feat(stdlib): add punycode encoding functions

[esensar]

This adds encode_punycode and decode_punycode functions. It also adds tests to confirm parse_url function behavior when it comes to punycode.

Fixes: #659

[esensar]

This adds punycode related functions. I have also realised that parse_url already handles that, so I have added a few tests to confirm it. The newly added crate for handling punycode was already a transitive dependency of the project (because of the use of url crate, and idna crate which handles punycode is a direct dependency of url and maintained in the same repository https://github.com/servo/rust-url).

[johnhtodd]

I haven't tested this patch, so forgive me if this lazy request could be answered that way: What happens if a string not containing UTF-8 characters is run through the encode_punycode process? Or a non-punycode string is run through decode_punycode?

If the strings remain un-mangled through both directions if they do not contain UTF-8, then the ability to detect if a string is UTF-8 or not is un-necessary since simple string evaluations can make that determination.

[esensar]

Right, I should have added some tests and examples to demonstrate that behavior. To answer your question, it will only encode whatever requires encoding, so for that reason the example (www.café.com) only encodes the café part.

I will add tests and examples with fully ASCII domains to make this more clear.

[johnhtodd]

Thanks - more test cases are always good. Very happy to see this feature. I suspect it will have a better chance of approval if it also had documentation in the main docs repository as part of the PR.

[pront]

Thank you @esensar, this looks good.

Can you also add a VRL test like this that exercises both encode and decode?

[jszwedko]

Thank you @esensar, this looks good.

Can you also add a VRL test like this that exercises both encode and decode?

Additionally, could we add these new functions to the benchmarks in benches/stdlib.rs?

[esensar]

Thank you @esensar, this looks good.
Can you also add a VRL test like this that exercises both encode and decode?

Additionally, could we add these new functions to the benchmarks in benches/stdlib.rs?

Yeah, I have added them now.

I have added both cases when there is nothing to encode/decode and with IDN strings, to confirm that cases when no encoding is needed are faster (an order of magnitude faster on my machine).

[esensar]

I have added VRL tests as well, demonstrating encoding, decoding and combination with parse_url as well. I realized that I have made the function infallible accidentally, so I have updated that too.

[pront]

Thank you for adding the tests!