Navigation

← Previous Changeset
Next Changeset →

Changeset 194

Timestamp:

14/09/06 14:40:29 (2 years ago)

Author:

rollercow

Message:

Adds more spam checking, for URLs using SURBL.org, start of sanity changes on blog.lib

Files:

blog.lib.php (modified) (6 diffs)
miscfunctions.lib.php (modified) (7 diffs)
two-level-tlds (added)

Legend:

: Unmodified
: Added
: Removed
: Modified
: Copied
: Moved

blog.lib.php

r188	r194
28	28	echo("<p class=\"errorinfo\">"._("Level ").$level._(" error - ").$error."</p>");
29	29	}
30
	30	//whats this? where is it used?
31	31	function error_exc($e) {
32	32	if (!isset($e->error) \|\| !isset($e->errormsg)) {
…	…
39	39	//Our Blogs Class
40	40	class blogs {
41		var $id; //Blog ID
42		var $userName; //Blogger's User Name
43		var $realName; //Blogger's Real Name
44		var $title; //Blog Title
45		var $description; //Blog Blurb
46		var $cssFile; //Blog CSS
47		var $shortDateFormat; //Short date format
48		var $longDateFormat; //Long date format
49		var $httpPath; //http Path for files
50		var $blogPath; //path to blog
51		var $basePath; //path to the display blog bits
52		var $adminPath; //path to blog admin page
53		var $commentError; //new comment errors
54		var $entryError; //new entry errors
55		var $currentEntry; // the shortsubject of the current entry (where applicable)
56		var $svnRevision; // the SVN revision number of the currently running blog
57
	41	//Blog ID
	42	var $id;
	43	//Blogger's Details
	44	var $userName;
	45	var $realName;
	46	//Blog Details
	47	var $title;
	48	var $description;
	49	var $cssFile;
	50	//Date formats
	51	var $shortDateFormat;
	52	var $longDateFormat;
	53	//Paths
	54	var $httpPath;
	55	var $blogPath;
	56	var $basePath;
	57	var $adminPath;
	58	//Errors
	59	var $error;
	60	//SVN Revision... the closest thing we've got to a version number
	61	var $svnRevision;
	62
58	63	//Constructor - checks we've been given a valid username, and pulls in generic blog info
59		function blogs($user)
60		{
	64	function blogs($user) {
	65	//set the error string first, so we dont wipe out any errors
	66	$this->error = '';
	67	//set the locale
61	68	setlocale(LC_ALL, 'en_GB');
	69	//check the username
62	70	if(!safeuname($user)) {
63		~~$this->error = 1~~;
64		~~$this->errormsg = "Bad username"~~;
	71	error(1,_("No such user"));
	72	return false;
65	73	} else {
66	74	$sql = db_query("SELECT id, name, title, description, css, moderate, editor from users where username = '".$user."' and enabled = true;");
67	75	$sqlNum = db_num_rows($sql);
68	76	if ($sqlNum != 1) {
69		$this->error = 1;
70		$this->errormsg = "No such user";
71		}
72		else {
	77	error(1,_("No such user"));
	78	return false;
	79	} else {
73	80	$sqlRow = db_getrow($sql);
74	81	$this->id = $sqlRow['id'];
…	…
95	102	$this->entryError = '';
96	103	$this->comment_moderation = ($sqlRow['moderate']=='t') ? TRUE : FALSE;
97		$this->editor = ($sqlRow['editor']=='t') ? TRUE : FALSE;
	104	$this->editor = ($sqlRow['editor']=='t') ? TRUE : FALSE;
98	105	$this->currentEntry = "";
99	106	$this->svnRevision = getSVNRevision();
100
101	107	// setup the session
102	108	session_name("BlogSession");
…	…
105	111	}
106	112	}
107
108	113	// print a blog entry
109	114	function printEntry($row, $commentLink = true, $titleLink = true)
…	…
557	562	$sqlNum = db_num_rows($sql);
558	563	if ($sqlNum != 1) {
559		~~$this->commenterror = _("Invalid blog entry, This entry may have been removed..?"); //why oh why do we need a global error string for this specific function?~~
560		~~//it seems silly to continue at this point and for that matter most cases where we spew an error.. but that will be left for another day/person~~
	564	error(1,_("Invalid blog entry, This entry may have been removed..?"));
	565	return;
561	566	}
562	567	//pull in the unadulterated subject for later on
…	…
579	584	$element = "comment";
580	585	}
581
582	586	//decided if the comment is likly to be spam
583		if (check~~RBL($host) or checkBody($comment~~)) {
584		$spam = ~~TRUE~~;
	587	if (checkSpam($host,$_POST['comment'])) {
	588	$spam = true;
585	589	//force this comment though moderation
586		$this->comment_moderation = ~~TRUE~~;
587		}
588		else {
589		$spam = ~~FALSE~~;
	590	$this->comment_moderation = true;
	591	}
	592	else {
	593	$spam = false;
590	594	}
591	595

miscfunctions.lib.php

r177	r194
67	67	}
68	68
69		//feeds a message body though LinkSleeve (http://www.linksleeve.org/) which at the time of testing seems realy rather good.
70		function checkBody ($comment) {
	69	/*
	70	*
	71	* Spam Checks
	72	*
	73	*/
	74	//Check the Spam URI Realtime Blocklist
	75	function checkSpamURLs($text) {
	76	$spam = false;
	77	//find urls, ugly but works
	78	while (ereg("http://[a-zA-Z0-9.-]+\.[a-zA-Z]{2,4}",$text,$match)) {
	79	$matches[] = $match[0];
	80	$text = ereg_replace($match[0],"",$text);
	81	}
	82	//pull in list of two level tlds, make an array from them. from http://spamcheck.freeapp.net/two-level-tlds
	83	$twoLevelTLD = file("./two-level-tlds");
	84	foreach($twoLevelTLD as $TLD) {
	85	$two_level_tlds[trim($TLD)] = true;
	86	}
	87	//for each url
	88	foreach ($matches as $url) {
	89	//break it down
	90	$urlBits = explode(".",substr($url, 7));
	91	//reverse the oder
	92	$bitsURL = array_reverse($urlBits);
	93	//if its a two level tld, we want the first 3 bits of the url.. if not just the first 2
	94	if ($two_level_tlds[($bitsURL[1].".".$bitsURL[0])]) {
	95	$URLstoTest[] = ($bitsURL[2].".".$bitsURL[1].".".$bitsURL[0]);
	96	} else {
	97	$URLstoTest[] = ($bitsURL[1].".".$bitsURL[0]);
	98	}
	99	}
	100	//actualy test each of he domains against the surbl
	101	foreach($URLstoTest as $url) {
	102	$result = gethostbyname($url.'.multi.surbl.org');
	103	if ($result != $url.'.multi.surbl.org') {
	104	$spam = true;
	105	}
	106	}
	107	return $spam;
	108	}
	109	//feeds a message body though LinkSleeve (http://www.linksleeve.org/) which at the time of testing seems quite good.
	110	function checkSpamLinkSleeve ($text) {
71	111	// Include the Pear XML-RPC Client Package
72	112	require_once 'XML/RPC.php';
73	113	// Build the XML-RPC message
74		$params = array(new XML_RPC_Value($~~commen~~t, 'string'));
	114	$params = array(new XML_RPC_Value($text, 'string'));
75	115	$msg = new XML_RPC_Message('slv', $params);
76	116	//Send the XML-RPC message
…	…
80	120	if (!$resp) {
81	121	echo 'Communication error: ' . $cli->errstr;
82		~~exit~~;
	122	return false;
83	123	}
84	124	//spam?
…	…
101	141
102	142	//checks an ip in several blacklists returns true if its present
103		function check~~RBL~~($ip) {
	143	function checkSpamIP($ip) {
104	144	$spam = false;
105	145	//reverse the ip
…	…
116	156	//CBL
117	157	if ($cbl) {
118		$spam = true;
	158	$spam = true;
119	159	}
120	160
…	…
159	199	//SpamCop
160	200	if ($scbl) {
161		$spam = true;
	201	$spam = true;
162	202	}
163	203
…	…
230	270	//SBL
231	271	if($sbl) {
232		$spam = true;
	272	$spam = true;
233	273	}
234	274	return $spam;
…	…
236	276
237	277	# General spam function combining all checks
238		function checkSpam($ip, $message) {
239		$spam=checkRBL($ip);
240		if (preg_match_all("#http#", $message, $out)>5) $spam=true;
241		return $spam;
242		}
	278	function checkSpam($ip, $text) {
	279	//Check LinkSleeve first, its a collaborative statistical thing, and will benefit from seeing all messages, spam or not
	280	if (checkSpamLinkSleeve($text)) {
	281	$spam = true;
	282	//Check any URL's the Spam URL Black List
	283	} elseif (checkSpamURLs($text)) {
	284	$spam = true;
	285	//If all else fails lookup the posting IP in all the normal IP Black Lists
	286	} elseif (checkSpamIP($ip)) {
	287	$spam = true;
	288	//Decide its probably not spam
	289	} else {
	290	$spam = false;
	291	}
	292	return $spam;
	293	}

Navigation

Changeset 194

Legend:

blog.lib.php

miscfunctions.lib.php

Download in other formats: