8 Practical PHP Regular Expressions - Page 2

Geert · 12-09-2007, 10:19 AM

Great topic. I love talking about all the tiny details of regular expressions. They do matter!

Nobody has made a comment on the very first regex though.

Quote:

Originally Posted by Matt83

Validating a Username:

PHP Code:


			
$string = "userNaME4234432_";

if (preg_match('/^[a-z\d_]{4,28}$/i', $string)) { 

echo "example 1 successful.";

}

I'm telling you this username regex is flawed. Have a good look at it. Think. Test it. Then read on for the explanation.

The Problem

Okay, the often made mistake is to think that $ matches at the end of a string. The truth is that by default $ matches immediately before the final character if it is a newline.

This means that usernames containing a newline character at the end would pass. You wouldn't want that, right?

The Solution

Simply alter the meaning of $ by applying the D modifier to your regex. This will make it match only at the real end of the string. Alternatively, you could use the \z metacharacter.

The example below should make it all crystal clear, I hope.

PHP Code:


			
$username = "mr_newline\n";



preg_match('/^[a-z\d_]{4,28}$/i',  $username); // TRUE

preg_match('/^[a-z\d_]{4,28}$/iD', $username); // FALSE

preg_match('/^[a-z\d_]{4,28}\z/i', $username); // FALSE

Further Reading

Jay · 12-09-2007, 12:47 PM

Interesting, thanks for the link to that blog post!

Wildhoney · 12-09-2007, 02:17 PM

I was under the impression that the problem you mentioned above, Geert, was only applicable when regex was in multi-line - if you were to switch it over to single-line mode then the $ would match the end of the string and not before the new-line?

Geert · 12-09-2007, 02:42 PM

Quote:

Originally Posted by Wildhoney

I was under the impression that the problem you mentioned above, Geert, was only applicable when regex was in multi-line - if you were to switch it over to single-line mode then the $ would match the end of the string and not before the new-line?

If you switch to multiline mode, by applying the m modifier, the D modifier is completely ignored. So $ will keep on matching before newlines as it does by default. However, you still could use \z.

PHP Code:


			
$str = "line1\nline2\n";



echo preg_match('/^line2$/m',  $str); // TRUE

echo preg_match('/^line2$/mD', $str); // TRUE (D is ignored)

echo preg_match('/^line2\z/m', $str); // FALSE

webosb · 12-16-2007, 01:52 AM

Quote:

Originally Posted by Matt83

Postal Codes:

PHP Code:


			
$string = "55324-4324";

if (preg_match('/^[0-9]{5,5}([- ]?[0-9]{4,4})?$/', $string)) { 

echo "example 4 successful.";

}

Looks like validating postal codes only work for US Postal Codes. Austalaia has 4 digit zip codes and UK has letters ...Is there a way to make the Postal Code validation work internationally?

Geert · 12-16-2007, 07:36 AM

Quote:

Originally Posted by webosb

Is there a way to make the Postal Code validation work internationally?

No, not in one regex at least. What if you enter only digits (valid Australian zip) but you live in the UK? The regex can't know.

So, first make sure you know what country the user is from. Then load the corresponding regex to validate zip.

obolus · 01-17-2008, 07:32 PM

Here's a cheat sheet for Regular Expressions: Regular Expressions Cheat Sheet - Cheat Sheets - ILoveJackDaniels.com

Wildhoney · 01-17-2008, 07:58 PM

Fantastic cheat sheet. Thanks a lot!

Alan @ CIT · 01-17-2008, 08:13 PM

Regular expressions still make about as much sense to me as Japanese but thanks for the cheat sheets - bookmarked for future use

Alan

Wildhoney · 01-17-2008, 08:51 PM

Code:

Regex is really (easy|difficult)

!

TlcAndres · 01-17-2008, 08:53 PM

Quote:

Originally Posted by Alan @ CIT

Regular expressions still make about as much sense to me as Japanese but thanks for the cheat sheets - bookmarked for future use

Alan

Sometimes it's the same for me and I understand Japanese,So the headache I aquire are easy to imagine lol.

Kalle · 02-10-2008, 01:21 PM

Great cheat sheet ;)

Andrial12 · 02-19-2008, 03:53 PM

If your based in the UK and stuck for Postcode and Telephone formatting ideas.

Try these

php Code:

// POSTCODES - will work for every UK postcode including the very strange GIRO Bank one.
$pattern = '/(^gir\s0aa$)|(^[a-pr-uwyz]((\d{1,2})|([a-hk-y]\d{1,2})|(\d[a-hjks-uw])|([a-hk-y]\d[abehmnprv-y]))\s\d[abd-hjlnp-uw-z]{2}$)/i';
if (!preg_match($pattern, $postcode)) {
    // Display Error
}


// Valids all uk numbers including +44's or anything like that
$telephone = str_replace(" ","",$telephone);
$telephone = str_replace("+","",$telephone);
$pattern = '/^01[1-9]{1}|02[03489]{1}|05[0-9]|07[056789]{1}|08[047]{1}|441[1-9]{1}|442[03489]{1}|445[0-9]|447[056789]{1}|448[047]{1}$/Ui';
if (!preg_match($pattern, $telephone)) {
    // Display Error
} else {
    if(substr($telephone, 0, 2) == "44") {
        $telephone = "0" . substr($telephone, 3, strlen($telephone));
    }
    if(strlen($telephone < 10)) {
        // Display Error
    }   
}

ReSpawN · 02-19-2008, 06:11 PM

Andrial, next time usetags for your message. It looks fine non the less!

ReSpawN · 02-19-2008, 06:13 PM

Andrial, next time use php tags (between []) tags for your message. It looks fine non the less!

delayedinsanity · 03-19-2008, 03:31 AM

Forgive me if this is a stupid question, but I've been a long time away from regular expressions. If you wanted to check a username and make sure it was all alphanumeric, but didn't start with a number, would the following be right?

/^[a-z]+[a-z\d_]{3,27}$/iD

Edit: er, maybe this would make more sense?
/^[a-z]{1}[a-z\d_]{3,27}$/iD

xenon · 03-19-2008, 09:39 PM

Code:

/^[a-z][a-zA-Z0-9_]{3,27}$/

{1} may be omitted in this case since you only need one letter to validate the beginning of the string, so the repetition doesn't make any sense.

delayedinsanity · 03-19-2008, 10:48 PM

Thank you. Just so everybody knows, I did test that before asking, I was just curious if it was the most efficient method. In that line of thinking, I also noticed you changed \d to 0-9 .. isn't it the same thing or is one way better than the other?
-m

xenon · 03-19-2008, 11:13 PM

I just used 0-9 instead of \d purely out of habit (and for consistency, I might add). \d and [0-9] are the same thing. Doesn't really matter which one of them you're using. I think the regex engine would have to first find the character group related to \d before matching, but that's fantasy already.