Parsing Indentation-based syntaxes in Haskell's Parsec

Question

I'm trying to parse an indentation-based language (think Python, Haskell itself, Boo, YAML) in Haskell using Parsec. I've seen the IndentParser library, and it looks like it's the perfect match, but what I can't figure out is how to make my TokenParser into an indentation parser. Here's the code I have so far:

import qualified Text.ParserCombinators.Parsec.Token as T
import qualified Text.ParserCombinators.Parsec.IndentParser.Token as IT

lexer = T.makeTokenParser mylangDef
ident = IT.identifier    lexer

This throws the error:

parser2.hs:29:28:
    Couldn't match expected type `IT.TokenParser st'
             against inferred type `T.GenTokenParser s u m'
    In the first argument of `IT.identifier', namely `lexer'
    In the expression: IT.identifier lexer
    In the definition of `ident': ident = IT.identifier lexer

What am I doing wrong? How should I create an IT.TokenParser? Or is IndentParser broken and to be avoided?

Answer 1

It looks like you're using Parsec 3 here, while IndentParser expects Parsec 2. Your example compiles for me with -package parsec-2.1.0.1.

So IndentParser isn't necessarily broken, but the author(s) should have been more specific about versions in the list of dependencies. It's possible to have both versions of Parsec installed, so there's no reason you shouldn't use IndentParser unless you're committed to using Parsec 3 for other reasons.

---

UPDATE: Actually no changes to the source are necessary to get IdentParser working with Parsec 3. The problem that both of us were having seems to be caused by the fact that cabal-install has a "soft preference" for Parsec 2. You can simply reinstall IndentParser with an explicit constraint on the Parsec version:

cabal install IndentParser --reinstall --constraint="parsec >= 3"

Alternatively you can download the source and build and install in the normal way.