OSDN Git Service

* Added spam_uri_removing_hocus_pocus().
[pukiwiki/pukiwiki_sandbox.git] / spam / spam.php
index 09a85b0..5681b51 100644 (file)
@@ -1,5 +1,5 @@
 <?php
-// $Id: spam.php,v 1.169 2007/06/07 13:07:06 henoheno Exp $
+// $Id: spam.php,v 1.179 2007/06/16 03:39:39 henoheno Exp $
 // Copyright (C) 2006-2007 PukiWiki Developers Team
 // License: GPL v2 or (at your option) any later version
 //
@@ -43,10 +43,12 @@ function preg_grep_invert($pattern = '//', $input = array())
 // NOTE: If the same data exists, it must be corrupted.
 function var_export_shrink($expression, $return = FALSE, $ignore_numeric_keys = FALSE)
 {
-       $result =preg_replace(
+       $result = var_export($expression, TRUE);
+
+       $result = preg_replace(
                // Remove a newline and spaces
                '# => \n *array \(#', ' => array (',
-               var_export($expression, TRUE)
+               $result
        );
 
        if ($ignore_numeric_keys) {
@@ -109,38 +111,41 @@ 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, $ignore_space = FALSE)
+function strings($binary = '', $min_len = 4, $ignore_space = FALSE, $multibyte = TRUE)
 {
+       // String only
+       $binary = (is_array($binary) || $binary === TRUE) ? '' : strval($binary);
+
+       $regex = $ignore_space ?
+               '[^[:graph:] \t\n]+' :          // Remove "\0" etc, and readable spaces
+               '[^[:graph:][:space:]]+';       // Preserve readable spaces if possible
+
+       $binary = $multibyte ?
+               mb_ereg_replace($regex,           "\n",  $binary) :
+               preg_replace('/' . $regex . '/s', "\n",  $binary);
+
        if ($ignore_space) {
                $binary = preg_replace(
                        array(
-                               '/(?:[^[:graph:] \t\n]|[\r])+/s',
                                '/[ \t]{2,}/',
                                '/^[ \t]/m',
                                '/[ \t]$/m',
                        ),
                        array(
-                               "\n",
                                ' ',
                                '',
                                ''
                        ),
                         $binary);
-       } else {
-               // Remove "\0" etc. Preserve readable spaces if possible.
-               $binary = preg_replace('/(?:[^[:graph:][:space:]]|[\r])+/s', "\n", $binary);
        }
 
        if ($min_len > 1) {
+               // The last character seems "\n" or not
+               $br = (! empty($binary) && $binary[strlen($binary) - 1] == "\n") ? "\n" : '';
+
                $min_len = min(1024, intval($min_len));
                $regex = '/^.{' . $min_len . ',}/S';
-               if (is_array($binary)) {
-                       foreach(array_keys($binary) as $key) {
-                               $binary[$key] = implode("\n", preg_grep($regex, explode("\n", $binary[$key])));
-                       }
-               } else {
-                       $binary = implode("\n", preg_grep($regex, explode("\n", $binary)));
-               }
+               $binary = implode("\n", preg_grep($regex, explode("\n", $binary))) . $br;
        }
 
        return $binary;
@@ -448,6 +453,27 @@ function _preg_replace_callback_domain_exposure($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);
+
+       // Remove words (has no '<>[]:') between spaces
+       $binary = preg_replace('/[ \t][\w.,()\ \t]+[ \t]/', ' ', $binary);
+
+       return $binary;
+}
+
 // Preprocess: rawurldecode() and adding space(s) and something
 // to detect/count some URIs _if possible_
 // NOTE: It's maybe danger to var_dump(result). [e.g. 'javascript:']
@@ -455,11 +481,12 @@ function _preg_replace_callback_domain_exposure($matches = array())
 // [OK] http://victim.example.org/nasty.example.org
 // [OK] http://victim.example.org/go?http%3A%2F%2Fnasty.example.org
 // [OK] http://victim.example.org/http://nasty.example.org
-function spam_uri_pickup_preprocess($string = '')
+function spam_uri_pickup_preprocess($string = '', $method = array())
 {
        if (! is_string($string)) return '';
 
-       $string = rawurldecode($string);
+       $string = spam_uri_removing_hocus_pocus(rawurldecode($string), $method);
+       //var_dump(htmlspecialchars($string));
 
        // Domain exposure (simple)
        // http://victim.example.org/nasty.example.org/path#frag
@@ -538,7 +565,7 @@ function spam_uri_pickup($string = '', $method = array())
                $method = check_uri_spam_method();
        }
 
-       $string = spam_uri_pickup_preprocess($string);
+       $string = spam_uri_pickup_preprocess($string, $method);
 
        $array  = uri_pickup($string);
 
@@ -1204,6 +1231,7 @@ function check_uri_spam($target = '', $method = array())
        foreach(array_keys($method) as $key) {
                if (! isset($sum[$key])) $sum[$key] = 0;
        }
+       if (! isset($sum['quantity'])) $sum['quantity'] = 0;
 
        if (is_array($target)) {
                foreach($target as $str) {
@@ -1409,8 +1437,10 @@ function array_flat_leaves($array, $unique = TRUE)
 }
 
 // An array() to an array leaf
-function array_leaf($array = array('A', 'B', 'C.D'), $stem = FALSE, $edge = array())
+function array_leaf($array = array('A', 'B', 'C.D'), $stem = FALSE, $edge = TRUE)
 {
+       if (! is_array($array)) return $array;
+
        $leaf = array();
        $tmp  = & $leaf;
        foreach($array as $arg) {
@@ -1490,52 +1520,68 @@ function summarize_detail_newtral($progress = array())
            empty($progress['hosts'])) return '';
 
        $result = '';
-       if (FALSE) {
-               // Sort by domain
-               $tmp = array();
-               foreach($progress['hosts'] as $value) {
-                       $tmp[delimiter_reverse($value)] = $value;
-               }
-               ksort($tmp, SORT_STRING);
-               $result = count($tmp) . ' (' .implode(', ', $tmp) . ')';
-       } else {
-               $tmp = array();
-               foreach($progress['hosts'] as $value) {
-                       $tmp = array_merge_recursive(
-                               $tmp,
-                               array_leaf(explode('.', delimiter_reverse($value) . '.'), TRUE, $value)
-                       );
-               }
 
-//var_dump($tmp);
-// TODO: IP address 1.2.3.4 => "0"-3-2-1 by array_daruma_otoshi()
+       // Generate a $trie
+       $trie = array();
+       foreach($progress['hosts'] as $value) {
 
-//var_export('<br>-------------<br>');
-//var_export($tmp);
-//var_export('<br>-------------<br>');
+               // Try to shorten (pre) -- array('example.com', 'bar', 'foo')
+               $resp = whois_responsibility($value);   // 'example.com'
+               $rest = rtrim(substr($value, 0, - strlen($resp)), '.'); // 'foo.bar'
+               if ($rest) {
+                       $parts = explode('.', delimiter_reverse('.' . $rest));
+                       array_unshift($parts, $resp);
+               } else {
+                       $parts = array($resp, $rest);
+               }
 
+               $trie = array_merge_recursive(
+                       $trie,
+                       array_leaf($parts, TRUE, $value)
+               );
+       }
 
-               array_daruma_otoshi($tmp, '.', TRUE); // "domain.tld"
-               array_joinbranch_leaf($tmp, '.', 0, TRUE);
-               foreach($tmp as $key => $value) {
-                       if (is_array($value)) {
-                               ksort($tmp[$key], SORT_STRING);
-                               $tmp[$key] = implode(', ', array_flat_leaves($value));
-                       }
-               }
-               ksort($tmp, SORT_STRING);
+       // Try to shorten (post, non-recursive) -- 'foo.bar.example.com'
+       array_joinbranch_leaf($trie, '.', 0, TRUE);
 
-               $result = var_export_shrink($tmp, TRUE, TRUE);
+       // Sort and flatten -- 'A.foo.bar.example.com, B.foo.bar.example.com'
+       foreach(array_keys($trie) as $key) {
+               if (is_array($trie[$key])) {
+                       ksort_by_domain($trie[$key]);
+                       $trie[$key] = implode(', ', array_flat_leaves($trie[$key]));
+               }
        }
 
-       return $result;
+       // TODO: ltrim('.') from $trie
+
+       ksort_by_domain($trie);
+
+       // TODO: from array('foobar' => 'foobar') to 'foobar'
+
+       return var_export_shrink($trie, TRUE, TRUE);
 }
 
+// ksort() by domain
+function ksort_by_domain(& $array)
+{
+       $sort = array();
+       foreach(array_keys($array) as $key) {
+               $sort[delimiter_reverse($key)] = $key;
+       }
+       ksort($sort, SORT_STRING);
+       $result = array();
+       foreach($sort as $key) {
+               $result[$key] = & $array[$key];
+       }
+       $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));
@@ -1567,91 +1613,423 @@ function array_joinbranch_leaf(& $array, $delim = '.', $limit = 0, $reverse = FA
 
        return $result;
 }
-//$a = array(array()); //=> array()
-//$a = array('F' => array('B' => array('C' => array('d' => array('' => '6')))));
-//$b = array('R' => array('S' => array('T' => array('U' => array('' => '7')))));
-//$a = array('I' => $a, '@' => $b);
-//$a = array('F' => array(5), 0 => array('H'));
-//echo "<br>";
-//var_dump(array_joinbranch_leaf($a, '#', 0, 0));
-//var_export($a);
-//echo "<br>";
-//echo "<br>";
-
-// array('A' => array('B' => 'C')) to
-// array('A.B' => 'C')
-// array(
-//     'A' => array(
-//             'B' => array(
-//                     'C' => array(
-//                             'D' => '1'
-//                     ),
-//             ),
-//     ),
-//     'G' => array(
-//             'H' => '2'
-//     ),
-// )
-// to
-// array (
-//     'G.H'     => '2',
-//     'A.B.C.D' => '1',
-// )
-function array_daruma_otoshi(& $array, $delim = '.', $reverse = FALSE, $recurse = FALSE)
+
+
+// Check responsibility-root of the FQDN
+// 'foo.bar.example.com'        => 'example.com'        (.com        has the last whois for it)
+// 'foo.bar.example.au'         => 'example.au'         (.au         has the last whois for it)
+// 'foo.bar.example.edu.au'     => 'example.edu.au'     (.edu.au     has the last whois for it)
+// 'foo.bar.example.act.edu.au' => 'example.act.edu.au' (.act.edu.au has the last whois for it)
+function whois_responsibility($fqdn = 'foo.bar.example.com', $parent = FALSE, $implicit = TRUE)
 {
-       $result = 0;
+       // 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,
+               ),
 
-       if (! is_array($array) || empty($array)) return $result;
+               // 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,
+               ),
 
-       foreach(array_keys($array) as $key) {
-               $branch = & $array[$key];
-               if (! is_array($branch) || empty($branch)) continue;
-
-               foreach(array_keys($branch) as $bkey) {
-                       $joinkey = $reverse ?
-                               $bkey . $delim . $key :
-                               $key  . $delim . $bkey;
-                       $array[$joinkey] = & $branch[$bkey];
-                       unset($array[$key]);
-                       ++$result;
+               // 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,
+               ),
+
+               // 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,
+               ),
+
+               // 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_string($fqdn)) return '';
+
+       $result  = array();
+       $dcursor = & $domain;
+       $array   = array_reverse(explode('.', $fqdn));
+       $i = 0;
+       while(TRUE) {
+               $acursor = $array[$i];
+               if (is_array($dcursor) && isset($dcursor[$acursor])) {
+                       $result[] = & $array[$i];
+                       $dcursor  = & $dcursor[$acursor];
+               } else {
+                       if (! $parent && isset($acursor)) {
+                               $result[] = & $array[$i];       // Whois servers must know this subdomain
+                       }
+                       break;
                }
+               ++$i;
        }
 
-       // Rescan (Recurse)
-       if ($recurse && $result) {
-               $result = array_daruma_otoshi($array, $delim, $reverse, $recurse);
+       // Implicit responsibility: Top-Level-Domains must not be yours
+       // 'bar.foo.something' => 'foo.something'
+       if ($implicit && count($result) == 1 && count($array) > 1) {
+               $result[] = & $array[1];
        }
 
-       return $result; // Tell me how many
+       return $result ? implode('.', array_reverse($result)) : '';
 }
-//$a = array (
-//     'edu' => array (
-//             'berkeley' => array (
-//                     'polisci' => array (
-//                             '' => 'polisci.berkeley.edu',
-//                     ),
-//             ),
-//             'cmich' => array (
-//                     'rso' => array (
-//                             '' => 'rso.cmich.edu',
-//                     ),
-//             ),
-//     ),
-//);
-//array_daruma_otoshi($a, '.', TRUE);
-//var_export($a);
-
-//$a = array (
-//     '4' => array (
-//             '5' => array (
-//                     '6' => array (
-//                             '' => '7.8.9',
-//                     ),
-//             ),
-//     ),
-//);
-//array_daruma_otoshi($a, '.', TRUE);
-//var_export($a);
 
 
 // ---------------------
@@ -1696,7 +2074,7 @@ function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method
        if (empty($progress['is_spam'])) {
                spam_dispose();
        } else {
-               $target = string($target, 0);   // Removing "\0" etc
+               $target = strings($target, 0);  // Removing "\0" etc
                pkwk_spamnotify($action, $page, $target, $progress, $method);
                spam_exit($exitmode, $progress);
        }