[RESOLVED] Is this an e-mail validator?

[bjhymans]

Hey all, I am trying to mod some code that does a validation check on an e-mail address field on a form. I have been looking all over the script for Javascript that was doing it and couldn't find it, then I came across this

function is_email($email){
	$rBool=false;

if(preg_match("/[\w\.\-]+@\w+[\w\.\-]*?\.\w{1,4}/", $email)){
	$rBool=true;
}
return $rBool;
 

I must admit I can't decipher this. Is this an e-mail validation function? And if so can some one give me a run down on what it is doing?

What i would LIKE to do is configure the validation to only allow submissions from a specific domain...
i.e. [e-mail]

Thanx for the help!!

[bpat1434]

Yes, it is an email validation script, but a crude one at that.

It basically does the following:
[\w.-]+
Look for any word character ("\w") any dot (".") or any hyphen ("-") that is of length 1 or more (no limit to maximum length).
@
Next there must be an "@" symbol

\w+
Now there must be at least one character as the domain, at least one!
[\w.-]*?
Any other domain valid characters (words, dots and hyphens) for any length (0 - infinity) but this can be optional!! That's what the ? means, optional.

.\w{1,4}
Next is a dot, followed by a word that is 1 to 4 characters long (the TLD).

Honestly, you should NOT use that regex, and should use a little more sophisticated one like:

[-a-z0-9!#$%&\'*+/=?^_`{|}~]+([\.[-a-z0-9!#$%&\'*+/=?^_`{|}~]+)*@(([a-z0-9]([-a-z0-9]*[a-z0-9]+)?){1,63}\.)+([a-z0-9]([-a-z0-9]*[a-z0-9]+)?){2,63}

That will validate all proper email addresses with any valid address. Whereas the one you posted wouldn't take an email address with an underscore in it 🙁

Oh, and to validate at only a specific domain, just use this:

[-a-z0-9!#$%&\'*+/=?^_`{|}~]+([\.[-a-z0-9!#$%&\'*+/=?^_`{|}~]+)*@specificdomain\.com

[bjhymans]

Thank you bpat, that is exactly the info I was looking for!

[Weedpacket]

bpat1434 wrote:

[\w.-]?
Any other domain valid characters (words, dots and hyphens) for any length (0 - infinity) but this can be optional!! That's what the ? means, optional.

Not quite; when ? immediately follows a quantifier (, +, {n,m} or ?), then ? is the ungreedy modifer: instead of "as many consecutive characters as possible", the quantifier instead matches "as few consecutive characters as possible". So [\w] will prefer to match five characters if it can instead of two, but [\w]? will prefer to match two instead of five. As long as it means the regexp as a whole can succeed, of course.

Why ? was reused for this I don't know. Perhaps Larry Wall didn't want to require yet another character to be "special".

[bpat1434]

Ah, thanks for clearin that up Weedpacket. I am no expert at regex...

But really the initial regex was very lax and not very good because not all true email addresses would validate.

[Weedpacket]

True; it would have quite a few false negatives. It's got a few false positives as well: ".@1.1" for example.

I'd actually prefer taking the whole process in stages: use strpos to check for the presence of @, for example, and basically build up to RFC2822 requirements (now where did I leave that Ultimate Email Regexp...?)

[bpat1434]

lol... isn't that what I posted... found in the [man]eregi/man manual by bobocop in the user-notes?

[Weedpacket]

Well, I read you as observing that it would miss some valid addresses; I was noting that it would allow some invalid ones....

Weedpacket wrote:

(now where did I leave that Ultimate Email Regexp...?)

Ah, thought so.... OOps: too long for a post. Okay, trimming out routing information (thus only implementing the "localpart@domain-spec" part) and option inline comments/whitespace it becomes (ignoring line breaks):

/([^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff]+(?![^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff])|"[^\\\x80-\xff\r\n"]*
(\\[^\x80-\xff]\x20[^\\\x80-\xff\r\n"]*)*")
(\.
([^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff]+(?![^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff])|"[^\\\x80-\xff\r\n"]*
(\\[^\x80-\xff]\x20[^\\\x80-\xff\r\n"]*)*"))*@
([^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff]+(?![^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff])|\[
([^\\\x80-\xff\r\n\[\]]|\\[^\x80-\xff])*\])
(\.
([^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff]+(?![^(\x20)<>@,;:".\\\[\]\x00-\x19\x80-\xff])|\[
([^\\\x80-\xff\r\n\[\]]|\\[^\x80-\xff])*\]))*/

NOTE: I haven't tested this.

[bpat1434]

wow.... wow... wow.... what's the deal with the hex characters? can't you just use regular ones? like a-z 0-9 or what not? Or are the spans you're using characters from ~ to )?