Factor out content set and childdef functionality to ContentSets. Remove redundant info suffix from attr_collections.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@725 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/AttrCollections.php

@@ -53,7 +53,7 @@ class HTMLPurifier_AttrCollections
         $info =& $this->info;
         // load extensions from the modules
         foreach ($modules as $module) {
-            foreach ($module->attr_collections_info as $coll_i => $coll) {
+            foreach ($module->attr_collections as $coll_i => $coll) {
                 foreach ($coll as $attr_i => $attr) {
                     if ($attr_i === 0 && isset($info[$coll_i][$attr_i])) {
                         // merge in includes

library/HTMLPurifier/ContentSets.php

@@ -0,0 +1,114 @@
+<?php
+
+// common defs that we'll support by default
+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
+{
+    
+    var $info = array();
+    var $lookup = array();
+    
+    var $keys = array();
+    var $values = array();
+    
+    function setup($modules) {
+        // populate content_sets based on module hints
+        foreach ($modules as $module_i => $module) {
+            foreach ($module->content_sets as $key => $value) {
+                if (isset($this->info[$key])) {
+                    // add it into the existing content set
+                    $this->info[$key] = $this->info[$key] . ' | ' . $value;
+                } else {
+                    $this->info[$key] = $value;
+                }
+            }
+        }
+        
+        // perform content_set expansions
+        $this->keys = array_keys($this->info);
+        foreach ($this->info as $i => $set) {
+            // only performed once, so infinite recursion is not
+            // a problem
+            $this->info[$i] =
+                str_replace(
+                    $this->keys,
+                    // must be recalculated each time due to
+                    // changing substitutions
+                    array_values($this->info),
+                $set);
+        }
+        $this->values = array_values($this->info);
+        
+        // generate lookup tables
+        foreach ($this->info as $name => $set) {
+            $this->lookup[$name] = $this->convertToLookup($set);
+        }
+    }
+    
+    function generateChildDef(&$def, $module) {
+        $content_model = $def->content_model;
+        if (is_string($content_model)) {
+            $def->content_model = str_replace(
+                $this->keys, $this->values, $content_model);
+        }
+        $def->child = $this->getChildDef($def, $module);
+    }
+    
+    /**
+     * Instantiates a ChildDef based on content_model and content_model_type
+     * member variables in HTMLPurifier_ElementDef
+     * @note This will also defer to modules for custom HTMLPurifier_ChildDef
+     *       subclasses that need content set expansion
+     * @param $def HTMLPurifier_ElementDef to have ChildDef extracted
+     * @return HTMLPurifier_ChildDef corresponding to ElementDef
+     */
+    function getChildDef($def, $module) {
+        $value = $def->content_model;
+        if (is_object($value)) return $value; // direct object, return
+        switch ($def->content_model_type) {
+            case 'required':
+                return new HTMLPurifier_ChildDef_Required($value);
+            case 'optional':
+                return new HTMLPurifier_ChildDef_Optional($value);
+            case 'empty':
+                return new HTMLPurifier_ChildDef_Empty();
+            case 'strictblockquote':
+                return new HTMLPurifier_ChildDef_StrictBlockquote($value);
+            case 'custom':
+                return new HTMLPurifier_ChildDef_Custom($value);
+        }
+        // defer to its module
+        if (!$module->defines_child_def) continue; // save a func call
+        $return = $module->getChildDef($def);
+        if ($return !== false) return $return;
+        // error-out
+        trigger_error(
+            'Could not determine which ChildDef class to instantiate',
+            E_USER_ERROR
+        );
+        return false;
+    }
+    
+    /**
+     * Converts a string list of elements separated by pipes into
+     * a lookup array.
+     * @param $string List of elements
+     * @return Lookup array of elements
+     */
+    function convertToLookup($string) {
+        $array = explode('|', str_replace(' ', '', $string));
+        $ret = array();
+        foreach ($array as $i => $k) {
+            $ret[$k] = true;
+        }
+        return $ret;
+    }
+    
+}
+
+?>

library/HTMLPurifier/HTMLDefinition.php

@@ -1,36 +1,25 @@
 <?php
 
+// components
 require_once 'HTMLPurifier/AttrTypes.php';
 require_once 'HTMLPurifier/AttrCollections.php';
+require_once 'HTMLPurifier/ContentSets.php';
 require_once 'HTMLPurifier/ElementDef.php';
 
-// we'll manage loading extremely commonly used attr definitions
 require_once 'HTMLPurifier/AttrDef.php';
-require_once 'HTMLPurifier/AttrDef/Enum.php';
+require_once 'HTMLPurifier/AttrDef/Enum.php'; // common
 
-// technically speaking, these includes would be more appropriate for
+// temporary: attribute transformations
-// other modules, but we're going to include all the common ones. A
-// custom one would have to be fed in as an actual object
-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';
-
-// handling attribute transformations until modules gain the capability
 require_once 'HTMLPurifier/AttrTransform.php';
 require_once 'HTMLPurifier/AttrTransform/Lang.php';
 require_once 'HTMLPurifier/AttrTransform/TextAlign.php';
 require_once 'HTMLPurifier/AttrTransform/BdoDir.php';
 require_once 'HTMLPurifier/AttrTransform/ImgRequired.php';
 
-// handling tag transformations until modules gain the capability
+// temporary: tag transformations
 require_once 'HTMLPurifier/TagTransform.php';
 
-// utility classes that are necessary (?)
+// default modules
-require_once 'HTMLPurifier/Generator.php';
-require_once 'HTMLPurifier/Token.php';
-
 require_once 'HTMLPurifier/HTMLModule.php';
 require_once 'HTMLPurifier/HTMLModule/Text.php';
 require_once 'HTMLPurifier/HTMLModule/Hypertext.php';
@@ -232,6 +221,11 @@ class HTMLPurifier_HTMLDefinition
      */
     var $setup = false;
     
+    /**
+     * Instance of HTMLPurifier_ContentSets
+     * @public
+     */
+    var $content_sets;
     
     
     /**
@@ -256,6 +250,7 @@ class HTMLPurifier_HTMLDefinition
         
         $this->attr_types       = new HTMLPurifier_AttrTypes();
         $this->attr_collections = new HTMLPurifier_AttrCollections();
+        $this->content_sets     = new HTMLPurifier_ContentSets();
         
         // some compat stuff, will be factored to modules
         
@@ -277,68 +272,43 @@ class HTMLPurifier_HTMLDefinition
     function setup($config) {
         
         // multiple call guard
-        if ($this->setup) return;
+        if ($this->setup) {return;} else {$this->setup = true;}
-        $this->setup = true;
         
-        // perform attribute collection substitutions
+        // would be nice if we could put each of these in their
+        // own object, would make this hookable too!
+        $this->setupInfo($config);
+        $this->setupAttrTransform($config);
+        $this->setupBlockWrapper($config);
+        $this->setupParent($config);
+        $this->setupCompat($config);
+        
+    }
+    
+    function setupInfo() {
         $this->attr_collections->setup($this->attr_types, $this->modules);
-        
+        $this->content_sets->setup($this->modules);
-        // populate content_sets based on module hints
+        $this->info_content_sets = $this->content_sets->lookup;
-        $content_sets = array();
-        foreach ($this->modules as $module_i => $module) {
-            foreach ($module->content_sets as $key => $value) {
-                if (isset($content_sets[$key])) {
-                    // add it into the existing content set
-                    $content_sets[$key] = $content_sets[$key] . ' | ' . $value;
-                } else {
-                    $content_sets[$key] = $value;
-                }
-            }
-        }
-        
-        // perform content_set expansions
-        foreach ($content_sets as $i => $set) {
-            // only performed once, so infinite recursion is not
-            // a problem, you'll just have a stray $Set lying around
-            // at the end
-            $content_sets[$i] =
-                str_replace(
-                    array_keys($content_sets),
-                    array_values($content_sets),
-                $set);
-        }
-        // define convenient variables
-        $content_sets_keys   = array_keys($content_sets);
-        $content_sets_values = array_values($content_sets);
-        foreach ($content_sets as $name => $set) {
-            $this->info_content_sets[$name] = $this->convertToLookup($set);
-        }
         
         foreach ($this->modules as $module_i => $module) {
             foreach ($module->info as $name => $def) {
                 $def =& $this->modules[$module_i]->info[$name];
                 
                 // attribute value expansions
-                
                 $this->attr_collections->performInclusions($def->attr);
                 $this->attr_collections->expandIdentifiers(
                     $def->attr, $this->attr_types);
                 
-                // perform content model expansions
+                // chameleon data, set descendants_are_inline
-                $content_model = $def->content_model;
+                if (is_string($def->content_model) &&
-                if (is_string($content_model)) {
+                strpos($def->content_model, 'Inline') !== false) {
-                    if (strpos($content_model, 'Inline') !== false) {
                     if ($name != 'del' && $name != 'ins') {
                         // this is for you, ins/del
                         $def->descendants_are_inline = true;
                     }
                 }
-                    $def->content_model = str_replace(
-                        $content_sets_keys, $content_sets_values, $content_model);
-                }
                 
-                // get child def from content model
+                // set child def from content model
-                $def->child = $this->getChildDef($def);
+                $this->content_sets->generateChildDef($def, $module);
                 
                 // setup info
                 $this->info[$name] = $def;
@@ -347,12 +317,6 @@ class HTMLPurifier_HTMLDefinition
                 }
             }
         }
-        
-        $this->setupAttrTransform($config);
-        $this->setupBlockWrapper($config);
-        $this->setupParent($config);
-        $this->setupCompat($config);
-        
     }
     
     /**
@@ -502,58 +466,6 @@ class HTMLPurifier_HTMLDefinition
         
     }
     
-    /**
-     * Instantiates a ChildDef based on content_model and content_model_type
-     * member variables in HTMLPurifier_ElementDef
-     * @note This will also defer to modules for custom HTMLPurifier_ChildDef
-     *       subclasses that need content set expansion
-     * @param $def HTMLPurifier_ElementDef to have ChildDef extracted
-     * @return HTMLPurifier_ChildDef corresponding to ElementDef
-     */
-    function getChildDef($def) {
-        $value = $def->content_model;
-        if (is_object($value)) return $value; // direct object, return
-        switch ($def->content_model_type) {
-            case 'required':
-                return new HTMLPurifier_ChildDef_Required($value);
-            case 'optional':
-                return new HTMLPurifier_ChildDef_Optional($value);
-            case 'empty':
-                return new HTMLPurifier_ChildDef_Empty();
-            case 'strictblockquote':
-                return new HTMLPurifier_ChildDef_StrictBlockquote($value);
-            case 'custom':
-                return new HTMLPurifier_ChildDef_Custom($value);
-        }
-        // defer to modules, see if they know what child_def to use
-        foreach ($this->modules as $module) {
-            if (!$module->defines_child_def) continue; // save a func call
-            $return = $module->getChildDef($def);
-            if ($return !== false) return $return;
-        }
-        // error-out
-        trigger_error(
-            'Could not determine which ChildDef class to instantiate',
-            E_USER_ERROR
-        );
-        return false;
-    }
-    
-    /**
-     * Converts a string list of elements separated by pipes into
-     * a lookup array.
-     * @param $string List of elements
-     * @return Lookup array of elements
-     */
-    function convertToLookup($string) {
-        $array = explode('|', str_replace(' ', '', $string));
-        $ret = array();
-        foreach ($array as $i => $k) {
-            $ret[$k] = true;
-        }
-        return $ret;
-    }
-    
 }
 
 ?>

library/HTMLPurifier/HTMLModule.php

@@ -46,12 +46,12 @@ class HTMLPurifier_HTMLModule
      * Associative array of attribute collection names to attribute
      * collection additions. More rarely used for adding attributes to
      * the global collections. Example is the StyleAttribute module adding
-     * the style attribute to the Core. Corresponds to attr_collections
+     * the style attribute to the Core. Corresponds to HTMLDefinition's
-     * attr_collections->info, as only one object with behavior is
+     * attr_collections->info, since the object's data is only info,
-     * necessary.
+     * with extra behavior associated with it.
      * @public
      */
-    var $attr_collections_info = array();
+    var $attr_collections = array();
     
     /**
      * Boolean flag that indicates whether or not getChildDef is implemented.

library/HTMLPurifier/HTMLModule/StyleAttribute.php

@@ -9,7 +9,7 @@ require_once 'HTMLPurifier/AttrDef/CSS.php';
  */
 class HTMLPurifier_HTMLModule_StyleAttribute extends HTMLPurifier_HTMLModule
 {
-    var $attr_collections_info = array(
+    var $attr_collections = array(
         // The inclusion routine differs from the Abstract Modules but
         // is in line with the DTD and XML Schemas.
         'Style' => array('style' => false), // see constructor
@@ -17,7 +17,7 @@ class HTMLPurifier_HTMLModule_StyleAttribute extends HTMLPurifier_HTMLModule
     );
     
     function HTMLPurifier_HTMLModule_StyleAttribute() {
-        $this->attr_collections_info['Style']['style'] = new HTMLPurifier_AttrDef_CSS();
+        $this->attr_collections['Style']['style'] = new HTMLPurifier_AttrDef_CSS();
     }
     
 }