[2.1.0] Further refactoring of AttrDef_URI, creation of new URIFilter and URIDefinition subsystems.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1335 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/AttrDef/URI.php

@@ -25,29 +25,6 @@ HTMLPurifier_ConfigSchema::define(
     'This directive has been available since 1.2.0.'
 );
 
-HTMLPurifier_ConfigSchema::define(
-    'URI', 'DisableExternal', false, 'bool',
-    'Disables links to external websites.  This is a highly effective '.
-    'anti-spam and anti-pagerank-leech measure, but comes at a hefty price: no'.
-    'links or images outside of your domain will be allowed.  Non-linkified '.
-    'URIs will still be preserved.  If you want to be able to link to '.
-    'subdomains or use absolute URIs, specify %URI.Host for your website. '.
-    'This directive has been available since 1.2.0.'
-);
-
-HTMLPurifier_ConfigSchema::define(
-    'URI', 'DisableExternalResources', false, 'bool',
-    'Disables the embedding of external resources, preventing users from '.
-    'embedding things like images from other hosts. This prevents '.
-    'access tracking (good for email viewers), bandwidth leeching, '.
-    'cross-site request forging, goatse.cx posting, and '.
-    'other nasties, but also results in '.
-    'a loss of end-user functionality (they can\'t directly post a pic '.
-    'they posted from Flickr anymore). Use it if you don\'t have a '.
-    'robust user-content moderation team. This directive has been '.
-    'available since 1.3.0.'
-);
-
 HTMLPurifier_ConfigSchema::define(
     'URI', 'DisableResources', false, 'bool',
     'Disables embedding resources, essentially meaning no pictures. You can '.
@@ -117,18 +94,35 @@ class HTMLPurifier_AttrDef_URI extends HTMLPurifier_AttrDef
         $uri = $this->parser->parse($uri);
         if ($uri === false) return false;
         
-        // generic validation
-        $context->register('EmbeddedURI', $this->embedsResource); // flag
-        $result = $uri->validate($config, $context);
-        $context->destroy('EmbeddedURI');
-        if (!$result) return false;
+        // add embedded flag to context for validators
+        $context->register('EmbeddedURI', $this->embedsResource); 
         
-        // scheme-specific validation 
-        $scheme_obj = $uri->getSchemeObj($config, $context);
-        if (!$scheme_obj) return false;
-        if ($this->embedsResource && !$scheme_obj->browsable) return false;
-        $result = $scheme_obj->validate($uri, $config, $context);
-        if (!$result) return false;
+        $ok = false;
+        do {
+            
+            // generic validation
+            $result = $uri->validate($config, $context);
+            if (!$result) break;
+            
+            // chained validation
+            $uri_def =& $config->getDefinition('URI');
+            $result = $uri_def->filter($uri, $config, $context);
+            if (!$result) break;
+            
+            // scheme-specific validation 
+            $scheme_obj = $uri->getSchemeObj($config, $context);
+            if (!$scheme_obj) break;
+            if ($this->embedsResource && !$scheme_obj->browsable) break;
+            $result = $scheme_obj->validate($uri, $config, $context);
+            if (!$result) break;
+            
+            // survived gauntlet
+            $ok = true;
+            
+        } while (false);
+        
+        $context->destroy('EmbeddedURI');
+        if (!$ok) return false;
         
         // back to string
         $result = $uri->toString();

library/HTMLPurifier/Config.php

@@ -5,6 +5,7 @@ require_once 'HTMLPurifier/ConfigSchema.php';
 // member variables
 require_once 'HTMLPurifier/HTMLDefinition.php';
 require_once 'HTMLPurifier/CSSDefinition.php';
+require_once 'HTMLPurifier/URIDefinition.php';
 require_once 'HTMLPurifier/Doctype.php';
 require_once 'HTMLPurifier/DefinitionCacheFactory.php';
 
@@ -313,6 +314,8 @@ class HTMLPurifier_Config
             $this->definitions[$type] = new HTMLPurifier_HTMLDefinition();
         } elseif ($type == 'CSS') {
             $this->definitions[$type] = new HTMLPurifier_CSSDefinition();
+        } elseif ($type == 'URI') {
+            $this->definitions[$type] = new HTMLPurifier_URIDefinition();
         } else {
             trigger_error("Definition of $type type not supported");
             $false = false;

library/HTMLPurifier/URI.php

@@ -1,6 +1,7 @@
 <?php
 
 require_once 'HTMLPurifier/URIParser.php';
+require_once 'HTMLPurifier/URIFilter.php';
 
 /**
  * HTML Purifier's internal representation of a URI
@@ -50,19 +51,7 @@ class HTMLPurifier_URI
         
         // validate host
         if (!is_null($this->host)) {
-            // remove URI if it's absolute and we disabled externals or
-            // if it's absolute and embedded and we disabled external resources
-            unset($our_host); // ensure this variable is not set
-            if (
-                $config->get('URI', 'DisableExternal') ||
-                (
-                    $config->get('URI', 'DisableExternalResources') &&
-                    $context->get('EmbeddedURI', true) // suppress errors
-                )
-            ) {
-                $our_host = $config->get('URI', 'Host');
-                if ($our_host === null) return false;
-            }
+            
             $host_def = new HTMLPurifier_AttrDef_URI_Host();
             $this->host = $host_def->validate($this->host, $config, $context);
             if ($this->host === false) $this->host = null;
@@ -70,16 +59,6 @@ class HTMLPurifier_URI
             // check host against blacklist
             if ($this->checkBlacklist($this->host, $config, $context)) return false;
             
-            // more lenient absolute checking
-            if (isset($our_host)) {
-                $host_parts = array_reverse(explode('.', $this->host));
-                // could be cached
-                $our_host_parts = array_reverse(explode('.', $our_host));
-                foreach ($our_host_parts as $i => $discard) {
-                    if (!isset($host_parts[$i])) return false;
-                    if ($host_parts[$i] != $our_host_parts[$i]) return false;
-                }
-            }
         }
         
         // munge scheme off if necessary

library/HTMLPurifier/URIDefinition.php

@@ -0,0 +1,53 @@
+<?php
+
+require_once 'HTMLPurifier/Definition.php';
+require_once 'HTMLPurifier/URIFilter.php';
+
+require_once 'HTMLPurifier/URIFilter/DisableExternal.php';
+require_once 'HTMLPurifier/URIFilter/DisableExternalResources.php';
+
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'DefinitionRev', 1, 'int', '
+<p>
+    Revision identifier for your custom definition. See
+    %HTML.DefinitionRev for details. This directive has been available
+    since 2.1.0.
+</p>
+');
+
+class HTMLPurifier_URIDefinition extends HTMLPurifier_Definition
+{
+    
+    var $type = 'URI';
+    var $filters = array();
+    var $registeredFilters = array();
+    
+    function HTMLPurifier_URIDefinition() {
+        $this->registerFilter(new HTMLPurifier_URIFilter_DisableExternal());
+        $this->registerFilter(new HTMLPurifier_URIFilter_DisableExternalResources());
+    }
+    
+    function registerFilter($filter) {
+        $this->registeredFilters[$filter->name] = $filter;
+    }
+    
+    function doSetup($config) {
+        foreach ($this->registeredFilters as $name => $filter) {
+            $conf = $config->get('URI', $name);
+            if ($conf !== false && $conf !== null) {
+                $this->filters[$name] = $filter;
+            }
+        }
+        foreach ($this->filters as $n => $x) $this->filters[$n]->prepare($config);
+        unset($this->registeredFilters);
+    }
+    
+    function filter(&$uri, $config, &$context) {
+        foreach ($this->filters as $name => $x) {
+            $result = $this->filters[$name]->filter($uri, $config, $context);
+            if (!$result) return false;
+        }
+        return true;
+    }
+    
+}

library/HTMLPurifier/URIFilter.php

@@ -0,0 +1,24 @@
+<?php
+
+/**
+ * Chainable filters for custom URI processing 
+ */
+class HTMLPurifier_URIFilter
+{
+    var $name;
+    
+    /**
+     * Performs initialization for the filter
+     */
+    function prepare($config) {}
+    
+    /**
+     * Filter a URI object
+     * @param &$uri Reference to URI object
+     * @param $config Instance of HTMLPurifier_Config
+     * @param &$context Instance of HTMLPurifier_Context
+     */
+    function filter(&$uri, $config, &$context) {
+        trigger_error('Cannot call abstract function', E_USER_ERROR);
+    }
+}

library/HTMLPurifier/URIFilter/DisableExternal.php

@@ -0,0 +1,34 @@
+<?php
+
+require_once 'HTMLPurifier/URIFilter.php';
+
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'DisableExternal', false, 'bool',
+    'Disables links to external websites.  This is a highly effective '.
+    'anti-spam and anti-pagerank-leech measure, but comes at a hefty price: no'.
+    'links or images outside of your domain will be allowed.  Non-linkified '.
+    'URIs will still be preserved.  If you want to be able to link to '.
+    'subdomains or use absolute URIs, specify %URI.Host for your website. '.
+    'This directive has been available since 1.2.0.'
+);
+
+class HTMLPurifier_URIFilter_DisableExternal extends HTMLPurifier_URIFilter
+{
+    var $name = 'DisableExternal';
+    var $ourHostParts = false;
+    function prepare($config) {
+        $our_host = $config->get('URI', 'Host');
+        if ($our_host !== null) $this->ourHostParts = array_reverse(explode('.', $our_host));
+    }
+    function filter(&$uri, $config, &$context) {
+        if (is_null($uri->host)) return true;
+        if ($this->ourHostParts === false) return false;
+        $host_parts = array_reverse(explode('.', $uri->host));
+        foreach ($this->ourHostParts as $i => $x) {
+            if (!isset($host_parts[$i])) return false;
+            if ($host_parts[$i] != $this->ourHostParts[$i]) return false;
+        }
+        return true;
+    }
+}
+

library/HTMLPurifier/URIFilter/DisableExternalResources.php

@@ -0,0 +1,26 @@
+<?php
+
+require_once 'HTMLPurifier/URIFilter/DisableExternal.php';
+
+HTMLPurifier_ConfigSchema::define(
+    'URI', 'DisableExternalResources', false, 'bool',
+    'Disables the embedding of external resources, preventing users from '.
+    'embedding things like images from other hosts. This prevents '.
+    'access tracking (good for email viewers), bandwidth leeching, '.
+    'cross-site request forging, goatse.cx posting, and '.
+    'other nasties, but also results in '.
+    'a loss of end-user functionality (they can\'t directly post a pic '.
+    'they posted from Flickr anymore). Use it if you don\'t have a '.
+    'robust user-content moderation team. This directive has been '.
+    'available since 1.3.0.'
+);
+
+class HTMLPurifier_URIFilter_DisableExternalResources extends HTMLPurifier_URIFilter_DisableExternal
+{
+    var $name = 'DisableExternalResources';
+    function filter(&$uri, $config, &$context) {
+        if (!$context->get('EmbeddedURI', true)) return true;
+        return parent::filter($uri, $config, $context);
+    }
+}
+