CODEWITHSUNDEEP
jsonparsinghaskellfunctional-programming
FrozenHawk
FrozenHawk

Reputation: 150

Haskell JSON Parser Improvements

For one of my courses last semester we were tasked with implementing a JSON parser that spits out formatted HTML. I ended up implementing it in what I felt was more of an iterative fashion than a functional one. Here's the main code snippet I have issues with:

stringToTokens (x:xs)
            .....
            | x == '\"' = StringLiteral (parseStringToken xs) : stringToTokens (advancePastString xs)
            | startsNumeric x = Numeric (processNumeric (x:xs) takeWhile) : stringToTokens (processNumeric xs dropWhile)
            | startsKeyword x = Keyword (processKeyword (x:xs) take) : stringToTokens (processKeyword (x:xs) drop)
            .....
            | otherwise = stringToTokens xs

....
advancePastString :: String -> String
advancePastString (x:xs)
            | x == '\"' = xs
            | x == '\\' = advancePastString (tail xs)
            | otherwise = advancePastString xs

My issue with how I went about handling this is that it feels as if I've taken how I would implement it in a procedural language and just replaced the loop with recursion and drop functions. While this does work I'm wondering what would be a more idiomatic functional way of handling this.

Upvotes: 2

Views: 144

Answers (1)

MCH
MCH

Reputation: 2214

I would try a parser combinator. This is a good place to start: Write You a Haskell: Parsing by Stephen Diehl. It gives you a simple example of what a parser combinator looks like. Then I would look at the available parser combinators like parsec, attoparsec and trifecta. I personally like attoparsec, but trifecta has a lot more features.

I've also written a small tutorial on attoparsec. It isn't complete but it should help you get started.

Upvotes: 2

Related Questions