Text.Parsing.Parser.ArrayBuffer

[jamesdbrock]

I want to parse the raw Uint8 bytes of a Javascript ArrayBuffer. You know, like the ByteString instance for Megaparsec Streams, or Data.Attoparsec.ByteString.

Is there a good way to do this in Purescript? I can't find it. If this doesn't exist, maybe we could add a new module to purescript-parsing called Text.Parsing.Parser.ArrayBuffer? It seems like purescript-parsing was designed to be extended in this way.

This module would be a peer of Text.Parsing.Parser.String and Text.Parsing.Parser.Token, and similarly constructed.

Or maybe this would be better as a separate library, since it would add some new dependencies:

https://pursuit.purescript.org/packages/purescript-arraybuffer
https://pursuit.purescript.org/packages/purescript-arraybuffer-types

Thoughts?

[jamesdbrock]

The fact that getUint8 is an Effect function might complicate this.

I would probably just unsafePerformEffect in the implementation and warn users not to mutate the ArrayBuffer while parsing.

[garyb]

I don't think the Effect is a problem, you can use the m part of ParserT to handle that. Something along the lines of:

import Control.Monad.State (get, modify_)
import Control.Monad.Trans.Class (lift)
import Data.ArrayBuffer.DataView (getUint8)
import Data.ArrayBuffer.Types (DataView)
import Data.Maybe (Maybe(..))
import Data.UInt (UInt)
import Effect.Class (class MonadEffect, liftEffect)
import Text.Parsing.Parser (ParseState(..), ParserT, fail)
import Text.Parsing.Parser.Pos (Position(..))

uint8 ∷ ∀ m. MonadEffect m ⇒ ParserT DataView m UInt
uint8 = do
  ParseState input (Position { column }) _ <- get
  lift (liftEffect (getUint8 input column)) >>= case _ of
    Nothing → fail "Unexpected EOF"
    Just i → do
      modify_ \(ParseState _ position _) ->
        ParseState input (Position { line: 0, column: column + 1 }) true
      pure i

I haven't tested that, so it might not behave quite right, but it compiles and shouldn't be too far off.

The Position hard coded into ParserT isn't really ideal for this purpose 😕, hence the fixed line: 0.

[jamesdbrock]

Good idea. I'll take a shot at a PR for this.

[jamesdbrock]

https://pursuit.purescript.org/packages/purescript-arraybuffer is not a purescript-contrib library, is it okay for purescript-parsing to depend on it?

[jamesdbrock]

purescript-arraybuffer is really nice because it handles endianness.

[garyb]

Hmm, good point - we try to avoid extra-contrib dependencies unless the library is maintaned by one of us in contrib/core (in the past we've had dependencies that end up blocking libraries that need updates).

Maybe this would be better off as a new library that depends on -arraybuffer and -parsing. I'd be happy to add a link to it in the readme here to help others find it, if you put something together.

[jamesdbrock]

Also transitively depends on non-contrib library https://pursuit.purescript.org/packages/purescript-typelevel