[Validator] more strict e-mail validation regex

[xabbuh]

Q	A
Bug fix?	yes
New feature?	no
BC breaks?	no
Deprecations?	no
Tests pass?	yes
Fixed tickets	#14848
License	MIT
Doc PR

[nicolas-grekas]

👍
Btw, I'm wondering if we ever considered using the W3's definition of an email as specified in HTML5?
This is much simpler than the full RFC, and still is already enforced by browsers for those using the "email" input type:

/^[a-zA-Z0-9.!#$%&’*+\/=?^_`{|}~-]+@[a-zA-Z0-9-]+(?:\.[a-zA-Z0-9-]+)*$/

[Tobion]

You sure this is according to the standard? From what I read, the domain may be bracketed.
RFC 2822, section 3.4.1

domain = dot-atom / domain-literal / obs-domain
domain-literal = [CFWS] "[" *([FWS] dcontent) [FWS] "]" [CFWS]

[nicolas-grekas]

See http://www.w3.org/TR/html-markup/input.email.html#input.email.attrs.value.single

[linaori]

http://en.wikipedia.org/wiki/Email_address

admin@mailserver1 (local domain name with no TLD)

I'm not sure about the RFC, but it seems that the simple test should allow the localhost example.

[xabbuh]

Btw, I'm wondering if we ever considered using the W3's definition of an email as specified in HTML5?

Why not? But that should imo be considered a new feature and therefore be introduced in Symfony 2.8.

[xabbuh]

Actually, @iltar is right about localhost being a valid domain-part. The HTML5 pattern would accept that. Does everybody agree with switching to that regular expression then?

[fabpot]

We had a very lengthy discussion about this a few months ago, changing anything in the current regex would be a BC break.

[xabbuh]

@fabpot Do you mean even the change forbidding whitespaces in the domain part would be a BC break?

[fabpot]

Not necessarily. I just wanted to say that it would be very easy to break BC if we change the current regex we agree upon a while ago.

[nicolas-grekas]

I'd be -1 on making the current rx even more weak. i.e. removing the "contains dot" constraint. It has been fine for months.

[xabbuh]

I didn't mean to remove the check for the period in the domain part. I meant that the HTML5 pattern that you suggested also addresses the issue @iltar raised.

[xabbuh]

So, what do we do here now?

[nicolas-grekas]

I think this one is conservative enough to be merges asis
ping @symfony/deciders

[Tobion]

👍

[fabpot]

Thank you @xabbuh.