miscfunctions.lib.php

<?php
// does the opposite of PHP's nl2br()
function br2nl($string) {
	$string = preg_replace("/(\r\n|\n|\r)/", "", $string);
	$string = preg_replace("/<br *\/?>/i", "\n", $string);
	return $string;
}

// generate a pseudo-word random password
function makePassword($length=8)
{
	$password = "";
	$vowels = "aeiouy";
	$consonants = "bcdfghjklmnprst";
	$cn = strlen($consonants)-1;
	$vn = strlen($vowels)-1;
	// Start on cons or vowel
	$alt = mt_rand(0, 1);
	// How many numbers
	$len = mt_rand($length-3,$length);
	//add the letters
	for ($i = 0; $i < $len; $i++)
	{
		if ($alt == 1)	{
			$password .= $consonants[ mt_rand(0,$cn) ];
			$alt = 0;
		}
		else	{
			$password .= $vowels[ mt_rand(0,$vn) ];
			$alt = 1;
		}
	}
	//add the numbers
	for ($i = 0; $i < $length-$len; $i++)
	{
		$password .= mt_rand(0,9);
	}
	return $password;
}

$revision = "unknown";
function startElement($parser, $name, $attrs)
{
	global $revision;
	if($name=="ENTRY" && $attrs['NAME']=="") {
		$revision = $attrs['REVISION'];
	}
}

function endElement($parser, $name){}

function getSVNRevision()
{
	global $revision;
	$xml_parser = xml_parser_create();
	xml_set_element_handler($xml_parser, "startElement", "endElement");
	if (!($fp = fopen(".svn/entries", "r"))) {
		return "unknown - couldn't open SVN XML file.";
	}
	while(($data = fread($fp, 1024)) && $revision=="unknown") {
		if (!xml_parse($xml_parser, $data, feof($fp))) {
			return "unknown - couldn't parse SVN XML file";
		}
	}
	xml_parser_free($xml_parser);
	return $revision;
}

/*
 *
 * Spam Checks
 *
 */
//Check the Spam URI Realtime Blocklist
function checkSpamURLs($text) {
	$spam = false;
	//find urls, ugly but works
	while (preg_match("/http://[a-zA-Z0-9.-]+\.[a-zA-Z]{2,4}/",$text,$match)) {
		$matches[] = $match[0];
		$text = preg_replace("/".$match[0]."/","",$text); //Can this be replaced with str_replace?
	}
	//pull in list of two level tlds, make an array from them. from http://spamcheck.freeapp.net/two-level-tlds
	$twoLevelTLD = file("/var/www/sucssite/lib/blog/two-level-tlds");
	foreach($twoLevelTLD as $TLD) {
		$two_level_tlds[trim($TLD)] = true;
	}
	if (!$matches)
		return;
	//for each url
	foreach ($matches as $url) {
		//break it down
		$urlBits = explode(".",substr($url, 7));
		//reverse the oder
		$bitsURL = array_reverse($urlBits);
		//if its a two level tld, we want the first 3 bits of the url.. if not just the first 2
		if ($two_level_tlds[($bitsURL[1].".".$bitsURL[0])]) {
			$URLstoTest[] = ($bitsURL[2].".".$bitsURL[1].".".$bitsURL[0]);
		} else {
			$URLstoTest[] = ($bitsURL[1].".".$bitsURL[0]);
		}
	}
	if (!$URLstoTest)
		return;
	//actualy test each of he domains against the surbl
	foreach($URLstoTest as $url) {
		$result = gethostbyname($url.'.multi.surbl.org');
		if ($result != $url.'.multi.surbl.org') {
			$spam = true;
		}
		elseif ($url == "blogspot.com") {
			$spam = true;
		}
	}
	return $spam;
}
//feeds a message body though LinkSleeve (http://www.linksleeve.org/) which at the time of testing seems quite good.
function checkSpamLinkSleeve ($text) {
	// Include the Pear XML-RPC Client Package
	require_once 'XML/RPC.php';
	// Build the XML-RPC message
	$params = array(new XML_RPC_Value($text, 'string'));
	$msg = new XML_RPC_Message('slv', $params);
	//Send the XML-RPC message
	$cli = new XML_RPC_Client('/slv.php', 'http://www.linksleeve.org');
	$resp = $cli->send($msg);
	//Check for a responce
	if (!$resp) {
		echo 'Communication error: ' . $cli->errstr;
		return false;
	}
	//spam?
	if (!$resp->faultCode()) {
		$val = $resp->value();
		if($val->scalarval()=='1') {
			$spam = false;
		}
		else {
			$spam = true;
		}
	} 
	//Handle Errors
	else {
		 echo 'Fault Code: ' . $resp->faultCode() . "\n";
		 echo 'Fault Reason: ' . $resp->faultString() . "\n";
	}
	return $spam;
}

//checks an ip in several blacklists returns true if its present
function checkSpamIP($ip) {
	$spam = false;
	//reverse the ip
	$ip = implode('.',array_reverse(explode('.',$ip)));
	//look up in various rbls
	$rbl = gethostbyname($ip.'.rbl-plus.mail-abuse.ja.net');
	$scbl = gethostbynamel($ip.'.bl.spamcop.net');
	$sorbs = gethostbynamel($ip.'.dnsbl.sorbs.net');
	$sbl = gethostbynamel($ip.'.sbl.spamhaus.org');
	$njabl = gethostbynamel($ip.'.dnsbl.njabl.org');
	$opm = gethostbyname($ip.'.opm.blitzed.org');
	$cbl = gethostbynamel($ip.'.cbl.abuseat.org');

	//CBL 
	if ($cbl) {
		$spam = true;
	}
	
	//OPM
	if ($opm != $ip.".opm.blitzed.org") {
		//this bl uses a decimal to represent one catagory of spam source
		$code = decbin(ip2long($opm));
		//check for WinGate
		if ($code[30])
			$spam = true;
		//check for SOCKS
		if ($code[29])
			$spam = true;
		//check for HTTP CONNECT
		if ($code[28])
			$spam = true;
		//check for Router
		if ($code[27])
			$spam = true;
		//check for HTTP POST
		if ($code[26])
			$spam = true;
	}

	//RBL+
	if ($rbl != $ip.".rbl-plus.mail-abuse.ja.net") 	{
		$code = decbin(ip2long($rbl));
		//check for rbl
		if ($code[30])
			$spam = true;
		//check for dul
		if ($code[29])
			//we dont care about dul
		//check for rss
		if ($code[28])
			$spam = true;
		//check for ops
		if ($code[27])
			$spam = true;
	}
	
	//SpamCop
	if ($scbl) {
		$spam = true;
	}
	
	//SORBS
	if ($sorbs) {
		foreach($sorbs as $result) {
			$result = explode('.',$result);
			//check for http
			if ($result[3] == 2)
				$spam = true;
			//check for socks
			if ($result[3] == 3)
				$spam = true;
			//check for misc
			if ($result[3] == 4)
				$spam = true;
			//check for smtp
			if ($result[3] == 5)
				$spam = true;
			//check for spam
			if ($result[3] == 6)
				$spam = true;
			//check for web
			if ($result[3] == 7)
				$spam = true;
			//check for block
			if ($result[3] == 8)
				$spam = true;
			//check for zombie
			if ($result[3] == 9)
				$spam = true;
			//check for dul
			if ($result[3] == 10)
				//dont care about dul
			//check for badconf
			if ($result[3] == 11)
				$spam = true;
			//check for nomail
			if ($result[3] == 12)
				$spam = true;
		}
	}

	//NJABL
	if ($njabl) {
		foreach($njabl as $result) {
			$result = explode('.',$result);
			//check for relay
			if ($result[3] == 2)
				$spam = true;
			//check for dul
			if ($result[3] == 3) {
				//dont care about dul
			}
			//check for spam
			if ($result[3] == 4)
				$spam = true;
			//check for relay
			if ($result[3] == 5)
				$spam = true;
			//check for web
			if ($result[3] == 8)
				$spam = true;
			//check for proxy
			if ($result[3] == 9)
				$spam = true;
		}
	}

	//SBL
	if($sbl) {
		$spam = true;
	}
	return $spam;
}

# General spam function combining all checks
function checkSpam($ip, $text) {
	//Check LinkSleeve first, its a collaborative statistical thing, and will benefit from seeing all messages, spam or not
	if (checkSpamLinkSleeve($text)) {
		$spam = true;
	//Check any URL's the Spam URL Black List
	} elseif (checkSpamURLs($text)) {
		$spam = true;
	//If all else fails lookup the posting IP in all the normal IP Black Lists
	} elseif (checkSpamIP($ip)) {
		$spam = true;
	//Decide its probably not spam
	} else {
		$spam = false;
	}
	return $spam;
}