Validate well formed email

**freakysid** · Jul 16, 2001 16:23

At the risk of upsetting sensitive penguins, I'm going to once again raise the issue of validating well formed email addresses. I've tried several regular expressions (some that I have found from going over old posts here). None of them seem to work for me.

I want to also test that the email ends in either a two letter country suffix or a three letter top level domain suffix.

Here is what I have tried, which doesn't seem to work for me:

eregi("^[a-z0-9\._-]+@[a-z0-9\._-]+\.+[a-z]{2,3}$", $email)

I have tried another couple of expression as well but I'm too confused right this minute to work out what they are

**Anarchos** · Jul 16, 2001 16:30

The - needs to be at the beginning.

**aspen** · Jul 16, 2001 17:00

Instead of doing a-z0-9 why not do [^]?

some people do have emails like this

peter.smith@domain.com

Anyways...

[^] = any number of characters

[^@] = any number of characters not an @

**Anarchos** · Jul 16, 2001 17:12

He's not using ^ because he only wants certain characters. Yes, some people do have emails like peter.smith@domain.com, and his script will work correctly with it.

**freddydoesphp** · Jul 16, 2001 17:18

I have been working on this same thing for awhile now, I mean I have used some of the expressions out there and they work for most cases, but I have yet to come across one that will fit all cases. So here is what I propose, we all put some email address down here and then we can try to come up with something that works for all scenarios. I think if we all contribute we can cover most cases.

john@domain.com
john.smith@domain.com
john.smith@domain.netcom.com
john.smith@domain.co.uk
john.smith@domain.co.au

What are some others that need to work?

**Anarchos** · Jul 16, 2001 17:22

Those will all work with freakysid's regex...

**freddydoesphp** · Jul 16, 2001 17:24

Anarchos thats my point, obviously freakysid is having trouble with some email addresses. So I thought if we all put down some variations we can come up with, then we can possibly write a better expression to cover all the possibilities, right?

**Anarchos** · Jul 16, 2001 17:25

Yeah I see your point, but the reason it wasn't working is because he didn't put the dash at the beginning. Otherwise, it should work for all valid email addresses.

**freakysid** · Jul 16, 2001 17:46

I think the thing that is not working at the moment is my brain, due to lack of sleep.

Here is another example not in the list (essentially it is all the domains of countries which do not use a "com" or "co" etc with their two letter country code:

eg:

john.smith@domain.to
smith@domain.to

**freakysid** · Jul 16, 2001 18:07

Thank you everyone!

Here is my validation function

PHP Code:


function validateWellFormedEmail($email) {

   return eregi("^[-a-z0-9\\._]+@[-a-z0-9\\._]+\\.+[a-z]{2,3}$", $email);

}

The following test cases all pass. Woo-hoo!

john@domain.com
john.smith@domain.com
john.smith@domain.netcom.com
john.smith@domain.co.uk
john.smith@domain.com.au
john.smith@domain.to
smith@domain.to
smith@subdomain.domain.to

I would just like to have my belief confirmed that an email address cannot be case sensitive. Is this right? for example,

Joe.Jones@foo.bar

is equivelent to

joe.jones@foo.bar

**DR_LaRRY_PEpPeR** · Jul 16, 2001 18:14

i notice there's no backslashes in front of the periods but i guess that's just stupid VB.

i wanted to ask about this part though.

\.+[a-z]{2,3}

why the + after the . if you only want ONE dot matched? the way it is now, it would match john@domain....com

see?

**DR_LaRRY_PEpPeR** · Jul 16, 2001 18:19

Originally posted by freakysid
I would just like to have my belief confirmed that an email address cannot be case sensitive. Is this right? for example,

Joe.Jones@foo.bar

is equivelent to

joe.jones@foo.bar

AFAIK, yes. i GUESS it's possible that certain mail servers could be configured to be case sensitive, but i've never heard of it. and i think it would be silly for one to be case sensitive. (i.e. non comp savvy people could type it wrong and it wouldn't work, etc.)

**NZ Joe** · Jul 16, 2001 18:19

I don't know about much regular expressions so this may be a stupid question, but does this work for underscores?
john_smith@domain.tld

**freakysid** · Jul 16, 2001 18:24

All these pass:

john_smith@domain.tld
_john_smith@domain.tld
john_smith@_domain.tld

Should the last case have passed? Is _domain.tld a valid domain name?

(looking at my post, vB doesn't seem to think so)

**NZ Joe** · Jul 16, 2001 18:27

_domain.tld is not valid, however I'm not sure about _subdomain.domain.tld

**freakysid** · Jul 16, 2001 18:40

Originally posted by DR_LaRRY_PEpPeR
i notice there's no backslashes in front of the periods but i guess that's just stupid VB.

i wanted to ask about this part though.

\.+[a-z]{2,3}

why the + after the . if you only want ONE dot matched? the way it is now, it would match john@domain....com

see?

That is correct john@domain....com passed when it should have failed.

So now we have two test cases which produce the wrong result:
john@domain....com
john_smith@_domain.tld

and we are not sure about whether this should pass or fail:
smith@_subdomain.domain.tld

Please help me out here. I am a pathetic copy-and-paste regex hacker. I have done quite a bit of searching on this and I am suprised that there does not seem to be a set of robust validation regex (that is, if it requires more than one) to test for a well formed email. Well, I have not found it yet

**padders** · Jul 16, 2001 19:28

to complicate it further, my email as uni is

christopher.padfield@new.ox.ac.uk

so it has 4 sections after the @ sign

**Anarchos** · Jul 16, 2001 20:12

The 4 sections doesn't matter, it'll still parse fine using the initial regex, and I don't see why anyone has a problem with that one. Regex validation cannot be used for anything more than user error, and the easiest way to do this is to either send a confirmation email or use a confirmation email input.

**freakysid** · Jul 17, 2001 00:46

Anarchos, I agree fully that regex cannot fill in my tax return or mow the lawn. In fact, I do send an email to new members and if the email bounces, Sendmail pipes the bounced email to a script which deals with matters from there. However, the audience of the site I am developing for cannot be assumed to know how to tie their shoe-laces. So, as much as possible I want to validate the user input while I have the opportunity of having the user re-enter bad data.

If you look at my last post above, you will see that vBulletin does not recognise the three emails I list as being valid email addresses. So vBulletin knows how to do this

**NZ Joe** · Jul 17, 2001 02:05

Found this in my travels today. It may be of use.
http://phpclasses.upperdesign.com/br...tml/package/13

**freakysid** · Jul 17, 2001 03:45

Thanks NZ Joe. I busted open that class and found the regex, which is below (regexB)

So now I have tested two regex

regexA:
"^[-a-z0-9\._]+@[-a-z0-9\._]+\.+[a-z]{2,3}$"

regexB:
"^([-!#\$%&'*+./0-9=?A-Z^_`a-z{|}~ ])+@([-!#\$%&'*+/0-9=?A-Z^_`a-z{|}~ ]+\\.)+[a-zA-Z]{2,4}\$"

Here is the test code (I just passed the value of $e in the URL)

PHP Code:


<?= eregi("^[-a-z0-9\\._]+@[-a-z0-9\\._]+\\.+[a-z]{2,3}$", $email) ? 'TRUE' : 'FALSE'; ?>

<br>

<?= eregi("^([-!#\$%&'*+./0-9=?A-Z^_`a-z{|}~ ])+@([-!#\$%&'*+/0-9=?A-Z^_`a-z{|}~ ]+\\.)+[a-zA-Z]{2,4}\$", $email) ? 'TRUE' : 'FALSE'; ?>

Here are the test cases I put them both through.

Test Cases (expected result: TRUE)
1. john@domain.com
2. john.smith@domain.com
3. john.smith@domain.netcom.com
4. john.smith@domain.co.uk
5. john.smith@domain.com.au
6. john.smith@domain.to
7. smith@subdomain.domain.to
8. smith@domain.to
9. john@some-domain.com
10. john_smith@some_domain.com
11.john@domain_42.com
12.christopher.padfield@new.ox.ac.uk

Test Cases (expected result: FALSE)

13.john@domain....com
14. john_smith@_domain.tld

Test Case (expected result unknown)
15. smith@_subdomain.domain.tld

Results regexA
Cases 1-12 all passed.
Cases 13 & 14 failed.
Case 15 produced a result of TRUE

Results regexB
Cases 1 - 13 passed.
Case 14 failed.
Case 15 produced a result of TRUE

So it seems that there is still an opportunity for a wise head to still make improvements to regexB.
Don't forget that the validation does not have to be achieved with just the one regex.
I can use multiple regex.

**MattR** · Jul 17, 2001 06:57

I have a regexp book which lists a 'failproof' regexp for e-mail addresses which is, oh, 600 characters long. If I have the time I will scan it in

and let you type it.

**techie2003** · Jul 17, 2001 07:06

freakysid: try this:

jng03@bls.boston.k12.ma.us

that's 5 i believe
most school addresses are like that in the US if they're using the ones provided in the city.

**freakysid** · Jul 17, 2001 11:09

Originally posted by MattR
I have a regexp book which lists a 'failproof' regexp for e-mail addresses which is, oh, 600 characters long. If I have the time I will scan it in and let you type it.

Perhaps they have a web site where code in the book is available for dl?

BTW, techie2003 jng03@bls.boston.k12.ma.us passes