I don't understand why you wouldn't use Tree Sitter's syntax highlighting for this. I mean it's not going to be as fast but that clearly isn't an issue here.
Is this a "no third party dependencies" thing?
Is this a "no third party dependencies" thing?
But fair enough - fewer dependencies is always nice, especially in C++ (which doesn't have a modern package manager) and in ML where an enormous janky Python installation is apparently a perfectly normal thing to require.
Microsoft is such an odd duck, sometimes, but I'm glad to take advantage of their "good years" while it lasts
And here - https://lobste.rs/s/9huy81/tbsp_tree_based_source_processing...
For new projects I use Chumsky. It's a pure Rust parser which is nice because it means you avoid the generated C, and it also gives you a fully parsed and natively typed output, rather than Tree Sitter's dynamically typed tree of nodes, which means there's no extra parsing step to do.
The main downside is it's more complicated to write the parser (some fairly extreme types). The API isn't stable yet either. But overall I like it more than Tree Sitter.