Little Help With Regular Expressions...

[KevinPHPDude]

I'm new here, having a little trouble with a basic regular expression. I'm pretty sure I am coding it right, but its not doing what I'm expecting.

I'm trying to write some code to check that the user names are only alphanumeric characters.

Here is the code:

$userNames = array("shouldwork","should not work","shouldwork2","badusername!");

foreach($userNames as $x => $strval)
{
	$valid = ereg('[[:alnum:]]', $strval);
	if(!$valid)
		echo "Index: {$x} Value: {$strval} is an invalid username!<br/> \n";

else
{
	echo "Index: {$x} Value: {$strval} is a valid username!<br/> \n";
}
}

Here is the output I am getting:

Index: 0 Value: shouldwork is a valid username!
Index: 1 Value: should not work is a valid username!
Index: 2 Value: shouldwork2 is a valid username!
Index: 3 Value: badusername! is a valid username!

So its telling me they are all valid. I was only expecting index 0 and 2 to work, since index 1 has spaces and index 3 has a '!'.

I am getting the reg exp I am using from a book. According to the book this is how you do it. But the more I think about it, maybe this reg exp is only looking to if there are any alphanumerics and not if they are all alphanumerics?? Maybe the book is wrong.

[laserlight]

I'm trying to write some code to check that the user names are only alphanumeric characters.

Skip the regex and use [man]ctype_alnum/man instead.

If you do use regex, use PCRE regex with say [man]preg_match/man. POSIX style regex with ereg() and company are deprecated.

[KevinPHPDude]

laserlight wrote:

Skip the regex and use [man]ctype_alnum/man instead.

If you do use regex, use PCRE regex with say [man]preg_match/man. POSIX style regex with ereg() and company are deprecated.

Thanks for your reply.

I figured regex wasn't the best way to do this, but the chapter I was reading was on regex so I wanted to do a few samples to make sure I understood what I was reading.

Also according to the book I'm reading, PCRE is not multi-byte enabled and as such, POSIX is preferred. The book is a few years old though, so maybe its a little out of date?

As far as the regex I used above.. the only thing I can think is that the book is wrong. When I add '!@#$' to the userNames array it says it is invalid. So I think the regex is returning true if there are any alphanumerics and not making sure they are all alphanumerics.

[NogDog]

To the best of my knowledge, neither the ereg nor preg functions provide multi-byte support yet, basically because PHP itself does not provide native support for mutli-byte strings (they may in PHP 6, though). However, the preg functions have access to the "u" pattern modifier to treat pattern strings as UTF-8. Another advantage of the preg functions is that they are "binary safe", which can have security implications in certain situations.

For true multi-byte support you will need to use the mutli-byte string functions, which include a set of mb_ereg* functions.

[KevinPHPDude]

NogDog wrote:

To the best of my knowledge, neither the ereg nor preg functions provide multi-byte support yet, basically because PHP itself does not provide native support for mutli-byte strings (they may in PHP 6, though). However, the preg functions have access to the "u" pattern modifier to treat pattern strings as UTF-8. Another advantage of the preg functions is that they are "binary safe", which can have security implications in certain situations.

For true multi-byte support you will need to use the mutli-byte string functions, which include a set of mb_ereg* functions.

You're correct about mb_ereg(). I used the mbstring.func_overload option in the php.ini file so that when I used ereg() it is actually using mb_ereg().

As far as my original question goes... I can see how it would be easy to just use php's built in string functions to ensure that all of the chars are alpha numeric and not use reg exps at all. However, what if the problem was a bit more complicated? For instance you wanted to ensure that a phone number was formatted correctly such as (###)###-####. The simplest solution here would be to dump all the chars that are not numbers, and then make sure you have the correct amount of digits. However, I'd still like to understand how to use a reg exp to do this.

Is it common to use regex to verify things like this? Or am I wasting my time trying to understand reg exps?

Thanks!

[bpat1434]

Well, for the PCRE it would be:

~(?[\d]{3})?\s?[\d]{3}\s?-?\s?[\d]{4}~

although, that means any of these would be allowed:

1234567890
(123)4567890
[phone]
[phone]
(123) 456 - 7890

I think it might just be best to ask them in 3 separate boxes: Area-code, phone number. Then just do ctype_digit() to make sure they're all numbers, and there ya have 3 parts to put together in any format you please.

[KevinPHPDude]

bpat1434 wrote:

Well, for the PCRE it would be:

~(?[\d]{3})?\s?[\d]{3}\s?-?\s?[\d]{4}~

although, that means any of these would be allowed:

1234567890
(123)4567890
[phone]
[phone]
(123) 456 - 7890

I think it might just be best to ask them in 3 separate boxes: Area-code, phone number. Then just do ctype_digit() to make sure they're all numbers, and there ya have 3 parts to put together in any format you please.

Thanks. I'm not familiar with PCRE but I was able to figure it out. Do you always put the tilde or does it have a meaning?

I tried a simpler problem for just a 7 digit phone number using posix. Actually this another sample from the same book. Not sure if I need to buy a new book or if I am just missing something. Here is what I tried:

$phoneNumbers = array("755-7727","7453-7727", "75-57727", "753-77277");

foreach($phoneNumbers as $x => $strval)
{
	$valid = ereg('[0-9]{3,3}[-. ]?[0-9]{4,4}', $strval);
	if(!$valid)
		echo "Index: {$x} Value: {$strval} is an invalid phone number!<br/> \n";

else
{
	echo "Index: {$x} Value: {$strval} is a valid phone number!<br/> \n";
}
}

Here is the output:

Index: 0 Value: 755-7727 is a valid phone number!
Index: 1 Value: 7453-7727 is a valid phone number!
Index: 2 Value: 75-57727 is an invalid phone number!
Index: 3 Value: 753-77277 is a valid phone number!

I don't understand why index 1 and 3 are valid. Is my regexp wrong or do I have to write code to count the # of digits to work in conjunction with my regex?

[bpat1434]

With PCRE, you have to have a starting and stopping character. Typically you see them as "/"; however, they can be any character. So if you're validating a URL and you don't want to escape every "/" you will have in your regex, you can change the delimiters to a character that is less likely to be used (like ~ or @). I use tilde just because I rarely use the ~ inside a regex.

As for your regex, remove the ",3".... it's not needed. Just use [0-9]{3} since that will limit it to exactly 3. Same thing with the {4,4}.

[KevinPHPDude]

bpat1434 wrote:

With PCRE, you have to have a starting and stopping character. Typically you see them as "/"; however, they can be any character. So if you're validating a URL and you don't want to escape every "/" you will have in your regex, you can change the delimiters to a character that is less likely to be used (like ~ or @). I use tilde just because I rarely use the ~ inside a regex.

As for your regex, remove the ",3".... it's not needed. Just use [0-9]{3} since that will limit it to exactly 3. Same thing with the {4,4}.

That makes sense, but are you suggesting that should fix the problem? I tried it and got the same output.

[NogDog]

You need to anchor your regexp with the start ('^') and end ('$') of string assertions:

'^[0-9]{3}[-. ]?[0-9]{4}$'