Highlight definition and control block structures

[rogancodes]

Motivation

Closes #2494

Implementation

If the cursor is on a definition or control node, both the start and end of the node will be highlighted.

Automated Tests

Amended existing test case.

Manual Tests

Place the cursor on one of the definition or control nodes.

Screencast.from.25-10-24.08.44.06.AM.IST.webm

[vinistock]

This is looking good!

[vinistock]

Thank you for the contribution!

[waynehoover]

Are there any editor.semanticTokenColorCustomizations rules we need to add to make this visible? I'm not seeing it using the github light color theme in vscode.

[jbigler]

This seems to be working great for me in Neovim.
I do see an issue when I am in an ERB file. While it is able to ignore the class= attribute of normal HTML tags, it gets confused when there is a class: argument in a ruby method such as a tag builder.

[vinistock]

AFAIK, all you need is for editor.occurrencesHighlight to not be turned off, which is already the default. The editor.semanticTokenColorCustomizations setting is for customizing your theme colours for semantic tokens, which is related to semantic highlighting, not document highlight.

[andyw8]

You can also verify that "Inspect Editor Tokens and Scopes" (via Command-P) matches this:

[waynehoover]

Thanks, my editor.occurrencesHighlight was off, which is why it wasn't showing.

Is there anyway to make the def/end highlighting continue to highlight when I'm inside the method? Like what Zed does, here is an example:

https://share.zight.com/yAugGqED

[vinistock]

That's possible, although not super conventional for the implementation of document highlight. The intent of the feature is to highlight occurrences of the thing under the cursor.

This feels more like highlighting the surrounding scope.

[jbigler]

Here's a visual example of my comment above:

My cursor is on the end statement, but it is confused by the class: argument and so isn't highlighting the corresponding do statement.

[vinistock]

@jbigler oh, that looks like a bug in our implementation. Could you please open an issue so that we can prioritize and look into it?