[1.7.0] Eliminated modes in favor for special-case "Tidy" modules

- Add $xml property to Doctype, make more serialize friendly in preparation for stuffing into Config object
- Add FIXME markers for areas of further development, code is hooked so this is easy
- Document what the new Tidy classes will be

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1065 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/Doctype.php

@@ -18,23 +18,27 @@ class HTMLPurifier_Doctype
     var $modules = array();
     
     /**
-     * Associative array of mode names to lists of modules; these are
+     * List of modules to use for tidying up code
-     * the modules added into the standard list if a particular mode
-     * is enabled, such as lenient or correctional.
      */
-    var $modulesForModes = array();
+    var $tidyModules = array();
+    
+    /**
+     * Is the language derived from XML (i.e. XHTML)?
+     */
+    var $xml = true;
     
     /**
      * List of aliases for this doctype
      */
     var $aliases = array();
     
-    function HTMLPurifier_Doctype($name = null, $modules = array(),
+    function HTMLPurifier_Doctype($name = null, $xml = true, $modules = array(),
-        $modules_for_modes = array(), $aliases = array()
+        $tidyModules = array(), $aliases = array()
     ) {
         $this->name         = $name;
+        $this->xml          = $xml;
         $this->modules      = $modules;
-        $this->modulesForModes = $modules_for_modes;
+        $this->tidyModules  = $tidyModules;
         $this->aliases      = $aliases;
     }
     
@@ -43,7 +47,7 @@ class HTMLPurifier_Doctype
      */
     function copy() {
         return new HTMLPurifier_Doctype(
-            $this->name, $this->modules, $this->modulesForModes, $this->aliases
+            $this->name, $this->xml, $this->modules, $this->tidyModules, $this->aliases
         );
     }
 }

library/HTMLPurifier/DoctypeRegistry.php

@@ -27,14 +27,15 @@ class HTMLPurifier_DoctypeRegistry
      * @param $aliases Alias names for doctype
      * @return Reference to registered doctype (usable for further editing)
      */
-    function &register($doctype, $modules = array(),
+    function &register($doctype, $xml = true, $modules = array(),
-        $modules_for_modes = array(), $aliases = array()
+        $tidy_modules = array(), $aliases = array()
     ) {
         if (!is_array($modules)) $modules = array($modules);
+        if (!is_array($tidy_modules)) $tidy_modules = array($tidy_modules);
         if (!is_array($aliases)) $aliases = array($aliases);
         if (!is_object($doctype)) {
             $doctype = new HTMLPurifier_Doctype(
-                $doctype, $modules, $modules_for_modes, $aliases
+                $doctype, $xml, $modules, $tidy_modules, $aliases
             );
         }
         $this->doctypes[$doctype->name] =& $doctype;
@@ -68,15 +69,16 @@ class HTMLPurifier_DoctypeRegistry
     /**
      * Creates a doctype based on a configuration object,
      * will perform initialization on the doctype
+     * @note Use this function to get a copy of doctype that config
+     *       can hold on to (this is necessary in order to tell
+     *       Generator whether or not the current document is XML
+     *       based or not).
      */
     function make($config) {
         $original_doctype = $this->get($this->getDoctypeFromConfig($config));
         $doctype = $original_doctype->copy();
-        // initialization goes here
+        // FIXME!!! Set $doctype into $config so others can use it,
-        foreach ($doctype->modulesForModes as $mode => $mode_modules) {
+        // you might need to use two copies
-            // TODO: test if $mode is active
-            $doctype->modules = array_merge($doctype->modules, $mode_modules);
-        }
         return $doctype;
     }
     

library/HTMLPurifier/HTMLModuleManager.php

@@ -115,33 +115,38 @@ class HTMLPurifier_HTMLModuleManager
         $transitional = array('Legacy', 'Target');
         
         $this->doctypes->register(
-            'HTML 4.01 Transitional',
+            'HTML 4.01 Transitional', false,
             array_merge($common, $transitional),
-            array('correctional' => array('TransformToStrict'))
+            array('TransformToStrict')
+            // Tidy: Transitional
         );
         
         $this->doctypes->register(
-            'XHTML 1.0 Transitional',
+            'HTML 4.01 Strict', false,
+            array_merge($common),
+            array('TransformToStrict')
+            // Tidy: Strict
+        );
+        
+        $this->doctypes->register(
+            'XHTML 1.0 Transitional', true,
             array_merge($common, $transitional),
-            array('correctional' => array('TransformToStrict'))
+            array('TransformToStrict')
+            // Tidy: Transitional, XHTML
         );
         
         $this->doctypes->register(
-            'HTML 4.01 Strict',
+            'XHTML 1.0 Strict', true,
             array_merge($common),
-            array('lenient' => array('TransformToStrict'))
+            array('TransformToStrict')
+            // Tidy: Strict, XHTML
         );
         
         $this->doctypes->register(
-            'XHTML 1.0 Strict',
+            'XHTML 1.1', true,
             array_merge($common),
-            array('lenient' => array('TransformToStrict'))
+            array('TransformToStrict', 'TransformToXHTML11')
-        );
+            // Tidy: Strict, XHTML1_1
-        
-        $this->doctypes->register(
-            'XHTML 1.1',
-            array_merge($common),
-            array('lenient' => array('TransformToStrict', 'TransformToXHTML11'))
         );
         
     }
@@ -247,15 +252,12 @@ class HTMLPurifier_HTMLModuleManager
         $modules = array_merge($modules, $this->userModules);
         
         foreach ($modules as $module) {
-            if (is_object($module)) {
+            $this->processModule($module);
-                $this->registeredModules[$module->name] = $module;
-                continue;
-            } else {
-                if (!isset($this->registeredModules[$module])) {
-                    $this->registerModule($module);
         }
-            }
+        
-            $this->modules[$module] = $this->registeredModules[$module];
+        foreach ($doctype->tidyModules as $module) {
+            $this->processModule($module);
+            // FIXME!!! initialize the tidy modules here
         }
         
         // setup lookup table based on all valid modules
@@ -284,6 +286,17 @@ class HTMLPurifier_HTMLModuleManager
         
     }
     
+    /**
+     * Takes a module and adds it to the active module collection,
+     * registering it if necessary.
+     */
+    function processModule($module) {
+        if (!isset($this->registeredModules[$module]) || is_object($module)) {
+            $this->registerModule($module);
+        }
+        $this->modules[$module] = $this->registeredModules[$module];
+    }
+    
     /**
      * Retrieves merged element definitions.
      * @return Array of HTMLPurifier_ElementDef

tests/HTMLPurifier/DoctypeRegistryTest.php

@@ -11,23 +11,22 @@ class HTMLPurifier_DoctypeRegistryTest extends UnitTestCase
         
         $d =& $registry->register(
             $name = 'XHTML 1.0 Transitional',
+            $xml = true,
             $modules = array('module-one', 'module-two'),
-            $modulesForModes = array(
+            $tidyModules = array('lenient-module'),
-                'lenient' => array('lenient-module'),
-            ),
             $aliases = array('X10T')
         );
         
-        $d2 = new HTMLPurifier_Doctype($name, $modules, $modulesForModes, $aliases);
+        $d2 = new HTMLPurifier_Doctype($name, $xml, $modules, $tidyModules, $aliases);
         
         $this->assertIdentical($d, $d2);
         $this->assertReference($d, $registry->get('XHTML 1.0 Transitional'));
         
         // test shorthand
         $d =& $registry->register(
-            $name = 'XHTML 1.0 Strict', 'module', array(), 'X10S'
+            $name = 'XHTML 1.0 Strict', true, 'module', 'Tidy', 'X10S'
         );
-        $d2 = new HTMLPurifier_Doctype($name, array('module'), array(), array('X10S'));
+        $d2 = new HTMLPurifier_Doctype($name, true, array('module'), array('Tidy'), array('X10S'));
         
         $this->assertIdentical($d, $d2);
         
@@ -52,22 +51,22 @@ class HTMLPurifier_DoctypeRegistryTest extends UnitTestCase
         
         $registry = new HTMLPurifier_DoctypeRegistry();
         
-        $d1 =& $registry->register('Doc1', array(), array(), array('1'));
+        $d1 =& $registry->register('Doc1', true, array(), array(), array('1'));
         
         $this->assertReference($d1, $registry->get('Doc1'));
         $this->assertReference($d1, $registry->get('1'));
         
-        $d2 =& $registry->register('Doc2', array(), array(), array('2'));
+        $d2 =& $registry->register('Doc2', true, array(), array(), array('2'));
         
         $this->assertReference($d2, $registry->get('Doc2'));
         $this->assertReference($d2, $registry->get('2'));
         
-        $d3 =& $registry->register('1', array(), array(), array());
+        $d3 =& $registry->register('1', true, array(), array(), array());
         
         // literal name overrides alias
         $this->assertReference($d3, $registry->get('1'));
         
-        $d4 =& $registry->register('One', array(), array(), array('1'));
+        $d4 =& $registry->register('One', true, array(), array(), array('1'));
         
         $this->assertReference($d4, $registry->get('One'));
         // still it overrides