DETAIL_NEUTRAL_HOST: Sort by domain

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

diff --git a/spam/spam.php b/spam/spam.php
index 0b2d91a..669315d 100644 (file)
--- a/spam/spam.php
+++ b/spam/spam.php
@@ -1,5 +1,5 @@
 <?php
-// $Id: spam.php,v 1.144 2007/05/03 15:30:46 henoheno Exp $
+// $Id: spam.php,v 1.150 2007/05/05 07:09:33 henoheno Exp $
 // Copyright (C) 2006-2007 PukiWiki Developers Team
 // License: GPL v2 or (at your option) any later version
 //
@@ -37,6 +37,27 @@ function preg_grep_invert($pattern = '//', $input = array())
        }
 }
 
+// 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 $array;
+}
+
 // Roughly strings(1) using PCRE
 // This function is useful to:
 //   * Reduce the size of data, from removing unprintable binary data
@@ -1091,7 +1112,14 @@ function check_uri_spam($target = '', $method = array())
                        // one or more spam will be included
                        // in this report
                ),
-               'remains' => array(
+               'blocked' => array(
+                       // Hosts blocked
+                       //'category' => array(
+                       //      'host',
+                       //)
+               ),
+               'hosts' => array(
+                       // Hosts not blocked
                ),
        );
 
@@ -1099,7 +1127,8 @@ function check_uri_spam($target = '', $method = array())
        $sum     = & $progress['sum'];
        $is_spam = & $progress['is_spam'];
        $progress['method'] = & $method;        // Argument
-       $remains = & $progress['remains'];
+       $blocked = & $progress['blocked'];
+       $hosts   = & $progress['hosts'];
        $asap    = isset($method['asap']);
 
        // Init
@@ -1129,33 +1158,21 @@ function check_uri_spam($target = '', $method = array())
                        // Merge $is_spam
                        $_is_spam = & $_progress['is_spam'];
                        foreach (array_keys($_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];
-                                               }
-                                       }
-                               } else {
-                                       $is_spam[$key] = TRUE;
-                                       if ($asap) break;
-                               }
+                               $is_spam[$key] = TRUE;
+                               if ($asap) break;
                        }
                        if ($asap && $is_spam) break;
 
-                       // Merge $remains
-                       foreach ($_progress['remains'] as $key=>$value) {
-                               foreach ($value as $_key=>$_value) {
-                                       if (is_int($_key)) {
-                                               $remains[$key][]      = $_value;
-                                       } else {
-                                               $remains[$key][$_key] = $_value;
-                                       }
-                               }
-                       }
+                       // Merge $blocked
+                       $blocked = array_merge_leaves($blocked, $_progress['blocked']);
+
+                       // Merge $hosts
+                       $hosts   = array_merge_leaves($hosts,   $_progress['hosts']);
                }
+
+               // Recount $sum['badhost']
+               $sum['badhost'] = array_count_leaves($blocked);
+
                return $progress;
        }
 
@@ -1260,7 +1277,6 @@ 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);
        $sum['uniqhost'] += count($hosts);
@@ -1285,23 +1301,7 @@ function check_uri_spam($target = '', $method = array())
                }
                unset($list);
 
-               if (! $asap && $hosts) {
-                       $remains['badhost'] = array();
-                       foreach ($hosts as $value) {
-                               $remains['badhost'][$value] = TRUE;
-                       }
-               }
-
-               if (! empty($blocked)) {
-
-                       //var_dump($blocked);   // BADHOST detail
-
-                       $sum['badhost'] += array_count_leaves($blocked);
-                       foreach(array_keys($blocked) as $keys) {
-                               $is_spam['badhost'][$keys] =
-                                       array_count_leaves($blocked[$keys]);
-                       }
-               }
+               if (! empty($blocked)) $is_spam['badhost'] = TRUE;
        }
 
        return $progress;
@@ -1320,12 +1320,19 @@ function array_count_leaves($array = array(), $count_empty = FALSE)
        return $count;
 }
 
-// Merge two leaves
-function array_merge_leaves($array1 = array(), $array2 = array())
+// Merge two leaves' value
+function array_merge_leaves(& $array1, & $array2, $unique_values = TRUE)
 {
-       return array_merge_recursive($array1, $array2);
+       // All NUMERIC keys are always renumbered from 0
+       $array = array_merge_recursive($array1, $array2);
+
+       // Redundant values (and keys) are vanished
+       if ($unique_values) $array = array_unique_recursive($array);
+
+       return $array;
 }
 
+
 // ---------------------
 // Reporting
 
@@ -1340,7 +1347,7 @@ 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 . '(' . $value . ')';
                                }
                        }
@@ -1350,23 +1357,43 @@ function summarize_spam_progress($progress = array(), $blockedonly = FALSE)
        return implode(', ', $tmp);
 }
 
+// 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)
+{
+       $result =preg_replace(
+               // Remove a newline and spaces
+               '# => \n *array \(#', ' => array (',
+               var_export($expression, TRUE)
+       );
+
+       if ($return) {
+               return $result;
+       } else {
+               echo   $result;
+               return NULL;
+       }
+}
+
 function summarize_detail_badhost($progress = array())
 {
        if (! isset($progress['is_spam']['badhost'])) return '';
 
-       $badhost = array();
-       foreach($progress['is_spam']['badhost'] as $glob=>$number) {
-               $badhost[] = $glob . '(' . $number . ')';
-       }
-       return implode(', ', $badhost);
+       return var_export_shrink($progress['blocked'], TRUE);
 }
 
 function summarize_detail_newtral($progress = array())
 {
-       if (! isset($progress['remains']['badhost'])) return '';
+       if (! isset($progress['hosts']) || ! is_array($progress['hosts'])) return '';
+
+       // Sort by domain
+       $tmp = array();
+       foreach($progress['hosts'] as $value) {
+               $tmp[strrev($value)] = $value;
+       }
+       ksort($tmp);
 
-       return count($progress['remains']['badhost']) .
-               ' (' . implode(', ', array_keys($progress['remains']['badhost'])) . ')';
+       return implode(', ', $tmp);
 }