OSDN Git Service

spam_uri_pickup_preprocess(): abstruction
[pukiwiki/pukiwiki_sandbox.git] / spam / spam.php
index 106ab90..0f32471 100644 (file)
@@ -1,5 +1,5 @@
 <?php
-// $Id: spam.php,v 1.178 2007/06/16 03:23:34 henoheno Exp $
+// $Id: spam.php,v 1.195 2007/06/29 15:35:53 henoheno Exp $
 // Copyright (C) 2006-2007 PukiWiki Developers Team
 // License: GPL v2 or (at your option) any later version
 //
@@ -7,7 +7,8 @@
 //
 // (PHP 4 >= 4.3.0): preg_match_all(PREG_OFFSET_CAPTURE): $method['uri_XXX'] related feature
 
-if (! defined('SPAM_INI_FILE')) define('SPAM_INI_FILE', 'spam.ini.php');
+if (! defined('SPAM_INI_FILE'))   define('SPAM_INI_FILE',   'spam.ini.php');
+if (! defined('DOMAIN_INI_FILE')) define('DOMAIN_INI_FILE', 'domain.ini.php');
 
 // ---------------------
 // Compat etc
@@ -111,7 +112,9 @@ function array_renumber_numeric_keys(& $array)
 // References:
 //   http://www.freebsd.org/cgi/man.cgi?query=strings (Man-page of GNU strings)
 //   http://www.pcre.org/pcre.txt
-function strings($binary = '', $min_len = 4, $ignore_space = FALSE, $multibyte = TRUE)
+// Note: mb_ereg_replace() is one of mbstring extension's functions
+//   and need to init its encoding.
+function strings($binary = '', $min_len = 4, $ignore_space = FALSE, $multibyte = FALSE)
 {
        // String only
        $binary = (is_array($binary) || $binary === TRUE) ? '' : strval($binary);
@@ -187,7 +190,7 @@ function uri_pickup($string = '')
                        // 3: Host
                        '\[[0-9a-f:.]+\]' . '|' .                               // IPv6([colon-hex and dot]): RFC2732
                        '(?:[0-9]{1,3}\.){3}[0-9]{1,3}' . '|' . // IPv4(dot-decimal): 001.22.3.44
-                       '[a-z0-9][a-z0-9.-]+[a-z0-9]' .                 // hostname(FQDN) : foo.example.org
+                       '[a-z0-9_-][a-z0-9_.-]+[a-z0-9_-]' .            // hostname(FQDN) : foo.example.org
                ')' .
                '(?::([0-9]*))?' .                                      // 4: Port
                '((?:/+[^\s<>"\'\[\]/\#]+)*/+)?' .      // 5: Directory path or path-info
@@ -424,7 +427,28 @@ function area_measure($areas, & $array, $belief = -1, $a_key = 'area', $o_key =
 // ---------------------
 // Spam-uri pickup
 
-// Domain exposure callback (See spam_uri_pickup_preprocess())
+// Preprocess: Removing uninterest part for URI detection
+function spam_uri_removing_hocus_pocus($binary = '', $method = array())
+{
+       $length = 4 ; // 'http'(1) and '://'(2) and 'fqdn'(1)
+       if (is_array($method)) {
+               // '<a'(2) or 'href='(5) or '>'(1) or '</a>'(4)
+               // '[uri'(4) or ']'(1) or '[/uri]'(6) 
+               if (isset($method['area_anchor']) || isset($method['uri_anchor']) ||
+                   isset($method['area_bbcode']) || isset($method['uri_bbcode']))
+                               $length = 1;    // Seems not effective
+       }
+
+       // Removing sequential spaces and too short lines
+       $binary = strings($binary, $length, TRUE, FALSE); // Multibyte NOT needed
+
+       // Remove words (has no '<>[]:') between spaces
+       $binary = preg_replace('/[ \t][\w.,()\ \t]+[ \t]/', ' ', $binary);
+
+       return $binary;
+}
+
+// Preprocess: Domain exposure callback (See spam_uri_pickup_preprocess())
 // http://victim.example.org/?foo+site:nasty.example.com+bar
 // => http://nasty.example.com/?refer=victim.example.org
 // NOTE: 'refer=' is not so good for (at this time).
@@ -453,27 +477,6 @@ function _preg_replace_callback_domain_exposure($matches = array())
        return $result;
 }
 
-// Preprocess: Removing uninterest part for URI detection
-function spam_uri_removing_hocus_pocus($binary = '', $method = array())
-{
-       $length = 4 ; // 'http'(1) and '://'(2) and 'fqdn'(1)
-       if (is_array($method)) {
-               // '<a'(2) or 'href='(5) or '>'(1) or '</a>'(4)
-               // '[uri'(4) or ']'(1) or '[/uri]'(6) 
-               if (isset($method['area_anchor']) || isset($method['uri_anchor']) ||
-                   isset($method['area_bbcode']) || isset($method['uri_bbcode']))
-                               $length = 1;    // Seems not effective
-       }
-
-       // Removing sequential spaces and too short lines
-       $binary = strings($binary, $length, TRUE, TRUE);
-
-       // Words between spaces
-       $binary = preg_replace('/[ \t][\w \t]+[ \t]/', ' ', $binary);
-
-       return $binary;
-}
-
 // Preprocess: rawurldecode() and adding space(s) and something
 // to detect/count some URIs _if possible_
 // NOTE: It's maybe danger to var_dump(result). [e.g. 'javascript:']
@@ -486,6 +489,7 @@ function spam_uri_pickup_preprocess($string = '', $method = array())
        if (! is_string($string)) return '';
 
        $string = spam_uri_removing_hocus_pocus(rawurldecode($string), $method);
+       //var_dump(htmlspecialchars($string));
 
        // Domain exposure (simple)
        // http://victim.example.org/nasty.example.org/path#frag
@@ -520,22 +524,27 @@ function spam_uri_pickup_preprocess($string = '', $method = array())
                $string
        );
 
-       // Domain exposure (See _preg_replace_callback_domain_exposure())
+       // Domain exposure (site:) See _preg_replace_callback_domain_exposure()
        $string = preg_replace_callback(
                array(
-                       '#(http)://' .
+                       '#(h?ttp)://' . // 1:Scheme
+                       // 2:Host
                        '(' .
+                               '(?:[a-z0-9_.-]+\.)?[a-z0-9_-]+\.[a-z0-9_-]+' .
                                // Something Google: http://www.google.com/supported_domains
-                               '(?:[a-z0-9.]+\.)?google\.[a-z]{2,3}(?:\.[a-z]{2})?' .
-                               '|' .
-                               // AltaVista
-                               '(?:[a-z0-9.]+\.)?altavista.com' .
-                               
+                               // AltaVista: http://es.altavista.com/web/results?q=site%3Anasty.example.org+foobar
+                               // Live Search: search.live.com
+                               // MySpace: http://sads.myspace.com/Modules/Search/Pages/Search.aspx?_snip_&searchString=site:nasty.example.org
+                               // (also searchresults.myspace.com)
+                               // alltheweb.com
+                               // search.bbc.co.uk
+                               // search.orange.co.uk
+                               // ...
                        ')' .
                        '/' .
-                       '([a-z0-9?=&.%_/\'\\\+-]+)' .                           // path/?query=foo+bar+
-                       '\bsite:([a-z0-9.%_-]+\.[a-z0-9.%_-]+)' .       // site:nasty.example.com
-                       //'()' .        // Preserve or remove?
+                       '([a-z0-9?=&.%_/\'\\\+-]+)' .                           // 3:path/?query=foo+bar+
+                       '\bsite:([a-z0-9.%_-]+\.[a-z0-9.%_-]+)' .       // 4:site:nasty.example.com
+                       '()' .                                                                          // 5:Preserve or remove?
                        '#i',
                ),
                '_preg_replace_callback_domain_exposure',
@@ -1518,46 +1527,47 @@ function summarize_detail_newtral($progress = array())
            ! is_array($progress['hosts']) ||
            empty($progress['hosts'])) return '';
 
-       $result = '';
-
-       // Generate a $trie
+       // Generate a responsible $trie
        $trie = array();
        foreach($progress['hosts'] as $value) {
-
-               // Try to shorten (pre) -- array('example.com', 'bar', 'foo')
+               // 'A.foo.bar.example.com'
                $resp = whois_responsibility($value);   // 'example.com'
-               $rest = rtrim(substr($value, 0, - strlen($resp)), '.'); // 'foo.bar'
-               if ($rest) {
-                       $parts = explode('.', delimiter_reverse('.' . $rest));
-                       array_unshift($parts, $resp);
+               if (empty($resp)) {
+                       // One or more test, or do nothing here
+                       $resp = strval($value);
+                       $rest = '';
                } else {
-                       $parts = array($resp, $rest);
+                       $rest = rtrim(substr($value, 0, - strlen($resp)), '.'); // 'A.foo.bar'
                }
-
-               $trie = array_merge_recursive(
-                       $trie,
-                       array_leaf($parts, TRUE, $value)
-               );
+               $trie = array_merge_recursive($trie, array($resp => array($rest => NULL)));
        }
 
-       // Try to shorten (post, non-recursive) -- 'foo.bar.example.com'
-       array_joinbranch_leaf($trie, '.', 0, TRUE);
-
-       // Sort and flatten -- 'A.foo.bar.example.com, B.foo.bar.example.com'
+       // Format: var_export_shrink() -like output
+       $result = array();
+       ksort_by_domain($trie);
        foreach(array_keys($trie) as $key) {
-               if (is_array($trie[$key])) {
-                       ksort_by_domain($trie[$key]);
-                       $trie[$key] = implode(', ', array_flat_leaves($trie[$key]));
+               ksort_by_domain($trie[$key]);
+               if (count($trie[$key]) == 1 && key($trie[$key]) == '') {
+                       // Just one 'responsibility.example.com'
+                       $result[] = '  \'' . $key . '\',';
+               } else {
+                       // One subdomain-or-host, or several ones
+                       $subs = array();
+                       foreach(array_keys($trie[$key]) as $sub) {
+                               if ($sub == '') {
+                                       $subs[] = $key;
+                               } else {
+                                       $subs[] = $sub . '.' . $key;
+                               }
+                       }
+                       $result[] = '  \'' . $key . '\' => \'' . implode(', ', $subs) . '\',';
                }
+               unset($trie[$key]);
        }
-
-       // TODO: ltrim('.') from $trie
-
-       ksort_by_domain($trie);
-
-       // TODO: from array('foobar' => 'foobar') to 'foobar'
-
-       return var_export_shrink($trie, TRUE, TRUE);
+       return
+               'array (' . "\n" .
+                       implode("\n", $result) . "\n" .
+               ')';
 }
 
 // ksort() by domain
@@ -1575,45 +1585,6 @@ function ksort_by_domain(& $array)
        $array = $result;
 }
 
-// array('F' => array('B' => array('C' => array('d' => array('' => 'foobar')))))
-// to
-// array('F.B.C.d.' => 'foobar')
-function array_joinbranch_leaf(& $array, $delim = '.', $limit = 0, $reverse = FALSE)
-{
-       $result = array();
-       if (! is_array($array)) return $result; // Nothing to do
-
-       $limit  = max(0, intval($limit));
-       $cstack = array();
-
-       foreach(array_keys($array) as $key) {
-               $kstack = array();
-               $k      = -1;
-
-               $single = array($key => & $array[$key]);        // Keep it single
-               $cursor = & $single;
-               while(is_array($cursor) && count($cursor) == 1) {       // Once
-                       ++$k;
-                       $kstack[] = key($cursor);
-                       $cursor   = & $cursor[$kstack[$k]];
-                       if ($limit != 0 && $k == $limit) break;
-               }
-
-               // Relink
-               if ($k != 0) {
-                       if ($reverse) $kstack = array_reverse($kstack);
-                       $joinkey = implode($delim, $kstack);
-
-                       unset($array[$key]);
-                       $array[$joinkey]  = & $cursor;
-                       $result[$joinkey] = $k + 1;     // Key seems not an single array => joined length
-               }
-       }
-
-       return $result;
-}
-
-
 // Check responsibility-root of the FQDN
 // 'foo.bar.example.com'        => 'example.com'        (.com        has the last whois for it)
 // 'foo.bar.example.au'         => 'example.au'         (.au         has the last whois for it)
@@ -1621,393 +1592,29 @@ function array_joinbranch_leaf(& $array, $delim = '.', $limit = 0, $reverse = FA
 // 'foo.bar.example.act.edu.au' => 'example.act.edu.au' (.act.edu.au has the last whois for it)
 function whois_responsibility($fqdn = 'foo.bar.example.com', $parent = FALSE, $implicit = TRUE)
 {
-       // Domains who have 2nd and/or 3rd level domains
-       static $domain = array(
-
-               // ccTLD: Australia
-               // http://www.auda.org.au/
-               // NIC  : http://www.aunic.net/
-               // Whois: http://www.ausregistry.com.au/
-               'au' => array(
-                       // .au Second Level Domains
-                       // http://www.auda.org.au/domains/
-                       'asn'   => TRUE,
-                       'com'   => TRUE,
-                       'conf'  => TRUE,
-                       'csiro' => TRUE,
-                       'edu'   => array(       // http://www.domainname.edu.au/
-                               // Geographic
-                               'act' => TRUE,
-                               'nt'  => TRUE,
-                               'nsw' => TRUE,
-                               'qld' => TRUE,
-                               'sa'  => TRUE,
-                               'tas' => TRUE,
-                               'vic' => TRUE,
-                               'wa'  => TRUE,
-                       ),
-                       'gov'   => array(
-                               // Geographic
-                               'act' => TRUE,  // Australian Capital Territory
-                               'nt'  => TRUE,  // Northern Territory
-                               'nsw' => TRUE,  // New South Wales
-                               'qld' => TRUE,  // Queensland
-                               'sa'  => TRUE,  // South Australia
-                               'tas' => TRUE,  // Tasmania
-                               'vic' => TRUE,  // Victoria
-                               'wa'  => TRUE,  // Western Australia
-                       ),
-                       'id'    => TRUE,
-                       'net'   => TRUE,
-                       'org'   => TRUE,
-                       'info'  => TRUE,
-               ),
-
-               // ccTLD: China
-               // NIC  : http://www.cnnic.net.cn/en/index/
-               // Whois: http://ewhois.cnnic.cn/
-               'cn' => array(
-                       // Provisional Administrative Rules for Registration of Domain Names in China
-                       // http://www.cnnic.net.cn/html/Dir/2003/11/27/1520.htm
-
-                       // Organizational
-                       'ac'  => TRUE,
-                       'com' => TRUE,
-                       'edu' => TRUE,
-                       'gov' => TRUE,
-                       'net' => TRUE,
-                       'org' => TRUE,
-
-                       // Geographic
-                       'ah' => TRUE,
-                       'bj' => TRUE,
-                       'cq' => TRUE,
-                       'fj' => TRUE,
-                       'gd' => TRUE,
-                       'gs' => TRUE,
-                       'gx' => TRUE,
-                       'gz' => TRUE,
-                       'ha' => TRUE,
-                       'hb' => TRUE,
-                       'he' => TRUE,
-                       'hi' => TRUE,
-                       'hk' => TRUE,
-                       'hl' => TRUE,
-                       'hn' => TRUE,
-                       'jl' => TRUE,
-                       'js' => TRUE,
-                       'jx' => TRUE,
-                       'ln' => TRUE,
-                       'mo' => TRUE,
-                       'nm' => TRUE,
-                       'nx' => TRUE,
-                       'qh' => TRUE,
-                       'sc' => TRUE,
-                       'sd' => TRUE,
-                       'sh' => TRUE,
-                       'sn' => TRUE,
-                       'sx' => TRUE,
-                       'tj' => TRUE,
-                       'tw' => TRUE,
-                       'xj' => TRUE,
-                       'xz' => TRUE,
-                       'yn' => TRUE,
-                       'zj' => TRUE,
-               ),
-
-               // ccTLD: South Korea
-               // NIC  : http://www.nic.or.kr/english/
-               // Whois: http://whois.nida.or.kr/english/
-               'kr' => array(
-                       // .kr domain policy [appendix 1] : Qualifications for Second Level Domains
-                       // http://domain.nida.or.kr/eng/policy.jsp
-
-                       // Organizational
-                       'co'  => TRUE,
-                       'ne ' => TRUE,
-                       'or ' => TRUE,
-                       're ' => TRUE,
-                       'pe'  => TRUE,
-                       'go ' => TRUE,
-                       'mil' => TRUE,
-                       'ac'  => TRUE,
-                       'hs'  => TRUE,
-                       'ms'  => TRUE,
-                       'es'  => TRUE,
-                       'sc'  => TRUE,
-                       'kg'  => TRUE,
-
-                       // Geographic
-                       'seoul'     => TRUE,
-                       'busan'     => TRUE,
-                       'daegu'     => TRUE,
-                       'incheon'   => TRUE,
-                       'gwangju'   => TRUE,
-                       'daejeon'   => TRUE,
-                       'ulsan'     => TRUE,
-                       'gyeonggi'  => TRUE,
-                       'gangwon'   => TRUE,
-                       'chungbuk'  => TRUE,
-                       'chungnam'  => TRUE,
-                       'jeonbuk'   => TRUE,
-                       'jeonnam'   => TRUE,
-                       'gyeongbuk' => TRUE,
-                       'gyeongnam' => TRUE,
-                       'jeju'      => TRUE,
-               ),
+       static $domain;
 
-               // ccTLD: Japan
-               // NIC  : http://jprs.co.jp/en/
-               // Whois: http://whois.jprs.jp/en/
-               'jp' => array(
-                       // Guide to JP Domain Name
-                       // http://jprs.co.jp/en/jpdomain.html
-
-                       // Organizational
-                       'ac' => TRUE,
-                       'ad' => TRUE,
-                       'co' => TRUE,
-                       'go' => TRUE,
-                       'gr' => TRUE,
-                       'lg' => TRUE,
-                       'ne' => TRUE,
-                       'or' => TRUE,
-
-                       // Geographic
-                       //
-                       // Examples for 3rd level domains
-                       //'kumamoto'  => array(
-                       //      // http://www.pref.kumamoto.jp/link/list.asp#4
-                       //      'amakusa'   => TRUE,
-                       //      'hitoyoshi' => TRUE,
-                       //      'jonan'     => TRUE,
-                       //      'kumamoto'  => TRUE,
-                       //      ...
-                       //),
-                       'aichi'     => TRUE,
-                       'akita'     => TRUE,
-                       'aomori'    => TRUE,
-                       'chiba'     => TRUE,
-                       'ehime'     => TRUE,
-                       'fukui'     => TRUE,
-                       'fukuoka'   => TRUE,
-                       'fukushima' => TRUE,
-                       'gifu'      => TRUE,
-                       'gunma'     => TRUE,
-                       'hiroshima' => TRUE,
-                       'hokkaido'  => TRUE,
-                       'hyogo'     => TRUE,
-                       'ibaraki'   => TRUE,
-                       'ishikawa'  => TRUE,
-                       'iwate'     => TRUE,
-                       'kagawa'    => TRUE,
-                       'kagoshima' => TRUE,
-                       'kanagawa'  => TRUE,
-                       'kawasaki'  => TRUE,
-                       'kitakyushu'=> TRUE,
-                       'kobe'      => TRUE,
-                       'kochi'     => TRUE,
-                       'kumamoto'  => TRUE,
-                       'kyoto'     => TRUE,
-                       'mie'       => TRUE,
-                       'miyagi'    => TRUE,
-                       'miyazaki'  => TRUE,
-                       'nagano'    => TRUE,
-                       'nagasaki'  => TRUE,
-                       'nagoya'    => TRUE,
-                       'nara'      => TRUE,
-                       'niigata'   => TRUE,
-                       'oita'      => TRUE,
-                       'okayama'   => TRUE,
-                       'okinawa'   => TRUE,
-                       'osaka'     => TRUE,
-                       'saga'      => TRUE,
-                       'saitama'   => TRUE,
-                       'sapporo'   => TRUE,
-                       'sendai'    => TRUE,
-                       'shiga'     => TRUE,
-                       'shimane'   => TRUE,
-                       'shizuoka'  => TRUE,
-                       'tochigi'   => TRUE,
-                       'tokushima' => TRUE,
-                       'tokyo'     => TRUE,
-                       'tottori'   => TRUE,
-                       'toyama'    => TRUE,
-                       'wakayama'  => TRUE,
-                       'yamagata'  => TRUE,
-                       'yamaguchi' => TRUE,
-                       'yamanashi' => TRUE,
-                       'yokohama'  => TRUE,
-               ),
-
-               // ccTLD: Ukraine
-               // NIC  : http://www.nic.net.ua/
-               // Whois: http://whois.com.ua/
-               'ua' => array(
-                       // policy for alternative 2nd level domain names (a2ld)
-                       // http://www.nic.net.ua/doc/a2ld
-                       // http://whois.com.ua/
-                       'cherkassy'  => TRUE,
-                       'chernigov'  => TRUE,
-                       'chernovtsy' => TRUE,
-                       'ck'         => TRUE,
-                       'cn'         => TRUE,
-                       'com'        => TRUE,
-                       'crimea'     => TRUE,
-                       'cv'         => TRUE,
-                       'dn'         => TRUE,
-                       'dnepropetrovsk' => TRUE,
-                       'donetsk'    => TRUE,
-                       'dp'         => TRUE,
-                       'edu'        => TRUE,
-                       'gov'        => TRUE,
-                       'if'         => TRUE,
-                       'ivano-frankivsk' => TRUE,
-                       'kh'         => TRUE,
-                       'kharkov'    => TRUE,
-                       'kherson'    => TRUE,
-                       'kiev'       => TRUE,
-                       'kirovograd' => TRUE,
-                       'km'         => TRUE,
-                       'kr'         => TRUE,
-                       'ks'         => TRUE,
-                       'lg'         => TRUE,
-                       'lugansk'    => TRUE,
-                       'lutsk'      => TRUE,
-                       'lviv'       => TRUE,
-                       'mk'         => TRUE,
-                       'net'        => TRUE,
-                       'nikolaev'   => TRUE,
-                       'od'         => TRUE,
-                       'odessa'     => TRUE,
-                       'org'        => TRUE,
-                       'pl'         => TRUE,
-                       'poltava'    => TRUE,
-                       'rovno'      => TRUE,
-                       'rv'         => TRUE,
-                       'sebastopol' => TRUE,
-                       'sumy'       => TRUE,
-                       'te'         => TRUE,
-                       'ternopil'   => TRUE,
-                       'uz'         => TRUE,
-                       'uzhgorod'   => TRUE,
-                       'vinnica'    => TRUE,
-                       'vn'         => TRUE,
-                       'zaporizhzhe' => TRUE,
-                       'zhitomir'   => TRUE,
-                       'zp'         => TRUE,
-                       'zt'         => TRUE,
-               ),
-
-               // ccTLD: United Kingdom
-               // NIC  : http://www.nic.uk/
-               'uk' => array(
-                       // Second Level Domains
-                       // http://www.nic.uk/registrants/aboutdomainnames/sld/
-                       'co'     => TRUE,
-                       'ltd'    => TRUE,
-                       'me'     => TRUE,
-                       'net'    => TRUE,
-                       'nic'    => TRUE,
-                       'org'    => TRUE,
-                       'plc'    => TRUE,
-                       'sch'    => TRUE,
-                       
-                       // Delegated Second Level Domains
-                       // http://www.nic.uk/registrants/aboutdomainnames/sld/delegated/
-                       'ac'     => TRUE,
-                       'gov'    => TRUE,
-                       'mil'    => TRUE,
-                       'mod'    => TRUE,
-                       'nhs'    => TRUE,
-                       'police' => TRUE,
-               ),
+       if ($fqdn === NULL) {
+               $domain = NULL; // Unset
+               return '';
+       }
+       if (! is_string($fqdn)) return '';
 
-               // ccTLD: United States of America
-               // NIC  : http://nic.us/
-               // Whois: http://whois.us/
-               'us' => array(
-                       // See RFC1480
-
-                       // Organizational
-                       'dni',
-                       'fed',
-                       'isa',
-                       'kids',
-                       'nsn',
-
-                       // Geographical
-                       // United States Postal Service: State abbreviations (for postal codes)
-                       // http://www.usps.com/ncsc/lookups/abbreviations.html
-                       'ak' => TRUE, // Alaska
-                       'al' => TRUE, // Alabama
-                       'ar' => TRUE, // Arkansas
-                       'as' => TRUE, // American samoa
-                       'az' => TRUE, // Arizona
-                       'ca' => TRUE, // California
-                       'co' => TRUE, // Colorado
-                       'ct' => TRUE, // Connecticut
-                       'dc' => TRUE, // District of Columbia
-                       'de' => TRUE, // Delaware
-                       'fl' => TRUE, // Florida
-                       'fm' => TRUE, // Federated states of Micronesia
-                       'ga' => TRUE, // Georgia
-                       'gu' => TRUE, // Guam
-                       'hi' => TRUE, // Hawaii
-                       'ia' => TRUE, // Iowa
-                       'id' => TRUE, // Idaho
-                       'il' => TRUE, // Illinois
-                       'in' => TRUE, // Indiana
-                       'ks' => TRUE, // Kansas
-                       'ky' => TRUE, // Kentucky
-                       'la' => TRUE, // Louisiana
-                       'ma' => TRUE, // Massachusetts
-                       'md' => TRUE, // Maryland
-                       'me' => TRUE, // Maine
-                       'mh' => TRUE, // Marshall Islands
-                       'mi' => TRUE, // Michigan
-                       'mn' => TRUE, // Minnesota
-                       'mo' => TRUE, // Missouri
-                       'mp' => TRUE, // Northern mariana islands
-                       'ms' => TRUE, // Mississippi
-                       'mt' => TRUE, // Montana
-                       'nc' => TRUE, // North Carolina
-                       'nd' => TRUE, // North Dakota
-                       'ne' => TRUE, // Nebraska
-                       'nh' => TRUE, // New Hampshire
-                       'nj' => TRUE, // New Jersey
-                       'nm' => TRUE, // New Mexico
-                       'nv' => TRUE, // Nevada
-                       'ny' => TRUE, // New York
-                       'oh' => TRUE, // Ohio
-                       'ok' => TRUE, // Oklahoma
-                       'or' => TRUE, // Oregon
-                       'pa' => TRUE, // Pennsylvania
-                       'pr' => TRUE, // Puerto Rico
-                       'pw' => TRUE, // Palau
-                       'ri' => TRUE, // Rhode Island
-                       'sc' => TRUE, // South Carolina
-                       'sd' => TRUE, // South Dakota
-                       'tn' => TRUE, // Tennessee
-                       'tx' => TRUE, // Texas
-                       'ut' => TRUE, // Utah
-                       'va' => TRUE, // Virginia
-                       'vi' => TRUE, // Virgin Islands
-                       'vt' => TRUE, // Vermont
-                       'wa' => TRUE, // Washington
-                       'wi' => TRUE, // Wisconsin
-                       'wv' => TRUE, // West Virginia
-                       'wy' => TRUE, // Wyoming
-               ),
-       );
+       if (is_ip($fqdn))       return $fqdn;
 
-       if (! is_string($fqdn)) return '';
+       if (! isset($domain)) {
+               $domain = array();
+               if (file_exists(DOMAIN_INI_FILE)) {
+                       include(DOMAIN_INI_FILE);       // Set
+               }
+       }
 
        $result  = array();
        $dcursor = & $domain;
        $array   = array_reverse(explode('.', $fqdn));
        $i = 0;
        while(TRUE) {
+               if (! isset($array[$i])) break;
                $acursor = $array[$i];
                if (is_array($dcursor) && isset($dcursor[$acursor])) {
                        $result[] = & $array[$i];
@@ -2038,14 +1645,15 @@ function whois_responsibility($fqdn = 'foo.bar.example.com', $parent = FALSE, $i
 function spam_dispose()
 {
        get_blocklist(NULL);
+       whois_responsibility(NULL);
 }
 
 // Common bahavior for blocking
 // NOTE: Call this function from various blocking feature, to disgueise the reason 'why blocked'
 function spam_exit($mode = '', $data = array())
 {
-
        $exit = TRUE;
+
        switch ($mode) {
                case '':
                        echo("\n");
@@ -2073,7 +1681,14 @@ function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method
        if (empty($progress['is_spam'])) {
                spam_dispose();
        } else {
-               $target = string($target, 0);   // Removing "\0" etc
+
+// TODO: detect encoding from $target for mbstring functions
+//             $tmp = array();
+//             foreach(array_keys($target) as $key) {
+//                     $tmp[strings($key, 0, FALSE, TRUE)] = strings($target[$key], 0, FALSE, TRUE);   // Removing "\0" etc
+//             }
+//             $target = & $tmp;
+
                pkwk_spamnotify($action, $page, $target, $progress, $method);
                spam_exit($exitmode, $progress);
        }