[1.7.0] Code audit

- Add set accessor, update access control on variables in AttrTypes
- Add warning notes to non-unit tested, out of date or unused code files
- Remove redundant include in EntityParser, expand string regexp to match all ASCII XML-style entities
- Remove obsolete hooks in HTMLModule

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1078 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/AttrTypes.php

@@ -16,12 +16,13 @@ class HTMLPurifier_AttrTypes
 {
     /**
      * Lookup array of attribute string identifiers to concrete implementations
-     * @public
+     * @protected
      */
     var $info = array();
     
     /**
-     * Constructs the info array
+     * Constructs the info array, supplying default implementations for attribute
+     * types.
      */
     function HTMLPurifier_AttrTypes() {
         $this->info['CDATA']    = new HTMLPurifier_AttrDef_Text();
@@ -40,15 +41,27 @@ class HTMLPurifier_AttrTypes
     
     /**
      * Retrieves a type
+     * @param $type String type name
+     * @return Object AttrDef for type
      */
     function get($type) {
-        // maybe some extra initialization could be done
+        // If $type is complicated, we may to clone the attribute
+        // definition and make custom changes
         if (!isset($this->info[$type])) {
             trigger_error('Cannot retrieve undefined attribute type ' . $type, E_USER_ERROR);
             return;
         }
         return $this->info[$type];
     }
+    
+    /**
+     * Sets a new implementation for a type
+     * @param $type String type name
+     * @param $impl Object AttrDef for type
+     */
+    function set($type, $impl) {
+        $this->info[$type] = $impl;
+    }
 }
 
 ?>

library/HTMLPurifier/ContentSets.php

@@ -6,6 +6,8 @@ require_once 'HTMLPurifier/ChildDef/Empty.php';
 require_once 'HTMLPurifier/ChildDef/Required.php';
 require_once 'HTMLPurifier/ChildDef/Optional.php';
 
+// NOT UNIT TESTED!!!
+
 class HTMLPurifier_ContentSets
 {
     

library/HTMLPurifier/EntityParser.php

@@ -24,8 +24,8 @@ class HTMLPurifier_EntityParser
      * @protected
      */                             
     var $_substituteEntitiesRegex =
-'/&(?:[#]x([a-fA-F0-9]+)|[#]0*(\d+)|([A-Za-z]+));?/';
+'/&(?:[#]x([a-fA-F0-9]+)|[#]0*(\d+)|([A-Za-z_:][A-Za-z0-9.\-_:]*));?/';
-//     1. hex             2. dec      3. string
+//     1. hex             2. dec      3. string (XML style)
     
     
     /**
@@ -97,7 +97,6 @@ class HTMLPurifier_EntityParser
         } else {
             if (isset($this->_special_ent2dec[$matches[3]])) return $entity;
             if (!$this->_entity_lookup) {
-                require_once 'HTMLPurifier/EntityLookup.php';
                 $this->_entity_lookup = HTMLPurifier_EntityLookup::instance();
             }
             if (isset($this->_entity_lookup->table[$matches[3]])) {

library/HTMLPurifier/HTMLModule.php

@@ -97,27 +97,6 @@ class HTMLPurifier_HTMLModule
      */
     function getChildDef($def) {return false;}
     
-    /**
-     * Hook method that lets module perform arbitrary operations on
-     * HTMLPurifier_HTMLDefinition before the module gets processed.
-     * @param $definition Reference to HTMLDefinition being setup
-     */
-    function preProcess(&$definition) {}
-    
-    /**
-     * Hook method that lets module perform arbitrary operations
-     * on HTMLPurifier_HTMLDefinition after the module gets processed.
-     * @param $definition Reference to HTMLDefinition being setup
-     */
-    function postProcess(&$definition) {}
-    
-    /**
-     * Hook method that is called when a module gets registered to
-     * the definition.
-     * @param $definition Reference to HTMLDefinition being setup
-     */
-    function setup(&$definition) {}
-    
     // -- Convenience -----------------------------------------------------
     
     /**

library/HTMLPurifier/Language.php

@@ -2,6 +2,8 @@
 
 require_once 'HTMLPurifier/LanguageFactory.php';
 
+// UNUSED
+
 class HTMLPurifier_Language
 {
     

library/HTMLPurifier/LanguageFactory.php

@@ -3,6 +3,8 @@
 require_once 'HTMLPurifier/Language.php';
 require_once 'HTMLPurifier/AttrDef/Lang.php';
 
+// UNUSED
+
 /**
  * Class responsible for generating HTMLPurifier_Language objects, managing
  * caching and fallbacks.

library/HTMLPurifier/Printer.php

@@ -4,6 +4,8 @@ require_once 'HTMLPurifier/Generator.php';
 require_once 'HTMLPurifier/Token.php';
 require_once 'HTMLPurifier/Encoder.php';
 
+// OUT OF DATE, NEEDS UPDATING!
+
 class HTMLPurifier_Printer
 {