picture of Sebastian
Have you ever wondered where people go when they leave your site? With php3 it is easy to trace this. Sure, several solutions of redirect-cgi's exist, but why bother with another language when you can keep it straight to php3 and make it more comfortable?
My setting for writing this from scratch was a site where I had multiple pages with identical links in a part of them, but I'd wanted to know the use of the external links in detail: Who left where from which page and when. So here is how it works:
When you read the documentation about php3 in detail, you may have come across this kind of one-liner: <?php Header("Location: http://www.php.net"); exit; ?> You find it in Section XIV. It uses a HTTP redirect command to make the users browser access a new location. This is a good example how simple things in PHP3 can be: We will expand it, so that we make a suitable tool for the curious ;).
First we make targeting more flexible, it is the 'where'. This one is easy, because you just add a parameter to the URL of the redirector page. I'll call it <?php '$u_target' ?> . (u_ stands for URI/URL, this is a personal way to determine where the variable belongs to that I am using. Others: db_ is for database , p_ for page , f_ for function and so on.)
Then we want to record from 'which' page the link was activated. Since the referrer-information is sometimes filtered out by proxies, browsers and firewalls, we will use the additional variable <?php '$u_referrer' ?> for this purpose. You can use simple keywords like 'homepage', the name of the page example.html or the full address "http://...". It doesn't really matter, but keywords are better to understand on first look. When you skip this parameter, <?php $HTTP_REFERER ?> is used automatically. Note that if it doesn't exist either, <?php '$u_referrer' ?> will be set to 'none'.
We will determine the "who" by using server-variables. We'll take <?php $REMOTE_ADDR ?> address for this purpose, since Apache and IIS deliver it in nearly 90% of all installations ;). <?php $REMOTE_HOST ?> is better but since DNS-lookup is switched off sometimes, you'll have to try out on your server yourself. (Use <?php phpinfo() ?> for this purpose. It's tip I got from Tim and others which lead also to this script. Thanks ;)
'When' is easy: <?php time() ?> is all we need.
Logging the information gathered to a database is the method of choice. Since the data-structure is simple and flat, you should be able to use any kind of solutions - even plain text files will do it. Since MySQL is nice for this job, we'll take it for our purpose.
Since this article focuses on redirecting, I will keep the part of information exchange with the database brief. You'll find more information on this topic here in phpbuilder.com and on the web.
Here is the SQL-Command for creating the table in MySQL:
# Definition
# Table structure for table \'redirect\'
#
CREATE TABLE redirect (
  target varchar(80) DEFAULT \'no target\' NOT NULL,
  referrer varchar(80) DEFAULT \'none\' NOT NULL,
  client varchar(40) DEFAULT \'no client\' NOT NULL,
  timeused datetime DEFAULT \'0000-00-00 00:00:00\' NOT NULL
);
You can easily expand the table for additional parameters like Keyword for the link, User ID's and more. What follows is the code for the php3-Page: redir.php3

<?php
 
// Where did they go today? Version 1.02 - redir.php3
// (c) 1998 by smk_at_well.com

// Set variables for Database Access - you'll have to set up your own
    
$db_server "127.0.0.1";
    
$db_user "redirector";
    
$db_pass "secret";
    
$db_data "redir";
    
$db_table "redirect";


// First, check target
if (!isset($u_target)) 
    die(
"Sorry, no link was set in \$u_target");

// Second, is the referrer set? Note: HTTP_REFERER and some other variables 
//have to be written uppercase or they won't work.

if (!isset($u_referrer)) {
    if (!isset(
$HTTP_REFERER)) {
        
$u_referrer "none";
    } else {
        
$u_referrer $HTTP_REFERER;
    };
};

// Now the client address

if (!isset($REMOTE_ADDR)) {
    
$p_remotehost "no client address";
} else {
    
$p_remotehost $REMOTE_ADDR;
};

// And what time is it now?
$p_timeused date("Y-m-d H:i:s",time());

// Get connected and transfer data
@mysql_connect("$db_server","$db_user","$db_pass") or 
    die(
"Couldn't connect to the database-server!");
@
mysql_select_db("$db_data") or 
    die(
"Can connect to the server, but not to the table or database itself");
@
mysql_db_query($db_data "insert into $db_table values('$u_target','$u_referrer',
    '$p_remotehost','$p_timeused')"
);

// And now she/he's gone with the wind ;)

Header("Location: $u_target");
exit; 

?>
When you have set up the database and placed the page right you can use it in different ways:
First, you can use it for standard <a href="redir.php3?u_target=http://www.zet.net"> -commands. Second, you can use it in special forms for easy (and logged ;) navigation. Look at this example:
 
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">

<html>
<head>
    <title>Where did they go today?</title>
</head>
<body>

<form action="redir.php3" method="POST" target="">

<select name="u_target" size="1">
    <option value="http://www.slashdot.org">News for Nerds</option>
       <option value="http://www.phpbuilder.com">News for cool Coders</option>
       <option value="http://www.hotbot.com">If you can't find it there...</option>
       <option value="http://community.zet.net">This is my baby...</option>
</select>
<input type="Submit">
</form>

</body>
</html>
Third: If you fear of getting 'copied' away with all of you links on your site, you can expand the script, so that the parameter '$u_refferer' is looked up in a separate database and replaced by an address. (Like the one in html: 1 is for www.slashdot.org, 2 is for www.phpbuilder.com, and so on)
Finally: Mail me an other ideas you have! Tell me if you want this code expanded with user-id's, symbolic names for links and so on. I will put this script on an own webpage soon and if I have the time, I will enhance it.
Sebastian