Would be nice if this library could expose intermediate methods for working with the parse tree -- in particular text offsets for model items (including malformed) and comments.

This would allow an API user to edit model content while maintaining a YAML source in synch bidirectionally, with comments and unrelated whitespace preserved. Also if YAML is malformed in places, the errors could be limited to those sections while nicely handling other sections. snakeyaml is a nice implementation of this (in Java).

I appreciate that this is a significant addition to the scope of this project but I think this is the most complete JS YAML implementation so I'm interested if it's on the roadmap or all or if not whether there might be appetite for it.