Validating Email addresses is hard, I mean really really hard! If you wrote your own function without reading all the relevant RFCs, you got it wrong. If you copied a short function found on Internet, it’s wrong. If you’re relying on a regular expression to validate Emails, it’s probably wrong too.
A validator should never rejects a valid Email address. It’s the Robustness principle: Be conservative in what you send; be liberal in what you accept. It's okay to accept invalid addresses, but rejecting a single valid one is wrong: it frustrates legitimate users.
I tried to write an Email validator a few weeks ago. After 4 hours of fiddling with it, here’s what I have:
'@' in address[1:-1]
It means that an Email address is valid if there’s an 'AT' character between the first and the last character. Everything else is either too restrictive or too complex. The equivalent regular expression would be:
I don’t think I can have something even a little bit better without a lot more code. That’s because I tested my validation function against the test suite of isemail. The author read the RFCs, and implemented his own function. On March 7 2011, its tests contain 278 Email addresses, valid and invalid. It’s just one function, but the file weights 52KBytes and is almost 1200 lines longs.
Did you know that those Emails were valid?
"first@last"@iana.org "first\\last"@iana.org ""@iana.org (foo)cal(bar)@(baz)iamcal.com(quux) first.last@[IPv6:a1:]
I’m pretty sure that most websites requiring an Email address reject those addresses. It’s sad.