[Haskell-beginners] Defining custom parser using Parsec

[Stephen Tetley]

On 18 October 2010 07:56, Magnus Therning <magnus at therning.org> wrote:

>
> AFAIK Parsec deals with String, not ByteString, have a look at the
> attoparsec library[1] instead.
>

Parsec 3.0 deals with ByteStrings, Parsec 3.1 and higher improve the
performance quite about.

>From the original post, the syntax of what is to be parsed looks
simple (essentially a single data type, no alternatives except the
AccessType tag), so it can probably easily get by without combinators
like sepBy and endBy. It would need a sample of the input data or a
grammar for it to affirm this, though. A working (simple) parser in
Parsec should be easy to convert to Attoparsec etc if it doesn't use
combinators like sepBy.


parseEvents :: Parser [Event]
parseEvents = many event

event :: Parser Event
event = do
    fn_ <- parseFn
    mft_num <- parseMftNum
    ft_     <- parseFt
    fs_     <- integer
    time_   <- integer
    at_     <- parseAccessType
    mt_     <- parseAccessType
    ct_     <- parseAccessType
    crt_    <- parseAccessType
    return $ Event
      { fn       = fn_
      , mftNum   = mft_num
      , ft       = ft_
      , fs       = fs_
      , time     = time_
      , at       = at_
      , mt       = mt_
      , ct       = ct_
      , crt      = crt_
      }

More parsers will need to be defined for parseMftNum etc. The
applicative style as used in RWH can get rid of the interim variables
which makes it much nicer for parsing than the monadic style. Parsec
3.0 and later can use the Applicative style. How the input data
separates fields will need taking care of.