X-Git-Url: http://git.osdn.net/view?a=blobdiff_plain;f=spam%2Fspam.php;h=0f324719a25cb16ec85e95338bbf84d1eec1e04e;hb=7542d07ac60ffe562656a0e1de76e644de4ab56e;hp=49458eba1bac90c7dd782371b3a1f9c5ce818495;hpb=ee258db3dd8f957e0571cd2adb3b2dcd122d1fc8;p=pukiwiki%2Fpukiwiki_sandbox.git
diff --git a/spam/spam.php b/spam/spam.php
index 49458eb..0f32471 100644
--- a/spam/spam.php
+++ b/spam/spam.php
@@ -1,5 +1,5 @@
= 4.3.0): preg_match_all(PREG_OFFSET_CAPTURE): $method['uri_XXX'] related feature
-if (! defined('SPAM_INI_FILE')) define('SPAM_INI_FILE', 'spam.ini.php');
+if (! defined('SPAM_INI_FILE')) define('SPAM_INI_FILE', 'spam.ini.php');
+if (! defined('DOMAIN_INI_FILE')) define('DOMAIN_INI_FILE', 'domain.ini.php');
// ---------------------
// Compat etc
@@ -37,6 +38,133 @@ function preg_grep_invert($pattern = '//', $input = array())
}
}
+// ----
+
+// Very roughly, shrink the lines of var_export()
+// NOTE: If the same data exists, it must be corrupted.
+function var_export_shrink($expression, $return = FALSE, $ignore_numeric_keys = FALSE)
+{
+ $result = var_export($expression, TRUE);
+
+ $result = preg_replace(
+ // Remove a newline and spaces
+ '# => \n *array \(#', ' => array (',
+ $result
+ );
+
+ if ($ignore_numeric_keys) {
+ $result =preg_replace(
+ // Remove numeric keys
+ '#^( *)[0-9]+ => #m', '$1',
+ $result
+ );
+ }
+
+ if ($return) {
+ return $result;
+ } else {
+ echo $result;
+ return NULL;
+ }
+}
+
+// Remove redundant values from array()
+function array_unique_recursive($array = array())
+{
+ if (! is_array($array)) return $array;
+
+ $tmp = array();
+ foreach($array as $key => $value){
+ if (is_array($value)) {
+ $array[$key] = array_unique_recursive($value);
+ } else {
+ if (isset($tmp[$value])) {
+ unset($array[$key]);
+ } else {
+ $tmp[$value] = TRUE;
+ }
+ }
+ }
+
+ return $array;
+}
+
+// Renumber all numeric keys from 0
+function array_renumber_numeric_keys(& $array)
+{
+ if (! is_array($array)) return $array;
+
+ $count = -1;
+ $tmp = array();
+ foreach($array as $key => $value){
+ if (is_array($value)) array_renumber_numeric_keys($array[$key]); // Recurse
+ if (is_numeric($key)) $tmp[$key] = ++$count;
+ }
+ array_rename_keys($array, $tmp);
+
+ return $array;
+}
+
+// Roughly strings(1) using PCRE
+// This function is useful to:
+// * Reduce the size of data, from removing unprintable binary data
+// * Detect _bare_strings_ from binary data
+// References:
+// http://www.freebsd.org/cgi/man.cgi?query=strings (Man-page of GNU strings)
+// http://www.pcre.org/pcre.txt
+// Note: mb_ereg_replace() is one of mbstring extension's functions
+// and need to init its encoding.
+function strings($binary = '', $min_len = 4, $ignore_space = FALSE, $multibyte = FALSE)
+{
+ // String only
+ $binary = (is_array($binary) || $binary === TRUE) ? '' : strval($binary);
+
+ $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) {
+ $binary = preg_replace(
+ array(
+ '/[ \t]{2,}/',
+ '/^[ \t]/m',
+ '/[ \t]$/m',
+ ),
+ array(
+ ' ',
+ '',
+ ''
+ ),
+ $binary);
+ }
+
+ if ($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';
+ $binary = implode("\n", preg_grep($regex, explode("\n", $binary))) . $br;
+ }
+
+ return $binary;
+}
+
+// Reverse $string with specified delimiter
+function delimiter_reverse($string = 'foo.bar.example.com', $from_delim = '.', $to_delim = '.')
+{
+ if (! is_string($string) || ! is_string($from_delim) || ! is_string($to_delim))
+ return $string;
+
+ // com.example.bar.foo
+ return implode($to_delim, array_reverse(explode($from_delim, $string)));
+}
+
+
// ---------------------
// URI pickup
@@ -54,7 +182,7 @@ function uri_pickup($string = '')
preg_match_all(
// scheme://userinfo@host:port/path/or/pathinfo/maybefile.and?query=string#fragment
// Refer RFC3986 (Regex below is not strict)
- '#(\b[a-z][a-z0-9.+-]{1,8}):/+' . // 1: Scheme
+ '#(\b[a-z][a-z0-9.+-]{1,8}):[/\\\]+' . // 1: Scheme
'(?:' .
'([^\s<>"\'\[\]/\#?@]*)' . // 2: Userinfo (Username)
'@)?' .
@@ -62,7 +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.-]+' . // 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
@@ -110,9 +238,9 @@ function uri_pickup_normalize(& $pickups, $destructive = TRUE)
if ($destructive) {
foreach (array_keys($pickups) as $key) {
$_key = & $pickups[$key];
- $_key['scheme'] = isset($_key['scheme']) ? scheme_normalize($_key['scheme']) : '';
+ $_key['scheme'] = isset($_key['scheme']) ? scheme_normalize($_key['scheme']) : '';
$_key['host'] = isset($_key['host']) ? host_normalize($_key['host']) : '';
- $_key['port'] = isset($_key['port']) ? port_normalize($_key['port'], $_key['scheme'], FALSE) : '';
+ $_key['port'] = isset($_key['port']) ? port_normalize($_key['port'], $_key['scheme'], FALSE) : '';
$_key['path'] = isset($_key['path']) ? strtolower(path_normalize($_key['path'])) : '';
$_key['file'] = isset($_key['file']) ? file_normalize($_key['file']) : '';
$_key['query'] = isset($_key['query']) ? query_normalize($_key['query']) : '';
@@ -121,14 +249,13 @@ function uri_pickup_normalize(& $pickups, $destructive = TRUE)
} else {
foreach (array_keys($pickups) as $key) {
$_key = & $pickups[$key];
- $_key['scheme'] = isset($_key['scheme']) ? scheme_normalize($_key['scheme']) : '';
- $_key['host'] = isset($_key['host']) ? strtolower($_key['host']) : '';
- $_key['port'] = isset($_key['port']) ? port_normalize($_key['port'], $_key['scheme'], FALSE) : '';
- $_key['path'] = isset($_key['path']) ? path_normalize($_key['path']) : '';
+ $_key['scheme'] = isset($_key['scheme']) ? scheme_normalize($_key['scheme']) : '';
+ $_key['host'] = isset($_key['host']) ? strtolower($_key['host']) : '';
+ $_key['port'] = isset($_key['port']) ? port_normalize($_key['port'], $_key['scheme'], FALSE) : '';
+ $_key['path'] = isset($_key['path']) ? path_normalize($_key['path']) : '';
}
}
-
return $pickups;
}
@@ -217,7 +344,7 @@ function area_pickup($string = '', $method = array())
// [OK] visit http://nasty.example.com/
// [OK] discount foobar
// [NG] visit http://ng.example.com _not_ended_
- $regex = '#]*\bhref\b[^>]*>.*?]*(>)#i';
+ $regex = '#]*\bhref\b[^>]*>.*?]*(>)#is';
if (isset($method['area_anchor'])) {
$areas = array();
$count = isset($method['asap']) ?
@@ -243,7 +370,7 @@ function area_pickup($string = '', $method = array())
// [OK] [link]http://nasty.example.com/[/link]
// [OK] [url=http://nasty.example.com]visit http://nasty.example.com/[/url]
// [OK] [link http://nasty.example.com/]buy something[/link]
- $regex = '#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#i';
+ $regex = '#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#is';
if (isset($method['area_bbcode'])) {
$areas = array();
$count = isset($method['asap']) ?
@@ -300,7 +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)) {
+ // ''(1) or ''(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 $binary;
+}
+
+// Preprocess: Domain exposure callback (See spam_uri_pickup_preprocess())
// http://victim.example.org/?foo+site:nasty.example.com+bar
// => http://nasty.example.com/?refer=victim.example.org
// NOTE: 'refer=' is not so good for (at this time).
@@ -332,31 +480,71 @@ function _preg_replace_callback_domain_exposure($matches = array())
// 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:']
+// [OK] http://victim.example.org/?site:nasty.example.org
+// [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
-// TODO: link.toolbot.com, urlx.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 (See _preg_replace_callback_domain_exposure())
+ // Domain exposure (simple)
+ // http://victim.example.org/nasty.example.org/path#frag
+ // => http://nasty.example.org/?refer=victim.example.org and original
+ $string = preg_replace(
+ '#h?ttp://' .
+ '(' .
+ 'ime\.nu' . '|' . // 2ch.net
+ 'ime\.st' . '|' . // 2ch.net
+ 'link\.toolbot\.com' . '|' .
+ 'urlx\.org' .
+ ')' .
+ '/([a-z0-9.%_-]+\.[a-z0-9.%_-]+)#i', // nasty.example.org
+ 'http://$2/?refer=$1 $0', // Preserve $0 or remove?
+ $string
+ );
+
+ // Domain exposure (gate-big5)
+ // http://victim.example.org/gate/big5/nasty.example.org/path
+ // => http://nasty.example.org/?refer=victim.example.org and original
+ $string = preg_replace(
+ '#h?ttp://' .
+ '(' .
+ 'big5.51job.com' . '|' .
+ 'big5.china.com' . '|' .
+ 'big5.xinhuanet.com' . '|' .
+ ')' .
+ '/gate/big5' .
+ '/([a-z0-9.%_-]+\.[a-z0-9.%_-]+)' .
+ '#i', // nasty.example.org
+ 'http://$2/?refer=$1 $0', // Preserve $0 or remove?
+ $string
+ );
+
+ // Domain exposure (site:) See _preg_replace_callback_domain_exposure()
$string = preg_replace_callback(
array(
- '#(http)://' .
+ '#(h?ttp)://' . // 1:Scheme
+ // 2:Host
'(' .
+ '(?:[a-z0-9_.-]+\.)?[a-z0-9_-]+\.[a-z0-9_-]+' .
// Something Google: http://www.google.com/supported_domains
- '(?:[a-z0-9.]+\.)?google\.[a-z]{2,3}(?:\.[a-z]{2})?' .
- '|' .
- // AltaVista
- '(?:[a-z0-9.]+\.)?altavista.com' .
-
+ // AltaVista: http://es.altavista.com/web/results?q=site%3Anasty.example.org+foobar
+ // Live Search: search.live.com
+ // MySpace: http://sads.myspace.com/Modules/Search/Pages/Search.aspx?_snip_&searchString=site:nasty.example.org
+ // (also searchresults.myspace.com)
+ // alltheweb.com
+ // search.bbc.co.uk
+ // search.orange.co.uk
+ // ...
')' .
'/' .
- '([a-z0-9?=&.%_/\'\\\+-]+)' . // path/?query=foo+bar+
- '\bsite:([a-z0-9.%_-]+\.[a-z0-9.%_-]+)' . // site:nasty.example.com
- //'()' . // Preserve or remove?
+ '([a-z0-9?=&.%_/\'\\\+-]+)' . // 3:path/?query=foo+bar+
+ '\bsite:([a-z0-9.%_-]+\.[a-z0-9.%_-]+)' . // 4:site:nasty.example.com
+ '()' . // 5:Preserve or remove?
'#i',
),
'_preg_replace_callback_domain_exposure',
@@ -385,7 +573,7 @@ function spam_uri_pickup($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);
@@ -701,14 +889,20 @@ function file_normalize($file = 'index.html.en')
if (isset($simple_defaults[$_file])) return '';
- // [Apache 2 Content-negotiation (type-map)]
- // Roughly removing language/character-set/encoding suffixes,
- // (See Apache 2 document about 'mod_mime' and 'mod_negotiation',
- // http://www.iana.org/assignments/character-sets, RFC3066, and ISO 639))
+ // Roughly removing language/character-set/encoding suffixes
+ // References:
+ // * Apache 2 document about 'Content-negotiaton', 'mod_mime' and 'mod_negotiation'
+ // http://httpd.apache.org/docs/2.0/content-negotiation.html
+ // http://httpd.apache.org/docs/2.0/mod/mod_mime.html
+ // http://httpd.apache.org/docs/2.0/mod/mod_negotiation.html
+ // * http://www.iana.org/assignments/character-sets
+ // * RFC3066: Tags for the Identification of Languages
+ // http://www.ietf.org/rfc/rfc3066.txt
+ // * ISO 639: codes of 'language names'
$suffixes = explode('.', $_file);
$body = array_shift($suffixes);
if ($suffixes) {
- // Remove the liast .gz/.z
+ // Remove the last .gz/.z
$last_key = end(array_keys($suffixes));
if (isset($encoding_suffix[$suffixes[$last_key]])) {
unset($suffixes[$last_key]);
@@ -736,6 +930,7 @@ function file_normalize($file = 'index.html.en')
// [OK] nothing==&eg=dummy&eg=padding&eg=foobar => eg=foobar
function query_normalize($string = '', $equal = TRUE, $equal_cutempty = TRUE, $stortolower = TRUE)
{
+ if (! is_string($string)) return '';
if ($stortolower) $string = strtolower($string);
$array = explode('&', $string);
@@ -801,6 +996,8 @@ function generate_glob_regex($string = '', $divider = '/')
// 23 => ']',
);
+ if (! is_string($string)) return '';
+
$string = str_replace($from, $mid, $string); // Hide
$string = preg_quote($string, $divider);
$string = str_replace($mid, $to, $string); // Unhide
@@ -833,6 +1030,8 @@ function is_ip($string = '')
// TODO: IPv4, CIDR?, IPv6
function generate_host_regex($string = '', $divider = '/')
{
+ if (! is_string($string)) return '';
+
if (mb_strpos($string, '.') === FALSE)
return generate_glob_regex($string, $divider);
@@ -857,10 +1056,15 @@ function generate_host_regex($string = '', $divider = '/')
function get_blocklist($list = '')
{
- static $regexs;
+ static $regexes;
- if (! isset($regexs)) {
- $regexs = array();
+ if ($list === NULL) {
+ $regexes = NULL; // Unset
+ return array();
+ }
+
+ if (! isset($regexes)) {
+ $regexes = array();
if (file_exists(SPAM_INI_FILE)) {
$blocklist = array();
include(SPAM_INI_FILE);
@@ -868,27 +1072,37 @@ function get_blocklist($list = '')
// '*.blogspot.com', // Blog services's subdomains (only)
// 'IANA-examples' => '#^(?:.*\.)?example\.(?:com|net|org)$#',
// );
- foreach(array('goodhost', 'badhost') as $_list) {
- if (! isset($blocklist[$list])) continue;
+ if (isset($blocklist['list'])) {
+ $regexes['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)) {
- $regexs[$_list][$key] = array();
+ $regexes[$_list][$key] = array();
foreach($value as $_key => $_value) {
- get_blocklist_add($regexs[$_list][$key], $_key, $_value);
+ get_blocklist_add($regexes[$_list][$key], $_key, $_value);
}
} else {
- get_blocklist_add($regexs[$_list], $key, $value);
+ get_blocklist_add($regexes[$_list], $key, $value);
}
}
+ unset($blocklist[$_list]);
}
}
}
- if ($list == '') {
- return $regexs; // ALL
- } else if (isset($regexs[$list])) {
- return $regexs[$list];
- } else {
+ if ($list === '') {
+ return $regexes; // ALL
+ } else if (isset($regexes[$list])) {
+ return $regexes[$list];
+ } else {
return array();
}
}
@@ -901,51 +1115,40 @@ 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)
+// Blocklist metrics: Separate $host, to $blocked and not blocked
+function blocklist_distiller(& $hosts, $keys = array('goodhost', 'badhost'), $asap = FALSE)
{
- $result = array();
if (! is_array($hosts)) $hosts = array($hosts);
- foreach(array_keys($hosts) as $key) {
- if (! is_string($hosts[$key])) unset($hosts[$key]);
- }
- if (empty($hosts)) return $result;
-
- foreach (get_blocklist('goodhost') as $regex) {
- $hosts = preg_grep_invert($regex, $hosts);
- }
- if (empty($hosts)) return $result;
-
- $tmp = array();
- 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]) && $asap) break;
+ if (! is_array($keys)) $keys = array($keys);
+
+ $list = get_blocklist('list');
+ $blocked = array();
+
+ foreach($keys as $key){
+ foreach (get_blocklist($key) as $label => $regex) {
+ if (is_array($regex)) {
+ foreach($regex as $_label => $_regex) {
+ $group = preg_grep($_regex, $hosts);
+ if ($group) {
+ $hosts = array_diff($hosts, $group);
+ $blocked[$key][$label][$_label] = $group;
+ if ($asap && $list[$key]) break;
+ }
+ }
+ } else {
+ $group = preg_grep($regex, $hosts);
+ if ($group) {
+ $hosts = array_diff($hosts, $group);
+ $blocked[$key][$label] = $group;
+ if ($asap && $list[$key]) break;
+ }
}
- if (empty($result[$label])) unset($result[$label]);
- } else {
- if (is_badhost_avail($label, $regex, $hosts, $result) && $asap) break;
}
}
- $remains = $hosts;
-
- return $result;
-}
-
-// Subroutine for is_badhost()
-function is_badhost_avail($label = '*.example.org', $regex = '/^.*\.example\.org$/', & $hosts, & $result)
-{
- $group = preg_grep($regex, $hosts);
- if ($group) {
- $result[$label] = & $group;
- $hosts = array_diff($hosts, $result[$label]);
- return TRUE;
- } else {
- return FALSE;
- }
+ return $blocked;
}
// Default (enabled) methods and thresholds (for content insertion)
@@ -988,70 +1191,92 @@ function check_uri_spam_method($times = 1, $t_area = 0, $rule = TRUE)
// Simple/fast spam check
function check_uri_spam($target = '', $method = array())
{
- if (! is_array($method) || empty($method)) {
- $method = check_uri_spam_method();
- }
+ // Return value
$progress = array(
+ 'method' => array(
+ // Theme to do => Dummy, optional value, or optional array()
+ //'quantity' => 8,
+ //'uniqhost' => TRUE,
+ //'non_uniqhost'=> 3,
+ //'non_uniquri' => 3,
+ //'badhost' => TRUE,
+ //'area_anchor' => 0,
+ //'area_bbcode' => 0,
+ //'uri_anchor' => 0,
+ //'uri_bbcode' => 0,
+ ),
'sum' => array(
- 'quantity' => 0,
- 'uniqhost' => 0,
- 'non_uniqhost'=> 0,
- 'non_uniquri' => 0,
- 'badhost' => 0,
- 'area_anchor' => 0,
- 'area_bbcode' => 0,
- 'uri_anchor' => 0,
- 'uri_bbcode' => 0,
+ // Theme => Volume found (int)
+ ),
+ 'is_spam' => array(
+ // Flag. If someting defined here,
+ // one or more spam will be included
+ // in this report
+ ),
+ 'blocked' => array(
+ // Hosts blocked
+ //'category' => array(
+ // 'host',
+ //)
+ ),
+ 'hosts' => array(
+ // Hosts not blocked
),
- 'is_spam' => array(),
- 'method' => & $method,
- 'remains' => array(),
- 'error' => array(),
);
+
+ // Aliases
$sum = & $progress['sum'];
$is_spam = & $progress['is_spam'];
- $remains = & $progress['remains'];
- $error = & $progress['error'];
+ $progress['method'] = & $method; // Argument
+ $blocked = & $progress['blocked'];
+ $hosts = & $progress['hosts'];
$asap = isset($method['asap']);
- // Recurse
+ // Init
+ if (! is_array($method) || empty($method)) {
+ $method = check_uri_spam_method();
+ }
+ foreach(array_keys($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) {
- // Recurse
- $_progress = check_uri_spam($str, $method);
- $_sum = & $_progress['sum'];
- $_is_spam = & $_progress['is_spam'];
- $_remains = & $_progress['remains'];
- $_error = & $_progress['error'];
+ if (! is_string($str)) continue;
+
+ $_progress = check_uri_spam($str, $method); // Recurse
+
+ // Merge $sum
+ $_sum = & $_progress['sum'];
foreach (array_keys($_sum) as $key) {
- $sum[$key] += $_sum[$key];
- }
- foreach (array_keys($_is_spam) as $key) {
- if (is_array($_is_spam[$key])) {
- // Marge keys (badhost)
- foreach(array_keys($_is_spam[$key]) as $_key) {
- if (! isset($is_spam[$key][$_key])) {
- $is_spam[$key][$_key] = $_is_spam[$key][$_key];
- } else {
- $is_spam[$key][$_key] += $_is_spam[$key][$_key];
- }
- }
+ if (! isset($sum[$key])) {
+ $sum[$key] = & $_sum[$key];
} else {
- $is_spam[$key] = TRUE;
+ $sum[$key] += $_sum[$key];
}
}
- foreach ($_remains as $key=>$value) {
- foreach ($value as $_key=>$_value) {
- if (is_int($_key)) {
- $remains[$key][] = $_value;
- } else {
- $remains[$key][$_key] = $_value;
- }
- }
+
+ // Merge $is_spam
+ $_is_spam = & $_progress['is_spam'];
+ foreach (array_keys($_is_spam) as $key) {
+ $is_spam[$key] = TRUE;
+ if ($asap) break;
}
- if (! empty($_error)) $error += $_error;
if ($asap && $is_spam) break;
+
+ // Merge only
+ $blocked = array_merge_recursive($blocked, $_progress['blocked']);
+ $hosts = array_merge_recursive($hosts, $_progress['hosts']);
}
+
+ // Unique values
+ $blocked = array_unique_recursive($blocked);
+ $hosts = array_unique_recursive($hosts);
+
+ // Recount $sum['badhost']
+ $sum['badhost'] = array_count_leaves($blocked);
+
return $progress;
}
@@ -1085,8 +1310,7 @@ function check_uri_spam($target = '', $method = array())
if ($asap && $is_spam) return $progress;
// URI: Pickup
- $pickups = spam_uri_pickup($target, $method);
- //$remains['uri_pickup'] = & $pickups;
+ $pickups = uri_pickup_normalize(spam_uri_pickup($target, $method));
// Return if ...
if (empty($pickups)) return $progress;
@@ -1134,8 +1358,6 @@ function check_uri_spam($target = '', $method = array())
// URI: Uniqueness (and removing non-uniques)
if ((! $asap || ! $is_spam) && isset($method['non_uniquri'])) {
- uri_pickup_normalize($pickups);
-
$uris = array();
foreach (array_keys($pickups) as $key) {
$uris[$key] = uri_pickup_implode($pickups[$key]);
@@ -1159,10 +1381,8 @@ function check_uri_spam($target = '', $method = array())
if ($asap && $is_spam) return $progress;
// Host: Uniqueness (uniq / non-uniq)
- $hosts = array();
foreach ($pickups as $pickup) $hosts[] = & $pickup['host'];
$hosts = array_unique($hosts);
- //$remains['uniqhost'] = & $hosts;
$sum['uniqhost'] += count($hosts);
if ((! $asap || ! $is_spam) && isset($method['non_uniqhost'])) {
$sum['non_uniqhost'] = $sum['quantity'] - $sum['uniqhost'];
@@ -1174,51 +1394,82 @@ function check_uri_spam($target = '', $method = array())
// Return if ...
if ($asap && $is_spam) return $progress;
- // URI: Bad host
+ // URI: Bad host (Separate good/bad hosts from $hosts)
if ((! $asap || ! $is_spam) && isset($method['badhost'])) {
- $__remains = array();
- $badhost = is_badhost($hosts, $asap, $__remains);
- if (! $asap) {
- if ($__remains) {
- $remains['badhost'] = array();
- foreach ($__remains as $value) {
- $remains['badhost'][$value] = TRUE;
- }
- }
- }
- unset($__remains);
- if (! empty($badhost)) {
- //var_dump($badhost); // BADHOST detail
- $sum['badhost'] += array_count_leaves($badhost);
- foreach(array_keys($badhost) as $keys) {
- $is_spam['badhost'][$keys] =
- array_count_leaves($badhost[$keys]);
- }
- unset($badhost);
+
+ // is_badhost()
+ $list = get_blocklist('list');
+ $blocked = blocklist_distiller($hosts, array_keys($list), $asap);
+ foreach($list as $key=>$type){
+ if (! $type) unset($blocked[$key]); // Ignore goodhost etc
}
+ unset($list);
+
+ if (! empty($blocked)) $is_spam['badhost'] = TRUE;
}
return $progress;
}
-// Count leaves
-function array_count_leaves($array = array(), $count_empty_array = FALSE)
+// Count leaves (A leaf = value that is not an array, or an empty array)
+function array_count_leaves($array = array(), $count_empty = FALSE)
{
- if (! is_array($array) || (empty($array) && $count_empty_array))
- return 1;
+ if (! is_array($array) || (empty($array) && $count_empty)) return 1;
// Recurse
- $result = 0;
+ $count = 0;
foreach ($array as $part) {
- $result += array_count_leaves($part, $count_empty_array);
+ $count += array_count_leaves($part, $count_empty);
}
- return $result;
+ return $count;
}
+// An array-leaves to a flat array
+function array_flat_leaves($array, $unique = TRUE)
+{
+ if (! is_array($array)) return $array;
+
+ $tmp = array();
+ foreach(array_keys($array) as $key) {
+ if (is_array($array[$key])) {
+ // Recurse
+ foreach(array_flat_leaves($array[$key]) as $_value) {
+ $tmp[] = $_value;
+ }
+ } else {
+ $tmp[] = & $array[$key];
+ }
+ }
+
+ return $unique ? array_values(array_unique($tmp)) : $tmp;
+}
+
+// An array() to an array leaf
+function array_leaf($array = array('A', 'B', 'C.D'), $stem = FALSE, $edge = TRUE)
+{
+ if (! is_array($array)) return $array;
+
+ $leaf = array();
+ $tmp = & $leaf;
+ foreach($array as $arg) {
+ if (! is_string($arg) && ! is_int($arg)) continue;
+ $tmp[$arg] = array();
+ $parent = & $tmp;
+ $tmp = & $tmp[$arg];
+ }
+ if ($stem) {
+ $parent[key($parent)] = & $edge;
+ } else {
+ $parent = key($parent);
+ }
+
+ return $leaf; // array('A' => array('B' => 'C.D'))
+}
+
+
// ---------------------
// Reporting
-// TODO: Don't show unused $method!
// Summarize $progress (blocked only)
function summarize_spam_progress($progress = array(), $blockedonly = FALSE)
{
@@ -1229,7 +1480,7 @@ function summarize_spam_progress($progress = array(), $blockedonly = FALSE)
$method = & $progress['method'];
if (isset($progress['sum'])) {
foreach ($progress['sum'] as $key => $value) {
- if (isset($method[$key])) {
+ if (isset($method[$key]) && $value) {
$tmp[] = $key . '(' . $value . ')';
}
}
@@ -1239,15 +1490,174 @@ function summarize_spam_progress($progress = array(), $blockedonly = FALSE)
return implode(', ', $tmp);
}
+function summarize_detail_badhost($progress = array())
+{
+ if (! isset($progress['blocked']) || empty($progress['blocked'])) return '';
+
+ // Flat per group
+ $blocked = array();
+ foreach($progress['blocked'] as $list => $lvalue) {
+ foreach($lvalue as $group => $gvalue) {
+ $flat = implode(', ', array_flat_leaves($gvalue));
+ if ($flat === $group) {
+ $blocked[$list][] = $flat;
+ } else {
+ $blocked[$list][$group] = $flat;
+ }
+ }
+ }
+
+ // Shrink per list
+ // From: 'A-1' => array('ie.to')
+ // To: 'A-1' => 'ie.to'
+ foreach($blocked as $list => $lvalue) {
+ if (is_array($lvalue) &&
+ count($lvalue) == 1 &&
+ is_numeric(key($lvalue))) {
+ $blocked[$list] = current($lvalue);
+ }
+ }
+
+ return var_export_shrink($blocked, TRUE, TRUE);
+}
+
+function summarize_detail_newtral($progress = array())
+{
+ if (! isset($progress['hosts']) ||
+ ! is_array($progress['hosts']) ||
+ empty($progress['hosts'])) return '';
+
+ // Generate a responsible $trie
+ $trie = array();
+ foreach($progress['hosts'] as $value) {
+ // 'A.foo.bar.example.com'
+ $resp = whois_responsibility($value); // 'example.com'
+ if (empty($resp)) {
+ // One or more test, or do nothing here
+ $resp = strval($value);
+ $rest = '';
+ } else {
+ $rest = rtrim(substr($value, 0, - strlen($resp)), '.'); // 'A.foo.bar'
+ }
+ $trie = array_merge_recursive($trie, array($resp => array($rest => NULL)));
+ }
+
+ // Format: var_export_shrink() -like output
+ $result = array();
+ ksort_by_domain($trie);
+ foreach(array_keys($trie) as $key) {
+ ksort_by_domain($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]);
+ }
+ return
+ 'array (' . "\n" .
+ implode("\n", $result) . "\n" .
+ ')';
+}
+
+// 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 = $result;
+}
+
+// Check responsibility-root of the FQDN
+// 'foo.bar.example.com' => 'example.com' (.com has the last whois for it)
+// 'foo.bar.example.au' => 'example.au' (.au has the last whois for it)
+// 'foo.bar.example.edu.au' => 'example.edu.au' (.edu.au has the last whois for it)
+// 'foo.bar.example.act.edu.au' => 'example.act.edu.au' (.act.edu.au has the last whois for it)
+function whois_responsibility($fqdn = 'foo.bar.example.com', $parent = FALSE, $implicit = TRUE)
+{
+ static $domain;
+
+ if ($fqdn === NULL) {
+ $domain = NULL; // Unset
+ return '';
+ }
+ if (! is_string($fqdn)) return '';
+
+ if (is_ip($fqdn)) return $fqdn;
+
+ if (! isset($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];
+ $dcursor = & $dcursor[$acursor];
+ } else {
+ if (! $parent && isset($acursor)) {
+ $result[] = & $array[$i]; // Whois servers must know this subdomain
+ }
+ break;
+ }
+ ++$i;
+ }
+
+ // Implicit responsibility: Top-Level-Domains must not be yours
+ // 'bar.foo.something' => 'foo.something'
+ if ($implicit && count($result) == 1 && count($array) > 1) {
+ $result[] = & $array[1];
+ }
+
+ return $result ? implode('.', array_reverse($result)) : '';
+}
+
+
// ---------------------
// Exit
+// Freeing memories
+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"); break;
+ case '':
+ echo("\n");
+ break;
case 'dump':
echo('' . "\n");
echo htmlspecialchars(var_export($data, TRUE));
@@ -1255,8 +1665,7 @@ function spam_exit($mode = '', $data = array())
break;
};
- // Force exit
- exit;
+ if ($exit) exit; // Force exit
}
@@ -1269,11 +1678,18 @@ function pkwk_spamfilter($action, $page, $target = array('title' => ''), $method
{
$progress = check_uri_spam($target, $method);
- if (! empty($progress['is_spam'])) {
- // Mail to administrator(s)
- pkwk_spamnotify($action, $page, $target, $progress, $method);
+ if (empty($progress['is_spam'])) {
+ spam_dispose();
+ } else {
+
+// 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 = & $tmp;
- // Exit
+ pkwk_spamnotify($action, $page, $target, $progress, $method);
spam_exit($exitmode, $progress);
}
}
@@ -1294,23 +1710,13 @@ function pkwk_spamnotify($action, $page, $target = array('title' => ''), $progre
if (! $asap) {
$summary['METRICS'] = summarize_spam_progress($progress);
}
- if (isset($progress['is_spam']['badhost'])) {
- $badhost = array();
- foreach($progress['is_spam']['badhost'] as $glob=>$number) {
- $badhost[] = $glob . '(' . $number . ')';
- }
- $summary['DETAIL_BADHOST'] = implode(', ', $badhost);
- }
- if (! $asap && $progress['remains']['badhost']) {
- $count = count($progress['remains']['badhost']);
- $summary['DETAIL_NEUTRAL_HOST'] = $count .
- ' (' .
- preg_replace(
- '/[^, a-z0-9.-]/i', '',
- implode(', ', array_keys($progress['remains']['badhost']))
- ) .
- ')';
- }
+
+ $tmp = summarize_detail_badhost($progress);
+ if ($tmp != '') $summary['DETAIL_BADHOST'] = $tmp;
+
+ $tmp = summarize_detail_newtral($progress);
+ if (! $asap && $tmp != '') $summary['DETAIL_NEUTRAL_HOST'] = $tmp;
+
$summary['COMMENT'] = $action;
$summary['PAGE'] = '[blocked] ' . (is_pagename($page) ? $page : '');
$summary['URI'] = get_script_uri() . '?' . rawurlencode($page);