I share your hesitancy. How hard would it be to extract the relevant info without tree-sitter? We do this for most compilers, though ad-hoc parsers have their downsides.

The problem is that, in the Cerberus Core output, annotations denoting the corresponding regions in the C code precede the Core language constructs (expressions, definitions, etc.) to which they apply. It is trivial to parse the annotation comments using simple regexps, but detecting the bounds of the Core expression (possibly spanning multiple lines) following the annotation requires deeper parsing of the language.

I am working on another patch that would allow the use of tree-sitter grammars for syntax highlighting and the use of existing tree-sitter grammars for other languages. Thus, the potential benefit of tree-sitter dependency would not be just for one language. It is a nice parsing framework that integrates well with both backend (Node.js) and frontend parsing (via wasm support and monaco-tree-sitter package). Tree-sitter-based parsers are also typically robust and suitable for real-time parsing and syntax highlighting while editing text in real time.

I understand your hesitancy to add another dependency. I am not familiar enough with the Node.js infrastructure, but perhaps it is possible to make this dependency optional, so it will only be required when the project is built with Cerberus compiler support?

From what I can tell, monaco-tree-sitter is very unmaintained and it would add a lot of complexity to attempt to bundle it on the client side (besides, that would be a second wasm build of tree sitter) https://github.com/Menci/monaco-tree-sitter

To clarify: I am not suggesting yet bringing in a monaco-tree-sitter. This is a discussion for another PR :) I just mentioned it as a potential future benefit.