Add support for hard exclusions that affect all child nodes.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@146 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/Definition.php

@@ -208,6 +208,16 @@ class HTMLPurifier_Definition
             $this->info[$name]->type = 'block';
         }
         
+        //////////////////////////////////////////////////////////////////////
+        // info[]->excludes : defines elements that aren't allowed in here
+        
+        // make sure you test using isset() and not !empty()
+        
+        $this->info['a']->excludes = array('a' => true);
+        $this->info['pre']->excludes = array_flip(array('img', 'big', 'small',
+            // technically in spec, but we don't allow em anyway
+            'object', 'applet', 'font', 'basefont'));
+        
         //////////////////////////////////////////////////////////////////////
         // info[]->attr : defines allowed attributes for elements
         
@@ -233,6 +243,8 @@ class HTMLPurifier_Definition
         //////////////////////////////////////////////////////////////////////
         // info[]->auto_close : tags that automatically close another
         
+        // make sure you test using isset() not !empty()
+        
         // these are all block elements: blocks aren't allowed in P
         $this->info['p']->auto_close = array_flip(array(
                 'address', 'blockquote', 'dd', 'dir', 'div', 'dl', 'dt',
@@ -265,6 +277,7 @@ class HTMLPurifier_ElementDef
     var $auto_close = array();
     var $child;
     var $type = 'unknown';
+    var $excludes = array();
     
 }
 

library/HTMLPurifier/Strategy/FixNesting.php

@@ -26,6 +26,9 @@ class HTMLPurifier_Strategy_FixNesting extends HTMLPurifier_Strategy
         // $stack[count($stack)-1] being the current parent
         $stack = array();
         
+        // stack that contains all elements that are excluded
+        $exclude_stack = array();
+        
         for ($i = 0, $size = count($tokens) ; $i < $size; ) {
             
             $child_tokens = array();
@@ -63,11 +66,31 @@ class HTMLPurifier_Strategy_FixNesting extends HTMLPurifier_Strategy
                 $context = 'unknown';
             }
             
-            // DEFINITION CALL
+            // determine whether or not element is excluded
-            $child_def = $this->definition->info[$tokens[$i]->name]->child;
+            $excluded = false;
+            if (!empty($exclude_stack)) {
+                foreach ($exclude_stack as $lookup) {
+                    if (isset($lookup[$tokens[$i]->name])) {
+                        $excluded = true;
+                        break;
+                    }
+                }
+            }
             
-            // have DTD child def validate children
+            if ($excluded) {
-            $result = $child_def->validateChildren($child_tokens, $context);
+                $result = false;
+            } else {
+                // DEFINITION CALL
+                $def = $this->definition->info[$tokens[$i]->name];
+                
+                $child_def = $def->child;
+                
+                // have DTD child def validate children
+                $result = $child_def->validateChildren($child_tokens, $context);
+                
+                // determine whether or not this element has any exclusions
+                $excludes = $def->excludes;
+            }
             
             // process result
             if ($result === true) {
@@ -77,6 +100,9 @@ class HTMLPurifier_Strategy_FixNesting extends HTMLPurifier_Strategy
                 // register start token as a parental node start
                 $stack[] = $i;
                 
+                // register exclusions if there are any
+                if (!empty($excludes)) $exclude_stack[] = $excludes;
+                
                 // move cursor to next possible start node
                 $i++;
                 
@@ -113,6 +139,9 @@ class HTMLPurifier_Strategy_FixNesting extends HTMLPurifier_Strategy
                 // register start token as a parental node start
                 $stack[] = $i;
                 
+                // register exclusions if there are any
+                if (!empty($excludes)) $exclude_stack[] = $excludes;
+                
                 // move cursor to next possible start node
                 $i++;
                 
@@ -124,8 +153,15 @@ class HTMLPurifier_Strategy_FixNesting extends HTMLPurifier_Strategy
             // Test if the token indeed is a start tag, if not, move forward
             // and test again.
             while ($i < $size and $tokens[$i]->type != 'start') {
-                // pop a token index off the stack if we ended a node
+                if ($tokens[$i]->type == 'end') {
-                if ($tokens[$i]->type == 'end') array_pop($stack);
+                    // pop a token index off the stack if we ended a node
+                    array_pop($stack);
+                    // pop an exclusion lookup off exclusion stack if
+                    // we ended node and that node had exclusions
+                    if ($this->definition->info[$tokens[$i]->name]->excludes) {
+                        array_pop($exclude_stack);
+                    }
+                }
                 $i++;
             }
             

tests/HTMLPurifier/Strategy/FixNestingTest.php

@@ -65,6 +65,10 @@ class HTMLPurifier_Strategy_FixNestingTest
         $inputs[11] = '<span><ins><div>Not allowed!</div></ins></span>';
         $expect[11] = '<span><ins>&lt;div&gt;Not allowed!&lt;/div&gt;</ins></span>';
         
+        // test exclusions
+        $inputs[12] = '<a><span><a>Not allowed</a></span></a>';
+        $expect[12] = '<a><span></span></a>';
+        
         $this->assertStrategyWorks($strategy, $inputs, $expect);
     }