[PHP] Little Regular Expression help

[+dave164 Subscriber¹]

Hey guys,

This is it for the day, i promise :p

I have the current code for a regular expression and replacement

$patterns = '/(.*?)(http:\/\/)*(www\S*)/'; 			// replace urls with CLICK
$replacements = '$1<a href="http://$3" target="blank">(Click)</A>';
$cjmsg = preg_replace($patterns, $replacements, $cjmsg);

This regex matches either http://www.* or www.* but i need it to match http://* as well.

I guess this means just iradicating http://www.* and having http://* and www.* seperately, but i have no idea how to do this..

Thanks,

Garnett

[+theblazingangel MVC]

'/(.*?)(((http:\/\/)*(www\S*))|(www\S*))/'

the | character means OR!

[+dave164 Subscriber¹]

Thanks, but that throws up quite a few problems

www.* is fine

http://www.* makes http://http//www.(whateverelse)

http://* doesnt do anything

However changing it to

'/(.*?)(((http:\/\/)*(www\S*))|(http:\/\/\S*))/' does the same as above... apart from http://* catches it but only puts a / after http://

Thanks for trying, any more help would be appreciated

Thanks,

Garnett

[rson451]

just add an additional preg_replace

$cjmsg = preg_replace('(.*)"www(.*)', '$1"http://www$2', $cjmsg);

[HeyRatFans]

<?php

	$test_data = "Test 1: http://www.google.co.uk\n";
	$test_data.= "Test 2: http://maps.google.co.uk\n";
	$test_data.= "Test 3: http://google.co.uk\n";
	$test_data.= "Test 4: www.google.co.uk\n";
	$test_data.= "Test 5: maps.google.co.uk\n";

	echo preg_replace('/(\s|\()((http:\/\/([^:\s]+:?[^@\s]+@)?[_\.0-9a-z-]*(:\d+)?([\/?#]\S*[^),\.\s])?)|(www\.[_\.0-9a-z-]*(:\d+)?([\/?#]\S*[^),\.\s])?))/iu', '$1<a href="$2">(Click)</a>', $test_data);
?>

Matches everything but the last maps.google.co.uk, which is what you want it do, right?

[+dave164 Subscriber¹]

It only works when you've put text before it, can we make it so that works at any time?

I wish i understood what that was doing..

Edited October 9, 2008 by dave164

[HeyRatFans]

Indeed. You can simply prefix the string with a space before running it through preg_replace then remove it afterwards. If you put the whole lot in a function you can make it entirely transparent to the calling script:

<?php

function links($str)
{
	$str = ' '. $str;

	$str = preg_replace('/(\s|\()((http:\/\/([^:\s]+:?[^@\s]+@)?[_\.0-9a-z-]*(:\d+)?([\/?#]\S*[^),\.\s])?)|(www\.[_\.0-9a-z-]*(:\d+)?([\/?#]\S*[^),\.\s])?))/iu', '$1<a href="$2">(Click)</a>', $str);

	return substr($str, 1);
}

$test_data = "http://www.google.co.uk should link to Google\n";
$test_data.= "Google Maps is here: http://maps.google.co.uk\n";
$test_data.= "http://google.co.uk\n";
$test_data.= "www.google.co.uk is also Google\n";
$test_data.= "Finally, maps.google.co.uk doesn't link anywhere\n";

echo links($test_data);

?>

[+dave164 Subscriber¹]

Thanks, that works good, however if i put in jhhd www.google.co.uk then the link doesn't work, it puts in the link of my website followed by the link put in

e.g. http://calltoarms.net/forums/GarnettsC2ASi...ww.google.co.uk

[HeyRatFans]

Here's version 3 :)

Had to split the regular expression into two separate calls, but it still does the same thing.

<?php

function links($str)
{
	$str = ' '. $str;

	$str = preg_replace('/(\s|\()(http:\/\/([^:\s]+:?[^@\s]+@)?[_\.0-9a-z-]*(:\d+)?([\/?#]\S*[^),\.\s])?)/iu', '$1<a href="$2">$2</a>', $str);
	$str = preg_replace('/(\s|\()(www\.[_\.0-9a-z-]*(:\d+)?([\/?#]\S*[^),\.\s])?)/iu', '$1<a href="http://$2">$2</a>', $str);

	return substr($str, 1);
}

$test_data = "http://www.google.co.uk should link to Google\n";
$test_data.= "Google Maps is here: [url="http://maps.google.co.uk\n";"]http://maps.google.co.uk\n";[/url]
$test_data.= "http://google.co.uk\n";
$test_data.= "www.google.co.uk is also Google\n";
$test_data.= "While maps.google.co.uk doesn't link anywhere\n";

echo links($test_data);

?>

[+dave164 Subscriber¹]

Terrific that works brilliantly.

Anychance that you could be a legend and do a brief description of what it does =]

[HeyRatFans]

(\s|\() - Matches any whitespace characters or an open bracket in case of user putting URL in brackets

(http:\/\/ Matches http://

([^:\s]+:?[^@\s]+@)? Matches optional username:password@ portion

[_\.0-9a-z-]* Matches 0 or more of _ . 0-9 a-z and -

(:\d+)? Matches optional port number that follows the domain.

([\/?#]\S*[^),\.\s])?) Matches optional slash or hash (for named achors), followed by any non-white-space character, followed by any character that's not a comma, full-stop or white-space character in the URL. This is basically everything that appears after the domain name (i.e. the path and filename). It stops at full-stop or comma as these should technically be encoded if present in the URL. This entire match is also optional as noted by the question mark just before the last closing bracket.

The second expression is a bit simpler. It doesn't include the username:password bit and changes the main match so that it is prefixed by "www.", otherwise it is exactly the same.

[+dave164 Subscriber¹]

Thank you very much you've been a saint!

David