KDL generator for fuzzing

[hkolbeck]

I wrote my first rust, a random valid KDL document generator for use in testing out parser implementations. It's very alpha-quality, but any reviews or feedback is quite welcome.

https://github.com/hkolbeck/kdl-gen

I think the first casualty is the spec itself. This rule in particular:

commented-block := '*/' | (multi-line-comment | '*' | '/' | [^*/]+) commented-block

The rule can resolve to * then / (or vice-versa), resulting in unbalanced multi-line comment boundaries.

There's probably a cleaner way, but this is the first thing that jumps into my head:

commented-block := '*/' | (multi-line-comment | '*[^/]' | '/[^*]' | [^*/]+) commented-block

If I'm missing something and this isn't a problem, let me know, otherwise I'll make an issue.

[Patitotective]

Looks nice, though, i think the best way to use it would be through GitHub actions, so rust and your library are installed automatically.
Looking forward to use it for https://github.com/Patitotective/kdl-nim ;]

[zkat]

I mean, won't it hit that first '*/' case before it hits a coincidental * followed by a /? I'm wondering if I'm thinking about it wrong, too, but I feel like this should be fine if the parser is going in the order specified.

[hkolbeck]

It's verging on a non-issue for sure. I don't think any parser will accept a document with mismatched boundaries, the issue is that the spec would technically say that such a document is legal KDL.

[zkat]

what I'm saying is that depending on how you interpret the grammar, it is, in fact, completely fine: The grammar was written with the assumption that clauses are checked in order, and with that assumption in place, you can't actually get unbalanced multiline comments.

[hkolbeck]

This has now found about 5 edge case bugs in https://github.com/hkolbeck/kdl4j. I've added options that I think will be useful for debugging, and I think this is more or less ready for usage by other parser authors.