From: henoheno <henoheno>
Date: Sun, 22 Apr 2007 08:04:19 +0000 (+0900)
Subject: List => Multiple lists
X-Git-Url: http://git.osdn.net/view?p=pukiwiki%2Fpukiwiki_sandbox.git;a=commitdiff_plain;h=0804eeeb75480d620b89e325f6daaa1b99bd0173;hp=85e9b5a6cceb7a2226ca2a2b22c574ef90f793ff

List => Multiple lists
---

diff --git a/spam/spam.ini.php b/spam/spam.ini.php
index cb95155..7729484 100644
--- a/spam/spam.ini.php
+++ b/spam/spam.ini.php
@@ -1,15 +1,31 @@
 <?php
-// $Id: spam.ini.php,v 1.82 2007/04/12 14:31:53 henoheno Exp $
+// $Id: spam.ini.php,v 1.83 2007/04/22 08:04:19 henoheno Exp $
 // Spam-related setting
 //
 // Reference:
 //   Spamdexing http://en.wikipedia.org/wiki/Spamdexing
 
-// NOTE for goodhost and badhost:
+// NOTE:
 //  'example.org' prohibits both 'example.org' AND 'www.example.org'.
 //  '*.example.org' prohibits all subdomains and hosts EXCEPT 'www.example.org'.
 //  '.example.org' prohibits ALL FQDN related 'example.org'.
 
+// NOTE for badhost (only):
+//  'unique_name' = array('a.example.org', 'b.example.org') is treated as a group 'unique_name'.
+
+
+// List of the lists
+$blocklist['list'] = array(
+	'goodhost'		=> FALSE,
+	'badhost A-1'	=> TRUE,
+	'badhost A-2'	=> TRUE,
+	'badhost B-1'	=> TRUE,
+	'badhost B-2'	=> TRUE,
+	'badhost C'		=> TRUE,
+	'badhost'		=> TRUE,
+);
+
+
 $blocklist['goodhost'] = array(
 	'IANA-examples' => '#^(?:.*\.)?example\.(?:com|net|org)$#',
 
@@ -25,12 +41,10 @@ $blocklist['goodhost'] = array(
 
 );
 
-// NOTE for badhost (only):
-//  'unique_name' = array('a.example.org', 'b.example.org') is treated as a group 'unique_name'.
-$blocklist['badhost'] = array(
+// A: Sample setting of
+// Existing URI redirection or masking services
 
-	// A: Sample setting of
-	// Existing URI redirection or masking services
+$blocklist['badhost A-1'] = array(
 
 	// A-1: General redirection services -- by HTML meta, HTML frame, JavaScript,
 	// web-based proxy, DNS subdomains, etc
@@ -1212,6 +1226,9 @@ $blocklist['badhost'] = array(
 	'*.zu5.net',
 	'zuso.tw',
 	'*.zwap.to',
+);
+
+$blocklist['badhost A-2'] = array(
 
 	// A-2: Dynamic DNS, Dynamic IP services, DNS vulnerabilities, or another DNS cases
 	//
@@ -1234,13 +1251,16 @@ $blocklist['badhost'] = array(
 	//'*.zenno.info',
 	//'.cm',	// 'Cameroon' ccTLD, sometimes used as typo of '.com'
 				// and all non-recorded domains redirect to 'agoga.com' now
+);
 
 
-	// B: Sample setting of:
-	// Jacked (taken advantage of) and cleaning-less sites
-	//
-	// Please notify us about this list with reason:
-	// http://pukiwiki.sourceforge.jp/dev/?BugTrack2%2F208
+// B: Sample setting of:
+// Jacked (taken advantage of) and cleaning-less sites
+//
+// Please notify us about this list with reason:
+// http://pukiwiki.sourceforge.jp/dev/?BugTrack2%2F208
+
+$blocklist['badhost B-1'] = array(
 
 	// B-1: Web spaces
 	//
@@ -1797,6 +1817,10 @@ $blocklist['badhost'] = array(
 		'*.your-websites.net',
 		'*.web-space.ws',
 	),
+);
+
+
+$blocklist['badhost B-2'] = array(
 
 	// B-2: Jacked contents, something implanted
 	// (e.g. some sort of blog comments, BBSes, forums, wikis)
@@ -1948,7 +1972,10 @@ $blocklist['badhost'] = array(
 	'yasushi.site.ne.jp',	// One of mixedmedia.net'
 	'youthpeer.org',
 	'*.zenburger.com',
+);
+
 
+$blocklist['badhost C'] = array(
 
 	// C: Sample setting of:
 	// Exclusive spam domains
@@ -3568,7 +3595,6 @@ $blocklist['badhost'] = array(
 	),
 
 
-
 	// C-2: Lonely domains (buddies not found yet)
 	'.0721-4404.com',
 	'.0nline-porno.info',	// by Timyr (timyr at narod.ru)
@@ -3721,7 +3747,10 @@ $blocklist['badhost'] = array(
 		'.nana.co.il',
 	),
 	'.mylexus.info',		// by Homer Simpson (simhomer12300 at mail.com), Redirect to Google
+);
+
 
+$blocklist['badhost'] = array(
 	// D: Sample setting of
 	// "third party in good faith"s
 	//
diff --git a/spam/spam.php b/spam/spam.php
index 547a481..0850859 100644
--- a/spam/spam.php
+++ b/spam/spam.php
@@ -1,5 +1,5 @@
 <?php
-// $Id: spam.php,v 1.130 2007/04/11 13:13:33 henoheno Exp $
+// $Id: spam.php,v 1.131 2007/04/22 08:04:19 henoheno Exp $
 // Copyright (C) 2006-2007 PukiWiki Developers Team
 // License: GPL v2 or (at your option) any later version
 //
@@ -878,7 +878,16 @@ function get_blocklist($list = '')
 			//		'*.blogspot.com',	// Blog services's subdomains (only)
 			//		'IANA-examples' => '#^(?:.*\.)?example\.(?:com|net|org)$#',
 			//	);
-			foreach(array('goodhost', 'badhost') as $_list) {
+			if (isset($blocklist['list'])) {
+				$regexs['list'] = & $blocklist['list'];
+			} else {
+				// Default
+				$blocklist['list'] = array(
+					'goodhost' => FALSE,
+					'badhost'  => TRUE,
+				);
+			}
+			foreach(array_keys($blocklist['list']) as $_list) {
 				if (! isset($blocklist[$_list])) continue;
 				foreach ($blocklist[$_list] as $key => $value) {
 					if (is_array($value)) {
@@ -890,6 +899,7 @@ function get_blocklist($list = '')
 						get_blocklist_add($regexs[$_list], $key, $value);
 					}
 				}
+				unset($blocklist[$_list]);
 			}
 		}
 	}
@@ -911,7 +921,7 @@ function get_blocklist_add(& $array, $key = 0, $value = '*.example.org')
 	} else {
 		$array[$value] = '/^' . generate_host_regex($value, '/') . '$/i';
 	}
-} 
+}
 
 function is_badhost($hosts = array(), $asap = TRUE, & $remains)
 {
@@ -924,29 +934,32 @@ function is_badhost($hosts = array(), $asap = TRUE, & $remains)
 	}
 	if (empty($hosts)) return $result;
 
-	foreach (get_blocklist('goodhost') as $regex) {
-		$hosts = preg_grep_invert($regex, $hosts);
-	}
-	if (empty($hosts)) return $result;
-
-	foreach (get_blocklist('badhost') as $label => $regex) {
-		if (is_array($regex)) {
-			$result[$label] = array();
-			foreach($regex as $_label => $_regex) {
-				if (is_badhost_avail($_label, $_regex, $hosts, $result[$label]) && $asap) {
-					break;
+	foreach(get_blocklist('list') as $key=>$value){
+		if ($value) {
+			foreach (get_blocklist($key) as $label => $regex) {
+				if (is_array($regex)) {
+					$result[$label] = array();
+					foreach($regex as $_label => $_regex) {
+						if (is_badhost_avail($_label, $_regex, $hosts, $result[$label]) && $asap) {
+							break;
+						}
+					}
+					if (empty($result[$label])) unset($result[$label]);
+				} else {
+					if (is_badhost_avail($label, $regex, $hosts, $result) && $asap) {
+						break;
+					}
 				}
 			}
-			if (empty($result[$label])) unset($result[$label]);
 		} else {
-			if (is_badhost_avail($label, $regex, $hosts, $result) && $asap) {
-				break;
+			foreach (get_blocklist($key) as $regex) {
+				$hosts = preg_grep_invert($regex, $hosts);
 			}
+			if (empty($hosts)) return $result;
 		}
 	}
 
 	$remains = $hosts;
-
 	return $result;
 }