Implementing Tatsu features (regex and AST generation with named nodes)

[badicsalex]

Hi all,

I'm currently migrating a project from Python to Rust. One main part is a non-trivial grammatical parsing of specific patterns in an otherwise natural language text. In Python, I used the Tatsu library with great success [sic], but I'm missing a few features from most Rust PEG parser libraries I've found:

• Regex (example)
• AST and Named captures: Tatsu can basically generate a namedtuple structure that corresponds to the captured AST. The main attraction there is you can give a name to your captures, so the AST manipulation is much more readable from the code.
• Left recursion: Depending on the parser structure, this can be as easy as modifying the memoization of left-recursive rules.

Since Tatsu is a very popular library, having an almost drop-in replacement in Rust would probably be a good thing for the community.

Currently I'm wondering whether I should send in pull requests to this project implementing these, or just roll my own PEG parser.

What do you think?

[badicsalex]

As a note, the pest AST "add-on" is almost good for the second point, all it needs is being able to name the structure fields.