<?php
// Backlinks script developped by Stephanie Booth - http://climbtothestars.org/ - July 2002
// http://climbtothestars.org/coding/backlinks/
// Feel free to modify and redistribute as long as no money is made and appropriate credit is given.
// Many thanks to all those who contributed to this script by answering my questions!


// grab the referrer and the current page name
$current_referrer=$HTTP_REFERER;
$current_ip=$REMOTE_ADDR;
$current_page=$REQUEST_URI;


// if the referrer is a search engine or anything uninteresting, forget about it - google gets special treatment
// ################# add your domain in this array (e.g. "climbtothestars") #####################

$uninteresting=array("search","daypop","mamma","extremetracking","suche","recherche","search1-2","askjeeves","aol","topsearcher","looksmart3","vachercher","radaruol","nedstatbasic","overture","szukaj","clickthru","hotbot","yaccs","bonzi","sitemeter","yahoo","mail","socnet","antibot","websearch","aolsearch","whoisreport","alltheweb","altavista","toile","mail","f2mail","excite");

$parsed_uri=parse_url($current_referrer);
$uri_parts=explode("/", $current_referrer);
$host_bits=explode(".", $uri_parts[2]);

$file_name=array_pop($uri_parts);

// ignore non http referrers (e.g. referrer blockers, etc)
if($uri_parts[0]!=="http:")
{
	$current_referrer="";
}

// bits of file names we want to avoid (referrer links, other search links)
$bad_file_names=array("referrer", "referer", "search", "linkrd");

// pick up google search terms
if(in_array("google", $host_bits))
{
	preg_match('!q=([^&]*)!', $current_referrer, $searches);
	$search=$searches[1];
	
	// decode url
	$search=urldecode($search);
	
	// this is for french characters etc... needs to be completed as it goes
	$nasty_stuff=array("Ã¨","Ã","Ã©","à©");
	$good_stuff=array("è","à","é","é");
	$search=str_replace($nasty_stuff, $good_stuff, $search);
	$current_referrer=strtolower($search);
}


// fix wiki alias
if(in_array("allmyfaqs", $host_bits))
{
	$current_referrer=str_replace("cgi-bin/wiki.pl", "faq.pl", $current_referrer);
}

// ignore blacklisted file names
foreach($bad_file_names as $key => $needle)
{
	if(stristr($file_name, $needle))
	{
		$current_referrer="";
	}
}

// ignore blacklisted domains
foreach($uninteresting as $key => $needle)
{
	if(in_array($needle, $host_bits)||in_array($needle, $uri_parts))
	{
		$current_referrer="";
	}
}

// remove index page names and any mess after them ### not good because of MT and dynamic pages ### (add .* at the end of the regex if this leaves too much mess)
$current_referrer=eregi_replace("(index|default|home)\.(php|php3|htm|html|asp|cfm|shtml)", "", $current_referrer);


// get rid of any stuff in the REQUEST_URI
// $current_page=eregi_replace("\?.*", "", $current_page);


// ################# enter your db info here ###################
// connect to database
$connect = mysql_pconnect("", "user", "pass")
		or die("Could not connect to mysql server.");
	mysql_select_db("db_name", $connect)
		or die ("Could not connect to database.");


// check this isn't a reload or click-through abuse (if the last referral through this link came from the same ip, ignore it)
	// pick out dates and ips of previous entries for this referrer
	$query="select ip, datetime from referrers where referrer='$current_referrer'";
	$request=mysql_query($query) or die("Testing: problem with $query");
	$previous_datetime="0";
	
	// check each date stamp against the previous ones to pick out the ip associated with the lastest referral
	while($row=mysql_fetch_array($request))
	{
		$ip=$row["ip"];
		$datetime=strtotime($row["datetime"]);
		if($datetime > $previous_datetime)
		{
			$last_ip=$ip;
			$previous_datetime=$datetime;
		}
	}
	// if the latest ip for this referrer is the same as the current one, trash this referrer as duplicate
	if($last_ip==$current_ip)
	{
		$current_referrer="";
	}


// what follows is identical, whether www or none


	// get rid of the www if possible, and grab the title of the page

if($fp=@fopen($current_referrer, "r"))
{
	while (!feof ($fp)) 
	{
    $line = fgets ($fp, 1024);
    /* This only works if the title and its tags are on one line */
    if(eregi ("<title>(.*)</title>", $line, $out)) 
		{
      $title_long = $out[1];
			// print("<p>Got the title for the long referrer, yay! $title_long</p>");
			$referrer_title=$title_long;
		}
		$counter_long++;
	}
	
	// if there is no www, $short_referrer will be the same as $current_referrer
	$short_referrer=str_replace("www.", "", $current_referrer);
	
	if($short_referrer!==$current_referrer)
	{
		// if short_referrer and current referrer are identical, it's useless to do all this again
		if($fp_short=@fopen($short_referrer, "r"))
		{
				while (!feof ($fp_short)) 
				{
    			$line = fgets ($fp_short, 1024);
  	  		/* This only works if the title and its tags are on one line */
	    		if(eregi ("<title>(.*)</title>", $line, $out))
					{
       		 	$title_short = $out[1];
						// print("<p>Got the title for the short referrer, yay! $title_short</p>");
					}
					$counter_short++;
				}
				
				if(($counter_long==$counter_short)&&($title_long==$title_short))
				{
					// we can do away with the www safely!!
					$use_www=false;				
				}
				else
				{
					// print("<p>$short_referrer exists, but titles ($title_long, $title_short) or number of lines ($counter_long, $counter_short) do not match. Use www!</p>");
					$use_www=true;
				}
		}
		else
		{
			// print("<p>$short_referrer does not exist.</p>");
			$use_www=true;
		}
	}
	else
	{
		// print("<p>There wasn't even a www in the initial url!</p>");
		$use_www=false;
	}
}
else
{
	// print("<p>Initial uri $current_referrer does not exist.</p>");
	// this gets rid of 404s and non-public pages
	$current_referrer="";
}

// different behaviour, depending on if we need to use www or not

if($use_www)
{
	// we need to use the www
	
	// clean up database
	$query_ref="update referrers set referrer='$current_referrer' where referrer='$short_referrer'";
	$query_tit="update titles set referrer='$current_referrer' where referrer='$short_referrer'";
	mysql_query($query_ref) or die("Problem updating referrer table.");
	mysql_query($query_tit) or die("Problem updating titles table.");
}
else
{
	// we can do away with www
	
	// clean up database
	$query_ref="update referrers set referrer='$short_referrer' where referrer='$current_referrer'";
	$query_tit="update titles set referrer='$short_referrer' where referrer='$current_referrer'";
	mysql_query($query_ref) or die("Problem updating referrer table.");
	mysql_query($query_tit) or die("Problem updating titles table.");
	
	$current_referrer=$short_referrer;	
}


if(!empty($current_referrer))
{
	// we don't want to do this if we are dealing with google search terms
	if(!isset($search))
	{
		$query="select title from titles where referrer='$current_referrer'";
		$result=mysql_query($query) or die("Problem with $query.");
		if(mysql_num_rows($result)==0)
		{
			// ||mysql_result($result, 0)=="" stick the title of the referring page in the db, as it isn't there yet

			$referrer_title=str_replace("'", "\\'", $referrer_title);
			$query="insert into titles (referrer, title) values ('$current_referrer', '$referrer_title')";
			mysql_query($query) or die("problem inserting title ($query)");
			
		}
	}


// stick this referrer in db
	$current_date=date("Y-m-d H:i:s");
	
	$stick_query="insert into referrers (page, referrer, datetime, ip) values('$current_page', '$current_referrer', '$current_date', '$current_ip')";

	mysql_query($stick_query) or die("Problem inserting referrer in database ($stick_query)");
}
?>