一个相当简单的问题,但我对Haskell很新,并且无法弄清楚如何完成这个功能。

我有这个问题:

  • 到目前为止我的方法是否正确?
  • 我怎样才能做到这一点?

1 个答案:

答案 0 :(得分:4)

我已经使用Parsec创建了您正在寻找的解析器,让您了解Parsec解析器的外观,因为您声明您对它没什么经验。

即使Haskell经验不足,也应该具有相当的可读性。我已经提供了一些关于特别需要注意的部分的评论。

import Text.Read (readMaybe)
import Data.Maybe (fromMaybe)
import Text.Parsec (parse, many, many1, digit, char, string, (<|>), choice, try)
import Text.Parsec.String (Parser)


data Compound
  = Monoatomic String Int
  | Poliatomic [Compound] Int
  deriving Show


-- Run the substance parser on "Ca(OH)2" and print the result which is
-- Right (Poliatomic [Monoatomic "Ca" 1,Poliatomic [Monoatomic "O" 1,Monoatomic "H" 1] 2] 1)
main = print (parse substance "" "Ca(OH)2")


-- parse the many parts which make out the top-level polyatomic compound
--
-- "many1" means "at least one"
substance :: Parser Compound
substance = do
  topLevel <- many1 part
  return (Poliatomic topLevel 1)


-- a single part in a substance is either a poliatomic compound or a monoatomic compound
part :: Parser Compound
part = poliatomic <|> monoatomic


-- a poliatomic compound starts with a '(', then has many parts inside, then
-- ends with ')' and has a number after it which indicates how many of it there
-- are.
poliatomic :: Parser Compound
poliatomic = do
  char '('
  inner <- many1 part
  char ')'
  amount <- many1 digit
  return (Poliatomic inner (read amount))


-- a monoatomic compound is one of the many element names, followed by an
-- optional digit. if omitted, the amount defaults to 1.
--
-- "try" is a little special, and required in this case. it means "if a parser
-- fails, try the next one from where you started, not from where the last one
-- failed."
--
-- "choice" means "try all parsers in this list, stop when one matches"
--
-- "many" means "zero or more"
monoatomic :: Parser Compound
monoatomic = do
  name <- choice [try nameParser | nameParser <- atomstrings]
  amount <- many digit
  return (Monoatomic name (fromMaybe 1 (readMaybe amount)))


-- a list of parser for atom names. it is IMPORTANT that the longest names
-- come first. the reason for that is that it makes the parser much simpler to
-- write, and it can execute much faster. it's common when designing parsers to
-- consider things like that when creating them.
atomstrings :: [Parser String]
atomstrings = map string (words "He Li Be Ne Na Mg Al Ca H B C N O F")

我试图以一种初学者应该至少可以合理访问的方式编写这段代码,但它可能不是很清楚,所以我很乐意回答任何有关此问题的问题。

上面的解析器就是你想要的。但是,如果我有自由缰绳的话,那不是我要写的那个。如果我想做但我想要,我会利用这个事实

Ca(OH)2

可以表示为

(Ca)1((O)1(H)1)2

这是一个更加统一的表示,反过来导致更简单的数据结构和具有更少样板的解析器。我想写的代码看起来像

import Text.Read (readMaybe)
import Data.Maybe (fromMaybe)
import Control.Applicative ((<$>), (<*>), pure)
import Text.Parsec (parse, many, many1, digit, char, string, (<|>), choice, try, between)
import Text.Parsec.String (Parser)


data Substance
  = Part [Substance] Int
  | Atom String
  deriving Show


main = print (parse substance "" "Ca(OH)2")
-- Right (Part [Part [Atom "Ca"] 1,Part [Part [Atom "O"] 1,Part [Atom "H"] 1] 2] 1)

substance :: Parser Substance
substance = Part <$> many1 part <*> pure 1

part :: Parser Substance
part = do
  inner <- polyatomic <|> monoatomic
  amount <- fromMaybe 1 . readMaybe <$> many digit
  return (Part inner amount)

polyatomic :: Parser [Substance]
polyatomic = between (char '(') (char ')') (many1 part)

monoatomic :: Parser [Substance]
monoatomic = (:[]) . Atom <$> choice (map (try . string) atomstrings)

atomstrings :: [String]
atomstrings = words "He Li Be Ne Na Mg Al Ca H B C N O F"

这在Haskell中使用了一些“高级”技巧(例如<$><*>运算符),所以可能对你不感兴趣,OP,但是我把它放在其他的可能是更高级的Haskell用户并了解Parsec的人。

如你所见,这个解析器只占用了大约半个页面,这就像Parsec这样的库的强大功能 - 它们使编写解析器变得简单而且有趣