help with regular expression interpretation

[hq4ever (at) 012 (dot) net (dot) il]

function testemail($email) {

$validEmailExpr =
"^[0-9a-z~!#$%&_-]([.]?[0-9a-z~!#$%&_-])*@[0-9a-z~!#$%&_-]([.]?[0-9a-z~!#$%&_-])*$";

return eregi($validEmailExpr, $email);
}

$email = "fo*@bar.gov.mil";
testmail($email); //return TRUE

$email = "fo*****@bar.gov.mil";
testmail($email); //return TRUE

$email = "foo..ba*@bar.gov.mil";
testmail($email); //return FALSE - why ??

$email = "[e-mail]";
testmail($email); //return FALSE - why ??

as i understand it : (steps)

1. accept only 1 char of group [0-9a-z~!#$%&_-]
2. require 0 or more chars of group ([.]?[0-9a-z~!#$%&_-])*
-- why foo..bar is not valid input ?,
-- shouldn't it be ([.]?[0-9a-z~!#$%&_-]*)* ?

thank you for your help.

p.s Sorry if my previous post "can i get the public key of client
machine using php" didn't fit right into USENET, i am very very new here
so go easy on me :)

[zYm3N]

Dnia Sun, 01 Aug 2004 21:03:44 +0000, hq4ever (at) 012 (dot) net (dot) il
sp³odzi³(a):

function testemail($email) {

$validEmailExpr =
"^[0-9a-z~!#$%&_-]([.]?[0-9a-z~!#$%&_-])*@[0-9a-z~!#$%&_-]([.]?[0-9a-z~!#$%&_-])*$";

return eregi($validEmailExpr, $email);
}

$email = "fo*@bar.gov.mil";
testmail($email); //return TRUE

$email = "fo*****@bar.gov.mil";
testmail($email); //return TRUE

$email = "foo..ba*@bar.gov.mil";
testmail($email); //return FALSE - why ?? "]([.]?[0-9"

zero or one '.'
$email = "[e-mail]";
testmail($email); //return FALSE - why ??

last char should be one of [0-9a-z~!#$%&_-]..

Try to use:

eregi(""^[a-z0-9_]+@([a-z0-9_]+.)+[a-z]{2,}$", $email);
--
zYm3N[@interia.pl]

..:: C++ | C | PHP | HTML | Delphi | Pascal
..:: >> http://zymen.cjb.net <<
..:: http://zymen.cjb.net/cytowanie.html

[Ian.H]

On Sun, 01 Aug 2004 20:49:48 +0200, zYm3N wrote:
[ re: e-mail syntax validation ]

Try to use:

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

fo*****@my-domain.com
would fall over drunk in your example.

Regards,

Ian
PS: _ (underscore) is not a valid char within a domain name.

--
Ian.H
digiServ Network
London, UK
http://digiserv.net/

[maxim vexler]

for what input will this "^([.]?[0-9a-z~!#$%&_-])*$" regular expression
be TRUE ?

as i understand, it should accept at most one time '.' & then one
character, no? and so it can be repeated n times, something like
..f.o.o.b.a.r which should evaluate as TRUE, why is it then that a string
such as .my-domain.root still evaluates TRUE ?

[Ian.H]

On Mon, 02 Aug 2004 02:37:40 +0000, maxim vexler <hq4ever (at) 012 (dot)
net (dot) il> wrote:

for what input will this "^([.]?[0-9a-z~!#$%&_-])*$" regular expression
be TRUE ?

as i understand, it should accept at most one time '.' & then one
character, no? and so it can be repeated n times, something like
.f.o.o.b.a.r which should evaluate as TRUE, why is it then that a string
such as .my-domain.root still evaluates TRUE ?

Maxim,

Although not 100% fail-safe (nothing will be with checks such as this), I
use this function I wrote normally:
function validate_email($addy, $return_mx_records = false) {
if (empty($addy)) return false;

if (!preg_match(
'/^[a-zA-Z0-9&\'\.\-_\+]+\@[a-zA-Z0-9.-]+\.+[a-zA-Z]{2,6}$/',
$addy
)) {
return false;
}

$mx_exists = false;
$mx_records = array();
if (getmxrr(array_pop(explode('@', $addy)), $mx_records)) {
$mx_exists = true;
}

if ($mx_exists) {
return ($return_mx_records) ? $mx_records : true;
} else {
unset($mx_records);
return false;
}
}

USAGE:

if (!validate_email($_POST['email'])) {
$email_valid = false;
/* Do fail stuff */
}

HTH =)

Regards,

Ian
PS: Watch for line wrapping.

--
Ian.H
digiServ Network
London, UK
http://digiserv.net/

[Ian.H]

On Mon, 02 Aug 2004 02:47:31 +0000, Ian.H wrote:

if (!preg_match(
'/^[a-zA-Z0-9&\'\.\-_\+]+\@[a-zA-Z0-9.-]+\.+[a-zA-Z]{2,6}$/',
$addy
)) {
return false;
}

Oops, to fix the very issue you mentioned, change the regex to:
'/^[a-zA-Z0-9&\'\.\-_\+]+\@[^\.][a-zA-Z0-9.-]+\.+[a-zA-Z]{2,6}$/'

Regards,

Ian

--
Ian.H
digiServ Network
London, UK
http://digiserv.net/

[Tim Van Wassenhove]

In article <pa****************************@bubbleboy.digiserv .net>, Ian.H wrote:

On Mon, 02 Aug 2004 02:47:31 +0000, Ian.H wrote:

if (!preg_match(
'/^[a-zA-Z0-9&\'\.\-_\+]+\@[a-zA-Z0-9.-]+\.+[a-zA-Z]{2,6}$/',
$addy
)) {
return false;
}

Oops, to fix the very issue you mentioned, change the regex to:
'/^[a-zA-Z0-9&\'\.\-_\+]+\@[^\.][a-zA-Z0-9.-]+\.+[a-zA-Z]{2,6}$/'

If it doesn't allow _all_ valid e-mail addresses it's useless.

A quick search in this newsgroup will direct you to phpclasses or
http://www.ex-parrot.com/~pdw/Mail-RFC822-Address.html

--
Tim Van Wassenhove <http://home.mysth.be/~timvw>

[maxim vexler]

you have sure given me an answer about "Howto e-mail validation" for
what i thank you.

still, could you explain why the expression in my question didn't
returned what i expected it to return, just for my own knowledge :

for what input will this "^([.]?[0-9a-z~!#$%&_-])*$" regular expression
be TRUE ?

as i understand, it should accept at most one time '.' & then one
character, no? and so it can be repeated n times, something like
.f.o.o.b.a.r which should evaluate as TRUE, why is it then that a string
such as .my-domain.root still evaluates TRUE ?
thank you.

Tim Van Wassenhove wrote: In article <pa****************************@bubbleboy.digiserv .net>, Ian.H wrote:

On Mon, 02 Aug 2004 02:47:31 +0000, Ian.H wrote:

if (!preg_match(
'/^[a-zA-Z0-9&\'\.\-_\+]+\@[a-zA-Z0-9.-]+\.+[a-zA-Z]{2,6}$/',
$addy
)) {
return false;
}

Oops, to fix the very issue you mentioned, change the regex to:
'/^[a-zA-Z0-9&\'\.\-_\+]+\@[^\.][a-zA-Z0-9.-]+\.+[a-zA-Z]{2,6}$/'

If it doesn't allow _all_ valid e-mail addresses it's useless.

A quick search in this newsgroup will direct you to phpclasses or
http://www.ex-parrot.com/~pdw/Mail-RFC822-Address.html

[Ian.H]

On Mon, 02 Aug 2004 12:08:53 +0000, maxim vexler <hq4ever (at) 012 (dot)
net (dot) il> wrote:

still, could you explain why the expression in my question didn't
returned what i expected it to return, just for my own knowledge :

> for what input will this "^([.]?[0-9a-z~!#$%&_-])*$" regular expression
> be TRUE ?

This would allow all your chosen chars above (assuming some of those chars
didn't require escaping, can't think off the top of my head) but I believe
your biggest issue with the above (actually 2):
It tries to validate _starting_ with a . as the first char (for no . at
the beginning, you'd need [^\.] rather than [.] (the ^ in this case,
negates things so "if not ." but don't get that confused with the first ^
(as you have it as that's the start anchor)).

The other issue by the looks of it, '(...)*' the * here specifying zero or
more occurrances, so 'foo@.com' would slide through too.

Something else that you might find useful:
<http://weitz.de/regex-coach/>
I use it normally for more complex regexs, but it's great also for
learning regex too IMO.
HTH =)

Regards,

Ian

--
Ian.H
digiServ Network
London, UK
http://digiserv.net/

[Michael Fesser]

.oO(maxim vexler <hq4ever (at) 012 (dot) net (dot) il>)

for what input will this "^([.]?[0-9a-z~!#$%&_-])*$" regular expression
be TRUE ?
foo
.foo
foo.bar

but not

foo.
..foo
foo..bar

It accepts every string with chars of the class [0-9a-z~!#$%&_-], if
there's a dot it has to be followed by at least one other char. There
can't be a dot at the end or directly followed by another dot.
as i understand, it should accept at most one time '.' & then one
character, no? and so it can be repeated n times, something like
.f.o.o.b.a.r which should evaluate as TRUE, why is it then that a string
such as .my-domain.root still evaluates TRUE ?

The dot is optional:

[.]?

means zero or one chars of the class [.],

\.?

would do the same in this case.

So the pattern could be read like this: Any number (zero or more) of
chars of the given class, where each char _may_ be preceded by one dot.

BTW: You should use the preg* functions (PCRE) instead of the old ereg*
functions, they're faster and much more flexible.

HTH
Micha

[maxim vexler]

what a wonderful tool, thank you very much.

Ian.H wrote:

Something else that you might find useful:
<http://weitz.de/regex-coach/>
I use it normally for more complex regexs, but it's great also for
learning regex too IMO.
HTH =)

Regards,

Ian

[maxim vexler]

now i see, i was wrong in the way i thought regular expression was
evaluating the text. thank you for your comments.
why do you say preg* is better then ereg*, i've seen the same tip on
php.net but it didn't said any thing about why is it better ? - i know
there are different forms of regular expression (POSIX & Perl no?)
- is it the case with this functions?
- will the syntax be different ?
- do they evaluate differently ?

thank you, you have been great help to me.

Michael Fesser wrote:

.oO(maxim vexler <hq4ever (at) 012 (dot) net (dot) il>)

for what input will this "^([.]?[0-9a-z~!#$%&_-])*$" regular expression
be TRUE ?

foo
.foo
foo.bar

but not

foo.
..foo
foo..bar

It accepts every string with chars of the class [0-9a-z~!#$%&_-], if
there's a dot it has to be followed by at least one other char. There
can't be a dot at the end or directly followed by another dot.

as i understand, it should accept at most one time '.' & then one
character, no? and so it can be repeated n times, something like
.f.o.o.b.a.r which should evaluate as TRUE, why is it then that a string
such as .my-domain.root still evaluates TRUE ?

The dot is optional:

[.]?

means zero or one chars of the class [.],

\.?

would do the same in this case.

So the pattern could be read like this: Any number (zero or more) of
chars of the given class, where each char _may_ be preceded by one dot.

BTW: You should use the preg* functions (PCRE) instead of the old ereg*
functions, they're faster and much more flexible.

HTH
Micha

[Michael Fesser]

.oO(maxim vexler <hq4ever (at) 012 (dot) net (dot) il>)

why do you say preg* is better then ereg*, i've seen the same tip on
php.net but it didn't said any thing about why is it better ?
I haven't done a benchmark, but according to the manual they are faster
in most cases. Additionally you can do much more things with them, there
are many very useful extensions and features that are not possible with
the ereg* functions, for example different modifiers to fine-control the
pattern matching, named and conditional subpatterns, assertions ...

This allows much more flexible and sometimes more reliable pattern
matching.
- i know
there are different forms of regular expression (POSIX & Perl no?)
- is it the case with this functions?
ereg* is POSIX, preg* is PCRE (Perl Compatible Regular Expressions)
- will the syntax be different ?

Slightly. The basic syntax is the same. The main difference is that a
PCRE pattern has to be enclosed in delimiters, i.e. a single char before
and after the pattern. The ending delimiter may then be followed by some
modifiers.

Micha