Pego aqui un texto de http://mpwebwizard.com/free_stuff/use_a_mirror que
puede resultar útil preservar para disponer de él en el futuro:

---
Use a mirror for hits from popular websites
Sun, 12/17/2006 - 00:20 — martin

information: this tutorial is a little old (i wrote it a while ago, the
code is something i wrote while learning php a few years back so it's not
so good. for a better script that integrates this and a server load script
see this post on the forums.

updated 17/12/2006: i've modified the code to what Seldaek suggested in the
comments, the code is basically the same however is better formatted. his
code also fixes a bug if the referer is www.domain.com instead of
domain.com.

so you wrote a gripping article or made the a out-of-this-world graphic
which has been added to a site (or multiple sites) such as digg, slashdot
or reddit and now your server is suffering the consequences of that graphic
intensive make-over you did on your site last week. you wish they'd stop
slamming your server to stop your host bugging you, or, to stop your server
from committing suicide.

caching services such as coral content distribution network provide an
alternative for when your site goes down, but why wait until it gets to
that stage? why not just send all your hits from high-end websites to the
coral cache version? well, there isn't a good reason not to, so i'll show
you how to do it automatically.

i'm going to break the code down for beginners however if you just want to
get the code skip to the end.

information
the idea for this script is based on that of digg defender except that it's
not just a plugin for wordpress. even though i've written the script from
scratch I do believe that Elliott Back deserves credit for it.

license and download
this tutorial/script is licensed under GNU GPL Version 2. to download the
source please scroll to the end of the article and click the link under
attachments (only available to members).

step 1 - let's identify some popular websites and put them in an array!
the first step will be to identify some websites likely to cause some
damage and put them in to an array, this will allow us to add new sites
when one comes along. we're going to some of the social bookmarking,
networking sites as well as some other popular ones. we actually only need
the domain name and tld part of the address (eg. it would just be
mpwebwizard.com for http://mpwebwizard.com) as we will split the referring
address down to be sure it's from a listed site.

$sites = array(
'digg.com',
'slashdot.org',
'reddit.com',
'fark.com',
'somethingawful.com',
'kuro5hin.org',
'engadget.com',
'boingboing.net',
'del.icio.us',
'netscape.com'
);

this code is pretty self explanatory, however, for the record, if you wish
to add a new site then all you have to do is add the following code to the
end of the list. in this code you should make sure every line ends with a
comma except the last one. and, obviously replace example.com with the site
you wish to list as popular.

'example.com'

step 2 - is the referrer on the list?
we now have to find out if the http referrer is on the list of domains. to
do this we will use the php superglobal variable $_SERVER['HTTP_REFERER']
and the php functions parse_url() and strstr().

if (isset ($_SERVER['HTTP_REFERER']) && !strstr ($_SERVER
['HTTP_USER_AGENT'], 'CoralWebPrx')) {

$referer = parse_url ($_SERVER['HTTP_REFERER']);
$referer = $referer['host'];

foreach ($sites as $site) {
if(stristr ($referer['host'], $site)) {
...
}
}
}

the code isset ($_SERVER['HTTP_REFERER']) makes sure that there is a http
referer and that this isn't a direct request and the code !strstr ($_SERVER
['HTTP_USER_AGENT'], 'CoralWebPrx') makes sure that it isn't the coral
cache service requesting the page (if this wasn't included an infinite loop
would occur).

step 3 - redirect to coral cache
now we need to redirect to coral cache. the beauty of the coral network is
that the urls are nearly identical except for the fact that your domain
ends in nyud.net:8080. to do that we need to get the php host name and the
request uri, which are both stored in the php superglobal $_SERVER. the
code to to get the coral cache uri of the current page is:

'http://'. $_SERVER['HTTP_HOST'] .'.nyud.net:8080'. $_SERVER['REQUEST_URI']

So, all we do is add this address as a redirect in the if statement for the
previous code block and it works! the code to redirect using http headers
requires the php header() function.

header ('http://'. $_SERVER['HTTP_HOST'] .'.nyud.net:8080'. $_SERVER
['REQUEST_URI']);

step 4 - all the code, you just need to include!
overall, the code we've created in this tutorial should look like:

$sites = array(
'digg.com',
'slashdot.org',
'reddit.com',
'fark.com',
'somethingawful.com',
'kuro5hin.org',
'engadget.com',
'boingboing.net',
'del.icio.us',
'netscape.com'
);

if (isset ($_SERVER['HTTP_REFERER']) && !strstr ($_SERVER
['HTTP_USER_AGENT'], 'CoralWebPrx')) {

$referer = parse_url ($_SERVER['HTTP_REFERER']);
$referer = $referer['host'];

foreach ($sites as $site) {
if(stristr ($referer['host'], $site)) {
header ('Location: http://'. $_SERVER['HTTP_HOST']
..'.nyud.net:8080'. $_SERVER['REQUEST_URI']);
exit;
}
}
}

all you need to do is include it into the top of your website and it's done
(in between php tags). you must make sure that there is no output to the
browser before this code runs, ie. it is at the top of your php script.
mpwebwizard is not responsible for any damage caused by the script however
if you need any support please ask in the forums.


--
Lex Sparrow

Pagina sobre gestión de las news:
http://www.pclandia.net/lexsparrow/usenet/

Press Play Then Any Key
http://www.pressplaythenanykey.com/

FAQ de es.comp.lenguajes.php:
http://www.pclandia.net/escomplenguajesphp