RegEx expression to find a href links and add NoFollow to them - regex

Question

I am trying to write a RegEx rule to find all a href HTML links on my webpage and add a 'rel="nofollow"' to them.
However, I have a list of URLs that must be excluded (for exmaple, ANY (wildcards) internal link (eg. pokerdiy.com) - so that any internal link that has my domain name in is excluded from this. I want to be able to specify exact URLs in the exclude list too - for example - http://www.example.com/link.aspx)
Here is what I have so far which is not working:
(]+)(href="http://.*?(?!(pokerdiy))[^>]+>)
If you need more background/info you can see the full thread and requirements here (skip the top part to get to the meat):
http://www.snapsis.com/Support/tabid/601/aff/9/aft/13117/afv/topic/afpgj/1/Default.aspx#14737

Answer 1

An improvement to James' regex:
(<a\s*(?!.*\brel=)[^>]*)(href="https?://)((?!(?:(?:www\.)?'.implode('|(?:www\.)?', $follow_list).'))[^"]+)"((?!.*\brel=)[^>]*)(?:[^>]*)>
This regex will matches links NOT in the string array $follow_list. The strings don't need a leading 'www'. :)
The advantage is that this regex will preserve other arguments in the tag (like target, style, title...). If a rel argument already exists in the tag, the regex will NOT match, so you can force follows on urls not in $follow_list
Replace the with:
$1$2$3"$4 rel="nofollow">
Full example (PHP):
function dont_follow_links( $html ) {
// follow these websites only!
$follow_list = array(
'google.com',
'mypage.com',
'otherpage.com',
);
return preg_replace(
'%(<a\s*(?!.*\brel=)[^>]*)(href="https?://)((?!(?:(?:www\.)?'.implode('|(?:www\.)?', $follow_list).'))[^"]+)"((?!.*\brel=)[^>]*)(?:[^>]*)>%',
'$1$2$3"$4 rel="nofollow">',
$html);
}
If you want to overwrite rel no matter what, I would use a preg_replace_callback approach where in the callback the rel attribute is replaced separately:
$subject = preg_replace_callback('%(<a\s*[^>]*href="https?://(?:(?!(?:(?:www\.)?'.implode('|(?:www\.)?', $follow_list).'))[^"]+)"[^>]*)>%', function($m) {
return preg_replace('%\srel\s*=\s*(["\'])(?:(?!\1).)*\1(\s|$)%', ' ', $m[1]).' rel="nofollow">';
}, $subject);

Answer 2

I've developed a slightly more robust version that can detect whether the anchor tag already has "rel=" in it, therefore not duplicating attributes.
(<a\s*(?!.*\brel=)[^>]*)(href="https?://)((?!blog.bandit.co.nz)[^"]+)"([^>]*)>
Matches
Google
<a title="Google" href="http://google.com">Google</a>
<a target="_blank" href="http://google.com">Google</a>
Google
But doesn't match
<a rel="nofollow" href="http://google.com">Google</a>
Google
Google
Google
Google
<a target="_blank" href="http://blog.bandit.co.nz">Bandit</a>
Replace using
$1$2$3"$4 rel="nofollow">
Hope this helps someone!
James

Answer 3

(<a href="https?://)((?:(?!\b(pokerdiy.com|www\.example\.com/link\.aspx)\b)[^"])+)"
would match the first part of any link that starts with http:// or https:// and doesn't contain pokerdiy.com or www.example.com/link.aspx anywhere in the href attribute. Replace that by
\1\2" rel="nofollow"
If a rel="nofollow" is already present, you'll end up with two of these. And of course, relative links or other protocols like ftp:// etc. won't be matched at all.
Explanation:
(?!\b(foo|bar)\b)[^"] matches any non-" character unless it it possible to match foo or bar at the current location. The \bs are there to make sure we don't accidentally trigger on rebar or foonly.
This whole contruct is repeated ((?: ... )+), and whatever is matched is preserved in backreference \2.
Since the next token to be matched is a ", the entire regex fails if the attribute contains foo or bar anywhere.