Although it can be done in javascript, I'm interested to how this can be done in PHP...

Basically, before you output an html document, you buffer it and remove all links which point to the same html document. Anyone who has worked in usability will know what i am talking about.

The idea is that a page should never link to itself.

Anyone have an idea?

By the way, for anyone who doesn't know what i'm talking about, a quick google has brought up http://snook.ca/archives/javascript/clear_links_to_1/ which does it in javascript.

Which of the moments is your problem, buffering the page or replacing links?

Buffering the page is rather simple, rtfm ob_start() and corresponding functions.

Replacing links is much more complicated, especially if you do not know much about the format of links output. Since link to the current page can be presented in one of following forms:
- http://full.domain.name/some-folder/url-to-page.html
- /path-from/site-root/to-page.html
- relative-path/from-current/folder-root.html
- only-page.html
- ?page=2&something=else
the task becomes really hard, because you have first to represent all links into the same form, e.g. into fully qualified URL and then filter out the link to current page. In JS this can be done much easier because the browser's JS engine automatically represents all urls within the page into such form.

The problem is with replacing links (thanks for the excellent answer).

Representing the links into fully qualified URL's is not a problem... its identifying them and removing them which is.

Any ideas?

Okay, i think i found the solution...

http://www.php.net/manual/en/functio...lace.php#64213

combined with

http://www.php.net/manual/en/functio...e-callback.php

Right?

Yes, that could help

i have a method if that does not work

Sorry i didn't respond any earlier... i didn't get any notifications.

idkwot, before i try to remake the wheel, what's your method?

I would use PHP's DOM extension.
http://www.php.net/manual/en/ref.dom.php

You grab the page content with ob_* functions, load the html as a dom document, and then, you can parse it like you do in javascript, with a getElementsByTagName('a').

Looks like the most flexible way to go.

Any examples?

Readily made, no...
I just found out this extension some weeks ago, but never used it.

But it should allow to do the same as you can in javascript, with the same calls (or almost).
And I always find that if you can avoid regexp to do an action, you should avoid it, may it be just for the readability of the code by someone else.
If it's not at the cost of too much load, of course. You have to balance that.

Thanks.

I've had a go and googled around for other's... not looking good i'm afraid!

I think i'm going to stick to js for now.