Question

I am trying get all html links within a string and replace them using preg_replace to another link (for link tracking etc)

It works fine on links like http://www.facebook.com but not those that do not have a ‘www’ within the string.
So the first URL would be fine, but the latter wouldn’t work – can anyone suggest how I alter my expression to allow BOTH links like this to work.

http://www.twitter.com
Posts by myusername

$message = preg_replace("/<a([^>]+)href=\"http\:\/\/([a-zA-Z0-9\-]+\.[a-zA-Z0-9]+\.[a-zA-Z]{2,3}(\/*)?)/", "<a$1href=\"http://www.site.com/system/link_tracker.php?URL=$2&ID={$ID}\"", $message);

Answer 1

http://([a-zA-Z0-9-]+.[a-zA-Z0-9]+.[a-zA-Z]{2,3}(/*)

This is by no means a URL regex. It might work for two or three cases, but you ignore the existence of:

• https,
• multiple labels in the domain name (foo15.cdn.amazon.com),
• dashes in the domain name,
• Internationalized domain names,
• TLDs existing of other than 2 or 3 letters (.info, .museum) or multiple parts (.co.uk) and most importantly
• deep links.

Users will especially not like the latter, because when they deeplink to some site, this link is rendered invalid using your regex. Use a common way, for example DOMDocument.replaceChild() to alter links in an HTML document (which I assume you use, since you’re capturing URL’s in <a> tags).

As said in this answer, that can be done with something like this, not tested:

$dom = new DOMDocument();  
$html = @$dom->load(...); // Load your html
$links = $dom->getElementsByTagName('a'); 

foreach ($links as $link)
{
    // Store original node
    $origLink = $link;

    // Find original href
    $href = $link->getAttribute('href');

    // Replace link
    $href = "http://www.site.com/system/link_tracker.php?URL=" . urlencode($href) . "&ID={$ID}";

    // Replace href
    $link->setAttribute('href', $href);

    // Replace child (don't know if this is required because you already alter $link)
    $dom->replaceChild($link, $origLink);

}