Capture groups with zero repetition and search_slots behavior

[keith-hall]

Hello, quick question. I am working on adding support for subroutine calls to the fancy-regex crate, which delegates to regex-automata to do the heavy lifting.

fancy-regex aims to support Oniguruma syntax. Oniguruma doesn't support defining subroutines using conditional syntax, so it is common to define subroutines as capture groups with zero-repetition, such that the capture group is not executed immediately, and only invoked via subroutine call. i.e. (ABC){0}

Now the problem I have comes when using search_slots on the regex_automata::meta::regex. It panics if the Vec passed in has space for the capture group - which will never be populated because it is inside a zero repetition.

https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=36195dbc5f4283f3400ec084c1ec2fd2

It seems like in general, regex-automata doesn't mind when the Vec has room for more slots than is required - except when the pattern contains this zero-repetition capture group... As can be seen in the Rust Playground link above, it panics... So my question is: is it expected, or is it a bug?

thread 'main' (40) panicked at /playground/.cargo/registry/src/index.crates.io-1949cf8c6b5b557f/regex-automata-0.4.13/src/dfa/onepass.rs:2576:33:
range end index 4 out of range for slice of length 2

It is legal to pass any number of slots to this routine. If the regex engine would otherwise write a slot offset that doesn’t fit in the provided slice, then it is simply skipped. In general though, there are usually three slice lengths you might want to use:

[BurntSushi]

This looks like a bug yeah.

[keith-hall]

Thanks, I have filed it as #1327, I hope it is okay 🙂

[BurntSushi]

Yes thank you!