<?php
/*******************************************************************************
 * LightPress - http://lightpress.org
 *
 * Copyright 2004-2005 Ludovico Magnocavallo.
 * Released under the GPL http://www.gnu.org/copyleft/gpl.html
 *
 * developers:
 *              Ludovico Magnocavallo <ludo@asiatica.org>
 *              Jerome Lavigne <jerome@vapourtrails.ca>
 *              Luca Lizzeri <luca.lizzeri@gmail.com>
 *
 * Adapted from Kitten's Spaminator
 * http://mookitty.co.uk/devblog/category/kittens-spaminator/
 *
 * $Id$
 ******************************************************************************/

class SpaminatorPort {
    var $strikes = 5;
    var $regexes = array(
        '/[0-9]{1,4}.*ewqrb2/i',
        '/^byob.*[0-9]{1,4}/i');
    var $nap_time = 2;
    var $crap_flood = 60;
    var $user_level = 3;
    var $comment_status = '1';
    
    var $_links_re = '/<a\s+[^>]*?href=/i';
    var $_frontend;
    var $_db;
    var $_options;
    var $_db_tables;
    var $_mail_helper;
    var $_strikes = 0;
    var $_max_links = 0;
    var $_word_list = array();
    var $_ip_list = array();
    var $_email_list = array();
    
    function SpaminatorPort(&$frontend) {
        $this->_frontend =& $frontend;
        $this->_db =& $frontend->db;
        $this->_options =& $frontend->options;
        $this->_db_tables =& $frontend->tables;
        $this->_mail_helper =& $frontend->getMailHelper();
        // get relevant options
        $q = "select option_name, option_value from {$this->_db_tables['options']} where option_name in ('moderation_keys', 'comment_max_links')";
        $this->_db->query($q);
        foreach ($this->_db->all() as $row) {
            switch ($row['option_name']) {
                case 'moderation_keys':
                    // build wordlist and iplist
                    $wordlist = str_replace("\r", '', $row['option_value']);
                    foreach (explode("\n", $row['option_value']) as $word) {
                        $word = strtolower(trim($word));
                        if (empty($word))
                            continue;
                        if (preg_match('/^([0-9]{1,3}\.?){2,4}/', $word))
                            $this->_ip_list[] = $word;
                        else
                            $this->_word_list[] = $word;
                    }
                    break;
                case 'comment_max_links':
                    $this->_max_links = (int)$row['option_value'];
                    break;
            }
        }
        // build approved emails list, MySQL specific
        $this->_db->query("select comment_author_email from {$this->_db_tables['comments']} where comment_approved='1' and comment_date_gmt < ('{$this->_frontend->where_date}' - interval 1 day)");
        $this->_email_list =& $this->_db->all();
    }
    
    function _addStrike($count) {
        $this->_strikes += $count;
        if ($this->_strikes >= $this->strikes) {
            return true;
        }
    }
    
    function _exit($user, $email, $url, $text, $xml_output=false) {
        // sleep for a bit
        sleep($this->nap_time);
        if ($this->_options['comment_spam_kill']) {
            // check if we have to send email
            if ($this->_options['comment_spam_mail']) {
                $body  = "A comment has been killed.\r\n\r\n";
                $body .= "The details:\r\n";
                $body .= "Strikes : " . $this->_strikes . "/" . $this->strikes . "\r\n";
                $body .= "IP Addr : ".$_SERVER['REMOTE_ADDR']."\r\n";
                $body .= "Referer : " . (isset($_SERVER['HTTP_REFERER']) ? $_SERVER['HTTP_REFERER'] : '') ."\r\n";
                $body .= "Client  : " . $_SERVER['HTTP_USER_AGENT']."\r\n";
                $body .= "Request : " . $_SERVER['REQUEST_METHOD']." ". $_SERVER['REQUEST_URI']."\r\n";
                $body .= "Email   : $email\r\n";
                $body .= "Author  : $user\r\n";
                $body .= "Body:\r\n";
                $body .= "$text\r\n\r\n";
                $body .= "--\r\nThis email has been sent because you have comment_spam_kill and comment_spam_mail set to true in your config\r\n";
                
                $users =& $this->_frontend->getUserPerms();
                $recipients = array();
                foreach ($users as $user_id=>$user_data) {
                    if ($user_id == $author_id)
                        $recipients[] = $user_data['user_email'];
                    else if ($user_data['user_level'] >= 3)
                        $recipients[] = $user_data['user_email'];
                    else if (isset($user_data['capabilities']['administrator']) || isset($user_data['capabilities']['editor']))
                        $recipients[] = $user_data['user_email'];
                }
                $to = implode(', ', $recipients);
                
                $shortname_hdr = $this->_mail_helper->encoded2hdr($this->_options['shortname']);
                
                $this->_mail_helper->send_mail(
                    $to,
                    "[$shortname_hdr]  Spammer caught!",
                    $body,
                    "From: " . $this->_options['mail_from']);
                
            }
            
            if (!$xml_output) {
                // Make sure the bot thinks that spam was posted:
                header( "HTTP/1.0 200 OK" );
                
                // Tell humans something else:
                echo "Sorry, you are banned from commenting on this blog.<p />";
                echo "Either your comment content was found to contain spam, or<br />";
                echo "your IP address (or a subnet of your IP address) has spammed this blog before.";
                echo "<p>Strike count: " . $this->_strikes . "</p>";
                exit;
            } else {
                header('Content-Type: text/xml');
                echo '<?xml version="1.0" encoding="' . $this->_options['charset'] . '"?>' .
                    "\n<response><error>1</error>" .
                    "<message>The system thinks your trackback is spam!</message>" .
                    "</response>\n";
                exit;
            }
        }
        return '0';
    }
    
    function isSpam($user, $email, $url, $text, $xml_output=false) {
        // comment attributes should have already been checked for non-emptyness
        $email = strtolower($email);
        // check if in whitelist
        $debug = 0;
        if (in_array($email, $this->_email_list))
            $this->_strikes -= 3;
        // check for regex match
        foreach ($this->regexes as $regex) {
            if (preg_match($regex, $text) && $this->_addStrike(10))
                return $this->_exit($user, $email, $url, $text, $xml_output);
        }
        // check referrer
        if (isset($_SERVER['HTTP_REFERER']) && strpos($_SERVER['HTTP_REFERER'], $this->_options['url']) === false && $this->_addStrike(2))
            return $this->_exit($user, $email, $url, $text, $xml_output);
        // check ip
        $ip = $_SERVER['REMOTE_ADDR'];
        foreach ($this->_ip_list as $spam_ip) {
            if (strpos($ip, $spam_ip) == 0 && $this->_addStrike(5))
                return $this->_exit($user, $email, $url, $text, $xml_output);
        }
        // check for flooding
        $q = "select max(UNIX_TIMESTAMP(comment_date)) as tstamp from " . 
            $this->_db_tables['comments'] . " where comment_author_IP='$ip'";
        $this->_db->query($q);
        $row = $this->_db->next();
        if (!empty($row)) {
            $t = (int)$row['tstamp'];
            if (($t + $this->crap_flood > time()) && $this->_addStrike(3))
                return $this->_exit($user, $email, $url, $text, $xml_output);
        }
        // check number of links
        //$links = preg_match_all($this->_links_re, $text, $matches);
        if ((count(explode('http://', $text)) - 1) >= $this->_max_links && $this->_addStrike(2))
            return $this->_exit($user, $email, $url, $text, $xml_output);
        // check for email, user, url, text in wordlist
        // start with smaller fields as they're faster to check
        foreach (array($user=>1, $email=>1, $url=>2, $text=>1) as $word => $strike_value) {
            foreach ($this->_word_list as $spam_word) {
                if (strlen($spam_word) < 3 || strlen($word)  == 0)
                    continue;
                if (strlen($spam_word) > strlen($word)) {
                    if (strpos($spam_word, $word) !== false && $this->_addStrike($strike_value))
                        return $this->_exit($user, $email, $url, $text, $xml_output);
                } else {
                    if (strpos($word, $spam_word) !== false && $this->_addStrike($strike_value))
                        return $this->_exit($user, $email, $url, $text, $xml_output);
                }
            }
        }
        // straight from spaminator
        // Useless numeric encoding is a pretty good spam indicator:
        // Extract entities:
        if (preg_match_all('/&#(\d+);/', $text, $chars)) {
            foreach ($chars[1] as $char) {
            // If it's an encoded char in the normal ASCII set, reject
                if ($char < 128 && $this->_addStrike(1))
                    return $this->_exit($user, $email, $url, $text, $xml_output);
            }
        }
        return $this->comment_status;
    }
}

require_once 'classes/LightPressPlugin.php';

class CommentSpam extends LightPressPlugin {
    
    var $constructor_args = array();
    var $default_context = 24; // LP_CONTEXT_POST | LP_CONTEXT_STATIC
    var $description = 'Check for spam on comment insertion';
    
    
    var $active = true;
    var $hooks = array('insert_comment');
    
    var $_spaminator;
    
    function CommentSpam(&$frontend, $args, $dummy_run=false) {
        $this->LightPressPlugin($frontend, $args, $dummy_run);
        
        if (!$dummy_run) {
            $this->_spaminator =& new SpaminatorPort($frontend);
        }
    }
    
    function run($hook, &$c) {
        $c['approved'] = $this->_spaminator->isSpam(
            $c['author'], $c['email'], $c['url'], $c['comment'], $c['type'] == 'trackback');
    }
}

?>