Length validators: Support all unicode characters

[DybekK]

Follow up to #3201. This PR adds optional support for counting unicode characters during validation.

[adamw]

There are some compilation failures. Some user-facing scaladoc on what the parameter means would also be nice.

Can you maybe also share what the specs (RFC) say about emojis as query/header/path values? Are they allowed, should they be encoded?

[kciesielski]

How about naming the PR "Length validators: Support all unicode characters", because "Add support for unicode characters" doesn't really say where exactly we're adding this support, and it suggest that unicode wasn't supported at all.

[DybekK]

The RFC 3986 specifies that URIs should consist of a limited set of characters: basic Latin alphabet letters, digits, and a few special characters. It doesn't explicitly mention emojis, but according to Section 2.1 of RFC 3986 any character outside the allowed set should be percent-encoded, meaning that emojis can be a part of URI syntax when they're properly encoded.

[kciesielski]

Emojis can be encoded, but then we get for example %F0%9F%98%8D to represent "😍" in an URI, so length of such text should equal 12, because for URI we treat it as length in ascii characters, right?

[adamw]

@DybekK thanks, I would put that in the docs - that the only scenario where this should be used is for body validators, where one may wish to count code points instead of characters (maybe giving an example of emojis ;) )

[adamw]

@kciesielski ah good point, we validate decoded strings. So this is also useful for queries etc., just depends on what you want to count.

[adamw]

Thanks!