Add support for polars

[deanm0000]

I think there's a really good synergy between display data tables and polars. Polars gives access to cloud data and to group by aggregations to only mention a few benefits.

Polars would be a good source especially for virtualized tables as they have lazy evaluation too.

[maccesch]

Oh this is an interesting one!

[maccesch]

How exactly would you want to use it though? It still would have to be downloaded from somewhere, right?
At the moment the rows get downloaded and cached internally simply as a vec of struct instances.

[deanm0000]

I'm thinking something like:

fn main() {
    mount_to_body(|| {
        let df = DataFrame::new(vec![
            Series::new("a",[1.0,2.0,3.0]),
            Series::new("b",[2.1,3.1,4.1]),
            Series::new("c",[0.0,1.2,2.2])
        ])
        .unwrap()

        view! {
            <table>
                <TableContent df />
            </table>
        }
    });
}

After looking at the paginated_rest_datasource, I think I can use the TableDataProvider to use a DataFrame. That said, I think polars would be a good synergy for if/when grouping is a feature that you'd support.

As a tangent, is there any way to have dynamically defined (not just types but the existence of) columns or can it only ever work with what's hard codes in a struct?

[maccesch]

I haven't tried to use it dynamically yet as this is not (yet?) the premise of this crate. But it could be possible if you implement TableDataProvideras well as TableRow manually without using the macro.

If get sth to work feel free to make a PR for a new example. Would be cool and could server as a basis for future development.

Grouping is certainly on the roadmap but probably not very soon.

[kstep]

I think it's a good idea to create a leptos-struct-table-polars crate and not put such integrations into the leptos-struct-table itself, otherwise we will end up with another bloated crate with a lot of integrations.

Also, I agree it should be doable with a custom TableDataProvider implementation, maybe some kind of wrapper type/newtype for polars dataframe?

[maccesch]

Could be a good idea. I haven't really thought about it yet.