How do I require whitespace?

[nathan-at-least]

Hi, new to chumsky but I've almost replaced a half-written homebrew parser for a simple (non-recursive) grammar in less than an hour! Really slick experience, so thank you!

I want to require whitespace between tokens, but chumsky::text::whitespace allows zero instances.

The example in chumsky::primitive::filter helped me write my own recipe that looks like this:

    let label = ident()
        .delimited_by(just('['), just(']'))
        .then(
            filter::<_, _, Simple<char>>(|c: &char| c.is_whitespace())
                .repeated()
                .at_least(1)
                .ignored(),
        )   
        .map(|(label, ())| label);

Is there a simpler approach I'm missing?

Would it be feasible and ergonomic to have the return type of whitespace behave similar to repeated, with modifier methods such as whitespace().at_least(1)?

[numToStr]

How about this?

let label = ident::<_, Simple<char>>()
    .delimited_by(just('['), just(']'))
    .then_ignore(just(' ').repeated().at_least(1));

[numToStr]

or (similar to your implementation)

let label = ident::<_, Simple<char>>()
    .delimited_by(just('['), just(']'))
    .then_ignore(filter(|c: &char| c.is_whitespace()).repeated().at_least(1));

[nathan-at-least]

This works! Thanks. Two bits I learned:

• Moving the type hint to ident is more concise.
• .then_ignore helps me avoid the custom .map to unwrap tuples.

Thanks!

[zesterer]

Hi! For now, that is the easiest setup. That said, whitespace internally uses Repeated so it should be possibly to simply reveal that fact to the caller, allowing it to invoke the usual methods. I'll open an issue.

[nathan-at-least]

I assume 1 or more whitespace is a common rule in many syntaxes, so it might be worth it. Thanks again, for a great tool.