<?
/*
* L I N K S M A N A G E R
* Author: Sarah King
* http://sarahk.pcpropertymanager.com
* March 2005
*
* Description: This script is designed to review the links on a page and identify
* if the links are already held in the tables of a links directory
* If not, then the link is presented in a submission form which can be customised
* for the users own directory system.
*
* The CURL option is fully debugged, was not able to fully test the fsockopen version
*
* Setup: The class has 4 variables used for connecting to the database.
* These need to be changed.
*
* $query needs to be updated with the actual query used to check for a links existence
*
* $form works for my old version of the wsnlinks directory. You need to change this
* so that it works for the directory system you use
*
* $useCurl should be true if curl is available. fsockopen functions haven't been fully tested
* $user_agent should be set to your site name
*
* getScratchPad is for static information you need to copy and paste from time to time
*
*/
class LinkManager
{
// user variables
var $dbhost = 'localhost';
var $dbuser = '';
var $dbname = 'test';
var $dbpass = '';
var $allowSubDomains = true; // treat subdomains as separate sites
var $myDomains = array('pcpropertymanager.com',
'https://secure.maxnet.co.nz/pcpm/',
'propertyinvestor.info'
); // I don't want to detect links on my domain
// the database query I use to see if the link exists and to get the category name where I'd find it.
// need to return something called 'name' so that it can be fed back out
var $query = "select cat.name
from `wsnlinks_links` link, `wsnlinks_categories` cat
where link.catid = cat.id
and link.url like '%{DOMAIN}%' ";
// the default form works on my directory system, you need to change this to the form needed on your system
var $form = "<form action='/wsnlinks/suggest.php?action=addlink&filled=1'
method='post' target='_blank'>
<table>
<tr><th>Category:</th><td><input type='text' name='catid' value='' size='5'><td></tr>
<tr><th>Title:</th><td><input type='text' name='title' size=40 value='{TITLE}'><td></tr>
<tr><th>URL:</th><td><input type='text' name='url' size='40' value='{URL}'><a href='{URL}' target='_blank'>Go</a><td></tr>
<tr><th>Description:</th><td><textarea cols='40' rows='4' NAME='description'></textarea><td></tr>
<tr><th>Reciprocal Link URL:</th><td><input type=text name='recipurl' size='40' value='http://'><td></tr>
<tr><th>Your E-mail:</th><td><input type=text name=email size=40><td></tr>
<tr><th>REW Username:</th><td><input type=text name='rewusername' size='20'></td></tr>
<tr><td> </td><td><input type=submit value='Submit Link'></td></tr>
</table>
</form>";
//curl option is fully debugged
var $useCurl = true; //set to false to use fsockopen
var $ch; //curl handler if we need it
//curl lets us be good citizens and declare our user agent
var $user_agent='http://www.pcpropertymanager.com/wsnlinks/ Property Investment Directory ';
//any words which mean a local page shouldn't be checked for a redirect
var $stopWords = array('archive', 'clicks', 'content', 'download', 'Your_Account', 'profile', 'private');
var $debug = false;
// internal variables
var $html;
var $links;
var $domain;
var $errors = array();
/**
* @return string
* @desc Fill this up with the info you need frequently when filling in link requests
*/
function getScratchPad()
{
$output = "<a href='http://www.pcpropertymanager.com/wsnlinks/' target='_blank'>Property Investment Directory</a><br /><br />
<b>Our Link Information</b><br>
PC Property Manager<br>
http://www.pcpropertymanager.com<br>
Rental Property Management Software saves you time, money and keeps you organized.<br>
More than a great bookkeeping system, designed for property management issues leaving you in control and stress free. Helps with tax returns and tracking rents.
///////////////////////////////////
// No changes to the class should be made below here
///////////////////////////////////
/**
* @return string
* @param array $item
* @desc Controls the presentation of the link form and allows it to be hidden
*/
function getLinkForm($item, $cnter)
{
$form = $this->form;
$form = str_replace('{URL}', $item['url'], $form);
$form = str_replace('{TITLE}', $item['title'], $form);
/**
* @return LinkManager
* @desc Constructor, just sets up the page
* if no url is passed check to see if the form has been submitted
*/
function LinkManager($url='')
{
global $_SERVER;
if (empty($url)) $url = $this->getGetVar('url');
$this->domain = $this->extractDomain($url);
echo $this->getHeader($url);
if (!empty($url))
{
if ($this->useCurl) $this->initialiseCurl();
$this->processPage($url, 0);
}
else
{
echo "<p>Enter a page to check</p>
<form name='formURL' action='linkmanager.php' method='get'>
<input type='text' name='url' size='35'> <input type='submit' name='cmdSave' value='Process'>
</form>";
}
echo $this->getFooter();
if ($this->useCurl) $this->closeCurl();
}//LinkManager
/**
* @return string
* @param string $url
* @desc Returns the main part of the domain for comparison in the database
*/
function extractDomain($url)
{
$bits = parse_url($url);
$domain = $bits['host'];
/**
* @return array
* @param string $data
* @desc returns an array of extracted links
*/
function extractLinks()
{
$data = $this->html;
unset($location);
$links = array();
$pos = 0;
$i = 0;
while (!(($pos = strpos($data,"<",$pos)) === false))
{
$pos++;
$curLink = array();
$endpos = strpos($data,">",$pos);
$tag = substr($data,$pos,$endpos-$pos);
$tag = trim($tag);
if (isset($location)) { // look for a </A>
if (!strcasecmp(strtok($tag," "),"/A"))
{
$link = substr($data, $linkpos, $pos-1-$linkpos);
$curLink['title'] = strip_tags($link);
$curLink['url'] = $location;
$curLink['status'] = 0;
$links[] = $curLink;
unset($location);
}
$pos = $endpos+1;
}
else
{ // look for a <A ...>
if (!strcasecmp(strtok($tag,' '),'A'))
{
$regs[] = array();
preg_match('/href\s*=\s*([\'"]?)([^\'">\s]+)\1/i', $tag, $regs);
if ($regs[2])
{ // Only use it if it seems to be reasonable
$location = $regs[2];
}
$pos = $endpos+1;
$linkpos = $pos;
} else $pos = $endpos+1;
}
$i++;
}
$this->links = $links;
} //extract_links
/**
* @return void
* @param string $url
* @desc Retrieves the html for the page in question
*/
function getPage($url)
{
$output = $this->getHTTPContent($url, 'GET');
$output = implode(' ', $output);
$output = ereg_replace("\n|\r", ' ', $output);
$this->html = $output;
}//getPage
/**
* @return string
* @desc Works through the links and displays according to status
*/
function showPossibleLinks()
{
$output = '';
for ($i = 0; $i < 3; $i++)
{
if ($i == 1) $output .= "<h2>Already Listed</h2><ul>\n";
if ($i == 2) $output .= "<h2>Potential Listings</h2>\n";
if ($i == 0 && $this->debug) $output .= "<h2>Rejected</h2><ul>\n";
foreach ($this->links as $k => $varray)
{
$label = $varray['title'];
if (empty($label)) $label = $varray['url'];
if ($varray['status'] == 1 && $i == 1)
{
$output .= "<li><A HREF='{$varray['url']}' target='_blank'>{$label}</A> - {$varray['category']}</li>\n";
}
elseif ($varray['status'] == 2 && $i == 2)
{
$output .= $this->getLinkForm($varray, $k);
}
elseif ($this->debug) $output .= "<li><A HREF='{$varray['url']}' target='_blank'>{$label}</A></li>\n";
}
if ($i == 1) $output .= "</ul>\n";
if ($i == 0 && $this->debug) $output .= "</ul>\n";
}
return $output;
}//showPossibleLinks
function checkLinks()
{
$this->dbConnect();
foreach ($this->links as $k => $varray)
{
// see if we have a redirect
if ( stristr($varray['url'], $this->domain))
{
$varray['url'] = $this->detectExternalLink($varray['url']);
$this->links[$k]['url'] = $varray['url'];
}
elseif ( substr($varray['url'],0,4) != 'http')
{
$varray['url'] = $this->detectExternalLink($this->domain . '/'. $varray['url']);
$this->links[$k]['url'] = $varray['url'];
}
// now check the link against the database
if ($this->validDomain($varray['url']))
{
$domain = $this->extractDomain( $varray['url']);
/**
* @return boolean
* @param string $url
* @desc Checks to see if this is a domain we want to think about linking to
*/
function validDomain($url)
{
if (empty($url)) return false;
if (substr($url,0,4) != 'http') return false;
if (stristr($url, $this->domain)) return false;
foreach($this->myDomains as $d)
{
if (stristr($url, $d)) return false;
}
return true;
}//validDomain
function detectExternalLink($url)
{
$newUrl = '';
if ($this->checkNoStopWords($url))
{
$header = $this->getHTTPContent($url);
foreach($header as $lines)
{
if (substr($lines, 0, 9) == 'Location:') $newUrl = trim( substr( $lines, 10));
}
}
return $newUrl;
}//detectExternalLink
/**
* @return boolean
* @param string $url
* @desc Check to see if the url has any stopwords
*/
function checkNoStopWords($url)
{
foreach($this->stopWords as $needle)
{
if (stristr($url, $needle)) return false;
}
return true;
}//checkNoStopWords
/**
* @return array
* @param string $url
* @param string $method
* @desc Returns the requested information in an array
*/
function getHTTPContent($url, $method = 'HEAD')
{
$output = array();
$info = parse_url($url);
if (!isset($info['host'])) $info['host'] = $this->domain;
$host = $info['host'];
if (!isset($info['port'])) $info['port'] = 80;
if (!isset($info['path'])) $info['path'] = '/';
$path = $info['path'];
if ($info['query']) {
$path .= '?' . $info['query'];
}
if (!stristr($path, $host)) $path = $host. $path;
