Supporting Internationalized Resource Identifier (RFC 3987)

[Xrampino]

URL validation right now fails with the widely-supported IRI (RFC-3987), for example https://www.example.com/michaël.

This library seems to support it, could it be possible to switch to this one for URL validation internally ? https://github.com/garycourt/uri-js

[WesTyler]

There's little chance that a new dependency will be added, since Joi is itself a dependency of hapi and those downstream deps are verrrry closely guarded.

That being said, that's not the only option for adding internationalization validation. If @Marsup gives the broad goal of supporting international characters in uri validation, I'll put this on my radar, right behind isemail; anyone else who wants to take a stab though has my full support! I am happy to help where I can.

[WesTyler]

Hm, actually, I think the decision whether or not to include IRIs by default in the validation comes down to the intended function of Joi.string().uri().

From RFC 3987 section 1.2.a -

in the HTTP protocol [RFC2616], the Request URI is defined
as a URI, which means that direct use of IRIs is not allowed in
HTTP requests

So, if Joi.string().uri() is being used to validate the endpoint for an HTTP request, IRIs should fail. But that's not the only use for URI validation.

[Xrampino]

I think they mean it should be normalized to punycode internally. But for the facing user https://www.examplé.com is still a valid domain, right?

Maybe we should talk about addingJoi.string().webAddress() to avoid confusion?

[WesTyler]

Admittedly I didn't have the chance to finish reading the RFC yet... :P

[DavidTPate]

@Xrampino This is where things become interesting because there's a number of different RFCs which apply to this. From what I've dug through you are correct that there are some specifications out there which do indeed support non-ASCII characters within the user's browser. These are then converted from Unicode to ASCII.

I agree with you that I don't think we should be modifying Joi.string().uri(). There are some different parsing mechanisms that are mentioned within the WhatWG work group for URL from what was specified in RFC-3986. These again are different from RFC-3987 which was also mentioned and handles IRIs.

[Marsup]

There hasn't been any demand for that in the past few years, nobody volunteered to implement it, and I don't see any joi extension doing it. Shall I close ?

[lock]

This thread has been automatically locked due to inactivity. Please open a new issue for related bugs or questions following the new issue template instructions.