Newly resurrected two $method: 'area_anchor' and 'area_bbcode'. They can capture...

diff --git a/spam.php b/spam.php
index ae03365..c3c25af 100644 (file)
--- a/spam.php
+++ b/spam.php
@@ -1,5 +1,5 @@
 <?php
-// $Id: spam.php,v 1.72 2006/12/16 02:59:26 henoheno Exp $
+// $Id: spam.php,v 1.73 2006/12/16 04:09:46 henoheno Exp $
 // Copyright (C) 2006 PukiWiki Developers Team
 // License: GPL v2 or (at your option) any later version
 
@@ -185,17 +185,22 @@ function area_pickup($string = '', $method = array())
 {
        $area = array();
 
-       // Anchor tag pair by preg_match_all()
+       // Anchor tag pair by preg_match and preg_match_all()
        // [OK] <a href></a>
        // [OK] <a href=  >Good site!</a>
        // [OK] <a href= "#" >test</a>
        // [OK] <a href="http://nasty.example.com">visit http://nasty.example.com/</a>
        // [OK] <a href=\'http://nasty.example.com/\' >discount foobar</a> 
        // [NG] <a href="http://ng.example.com">visit http://ng.example.com _not_ended_
+       $regex = '#<a\b[^>]*\bhref\b[^>]*>.*?</a\b[^>]*(>)#i';
+       if (isset($method['area_anchor'])) {
+               if (preg_match($regex, $string)) {
+                       $area['area_anchor'] = TRUE;
+               }
+       }
        if (isset($method['uri_anchor'])) {
                $areas = array();
-               preg_match_all('#<a\b[^>]*\bhref\b[^>]*>.*?</a\b[^>]*(>)#i',
-                        $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
+               preg_match_all($regex, $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
                foreach(array_keys($areas) as $_area) {
                        $areas[$_area] =  array(
                                $areas[$_area][0][1], // Area start (<a href>)
@@ -205,16 +210,21 @@ function area_pickup($string = '', $method = array())
                if (! empty($areas)) $area['uri_anchor'] = $areas;
        }
 
-       // phpBB's "BBCode" pair by preg_match_all()
+       // phpBB's "BBCode" pair by preg_match and preg_match_all()
        // [OK] [url][/url]
        // [OK] [url]http://nasty.example.com/[/url]
        // [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';
+       if (isset($method['area_bbcode'])) {
+               if (preg_match($regex, $string)) {
+                       $area['area_bbcode'] = TRUE;
+               }
+       }
        if (isset($method['uri_bbcode'])) {
                $areas = array();
-               preg_match_all('#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#i',
-                        $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
+               preg_match_all($regex, $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
                foreach(array_keys($areas) as $_area) {
                        $areas[$_area] = array(
                                $areas[$_area][0][1], // Area start ([url])
@@ -342,19 +352,23 @@ function spam_uri_pickup($string = '', $method = array())
 
        // Area elevation for '(especially external)link' intension
        if (! empty($array)) {
-               $areas = area_pickup($string, $method);
+               $_method = array();
+               if (isset($method['uri_anchor'])) $_method['uri_anchor'] = & $method['uri_anchor'];
+               if (isset($method['uri_bbcode'])) $_method['uri_bbcode'] = & $method['uri_bbcode'];
+
+               $areas = area_pickup($string, $_method, TRUE);
                if (! empty($areas)) {
                        $area_shadow = array();
-                       foreach(array_keys($array) as $key){
+                       foreach (array_keys($array) as $key) {
                                $area_shadow[$key] = & $array[$key]['area'];
-                               $area_shadow[$key]['uri_anchor'] = 0;
-                               $area_shadow[$key]['uri_bbcode'] = 0;
-                       }
-                       if (isset($areas['uri_anchor'])) {
-                               area_measure($areas['uri_anchor'], $area_shadow, 1, 'uri_anchor');
+                               foreach (array_keys($_method) as $_key) {
+                                       $area_shadow[$key][$_key] = 0;
+                               }
                        }
-                       if (isset($areas['uri_bbcode'])) {
-                               area_measure($areas['uri_bbcode'], $area_shadow, 1, 'uri_bbcode');
+                       foreach (array_keys($_method) as $_key) {
+                               if (isset($areas[$_key])) {
+                                       area_measure($areas[$_key], $area_shadow, 1, $_key);
+                               }
                        }
                }
        }
@@ -678,10 +692,10 @@ function check_uri_spam_method($times = 1, $t_area = 0, $rule = TRUE)
                'non_uniq'    => 3 * $times,    // Allow N duped (and normalized) URIs
 
                // Areas
-               //'area_anchor' => $t_area,     // Using <a href> HTML tag
-               //'area_bbcode' => $t_area,     // Using [url] or [link] BBCode
-               'uri_anchor'  => $t_area,       // URI inside <a href> HTML tag
-               'uri_bbcode'  => $t_area,       // URI inside [url] or [link] BBCode
+               'area_anchor' => $t_area,       // Using <a href> HTML tag
+               'area_bbcode' => $t_area,       // Using [url] or [link] BBCode
+               //'uri_anchor'  => $t_area,     // URI inside <a href> HTML tag
+               //'uri_bbcode'  => $t_area,     // URI inside [url] or [link] BBCode
        );
        if ($rule) {
                $bool = array(
@@ -741,12 +755,32 @@ function check_uri_spam($target = '', $method = array())
                }
                return $progress;
        }
+
+       // Area: There's HTML anchor tag
+       if ((! $asap || ! $is_spam) && isset($method['area_anchor'])) {
+               $key = 'area_anchor';
+               if (area_pickup($target, array($key => TRUE))) {
+                       $sum[$key]    += 1;
+                       $is_spam[$key] = TRUE;
+               }
+       }
+
+       // Area: There's 'BBCode' linking tag
+       if ((! $asap || ! $is_spam) && isset($method['area_bbcode'])) {
+               $key = 'area_bbcode';
+               if (area_pickup($target, array($key => TRUE))) {
+                       $sum[$key]    += 1;
+                       $is_spam[$key] = TRUE;
+               }
+       }
+
+       // URI Init
        $pickups = spam_uri_pickup($target, $method);
        if (empty($pickups)) {
                return $progress;
        }
 
-       // Check quantity
+       // URI: Check quantity
        $sum['quantity'] += count($pickups);
                // URI quantity
        if ((! $asap || ! $is_spam) && isset($method['quantity']) &&
@@ -754,7 +788,7 @@ function check_uri_spam($target = '', $method = array())
                $is_spam['quantity'] = TRUE;
        }
 
-       // URI: Using invalid area: anchor
+       // URI: used inside HTML anchor tag pair
        if ((! $asap || ! $is_spam) && isset($method['uri_anchor'])) {
                $key = 'uri_anchor';
                foreach($pickups as $pickup) {
@@ -770,7 +804,7 @@ function check_uri_spam($target = '', $method = array())
                }
        }
 
-       // URI: Using invalid area: bbcode
+       // URI: used inside 'BBCode' pair
        if ((! $asap || ! $is_spam) && isset($method['uri_bbcode'])) {
                $key = 'uri_bbcode';
                foreach($pickups as $pickup) {
@@ -786,7 +820,7 @@ function check_uri_spam($target = '', $method = array())
                }
        }
 
-       // URI uniqueness (and removing non-uniques)
+       // URI: Uniqueness (and removing non-uniques)
        if ((! $asap || ! $is_spam) && isset($method['non_uniq'])) {
 
                // Destructive normalize of URIs
@@ -811,13 +845,13 @@ function check_uri_spam($target = '', $method = array())
                unset($uris);
        }
 
-       // Unique host
+       // URI: Unique host
        $hosts = array();
        foreach ($pickups as $pickup) $hosts[] = & $pickup['host'];
        $hosts = array_unique($hosts);
        $sum['uniqhost'] += count($hosts);
 
-       // Bad host
+       // URI: Bad host
        if ((! $asap || ! $is_spam) && isset($method['badhost'])) {
                $count = array_count_leaves(is_badhost($hosts, $asap));
                $sum['badhost'] += $count;

diff --git a/spam/spam.php b/spam/spam.php
index ae03365..c3c25af 100644 (file)
--- a/spam/spam.php
+++ b/spam/spam.php
@@ -1,5 +1,5 @@
 <?php
-// $Id: spam.php,v 1.72 2006/12/16 02:59:26 henoheno Exp $
+// $Id: spam.php,v 1.73 2006/12/16 04:09:46 henoheno Exp $
 // Copyright (C) 2006 PukiWiki Developers Team
 // License: GPL v2 or (at your option) any later version
 
@@ -185,17 +185,22 @@ function area_pickup($string = '', $method = array())
 {
        $area = array();
 
-       // Anchor tag pair by preg_match_all()
+       // Anchor tag pair by preg_match and preg_match_all()
        // [OK] <a href></a>
        // [OK] <a href=  >Good site!</a>
        // [OK] <a href= "#" >test</a>
        // [OK] <a href="http://nasty.example.com">visit http://nasty.example.com/</a>
        // [OK] <a href=\'http://nasty.example.com/\' >discount foobar</a> 
        // [NG] <a href="http://ng.example.com">visit http://ng.example.com _not_ended_
+       $regex = '#<a\b[^>]*\bhref\b[^>]*>.*?</a\b[^>]*(>)#i';
+       if (isset($method['area_anchor'])) {
+               if (preg_match($regex, $string)) {
+                       $area['area_anchor'] = TRUE;
+               }
+       }
        if (isset($method['uri_anchor'])) {
                $areas = array();
-               preg_match_all('#<a\b[^>]*\bhref\b[^>]*>.*?</a\b[^>]*(>)#i',
-                        $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
+               preg_match_all($regex, $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
                foreach(array_keys($areas) as $_area) {
                        $areas[$_area] =  array(
                                $areas[$_area][0][1], // Area start (<a href>)
@@ -205,16 +210,21 @@ function area_pickup($string = '', $method = array())
                if (! empty($areas)) $area['uri_anchor'] = $areas;
        }
 
-       // phpBB's "BBCode" pair by preg_match_all()
+       // phpBB's "BBCode" pair by preg_match and preg_match_all()
        // [OK] [url][/url]
        // [OK] [url]http://nasty.example.com/[/url]
        // [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';
+       if (isset($method['area_bbcode'])) {
+               if (preg_match($regex, $string)) {
+                       $area['area_bbcode'] = TRUE;
+               }
+       }
        if (isset($method['uri_bbcode'])) {
                $areas = array();
-               preg_match_all('#\[(url|link)\b[^\]]*\].*?\[/\1\b[^\]]*(\])#i',
-                        $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
+               preg_match_all($regex, $string, $areas, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
                foreach(array_keys($areas) as $_area) {
                        $areas[$_area] = array(
                                $areas[$_area][0][1], // Area start ([url])
@@ -342,19 +352,23 @@ function spam_uri_pickup($string = '', $method = array())
 
        // Area elevation for '(especially external)link' intension
        if (! empty($array)) {
-               $areas = area_pickup($string, $method);
+               $_method = array();
+               if (isset($method['uri_anchor'])) $_method['uri_anchor'] = & $method['uri_anchor'];
+               if (isset($method['uri_bbcode'])) $_method['uri_bbcode'] = & $method['uri_bbcode'];
+
+               $areas = area_pickup($string, $_method, TRUE);
                if (! empty($areas)) {
                        $area_shadow = array();
-                       foreach(array_keys($array) as $key){
+                       foreach (array_keys($array) as $key) {
                                $area_shadow[$key] = & $array[$key]['area'];
-                               $area_shadow[$key]['uri_anchor'] = 0;
-                               $area_shadow[$key]['uri_bbcode'] = 0;
-                       }
-                       if (isset($areas['uri_anchor'])) {
-                               area_measure($areas['uri_anchor'], $area_shadow, 1, 'uri_anchor');
+                               foreach (array_keys($_method) as $_key) {
+                                       $area_shadow[$key][$_key] = 0;
+                               }
                        }
-                       if (isset($areas['uri_bbcode'])) {
-                               area_measure($areas['uri_bbcode'], $area_shadow, 1, 'uri_bbcode');
+                       foreach (array_keys($_method) as $_key) {
+                               if (isset($areas[$_key])) {
+                                       area_measure($areas[$_key], $area_shadow, 1, $_key);
+                               }
                        }
                }
        }
@@ -678,10 +692,10 @@ function check_uri_spam_method($times = 1, $t_area = 0, $rule = TRUE)
                'non_uniq'    => 3 * $times,    // Allow N duped (and normalized) URIs
 
                // Areas
-               //'area_anchor' => $t_area,     // Using <a href> HTML tag
-               //'area_bbcode' => $t_area,     // Using [url] or [link] BBCode
-               'uri_anchor'  => $t_area,       // URI inside <a href> HTML tag
-               'uri_bbcode'  => $t_area,       // URI inside [url] or [link] BBCode
+               'area_anchor' => $t_area,       // Using <a href> HTML tag
+               'area_bbcode' => $t_area,       // Using [url] or [link] BBCode
+               //'uri_anchor'  => $t_area,     // URI inside <a href> HTML tag
+               //'uri_bbcode'  => $t_area,     // URI inside [url] or [link] BBCode
        );
        if ($rule) {
                $bool = array(
@@ -741,12 +755,32 @@ function check_uri_spam($target = '', $method = array())
                }
                return $progress;
        }
+
+       // Area: There's HTML anchor tag
+       if ((! $asap || ! $is_spam) && isset($method['area_anchor'])) {
+               $key = 'area_anchor';
+               if (area_pickup($target, array($key => TRUE))) {
+                       $sum[$key]    += 1;
+                       $is_spam[$key] = TRUE;
+               }
+       }
+
+       // Area: There's 'BBCode' linking tag
+       if ((! $asap || ! $is_spam) && isset($method['area_bbcode'])) {
+               $key = 'area_bbcode';
+               if (area_pickup($target, array($key => TRUE))) {
+                       $sum[$key]    += 1;
+                       $is_spam[$key] = TRUE;
+               }
+       }
+
+       // URI Init
        $pickups = spam_uri_pickup($target, $method);
        if (empty($pickups)) {
                return $progress;
        }
 
-       // Check quantity
+       // URI: Check quantity
        $sum['quantity'] += count($pickups);
                // URI quantity
        if ((! $asap || ! $is_spam) && isset($method['quantity']) &&
@@ -754,7 +788,7 @@ function check_uri_spam($target = '', $method = array())
                $is_spam['quantity'] = TRUE;
        }
 
-       // URI: Using invalid area: anchor
+       // URI: used inside HTML anchor tag pair
        if ((! $asap || ! $is_spam) && isset($method['uri_anchor'])) {
                $key = 'uri_anchor';
                foreach($pickups as $pickup) {
@@ -770,7 +804,7 @@ function check_uri_spam($target = '', $method = array())
                }
        }
 
-       // URI: Using invalid area: bbcode
+       // URI: used inside 'BBCode' pair
        if ((! $asap || ! $is_spam) && isset($method['uri_bbcode'])) {
                $key = 'uri_bbcode';
                foreach($pickups as $pickup) {
@@ -786,7 +820,7 @@ function check_uri_spam($target = '', $method = array())
                }
        }
 
-       // URI uniqueness (and removing non-uniques)
+       // URI: Uniqueness (and removing non-uniques)
        if ((! $asap || ! $is_spam) && isset($method['non_uniq'])) {
 
                // Destructive normalize of URIs
@@ -811,13 +845,13 @@ function check_uri_spam($target = '', $method = array())
                unset($uris);
        }
 
-       // Unique host
+       // URI: Unique host
        $hosts = array();
        foreach ($pickups as $pickup) $hosts[] = & $pickup['host'];
        $hosts = array_unique($hosts);
        $sum['uniqhost'] += count($hosts);
 
-       // Bad host
+       // URI: Bad host
        if ((! $asap || ! $is_spam) && isset($method['badhost'])) {
                $count = array_count_leaves(is_badhost($hosts, $asap));
                $sum['badhost'] += $count;