<?php

 /**

 * File: SLVbase.class.php

 * Spam Link Verification (SLV) Base Class

 *

 * Copyright (C) 2006 by the following authors:

 * Author        Dirk Haun       dirk AT haun-online DOT de

 *

 * Licensed under the GNU General Public License

 *

 * @package Spam-X

 * @subpackage Modules

 */

 if (strpos(strtolower($_SERVER['PHP_SELF']), 'slvbase.class.php') !== false) {

     die('This file can not be used on its own!');

 }

 /**

 * Sends posts to SLV (linksleeve.org) for examination

 *

 * @author Dirk Haun     dirk AT haun-online DOT de

 * based on the works of Tom Willet (Spam-X) and Russ Jones (SLV)

 * @package Spam-X

 *

 */

 class SLVbase {

     var $_debug = false;

     var $_verbose = false;

     /**

     * Constructor

     */

     function SLVbase()

     {

         $this->_debug = false;

         $this->_verbose = false;

     }

     /**

     * Check for spam links

     *

     * @param    string  $post   post to check for spam

     * @return   boolean         true = spam found, false = no spam

     *

     * Note: Also returns 'false' in case of problems communicating with SLV.

     *       Error messages are logged in Geeklog's error.log

     *

     */

     function CheckForSpam ($post)

     {

         global $_SPX_CONF;

         require_once ('XML/RPC.php');

         $retval = false;

         if (empty ($post)) {

             return $retval;

         }

         $links = $this->prepareLinks ($post);

         if (empty ($links)) {

             return $retval;

         }

         if (!isset ($_SPX_CONF['timeout'])) {

             $_SPX_CONF['timeout'] = 5; // seconds

         }

         if ($this->_verbose) {

             SPAMX_log ("Sending to SLV: $links");

         }

         $params = array (new XML_RPC_Value ($links, 'string'));

         $msg = new XML_RPC_Message ('slv', $params);

         $cli = new XML_RPC_Client ('/slv.php', 'http://www.linksleeve.org');

         if ($this->_debug) {

             $client->setDebug (1);

         }

         $resp = $cli->send ($msg, $_SPX_CONF['timeout']);

         if (!$resp) {

             COM_errorLog ('Error communicating with SLV: ' . $cli->errstr

                           . '; Message was ' . $msg->serialize());

         } else if ($resp->faultCode ()) {

             COM_errorLog ('Error communicating with SLV. Fault code: '

                           . $resp->faultCode() . ', Fault reason: '

                           . $resp->faultString() . '; Message was '

                           . $msg->serialize());

         } else {

             $val = $resp->value();

             // note that SLV returns '1' for acceptable posts and '0' for spam

             if ($val->scalarval() != '1') {

                 $retval = true;

                 SPAMX_log ("SLV: spam detected");

             } else if ($this->_verbose) {

                 SPAMX_log ("SLV: no spam detected");

             }

         }

         return $retval;

     }

     /**

     * Check whitelist

     *

     * Check against our whitelist of sites not to report to SLV. Note that

     * URLs starting with $_CONF['site_url'] have already been removed earlier.

     *

     * @param    array   &$links     array of URLs from a post

     * @return   void ($links is passed by reference and modified in place)

     *

     */

     function checkWhitelist (&$links)

     {

         global $_TABLES;

         $result = DB_query ("SELECT value FROM {$_TABLES['spamx']} WHERE name='SLVwhitelist'", 1);

         $nrows = DB_numRows ($result);

         for ($i = 0; $i < $nrows; $i++) {

             $A = DB_fetchArray ($result);

             $val = $A['value'];

             $val = str_replace ('#', '\\#', $val);

             foreach ($links as $key => $link) {

                 if (!empty ($link)) {

                     if (preg_match ("#$val#i", $link)) {

                         $links[$key] = '';

                     }

                 }

             }

         }

     }

     /**

     * Extract links

     *

     * Extracts all the links from a post; expects HTML links, i.e. <a> tags

     *

     * @param    string  $comment    The post to check

     * @return   array               All the URLs in the post

     *

     */

     function getLinks ($comment)

     {

         global $_CONF;

         $links = array();

         preg_match_all( "/<a[^>]*href=[\"']([^\"']*)[\"'][^>]*>(.*?)<\/a>/i",

                         $comment, $matches );

         for ($i = 0; $i < count ($matches[0]); $i++) {

             $url = $matches[1][$i];

             if (!empty ($_CONF['site_url']) &&

                     strpos ($url, $_CONF['site_url']) === 0) {

                 // skip links to our own site

                 continue;

             } else {

                 $links[] = $url;

             }

         }

         return $links;

     }

     /**

     * Extract only the links from the post

     *

     * SLV has a problem with non-ASCII character sets, so we feed it the URLs

     * only. We also remove all URLs containing our site's URL.

     *

     * Since we don't know if the post is in HTML or plain ASCII, we run it

     * through getLinks() twice.

     *

     * @param    string  $comment    The post to check

     * @return   string              All the URLs in the post, sep. by linefeeds

     *

     */

     function prepareLinks ($comment)

     {

         $links = array();

         $linklist = '';

         // some spam posts have extra backslashes

         $comment = stripslashes ($comment);

         // some spammers have yet to realize that we're not supporting BBcode

         // but since we want the URLs, convert it here ...

         $comment = preg_replace ('/\[url=([^\]]*)\]/i', '<a href="\1">',

                                  $comment);

         $comment = str_replace (array ('[/url]', '[/URL]'),

                                 array ('</a>',   '</a>'  ), $comment);

         // get all links from <a href="..."> tags

         $links = $this->getLinks ($comment);

         // strip all HTML, then get all the plain text links

         $comment = COM_makeClickableLinks (strip_tags ($comment));

         $links += $this->getLinks ($comment);

         if (count ($links) > 0) {

             $this->checkWhitelist ($links);

             $linklist = implode ("\n", $links);

         }

         return trim ($linklist);

     }

 }

 ?>