purescript-contrib/purescript-parsing

match combinator

jamesdbrock opened this issue · 3 comments

We should have the super-useful match combinator.

To write this combinator for Text.Parsing.Parser.String, I think we will need some more members of StringLike.

class StringLike s where
drop :: Int -> s -> s
indexOf :: Pattern -> s -> Maybe Int
null :: s -> Boolean
uncons :: s -> Maybe { head :: Char, tail :: s }

Maybe if we have

length :: s -> Int
length = Data.String.length

take :: Int -> s -> s
take = Data.String.take

Then we can write

match :: forall m s a.  ParserT s m a -> ParserT s m (Tuple a s)
match p = do
  ParseState input1 _ _ <- get
  x <- p
  ParseState input2 _ _ <- get
  pure $ Tuple x $ take (length input1 - length input2) input1

Actually we should use the Javascript length, which is in units of “code units”, instead of Data.String.length. which is in units of “code points”.

And likewise for take.

This would be more efficient, and it will be correct as long as we can assume that the ParseState input always begins on a code point boundary.

Done in v7.0.0