CODEWITHSUNDEEP
parsinghaskellghcghcihaskell-src-exts
Andrew Gibiansky
Andrew Gibiansky

Reputation: 233

Haskell Src Exts Parsing Multiple Statements

I would like to parse a list of Haskell statements. For instance, suppose I have the following code:

let a = b
    c = e
out <- return 3

I'd like a function, for instance parseStmts, which can return this in some parsed format.

I've looked into haskell-src-exts and saw parseStmt. This works for a single statement. It has type parseStmt :: String -> ParseResult Stmt, and if you try parseStmt "let a = 3", the result is a successful ParseOk. However, if you provide multiple statements, this function complains because there is more than one statement in the string.

How do I parse multiple statements, without wrapping them in a do block? Alternatively, how can I find the places in a string which are separations of Haskell statements, so I can separate them and then use parseStmt from haskell-src-exts?

Thanks!

Upvotes: 1

Views: 279

Answers (2)

Daniel Wagner
Daniel Wagner

Reputation: 153342

I don't think haskell-src-exts offers a ready-made function doing what you want, so one way or another you're going to have to write some of your own parsing code. That said, not all is lost. You may have to hack on haskell-src-exts itself to expose a few of its internals, but it should not be unduly difficult to throw together -- a few hours of work to get something decent if you're already familiar with whatever parsing technology it uses (alex/happy, I think?), or double it if you have to learn the parsing technology, too.

I'm sure some patches to the package to make this kind of thing easier would be welcomed with open arms, as well.

Upvotes: 0

bheklilr
bheklilr

Reputation: 54078

You're looking for parseExp, although the output is a bit large:

> :m + Language.Haskell.Exts.Parser
> parseExp "do\n  let a = b\n      c = e\n  out <- return 3\n  return $ a + c + out"
ParseOk (Do [LetStmt (BDecls [PatBind (SrcLoc {srcFilename = "<unknown>.hs", srcLine = 2, srcColumn = 7}) (PVar (Ident "a")) Nothing (UnGuardedRhs (Var (UnQual (Ident "b")))) (BDecls []),PatBind (SrcLoc {srcFilename = "<unknown>.hs", srcLine = 3, srcColumn = 7}) (PVar (Ident "c")) Nothing (UnGuardedRhs (Var (UnQual (Ident "e")))) (BDecls [])]),Generator (SrcLoc {srcFilename = "<unknown>.hs", srcLine = 4, srcColumn = 3}) (PVar (Ident "out")) (App (Var (UnQual (Ident "return"))) (Lit (Int 3))),Qualifier (InfixApp (Var (UnQual (Ident "return"))) (QVarOp (UnQual (Symbol "$"))) (InfixApp (InfixApp (Var (UnQual (Ident "a"))) (QVarOp (UnQual (Symbol "+"))) (Var (UnQual (Ident "c")))) (QVarOp (UnQual (Symbol "+"))) (Var (UnQual (Ident "out")))))])

I had to add the return $ a + c + out to the end or else it throws an error, since it wouldn't be considered a valid do block otherwise.

Upvotes: 2

Related Questions