<?php
-// $Id: spam.php,v 1.165 2007/05/20 02:22:56 henoheno Exp $
+// $Id: spam.php,v 1.176 2007/06/14 14:57:37 henoheno Exp $
// Copyright (C) 2006-2007 PukiWiki Developers Team
// License: GPL v2 or (at your option) any later version
//
),
$binary);
} else {
+ // Remove "\0" etc. Preserve readable spaces if possible.
$binary = preg_replace('/(?:[^[:graph:][:space:]]|[\r])+/s', "\n", $binary);
}
if ($min_len > 1) {
$min_len = min(1024, intval($min_len));
- $binary =
- implode("\n",
- preg_grep('/^.{' . $min_len . ',}/S',
- explode("\n", $binary)
- )
- );
+ $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)));
+ }
}
return $binary;
function get_blocklist($list = '')
{
- static $f_dispose = FALSE, $regexes;
+ static $regexes;
if ($list === NULL) {
- $f_dispose = TRUE;
- $regexes = NULL; // Unset
+ $regexes = NULL; // Unset
return array();
}
if (! isset($regexes)) {
- if ($f_dispose === TRUE) die(__FUNCTION__ . '(): Memory already disposed');
-
$regexes = array();
if (file_exists(SPAM_INI_FILE)) {
$blocklist = array();
}
// An array() to an array leaf
-function array_leaf($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 $arg) {
foreach($progress['hosts'] as $value) {
$tmp = array_merge_recursive(
$tmp,
- array_leaf(explode('.', delimiter_reverse($value) . '.'), TRUE, $value)
+ array_leaf(explode('.', delimiter_reverse($value)), TRUE, $value)
);
}
- ksort($tmp, SORT_STRING);
- separate_and_joinkey_leaves($tmp, '.', TRUE, TRUE);
- separate_and_joinkey_leaves($tmp, '.', TRUE, FALSE);
- separate_and_joinkey_leaves($tmp, '.', TRUE, FALSE);
- //separate_and_joinkey_leaves($tmp, '.', TRUE, FALSE);
+//var_dump($tmp);
+// TODO: IP address 1.2.3.4 => "0"-3-2-1 by array_shrinkbranch_leaves()
+ array_shrinkbranch_leaves($tmp, '.', TRUE); // "domain.tld"
+ array_joinbranch_leaf($tmp, '.', 0, TRUE);
foreach($tmp as $key => $value) {
if (is_array($value)) {
- ksort($tmp[$key]);
- // $tmp[$key] = implode(', ', array_flat_leaves($value));
+ ksort($tmp[$key], SORT_STRING);
+ $tmp[$key] = implode(', ', array_flat_leaves($value));
}
}
+ ksort($tmp, SORT_STRING);
$result = var_export_shrink($tmp, TRUE, TRUE);
}
return $result;
}
-function separate_and_joinkey_leaves(
- & $array, // array('A' => array('B' => 'C.D')),
- $delim = '.', $reversejoin = FALSE, $allowmulti = FALSE)
+// 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, $reverse = FALSE)
{
- if (! is_array($array)) return $array;
-
$result = array();
+ if (! is_array($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;
+}
+
+
+// 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_shrinkbranch_leaves(& $array, $delim = '.', $reverse = FALSE, $recurse = FALSE)
+{
+ $result = 0;
+ if (! is_array($array) || empty($array)) return $result;
+
foreach(array_keys($array) as $key) {
- if (! is_array($array[$key]) || (! $allowmulti && count($array[$key]) > 1)) {
- $result[$key] = & $array[$key]; // Do nothing
+ $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]);
+ ++$result;
+ }
+ }
+
+ // Rescan (Recurse)
+ if ($recurse && $result) {
+ $result = array_shrinkbranch_leaves($array, $delim, $reverse, $recurse);
+ }
+
+ return $result; // Tell me how many
+}
+//$a = array (
+// 'edu' => array (
+// 'berkeley' => array (
+// 'polisci' => array (
+// '' => 'polisci.berkeley.edu',
+// ),
+// ),
+// 'cmich' => array (
+// 'rso' => array (
+// '' => 'rso.cmich.edu',
+// ),
+// ),
+// ),
+//);
+//array_shrinkbranch_leaves($a, '.', TRUE);
+//var_export($a);
+
+//$a = array (
+// '4' => array (
+// '5' => array (
+// '6' => array (
+// '' => '7.8.9',
+// ),
+// ),
+// ),
+//);
+//array_shrinkbranch_leaves($a, '.', TRUE);
+//var_export($a);
+
+
+
+// Check responsibility-root of the FQDN
+// 'foo.bar.example.com' => 'example.com' (.com has the last whois for it)
+// 'foo.bar.example.au' => 'example.au' (.au has the last whois for it)
+// 'foo.bar.example.edu.au' => 'example.edu.au' (.edu.au has the last whois for it)
+// 'foo.bar.example.act.edu.au' => 'example.act.edu.au' (.act.edu.au has the last whois for it)
+function whois_responsibility($fqdn = 'foo.bar.example.com', $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/
+ 'act' => TRUE,
+ 'nt' => TRUE,
+ 'nsw' => TRUE,
+ 'qld' => TRUE,
+ 'sa' => TRUE,
+ 'tas' => TRUE,
+ 'vic' => TRUE,
+ 'wa' => TRUE,
+ ),
+ 'gov' => array(
+ '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,
+ ),
+
+ // 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, // www.cherkassy.ua
+ '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(
+ // RFC1480
+
+ // 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
+
+ // Others
+ 'dni',
+ 'fed',
+ 'isa',
+ 'kids',
+ 'nsn',
+ ),
+ );
+
+ 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 {
- foreach(array_keys($array[$key]) as $_key) {
- $joinkey = $reversejoin ?
- $_key . $delim . $key :
- $key . $delim . $_key;
- $result[$joinkey] = & $array[$key][$_key];
+ if (isset($acursor)) {
+ $result[] = & $array[$i]; // Whois servers must know this subdomain
}
+ break;
}
+ ++$i;
}
- $array = & $result;
+ // Implicit responsibility: Top-Level-Domains must not be yours
+ // 'bar.foo.something' => 'foo.something'
+ if ($implicit && count($result) == 1 && count($array) > 1) {
+ $result[] = & $array[1];
+ }
- return $result; // array('A.B' => 'C.D')
+ return $result ? implode('.', array_reverse($result)) : '';
}
// ---------------------
// Exit
+// Freeing memories
+function spam_dispose()
+{
+ get_blocklist(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())
{
- // Dispose
- get_blocklist(NULL);
$exit = TRUE;
switch ($mode) {
{
$progress = check_uri_spam($target, $method);
- if (! empty($progress['is_spam'])) {
- // Mail to administrator(s)
+ if (empty($progress['is_spam'])) {
+ spam_dispose();
+ } else {
+ $target = string($target, 0); // Removing "\0" etc
pkwk_spamnotify($action, $page, $target, $progress, $method);
-
- // Exit
spam_exit($exitmode, $progress);
}
}
OSDN Git Service
