[2.1.0] Fix bug in auto-paragraphing: empty tags should be treated like start tags too.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1269 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/Injector.php

@@ -103,9 +103,9 @@ class HTMLPurifier_Injector
     function handleText(&$token) {}
     
     /**
-     * Handler that is called when a start token is processed
+     * Handler that is called when a start or empty token is processed
      */
-    function handleStart(&$token) {}
+    function handleElement(&$token) {}
     
 }
 

library/HTMLPurifier/Injector/AutoParagraph.php

@@ -87,7 +87,7 @@ class HTMLPurifier_Injector_AutoParagraph extends HTMLPurifier_Injector
         
     }
     
-    function handleStart(&$token) {
+    function handleElement(&$token) {
         // check if we're inside a tag already
         if (!empty($this->currentNesting)) {
             if ($this->allowsElement('p')) {
@@ -113,8 +113,8 @@ class HTMLPurifier_Injector_AutoParagraph extends HTMLPurifier_Injector
                 $ok = false;
                 // maintain a mini-nesting counter, this lets us bail out
                 // early if possible
-                $j = 2; // current nesting, is two due to parent and this start
+                $j = 1; // current nesting, one is due to parent (we recalculate current token)
-                for ($i = $this->inputIndex + 1; isset($this->inputTokens[$i]); $i++) {
+                for ($i = $this->inputIndex; isset($this->inputTokens[$i]); $i++) {
                     if ($this->inputTokens[$i]->type == 'start') $j++;
                     if ($this->inputTokens[$i]->type == 'end') $j--;
                     if ($this->inputTokens[$i]->type == 'text') {

library/HTMLPurifier/Strategy/MakeWellFormed.php

@@ -127,26 +127,24 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
             
             $info = $definition->info[$token->name]->child;
             
-            // quick checks:
+            // quick tag checks: anything that's *not* an end tag
-            // test if it claims to be a start tag but is empty
+            $ok = false;
             if ($info->type == 'empty' && $token->type == 'start') {
-                $result[] = new HTMLPurifier_Token_Empty($token->name, $token->attr);
+                // test if it claims to be a start tag but is empty
-                continue;
+                $token = new HTMLPurifier_Token_Empty($token->name, $token->attr);
-            }
+                $ok = true;
-            // test if it claims to be empty but really is a start tag
+            } elseif ($info->type != 'empty' && $token->type == 'empty' ) {
-            if ($info->type != 'empty' && $token->type == 'empty' ) {
+                // claims to be empty but really is a start tag
-                $result[] = new HTMLPurifier_Token_Start($token->name, $token->attr);
+                $token = array(
-                $result[] = new HTMLPurifier_Token_End($token->name);
+                    new HTMLPurifier_Token_Start($token->name, $token->attr),
-                continue;
+                    new HTMLPurifier_Token_End($token->name)
-            }
+                );
-            // automatically insert empty tags
+                $ok = true;
-            if ($token->type == 'empty') {
+            } elseif ($token->type == 'empty') {
-                $result[] = $token;
+                // real empty token
-                continue;
+                $ok = true;
-            }
+            } elseif ($token->type == 'start') {
-            
+                // start tag
-            // start tags have precedence, so they get passed through...
-            if ($token->type == 'start') {
                 
                 // ...unless they also have to close their parent
                 if (!empty($this->currentNesting)) {
@@ -168,16 +166,18 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
                     
                     $this->currentNesting[] = $parent; // undo the pop
                 }
-                
+                $ok = true;
-                // injector handler code; duplicated for performance reasons
+            }
+            
+            // injector handler code; duplicated for performance reasons
+            if ($ok) {
                 foreach ($this->injectors as $i => $x) {
-                    if (!$x->skip) $x->handleStart($token);
+                    if (!$x->skip) $x->handleElement($token);
                     if (is_array($token)) {
                         $this->currentInjector = $i;
                         break;
                     }
                 }
-                
                 $this->processToken($token, $config, $context);
                 continue;
             }
@@ -285,9 +285,11 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
             array_splice($this->inputTokens, $this->inputIndex--, 1, $token);
             
             // adjust the injector skips based on the array substitution
-            $offset = count($token) + 1;
+            if ($this->injectors) {
-            for ($i = 0; $i <= $this->currentInjector; $i++) {
+                $offset = count($token) + 1;
-                $this->injectors[$i]->skip += $offset;
+                for ($i = 0; $i <= $this->currentInjector; $i++) {
+                    $this->injectors[$i]->skip += $offset;
+                }
             }
         } elseif ($token) {
             // regular case

tests/HTMLPurifier/Injector/AutoParagraphTest.php

@@ -237,6 +237,11 @@ Par1
 '<p>Par1</p><p><b>Par2</b></p>'
         );
         
+        $this->assertResult(
+'<img /> Foo',
+'<p><img /> Foo</p>'
+        );
+        
     }
     
     function testInlineRootNode() {