Using Lezer/CodeMirror to power Highlight.js grammars

Considering using Lezer (or at least providing a plugin to use it) with Highlight.js for “super high fidelity” language highlighting. It’s been very hard for our parser to properly parse some complex languages like JSX (mostly because our goal is not to be a full parser) - and having a “real” parser (that understands the entire language) is a whole new level of fidelity that we’re not really familiar with.

Any one have any thoughts on this? Good idea? Bad idea? Does the Lezer community welcome people using Lezer and it’s grammars outside the CodeMirror system itself for syntax classification/highlighting purposes?

Related: Question: How to query "syntax tags" for a given Node/NodeType? · Issue #29 · lezer-parser/lezer · GitHub

So far this is proving a little harder than I expected because Lezer doesn’t seem to have a way to directly expose the actual syntax tags for a given node (only it’s raw “type name”) - that’s all buried deep inside Highlighting abstractions (that I don’t need). I just want semi-cooked (not fully raw) syntax tag and classification data.