Support modal interactions

[LukasDoesDev]

Modal support would be really convenient. A basic struct for modals would be great.

#[derive(ModalDerive)]
#[modal(title="my modal's title", custom_id="this_will_be_used_in_the_parser_too"]
struct MyModal {
    // The fields will automatically get wrapped with actionrows maybe?
    #[modal_field(custom_id="foo", label="Foo", value=1_u8, type=select_menu)]
    my_modal_field: u8,
    #[modal_field(custom_id="bar", label="Bar", max_length=20, text_input_ style=short, type=text_input)]
    my_text_input_field: String
}

[baptiste0928]

Hello! Modals are much simpler than commands, I don't think that derive macros are necessary for this (especially since it is complicated to maintain). What do you think about a method similar to CommandInputData::parse_field instead?

[LukasDoesDev]

A parser is really the thing I'm looking for

[randomairborne]

A parser is what I would be looking for too. Creating them with a macro would be nice, parsing them with a macro would be amazing.

[lara-rium]

modals work with components, why not make it so that it works with components and modals
here's a PoC:

#[derive(ComponentModel, CreateComponent)]
struct MyComponents {
    #[component(label = "just the beginning", kind = ComponentType::Button)]
    // custom_id is my_component
    my_component: Vec<String>, // values, can be anything that impl parse from String, empty for buttons etc
}

// create usage
interaction_client.create_response(.., &Response {
        ...
    }.add_component::<MyComponents>::() // provided by CreateComponent trait
)

// parse usage
MyComponents::from_components(
    ComponentInputData::from(message_component_interaction_data)
)?

[gpluscb]

I am working on an implementation for myself currently based on the original derive macro idea, but with two derive traits ModalModel and CreateModal (https://github.com/gpluscb/twilight-interactions).

@baptiste0928 since you said macros might be overkill and it's complicated to maintain, should I still clean it up and PR it in the future?
For the record, my implementation currently assumes only text inputs are possible and that each component is wrapped in a ActionRow.
Not sure if Discord is planning on changing those limitations anytime soon.

[baptiste0928]

Hi @gpluscb! If you have a working implementation, please open a PR and I'll take a look. Breaking changes are less frequent on the Twilight side these days (due to the slower development pace and the library becoming more mature), so I don't think this will add much to the maintenance burden now.