Just wondering if anyone has any suggestions for detecting affiliate links with PHP. As an example, say that I'm creating a comments section for my website and I want to allow regular links in the comments, but not affiliate links. What would be the best way to go about that?

From an SEO perspective, it would probably be wise to 'no-follow' all user-submitted links anyways, but I would like to try to weed out the affiliates if I can as well.

I suspect that regular expressions could help, but I'm not familiar with all of the different affiliate link formats. Also, I would worry a little about false positives. I could probably also check the url against an array of common affiliate sites like clickbank, etc. However, using that method would require that a large list of affiliate sites be maintained.

I'll probably end up using a combination of the above, but I would love to get some more input if anyone else has ideas as well.

Most affiliate links I've seen just use url parameters. This might break user's abilities to post certain non-affliliate links, but you could just cut off everything after the ? in a url. By this I mean:

http://mydomain.com/index.php?affiliateRef=11111
//turns into
http://mydomain.com/index.php

All you would have to do is use explode to split the link.

PHP Code:

$link = 'http://mydomain.com/index.php?affiliateRef=11111';
$linkArray = explode('?', $link);
$newLink = $linkArray[0]; 


Hope that helps

Thanks, NullPointer. I did consider that, but there are so many non-affiliate sites that use url parameters that I didn't want to go down that route if I could help it.

I think NullPointer basically covered what you need to do. Personally I would use a preg_replace ( http://uk2.php.net/manual/en/function.preg-replace.php ) though, and possibly encode the URL's so that when they leave they go via a "Your now leaving my website" page.

there are so many ways to cloak affiliate links though. it would be impossible to catch them. many people use url shortening services, some have scripts running on their websites so that the url looks normal. it's more trouble than it's worth.