[1.5.0] Implement TransformToStrict proprietary module

- Factored out strictblockquote from the common definition
- Text module now follows "strict" rules by default
- attr_transform_* now are indexed with string keys, to allow overloading
- Implement ElementDef mergin, and add standalone class variable to ElementDef to prevent half-baked element definitions from masquerading as full ones
- Implement merging global attributes from modules, namely info_attr_transform_post, info_attr_transform_pre and info_tag_transform
- Rename setupInfo() to processModules()
- Fix typo in HTMLModule/Bdo.php

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@731 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/ContentSets.php

@@ -5,7 +5,6 @@ require_once 'HTMLPurifier/ChildDef.php';
 require_once 'HTMLPurifier/ChildDef/Empty.php';
 require_once 'HTMLPurifier/ChildDef/Required.php';
 require_once 'HTMLPurifier/ChildDef/Optional.php';
-require_once 'HTMLPurifier/ChildDef/StrictBlockquote.php'; // transform
 
 class HTMLPurifier_ContentSets
 {
@@ -111,8 +110,6 @@ class HTMLPurifier_ContentSets
                 return new HTMLPurifier_ChildDef_Optional($value);
             case 'empty':
                 return new HTMLPurifier_ChildDef_Empty();
-            case 'strictblockquote':
-                return new HTMLPurifier_ChildDef_StrictBlockquote($value);
             case 'custom':
                 return new HTMLPurifier_ChildDef_Custom($value);
         }

library/HTMLPurifier/ElementDef.php

@@ -7,6 +7,12 @@
 class HTMLPurifier_ElementDef
 {
     
+    /**
+     * Does the definition work by itself, or is it created solely
+     * for the purpose of merging into another definition?
+     */
+    var $standalone = true;
+    
     /**
      * Associative array of attribute name to HTMLPurifier_AttrDef
      * @note Before being processed by HTMLPurifier_AttrCollections
@@ -22,13 +28,13 @@ class HTMLPurifier_ElementDef
     var $attr = array();
     
     /**
-     * List of tag's HTMLPurifier_AttrTransform to be done before validation
+     * Indexed list of tag's HTMLPurifier_AttrTransform to be done before validation
      * @public
      */
     var $attr_transform_pre = array();
     
     /**
-     * List of tag's HTMLPurifier_AttrTransform to be done after validation
+     * Indexed list of tag's HTMLPurifier_AttrTransform to be done after validation
      * @public
      */
     var $attr_transform_post = array();
@@ -79,6 +85,27 @@ class HTMLPurifier_ElementDef
      */
     var $excludes = array();
     
+    /**
+     * Merges the values of another element definition into this one.
+     * Values from the new element def take precedence if a value is
+     * not mergeable.
+     */
+    function mergeIn($def) {
+        
+        // later keys takes precedence
+        foreach($def->attr                  as $k => $v) $this->attr[$k]                = $v;
+        foreach($def->attr_transform_pre    as $k => $v) $this->attr_transform_pre[$k]  = $v;
+        foreach($def->attr_transform_post   as $k => $v) $this->attr_transform_post[$k] = $v;
+        foreach($def->auto_close            as $k => $v) $this->auto_close[$k]          = $v;
+        foreach($def->excludes              as $k => $v) $this->excludes[$k]            = $v;
+        
+        if(!is_null($def->child)) $this->child = $def->child;
+        if(!empty($def->content_model)) $this->content_model .= ' | ' . $def->content_model;
+        if(!empty($def->content_model_type)) $this->content_model_type = $def->content_model_type;
+        if(!is_null($def->descendants_are_inline)) $this->descendants_are_inline = $def->descendants_are_inline;
+        
+    }
+    
 }
 
 ?>

library/HTMLPurifier/HTMLDefinition.php

@@ -34,6 +34,9 @@ require_once 'HTMLPurifier/HTMLModule/Tables.php';
 require_once 'HTMLPurifier/HTMLModule/Image.php';
 require_once 'HTMLPurifier/HTMLModule/StyleAttribute.php';
 
+// compat modules
+require_once 'HTMLPurifier/HTMLModule/TransformToStrict.php';
+
 HTMLPurifier_ConfigSchema::define(
     'HTML', 'EnableAttrID', false, 'bool',
     'Allows the ID attribute in HTML.  This is disabled by default '.
@@ -168,19 +171,19 @@ class HTMLPurifier_HTMLDefinition
     /**
      * Associative array of deprecated tag name to HTMLPurifier_TagTransform
      * @public
-     */ // use + operator
+     */
     var $info_tag_transform = array();
     
     /**
-     * List of HTMLPurifier_AttrTransform to be performed before validation.
+     * Indexed list of HTMLPurifier_AttrTransform to be performed before validation.
      * @public
-     */ // use array_merge or a foreach loop
+     */
     var $info_attr_transform_pre = array();
     
     /**
-     * List of HTMLPurifier_AttrTransform to be performed after validation.
+     * Indexed list of HTMLPurifier_AttrTransform to be performed after validation.
      * @public
-     */ // use array_merge or a foreach loop
+     */
     var $info_attr_transform_post = array();
     
     /**
@@ -241,6 +244,7 @@ class HTMLPurifier_HTMLDefinition
         // this will eventually influence module loading
         $this->strict = $config->get('HTML', 'Strict');
         
+        // order is important!
         $this->modules['Text']          = new HTMLPurifier_HTMLModule_Text();
         $this->modules['Hypertext']     = new HTMLPurifier_HTMLModule_Hypertext();
         $this->modules['List']          = new HTMLPurifier_HTMLModule_List();
@@ -251,6 +255,8 @@ class HTMLPurifier_HTMLDefinition
         $this->modules['Image']         = new HTMLPurifier_HTMLModule_Image();
         $this->modules['StyleAttribute']= new HTMLPurifier_HTMLModule_StyleAttribute();
         
+        $this->modules['TransformToStrict'] = new HTMLPurifier_HTMLModule_TransformToStrict($config);
+        
         $this->attr_types       = new HTMLPurifier_AttrTypes();
         $this->attr_collections = new HTMLPurifier_AttrCollections();
         $this->content_sets     = new HTMLPurifier_ContentSets();
@@ -279,7 +285,7 @@ class HTMLPurifier_HTMLDefinition
         
         // would be nice if we could put each of these in their
         // own object, would make this hookable too!
-        $this->setupInfo($config);
+        $this->processModules($config);
         $this->setupAttrTransform($config);
         $this->setupBlockWrapper($config);
         $this->setupParent($config);
@@ -288,24 +294,39 @@ class HTMLPurifier_HTMLDefinition
     }
     
     /**
-     * Sets up the info array.
+     * Processes the modules, setting up related info variables
      * @param $config Instance of HTMLPurifier_Config
      */
-    function setupInfo($config) {
+    function processModules($config) {
         $this->attr_collections->setup($this->attr_types, $this->modules);
         $this->content_sets->setup($this->modules);
         $this->info_content_sets = $this->content_sets->lookup;
         
         foreach ($this->modules as $module_i => $module) {
+            // process element-wise definitions
             foreach ($module->info as $name => $def) {
-                $def =& $this->modules[$module_i]->info[$name];
+                // setup info
+                if (!isset($this->info[$name])) {
+                    if ($def->standalone) {
+                        $this->info[$name] = $this->modules[$module_i]->info[$name];
+                    } else {
+                        // attempting to merge into an element that doesn't
+                        // exist, ignore it
+                        continue;
+                    }
+                } else {
+                    $this->info[$name]->mergeIn($this->modules[$module_i]->info[$name]);
+                }
+                
+                // process info
+                $def = $this->info[$name];
                 
                 // attribute value expansions
                 $this->attr_collections->performInclusions($def->attr);
                 $this->attr_collections->expandIdentifiers(
                     $def->attr, $this->attr_types);
                 
-                // chameleon data, set descendants_are_inline
+                // descendants_are_inline, for ChildDef_Chameleon
                 if (is_string($def->content_model) &&
                 strpos($def->content_model, 'Inline') !== false) {
                     if ($name != 'del' && $name != 'ins') {
@@ -317,13 +338,16 @@ class HTMLPurifier_HTMLDefinition
                 // set child def from content model
                 $this->content_sets->generateChildDef($def, $module);
                 
-                // setup info
                 $this->info[$name] = $def;
-                if ($this->info_parent == $name) {
+                
-                    $this->info_parent_def = $this->info[$name];
-                }
             }
+            
+            // merge in global info variables from module
+            foreach($module->info_tag_transform         as $k => $v) $this->info_tag_transform[$k]      = $v;
+            foreach($module->info_attr_transform_pre    as $k => $v) $this->info_attr_transform_pre[$k] = $v;
+            foreach($module->info_attr_transform_post   as $k => $v) $this->info_attr_transform_post[$k]= $v;
         }
+        
     }
     
     /**
@@ -369,17 +393,13 @@ class HTMLPurifier_HTMLDefinition
      */
     function setupCompat($config) {
         
+        // convenience for compat
         $e_Inline = new HTMLPurifier_ChildDef_Optional(
                         $this->info_content_sets['Inline'] +
                         array('#PCDATA' => true));
         
-        // blockquote changes, implement in TransformStrict and Legacy
+        // blockquote alt child def, implement in Legacy
-        if ($this->strict) {
+        if (!$this->strict) {
-            $this->info['blockquote']->child =
-                new HTMLPurifier_ChildDef_StrictBlockquote(
-                    $this->info_content_sets['Block'] +
-                    array('#PCDATA' => true));
-        } else {
             $this->info['blockquote']->child =
                 new HTMLPurifier_ChildDef_Optional(
                     $this->info_content_sets['Flow'] +
@@ -409,8 +429,7 @@ class HTMLPurifier_HTMLDefinition
                 array('#PCDATA' => true, 'p' => true));
         }
         
-        // custom, not sure where to implement, because it's not
+        // deprecated config setting, implement in DisableURI module
-        // just /one/ module
         if ($config->get('Attr', 'DisableURI')) {
             $this->info['a']->attr['href'] =
             $this->info['img']->attr['longdesc'] =
@@ -427,28 +446,7 @@ class HTMLPurifier_HTMLDefinition
             $this->info['ol']->attr['start'] = new HTMLPurifier_AttrDef_Integer();
         }
         
-        // deprecated elements transforms, implement in TransformToStrict
+        // setup allowed elements, SubtractiveWhitelist module
-        $this->info_tag_transform['font']   = new HTMLPurifier_TagTransform_Font();
-        $this->info_tag_transform['menu']   = new HTMLPurifier_TagTransform_Simple('ul');
-        $this->info_tag_transform['dir']    = new HTMLPurifier_TagTransform_Simple('ul');
-        $this->info_tag_transform['center'] = new HTMLPurifier_TagTransform_Center();
-        
-        // deprecated attribute transforms, implement in TransformToStrict
-        $this->info['h1']->attr_transform_pre[] =
-        $this->info['h2']->attr_transform_pre[] =
-        $this->info['h3']->attr_transform_pre[] =
-        $this->info['h4']->attr_transform_pre[] =
-        $this->info['h5']->attr_transform_pre[] =
-        $this->info['h6']->attr_transform_pre[] =
-        $this->info['p'] ->attr_transform_pre[] = 
-                    new HTMLPurifier_AttrTransform_TextAlign();
-        
-        // xml:lang <=> lang mirroring, implement in TransformToStrict?
-        $this->info_attr_transform_post[] = new HTMLPurifier_AttrTransform_Lang();
-        $this->info_global_attr['lang'] = new HTMLPurifier_AttrDef_Lang();
-        
-        // setup allowed elements, obsoleted by Modules? (does offer
-        // different functionality)
         $allowed_elements = $config->get('HTML', 'AllowedElements');
         if (is_array($allowed_elements)) {
             foreach ($this->info as $name => $d) {

library/HTMLPurifier/HTMLModule.php

@@ -53,6 +53,24 @@ class HTMLPurifier_HTMLModule
      */
     var $attr_collections = array();
     
+    /**
+     * Associative array of deprecated tag name to HTMLPurifier_TagTransform
+     * @public
+     */
+    var $info_tag_transform = array();
+    
+    /**
+     * List of HTMLPurifier_AttrTransform to be performed before validation.
+     * @public
+     */
+    var $info_attr_transform_pre = array();
+    
+    /**
+     * List of HTMLPurifier_AttrTransform to be performed after validation.
+     * @public
+     */
+    var $info_attr_transform_post = array();
+    
     /**
      * Boolean flag that indicates whether or not getChildDef is implemented.
      * For optimization reasons: may save a call to a function. Be sure
@@ -72,6 +90,7 @@ class HTMLPurifier_HTMLModule
      * @public
      */
     function getChildDef($def) {return false;}
+    
 }
 
 ?>

library/HTMLPurifier/HTMLModule/Bdo.php

@@ -13,13 +13,13 @@ class HTMLPurifier_HTMLModule_Bdo extends HTMLPurifier_HTMLModule
     var $elements = array('bdo');
     var $info = array();
     var $content_sets = array('Inline' => 'bdo');
-    var $attr_collections_info = array(
+    var $attr_collections = array(
         'I18N' => array('dir' => false)
     );
     
     function HTMLPurifier_HTMLModule_Bdo() {
         $dir = new HTMLPurifier_AttrDef_Enum(array('ltr','rtl'), false);
-        $this->attr_collections_info['I18N']['dir'] = $dir;
+        $this->attr_collections['I18N']['dir'] = $dir;
         $this->info['bdo'] = new HTMLPurifier_ElementDef();
         $this->info['bdo']->attr = array(
             0 => array('Core', 'Lang'),
@@ -33,7 +33,7 @@ class HTMLPurifier_HTMLModule_Bdo extends HTMLPurifier_HTMLModule
         $this->info['bdo']->content_model = '#PCDATA | Inline';
         $this->info['bdo']->content_model_type = 'optional';
         // provides fallback behavior if dir's missing (dir is required)
-        $this->info['bdo']->attr_transform_post[] =
+        $this->info['bdo']->attr_transform_post['required-dir'] =
             new HTMLPurifier_AttrTransform_BdoDir();
     }
     

library/HTMLPurifier/HTMLModule/Text.php

@@ -45,7 +45,7 @@ class HTMLPurifier_HTMLModule_Text extends HTMLPurifier_HTMLModule
                 $this->info[$element]->content_model_type = 'empty';
             } elseif ($element == 'blockquote') {
                 $this->info[$element]->content_model = 'Heading | Block | List';
-                $this->info[$element]->content_model_type = 'strictblockquote';
+                $this->info[$element]->content_model_type = 'optional';
             } elseif ($element == 'div') {
                 $this->info[$element]->content_model = '#PCDATA | Flow';
                 $this->info[$element]->content_model_type = 'optional';

library/HTMLPurifier/HTMLModule/TransformToStrict.php

@@ -1,5 +1,7 @@
 <?php
 
+require_once 'HTMLPurifier/ChildDef/StrictBlockquote.php';
+
 /**
  * Proprietary module that transforms deprecated elements into Strict
  * HTML (see HTML 4.01 and XHTML 1.0) when possible.
@@ -8,7 +10,62 @@
 class HTMLPurifier_HTMLModule_TransformToStrict extends HTMLPurifier_HTMLModule
 {
     
-    // unimplemented
+    // we're actually modifying these elements, not defining them
+    var $elements = array('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'blockquote');
+    
+    var $info_tag_transform = array(
+        // placeholders, see constructor for definitions
+        'font'  => false,
+        'menu'  => false,
+        'dir'   => false,
+        'center'=> false
+    );
+    
+    var $attr_collections = array(
+        'Lang' => array(
+            'lang' => false // placeholder
+        )
+    );
+    
+    function HTMLPurifier_HTMLModule_TransformToStrict($config) {
+        // deprecated tag transforms
+        $this->info_tag_transform['font']   = new HTMLPurifier_TagTransform_Font();
+        $this->info_tag_transform['menu']   = new HTMLPurifier_TagTransform_Simple('ul');
+        $this->info_tag_transform['dir']    = new HTMLPurifier_TagTransform_Simple('ul');
+        $this->info_tag_transform['center'] = new HTMLPurifier_TagTransform_Center();
+        
+        foreach ($this->elements as $name) {
+            $this->info[$name] = new HTMLPurifier_ElementDef();
+            $this->info[$name]->standalone = false;
+        }
+        
+        // deprecated attribute transforms
+        $this->info['h1']->attr_transform_pre['align'] =
+        $this->info['h2']->attr_transform_pre['align'] =
+        $this->info['h3']->attr_transform_pre['align'] =
+        $this->info['h4']->attr_transform_pre['align'] =
+        $this->info['h5']->attr_transform_pre['align'] =
+        $this->info['h6']->attr_transform_pre['align'] =
+        $this->info['p'] ->attr_transform_pre['align'] = 
+                    new HTMLPurifier_AttrTransform_TextAlign();
+        
+        // xml:lang <=> lang mirroring, implement in TransformToStrict,
+        // this is overridden in TransformToXHTML11
+        $this->info_attr_transform_post['lang'] = new HTMLPurifier_AttrTransform_Lang();
+        $this->attr_collections['Lang']['lang'] = new HTMLPurifier_AttrDef_Lang();
+        
+        if ($config->get('HTML', 'Strict')) {
+            $this->info['blockquote']->content_model_type = 'strictblockquote';
+            $this->info['blockquote']->child = false; // recalculate please!
+        }
+        
+    }
+    
+    var $defines_child_def = true;
+    function getChildDef($def) {
+        if ($def->content_model_type != 'strictblockquote') return false;
+        return new HTMLPurifier_ChildDef_StrictBlockquote($def->content_model);
+    }
     
 }
 

library/HTMLPurifier/Printer/HTMLDefinition.php

@@ -130,7 +130,9 @@ class HTMLPurifier_Printer_HTMLDefinition extends HTMLPurifier_Printer
             $elements = array();
             $attr = array();
             if (isset($def->elements)) {
-                if ($def->type == 'strictblockquote') $def->validateChildren(array(), $this->config, $context);
+                if ($def->type == 'strictblockquote') {
+                    $def->validateChildren(array(), $this->config, $context);
+                }
                 $elements = $def->elements;
             } elseif ($def->type == 'chameleon') {
                 $attr['rowspan'] = 2;