Simple whois_responsibility() kills complex array_joinbranch_leaf()

[pukiwiki/pukiwiki_sandbox.git] / spam / spam.php

diff --git a/spam/spam.php b/spam/spam.php
index 106ab90..858be20 100644 (file)
--- a/spam/spam.php
+++ b/spam/spam.php
@@ -1,5 +1,5 @@
 <?php
-// $Id: spam.php,v 1.178 2007/06/16 03:23:34 henoheno Exp $
+// $Id: spam.php,v 1.182 2007/06/16 13:38:00 henoheno Exp $
 // Copyright (C) 2006-2007 PukiWiki Developers Team
 // License: GPL v2 or (at your option) any later version
 //
@@ -111,7 +111,9 @@ 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);
@@ -466,10 +468,10 @@ function spam_uri_removing_hocus_pocus($binary = '', $method = array())
        }
 
        // Removing sequential spaces and too short lines
-       $binary = strings($binary, $length, TRUE, TRUE);
+       $binary = strings($binary, $length, TRUE, FALSE); // Multibyte NOT needed
 
-       // Words between spaces
-       $binary = preg_replace('/[ \t][\w \t]+[ \t]/', ' ', $binary);
+       // Remove words (has no '<>[]:') between spaces
+       $binary = preg_replace('/[ \t][\w.,()\ \t]+[ \t]/', ' ', $binary);
 
        return $binary;
 }
@@ -486,6 +488,7 @@ function spam_uri_pickup_preprocess($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
@@ -1524,7 +1527,7 @@ function summarize_detail_newtral($progress = array())
        $trie = array();
        foreach($progress['hosts'] as $value) {
 
-               // Try to shorten (pre) -- array('example.com', 'bar', 'foo')
+               // array('example.com', 'bar', 'foo')
                $resp = whois_responsibility($value);   // 'example.com'
                $rest = rtrim(substr($value, 0, - strlen($resp)), '.'); // 'foo.bar'
                if ($rest) {
@@ -1540,24 +1543,28 @@ function summarize_detail_newtral($progress = array())
                );
        }
 
-       // Try to shorten (post, non-recursive) -- 'foo.bar.example.com'
-       array_joinbranch_leaf($trie, '.', 0, TRUE);
-
        // Sort and flatten -- 'A.foo.bar.example.com, B.foo.bar.example.com'
        foreach(array_keys($trie) as $key) {
                if (is_array($trie[$key])) {
                        ksort_by_domain($trie[$key]);
+
                        $trie[$key] = implode(', ', array_flat_leaves($trie[$key]));
+                       //$trie[$key] = var_export($trie[$key], TRUE);  // DEBUG
                }
        }
 
-       // TODO: ltrim('.') from $trie
-
        ksort_by_domain($trie);
 
-       // TODO: from array('foobar' => 'foobar') to 'foobar'
+       // Format: From array('foobar' => 'foobar') to 'foobar'
+       $tmp = array();
+       foreach($trie as $key => $value) {
+               $tmp[] = '  \'' .
+                       (($trie[$key] == $key) ? $key : $key . '\' => \'' . $trie[$key])
+                       . '\',';
+               unset($trie[$key]);
+       }
 
-       return var_export_shrink($trie, TRUE, TRUE);
+       return 'array (' . "\n" . implode("\n", $tmp) . "\n" . ')';
 }
 
 // ksort() by domain
@@ -1575,45 +1582,6 @@ 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 $result;
-}
-
-
 // Check responsibility-root of the FQDN
 // 'foo.bar.example.com'        => 'example.com'        (.com        has the last whois for it)
 // 'foo.bar.example.au'         => 'example.au'         (.au         has the last whois for it)
@@ -2073,7 +2041,14 @@ function pkwk_spamfilter($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);
        }