8 Practical PHP Regular Expressions - Page 3

kikloo · 04-25-2008, 07:03 AM

Hi,

PHP Code:


			
// Check Valid E-Mail

if (!eregi("^[_a-z0-9-]+(\.[_a-z0-9-]+)*@[a-z0-9-]+(\.[a-z0-9-]+)*(\.[a-z]{2,3})$", $_POST["email"])) { echo "Email Address is not valid."; }



// Check valid URL

if (!preg_match('/^(http|https):\/\/(([A-Z0-9][A-Z0-9_-]*)(\.[A-Z0-9][A-Z0-9_-]*)+)(:(\d+))?\//i', $_POST["url"])) { echo "Website URL is not valid."; }

Thanks.

delayedinsanity · 05-29-2008, 12:55 AM

Multiline regular expressions... is it best just to use the s modifier? For example, if I wanted to remove all PHP style comments from a file (no, I'm not trying to make a compression function like Kalle's, just playing around with different methods of templating), I'm currently using;

~/\*.*?\*/~s

What about without the s? I got this to work:

~/\*(.*?\r?\n?)+?\*/~

Can you think of a more efficient way?
-m

Geert · 05-31-2008, 11:24 AM

Your second regex is a nice experiment, however, my guess is that it is horribly inefficient because of the combination of (lazy) quantifiers.

Note that there is nothing wrong at all with using the s modifier.

In Friedl's book you'll read about a more efficient way. I believe it also takes into account situations like /* echo '*/'; */.

delayedinsanity · 05-31-2008, 02:31 PM

I reduced the lazyness a bit by changing it to this: ~/\*(.|[\r\n])*?\*/~ so now we're down to just one. I don't know much about lookahead and look behind assertations which may be incorporated in the friedl example if it's anything like his example for matching html style tags. Maybe you could post it here so we could dissect it?
-m

delayedinsanity · 05-31-2008, 03:14 PM

How about this? It doesn't count for nested comments, which means it'll still break using your haystack Geert, but it doesn't use any (I repeat any!) lazy operators. Based off an example I posted elsewhere yesterday, I've updated it - so we're looking for an opening and closing tag, but the in between is a little smarter than before where it just looked for anything and everything. It looks for any character except for another star, or a star but only if it's not the closing tag, or a new line.

PHP Code:


			
$szTest = <<<EOF
/*** this
is*** the
test **/
/* test*/
booga!

/* more
new lines
in this
comment
*/
EOF;

$szTest = preg_replace("~/\*([^*]|\*+[^*/]|[\r\n])*\*+/~", 'booga!', $szTest);
//$szTest = preg_replace("~/\*(.|[\r\n])*?\*/~", 'booga!', $szTest);
echo $szTest;

/\*([^*]|\*+[^*/]|[\r\n])*\*+/
/\*(.|[\r\n])*?\*/

...but which one is more efficient? The second one has a lazy operator, but it matches the same as the first one. I like the second one because it's more readable, but I'm prone to using things like the first one because it's less readable and people looking at my work may think I'm actually smarter than I really am.

Other than the ego, I also like the first one because it's job description is much more defined. Despite accomplishing the same task, it's doing what it's meant to do, whereas the second, although accomplishing the same task for now, has more ability to let something slide through that it shouldn't in the future. Or will it... hmmmmmmmmmmmmm.
-m

edit: Here's a hackish version that matches newlines without mentioning newlines: /\*[\w\W]*?\*/

xcasio · 07-31-2008, 09:47 AM

Hey,

Just a small tip. A perfectly valid e-mail address (example@example.museum) would not validate in this way due to the 4 character limit on the top level domain.