29  Collecting extendr types

Objective

Learn to collect iterators into extendr types.

We’ve been using .collect() to create vectors of a return type. However, these are not native R types. Using Vec<T> requires allocating a new (R) vector and filling the values.

Instead, we can collect into the wrapper types. To do so, the iterator must be a scalar type—e.g. Rfloat, Rint, or Rstr.

Rather than .collect::<Vec<_>>() we specify the wrapper type we want.

Example

Using the previous parallel sum, we can instead return Doubles using the turbo-fish syntax

#[extendr]
fn psum(x: Doubles, y: Doubles) -> Doubles {
    x.into_iter()
        .zip(y.into_iter())
        .map(|(xi, yi)| Rfloat::from(xi.inner() + yi.inner()))
        .collect::<Doubles>()
}

Exercise

• Modify gh_encode() so that it returns Strings
• Ensure that each iteration returns an Rstr instead of String

View solution

#[extendr]
fn gh_encode(x: Doubles, y: Doubles, length: usize) -> Strings {
    if length == 0 || length > 12 {
        throw_r_error("`length` must be between 1 and 12");
    }

    x.into_iter()
        .zip(y.into_iter())
        .map(|(xi, yi)| {
            let coord = Coord {
                x: xi.inner(),
                y: yi.inner(),
            };
            let gh = encode(coord, length).unwrap();
            Rstr::from(gh)
        })
        .collect::<Strings>()
}