After further debugging I have discovered that commenting out line 1251 in Sanitizer.php in the function cleanURL will make the problem go away. This is a call to preg_replace that is attempting to strip out invalid UTF-8 characters from a URL. For some reason the version of PCRE that I compiled with UTF-8 support enabled is not being used, in spite of the fact that I recompiled PHP to use it. So this is really a PHP problem and not a mediawiki problem. So off to the PHP support forums....
On Mon, 2 Apr 2007, Chuck Harding wrote:
An additional data point: one of our users discovered that by doubling the '/'s in the URL, it gets generated correctly: i.e., http:////www.harding-family.org It looks ugly but it works. I hope this provides more information so that this can be tracked down. I'd hate to have to edit every page that has an external link on it to change http(s)://whatever to http(s):////whatever....
On Thu, 29 Mar 2007, Chuck Harding wrote:
On Thu, 29 Mar 2007, Jim Wilson wrote:
That seems very strange - can you send us a snippet from a page (the wikitext) that's causing the problem?
-- Jim R. Wilson (jimbojw)
=== How to edit a Wiki page ===
: A useful link on How to edit a Wiki page can be found [http://en.wikipedia.org/wiki/Wikipedia:How_to_edit_a_page here]
gets rendered as:
<a name="How_to_edit_a_Wiki_page"></a><h3> <span class="mw-headline"> How to edit a Wiki page </span></h3>
<dl><dd> A useful link on How to edit a Wiki page can be found <a href="http:/wiki/Wikipedia:How_to_edit_a_page" class="external text" title="http:/wiki/Wikipedia:How_to_edit_a_page" rel="nofollow">here</a> </dd></dl>
notice that en.wikipedia.org is left out of the URL. The wierd thing is that when I click on the link on the rendered page, it actually does take me to a wikipedia article but not on how to edit a page, it takes me to the article on Wiki. Just to make sure it wasn't anything wierd about the URLs, here are some other messed up links:
==== Favorite links ====
:* [http://www.harding-family.org My family web pages] :* [http://k6ckt.home.comcast.net My Ham Radio web page] :* [http://nimrodvideo.home.comcast.net My Video/Multimedia Production Company web page] :* [http://kofc4588.home.comcast.net Knights of Columbus Fr. Patric Power Council #4588 web site] :* [http://www.arrl.org American Radio Relay League] :* [http://www.hello-radio.org Hello Ham Radio] :* [http://www.emergency-radio.org/ Emergancy Radio] :* [http://www.livermoreark.org '''L'''ivermore '''A'''mateur '''R'''adio '''K'''lub]
and the resulting HTML:
<a name="Favorite_links"></a><h4><span class="editsection">[<a href="/wiki/index.php?title=User:Charding&action=edit&section=1" title="Edit section: Favorite links">edit</a>]</span> <span class="mw-headline"> Favorite links </span></h4>
<dl><dd><ul><li> <a href="http:" class="external text" title="http:" rel="nofollow">My family web pages</a>
</li><li> <a href="http:" class="external text" title="http:" rel="nofollow">My Ham Radio web page</a> </li><li> <a href="http:" class="external text" title="http:" rel="nofollow">My Video/Multimedia Production Company web page</a> </li><li> <a href="http:" class="external text" title="http:" rel="nofollow">Knights of Columbus Fr. Patric Power Council #4588 web site</a> </li><li> <a href="http:" class="external text" title="http:" rel="nofollow">American Radio Relay League</a> </li><li> <a href="http:" class="external text" title="http:" rel="nofollow">Hello Ham Radio</a> </li><li> <a href="http:/" class="external text" title="http:/" rel="nofollow">Emergancy Radio</a>
</li><li> <a href="http:" class="external text" title="http:" rel="nofollow"><b>L</b>ivermore <b>A</b>mateur <b>R</b>adio <b>K</b>lub</a> </li></ul> </dd></dl>
Even though I upgraded mediawiki from 1.6.7 to 1.9.3, the rest of the supporting packages were not touched - apache, php, and mysql are all the same.
* MediaWiki: 1.9.3 * PHP: 5.1.2 (apache2handler) * MySQL: 5.0.18# httpd -v Server version: Apache/2.2.2 Server built: Jun 29 2006 12:13:15