Main: implemented regexp-style validation for complicated child definitions

Also:
* Updated spec with some extra comments
* Trigger error if HTMLDTD_ChildDef_Simple has validateChildren called
* Factor out definition assertion in test class

git-svn-id: http://htmlpurifier.org/svnroot/html_purifier/trunk@50 48356398-32a2-884e-a903-53898d9a118a

PureHTMLDefinition.php

@@ -352,13 +352,58 @@ class HTMLDTD_Element
 // true = leave nodes as is
 // false = delete parent node and all children
 // array(...) = replace children nodes with these
+
+// this is the hardest one to implement. We'll use fancy regexp tricks
+// right now, we only expect it to return TRUE or FALSE (it won't attempt
+// to fix the tree)
+
+// we may end up writing custom code for each HTML case
+// in order to make it self correcting
 class HTMLDTD_ChildDef
 {
     var $dtd_regex;
+    var $_pcre_regex;
     function HTMLDTD_ChildDef($dtd_regex) {
         $this->dtd_regex = $dtd_regex;
+        $this->_compileRegex();
+    }
+    function _compileRegex() {
+        $raw = str_replace(' ', '', $this->dtd_regex);
+        if ($raw{0} != '(') {
+            $raw = "($raw)";
+        }
+        $reg = str_replace(',', ',?', $raw);
+        $reg = preg_replace('/([#a-zA-Z0-9_.-]+)/', '(,?\\0)', $reg);
+        $this->_pcre_regex = $reg;
+    }
+    function validateChildren($tokens_of_children) {
+        $list_of_children = '';
+        $nesting = 0; // depth into the nest
+        foreach ($tokens_of_children as $token) {
+            if (!empty($token->is_whitespace)) continue;
+            
+            $is_child = ($nesting == 0); // direct
+            
+            if (is_a($token, 'MF_StartTag')) {
+                $nesting++;
+            } elseif (is_a($token, 'MF_EndTag')) {
+                $nesting--;
+            }
+            
+            if ($is_child) {
+                $list_of_children .= $token->name . ',';
+            }
+        }
+        $list_of_children = rtrim($list_of_children, ',');
+        
+        $okay =
+            preg_match(
+                '/^'.$this->_pcre_regex.'$/',
+                $list_of_children
+            );
+        
+        return (bool) $okay;
     }
-    function validateChildren($tokens_of_children) {}
 }
 class HTMLDTD_ChildDef_Simple extends HTMLDTD_ChildDef
 {
@@ -373,6 +418,9 @@ class HTMLDTD_ChildDef_Simple extends HTMLDTD_ChildDef
         $this->elements = $elements;
         $this->gen = new HTML_Generator();
     }
+    function validateChildren() {
+        trigger_error('Cannot call abstract function!', E_USER_ERROR);
+    }
 }
 class HTMLDTD_ChildDef_Required extends HTMLDTD_ChildDef_Simple
 {

docs/spec.txt

@@ -153,6 +153,8 @@ So... here's the interesting code:
 
 --
 
+// # This actually does the validation
+
 // Validate the order of the children
 if (!$was_error && count($dtd_children)) {
     $children_list = implode(',', $children);
@@ -166,6 +168,8 @@ if (!$was_error && count($dtd_children)) {
 
 --
 
+// # This figures out the PcreRegex
+
 //$ch is a string of the allowed childs
 $children = preg_split('/([^#a-zA-Z0-9_.-]+)/', $ch, -1, PREG_SPLIT_NO_EMPTY);
 // check for parsed character data special case
@@ -222,7 +226,8 @@ form
     must not contain other form elements. 
 
 Normative exclusions straight from the horses mouth. These are SGML style,
-not XML style, so we need to modify the ruleset slightly.
+not XML style, so we need to modify the ruleset slightly. However, the DTD
+may have done this for us already.
 
 --
 

tests/PureHTMLDefinition.php

@@ -190,8 +190,61 @@ class Test_PureHTMLDefinition extends UnitTestCase
 class Test_HTMLDTD_ChildDef extends UnitTestCase
 {
     
+    function assertSeries($inputs, $expect, $def) {
+        foreach ($inputs as $i => $input) {
+            $result = $def->validateChildren($input);
+            if (is_bool($expect[$i])) {
+                $this->assertIdentical($expect[$i], $result);
+            } else {
+                $this->assertEqual($expect[$i], $result);
+                paintIf($result, $result != $expect[$i]);
+            }
+        }
+    }
+    
+    function test_complex() {
+        
+        // the table definition
+        $def = new HTMLDTD_ChildDef(
+            '(caption?, (col*|colgroup*), thead?, tfoot?, (tbody+|tr+))');
+        
+        $inputs[0] = array();
+        $expect[0] = false;
+        
+        // we really don't care what's inside, because if it turns out
+        // this tr is illegal, we'll end up re-evaluating the parent node
+        // anyway.
+        $inputs[1] = array(
+            new MF_StartTag('tr')       ,new MF_EndTag('tr')
+            );
+        $expect[1] = true;
+        
+        $inputs[2] = array(
+            new MF_StartTag('caption')  ,new MF_EndTag('caption')
+           ,new MF_StartTag('col')      ,new MF_EndTag('col')
+           ,new MF_StartTag('thead')    ,new MF_EndTag('thead')
+           ,new MF_StartTag('tfoot')    ,new MF_EndTag('tfoot')
+           ,new MF_StartTag('tbody')    ,new MF_EndTag('tbody')
+            );
+        $expect[2] = true;
+        
+        $inputs[3] = array(
+            new MF_StartTag('col')      ,new MF_EndTag('col')
+           ,new MF_StartTag('col')      ,new MF_EndTag('col')
+           ,new MF_StartTag('col')      ,new MF_EndTag('col')
+           ,new MF_StartTag('tr')       ,new MF_EndTag('tr')
+            );
+        $expect[3] = true;
+        
+        $this->assertSeries($inputs, $expect, $def);
+        
+    }
+    
     function test_simple() {
         
+        // simple is actually an abstract class
+        // but we're unit testing some of the conv. functions it gives
+        
         $def = new HTMLDTD_ChildDef_Simple('foobar | bang |gizmo');
         $this->assertEqual($def->elements,
           array(
@@ -274,15 +327,7 @@ class Test_HTMLDTD_ChildDef extends UnitTestCase
             );
         $expect[5] = false;
         
-        foreach ($inputs as $i => $input) {
+        $this->assertSeries($inputs, $expect, $def);
-            $result = $def->validateChildren($input);
-            if (is_bool($expect[$i])) {
-                $this->assertIdentical($expect[$i], $result);
-            } else {
-                $this->assertEqual($expect[$i], $result);
-                paintIf($result, $result != $expect[$i]);
-            }
-        }
         
     }