Problem with MenuItemAction vs. Selectors

[Daniel Faria Sampaio]

I'm working on a fork of FolioReaderKit and your library is used there.

As a way of "filtering" which actions will be displayed on each scenario it was chosen to override the function canPerformAction of the UIWebView. There they verify each action (which is a Selector) and see if the item will end up displaying on the screen.

Like this:

open override func canPerformAction(_ action: Selector, withSender sender: Any?) -> Bool {
    if action == #selector(highlight(_:))
        || action == #selector(createAnnotation(_:))
        || (action == #selector(define(_:)) && isOneWord)
        || (action == #selector(play(_:)) && (book.hasAudio() || readerConfig.enableTTS))
        || (action == #selector(share(_:)) && readerConfig.allowSharing)
        || action == #selector(copyText(_:)) {
        return true
    } else {
        return false
    }
}

The thing is that most of those actions are actually a Selector and were created using the default API like this:

let highlightItem = UIMenuItem(title: readerConfig.localizedHighlightMenu, action: #selector(highlight(_:)))

When using your API for creating items with images we end up sending MenuItemAction instead of Selectors:

let discussion = UIImage(readerImageNamed: "discussion-marker")
let discussionItem = UIMenuItem(title: "D", image: discussion!) { [weak self] _ in   
    self?.createHighlight()
}

Thus it's very hard to work combined with canPerformAction (which has a Selector as parameter)

Any suggestions of filtering which UIMenuItem(created with images, using MenuItemAction) will appear on a specific scenario?

[Zayin Krige]

We built an epub reader using WKWebView (uiwebview leaks memory), and we have a similar canPerformAction block as to what you described.

UIMenuKit creates actions with a fixed name, so you can filter them like this

    NSString *selectorName = NSStringFromSelector(action);
    if ([selectorName hasPrefix:@"_menuitemkit_block"]) {
        return YES;
    }

essentially I'm allowing all the menuitemkit actions through. You could possibly do custom naming of actions instead of random names

the constructor could be changed to inject an identifier - see here self.init(title: title, action: Selector(blockIdentifierPrefix + UUID.stripedString + ":"))

[realazy]

Now exposing this method from MenuItemKit, version 2.0.2.

[Zayin Krige]

I dont think this is quite what @danfsd was looking for :)

this only tells us if a selector is a menuitemkit selector - it doesnt tell us what the selector is. We need someway to identify a specific selector's functionality.

In his example he was checking if action == #selector(highlight(_:)), which tells us if the selector is the highlight selector.

In order to do this we would need to be able to identify UIMenuItemAction's with a specific identifier

[Saleel Karkhanis]

Can you explain me how can we check if a selector is a menuitemkit selector in CanPerformAction ?

[Daniel Faria Sampaio]

One temporary solution is to create a function which creates a MenuItemAction that accepts a selector instead of a function as a parameter.

We can argue whether comparing the action in the WKWebView.canPerformAction(_:) is the right way to selectively show or hide the action (based on a scenario), but this solution I give should fix the things for now.

[realazy]

Considering adding this way in future version.

[Saleel Karkhanis]

Actully My problem is CanPerformAction method is not getting called for custom menuitem selectors using WebView.

[realazy]

Now you can use UIMenuController.installTo(responder:shouldShowForAction:) introduced in 3.1.0.