On Wed, Apr 20, 2011 at 1:03 PM, Sergey Mironov
Hello cafe.
Haskell wiki told me about continuation-based parser Data.Binary.IncrementalGet [1] from binary-strict package. I found the idea very useful and tried to use it. Original library by Lennart Kolmodin raises some questions. The lib's main data structures are:
Lennart Kolmodin has a branch of Binary with incremental get which supports lookAhead: https://github.com/kolmodin/binary/tree/cps I don't have performance measurements, but if you look-ahead too far it obviously isn't good for memory consumption. Antoine
data IResult a = IFailed S String | IFinished S a | IPartial (B.ByteString -> IResult a)
newtype Get r a = Get { unGet :: S -> (a -> S -> IResult r) -> IResult r }
instance Monad (Get r) where return a = Get (\s -> \k -> k a s) m >>= k = Get (\s -> \cont -> unGet m s (\a -> \s' -> unGet (k a) s' cont)) fail err = Get (\s -> const $ IFailed s err)
Here, "S" is parser's state. It works well, but currently doesn't support lookAhead. I tried to add such function and gave up to do it having current implementation, but write simpler one instead. Please see IncrementalGet2.hs (link [2] below). Remake is briefly tested, has no ghc-specific optimizations, but allows user to peek data from stream.
What bothering me is the fact that I could actually miss idea (or part of idea) of the original. If it is so, please let me know :) For example, what is the point of using separate result type r in original Get r a?
[1] - Original IncrementalGet lib. http://www.haskell.org/haskellwiki/DealingWithBinaryData#Incremental_parsing [2] - Main file of remake https://github.com/ierton/binary-incremental-get2/blob/a9f9afaa0cbc0435a8ace... [3] - whole github project https://github.com/ierton/binary-incremental-get2 -- Sergey
sorry for weak English
_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe