linkify config

skysbird · Jul 30, 2013 · 0f9251a · 0f9251a
1 parent cf8ed38
commit 0f9251a
Show file tree

Hide file tree

Showing 5 changed files with 39 additions and 100 deletions.
diff --git a/configdoc/usage.xml b/configdoc/usage.xml
@@ -446,6 +446,11 @@
    <line>277</line>
   </file>
  </directive>
+ <directive id="Linkify.Hostlist">
+  <file name="HTMLPurifier/Filter/Linkify.php">
+   <line>26</line>
+  </file>
+ </directive>
  <directive id="HTML.SafeIframe">
   <file name="HTMLPurifier/HTMLModule/Iframe.php">
    <line>17</line>

diff --git a/library/HTMLPurifier/ConfigSchema/schema.ser b/library/HTMLPurifier/ConfigSchema/schema.ser
diff --git a/library/HTMLPurifier/ConfigSchema/schema/Linkify.Hostlist.txt b/library/HTMLPurifier/ConfigSchema/schema/Linkify.Hostlist.txt
@@ -0,0 +1,8 @@
+Linkify.Hostlist
+TYPE: list
+VERSION: 1.3.0
+DEFAULT: array()
+--DESCRIPTION--
+List of strings that will auto transform to a link with 
+so <tt>http://www.taobao.com</tt> will transform to  <tt><a href="http://www.taobao.com">http://www.taobao.com</a></tt>.
+--# vim: et sw=4 sts=4
diff --git a/library/HTMLPurifier/Filter/Linkify.php b/library/HTMLPurifier/Filter/Linkify.php
@@ -4,121 +4,46 @@ class HTMLPurifier_Filter_Linkify extends HTMLPurifier_Filter
 {
 
     public $name = 'Linkify';
-
+
+    protected $host_list = array();
     protected $config;
 
+
+    protected function is_safe($uri){
+       $rel = false;
+       foreach($this->host_list as $host){
+           if (fnmatch($host,$uri)) { 
+                $rel = true;
+                break;
+           }
+       }
+
+       return $rel;
+    }
+
     public function postFilter($html, $config, $context) {
+
+        $this->host_list = $config->get('Linkify.Hostlist');
         $this->config = $config;
         $doc = new DOMDocument();
         $doc->loadHTML($html);  
-        $doc->strictErrorChecking = false;
-        libxml_use_internal_errors(true);
         $xpath = new DOMXPath($doc);
         $textnodes = $xpath->query('//text()[not(ancestor::a) and normalize-space()]');
         foreach($textnodes as $node){
-            $parent = $node->parentNode;
             $v = $node->nodeValue;
-            $a = $doc->createElement('a');
-            $a->nodeValue = $v;
-            $a->setAttribute('href',$v);
-            $parent->replaceChild($a,$node);
+            if($this->is_safe($v)){
+                $parent = $node->parentNode;
+                $a = $doc->createElement('a');
+                $a->nodeValue = $v;
+                $a->setAttribute('href',$v);
+                $parent->replaceChild($a,$node);
+            }
         }
         $b = $doc->getElementsByTagName('body')->item(0);
         //print_r($token_list); */
         return $doc->saveHtml($b->firstChild);
     }
 
-    protected function textCb($matches){
-        $text = $matches[1];
-
-        return preg_replace_callback('#((?:https?|ftp)://[^\s\'"<>()]+)#S', array($this,'linkCb'),$text);
-    }
-
-    protected function linkCb($matches){
-        print_r($matches);
-        return "xx";
-        $obj_regex = '#(<object[^>]+>)(.*?)(</object>)#s';
-        $obj_html = $matches[0];
-
-        //check allowScriptAccess
-        $ret = preg_match($this->allowScriptAccess_regex,$matches[0]);
-        if(!$ret){
-            //add allowScriptAccess
-            $obj_html = preg_replace_callback($obj_regex,array($this, 'addScriptAccess'),$obj_html);
-        }
-
-        //check allowNetworking
-        $ret = preg_match($this->allowNetworking_regex,$matches[0]);
-        if(!$ret){
-            //add allowNetworking
-            $obj_html = preg_replace_callback($obj_regex,array($this, 'addNetworking'),$obj_html);
-        }
-
-        //check allowFullScreen
-        $ret = preg_match($this->allowFullScreen_regex,$matches[0]);
-        if(!$ret){
-            //add allowFullScreen
-            $obj_html = preg_replace_callback($obj_regex,array($this, 'addFullScreen'),$obj_html);
-        }
-
-        $param_regex = '#<param[^>]+/>#s';
-
-        return preg_replace_callback($param_regex,array($this, 'paramCb'),$obj_html);
-    }
-
-    protected function addScriptAccess($matches){
-        $to_add = '<param name="allowScriptAccess" value="never" />';
-        return $matches[1].$to_add.$matches[2].$matches[3];
-
-    }
-
-    protected function addNetworking($matches){
-        $to_add = '<param name="allowNetworking" value="internal" />';
-        return $matches[1].$to_add.$matches[2].$matches[3];
-
-    }
-
-
-    protected function addFullScreen($matches){
-        $allow = $this->config->get('HTML.FlashAllowFullScreen');
-        if($allow){
-            $to_add = '<param name="allowFullScreen" value="true" />';
-            return $matches[1].$to_add.$matches[2].$matches[3];
-        }
-        else {
-            return $matches[0];
-        }
-
-    }
-
-    protected function paramCb($matches){
-        $ret = preg_match($this->allowFullScreen_regex,$matches[0]);
-        $html = $matches[0];
-        if($ret){
-            $allow = $this->config->get('HTML.FlashAllowFullScreen');
-            if($allow){
-                $html = '<param name="allowFullScreen" value="true" />';
-            }
-            else{
-                $html = '';
-            }
-        }
-
-        $ret = preg_match($this->allowScriptAccess_regex,$matches[0]);
-        if($ret){
-            $html = '<param name="allowScriptAccess" value="never" />';
-        }
-
-        $ret = preg_match($this->allowNetworking_regex,$matches[0]);
-        if($ret){
-            $html = '<param name="allowNetworking" value="internal" />';
-        }
-
-        return $html;
-    }
-    protected function armorUrl($url) {
-        return str_replace('--', '-&#45;', $url);
-    }
 
 }
 

diff --git a/tests/HTMLPurifierTest.php b/tests/HTMLPurifierTest.php
@@ -119,6 +119,7 @@ function test_linkify(){
 //        $this->config->set('Filter.YouTube', true);
 
         /* use the custom FlashObject Filter to instead of the SafeObject Inector */
+        $this->config->set('Linkify.Hostlist',array('http://*.taobao.com'));
         $this->config->set('Filter.Custom',array(new HTMLPurifier_Filter_Linkify(),new HTMLPurifier_Filter_FlashObject()));
 
         /* use custom Whitelist URIFilter to Filter the unsafe URL */
@@ -130,7 +131,7 @@ function test_linkify(){
         $hm->manager->addModule(new HTMLPurifier_HTMLModule_SafeFlashObject());
 
 
-        $content = "<a><span>http://www.taobao2.com</span></a>";
+        $content = "<span>http://www.taobao.com</span>";
 
         $after = $this->purifier->purify($content,$this->config);