Auto closing tags if they allows only inline nodes as child nodes

Fixes #68
Masterminds · Dec 14, 2014 · 025f331 · 025f331
1 parent c158103
commit 025f331
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 2 deletions.
diff --git a/src/HTML5/Elements.php b/src/HTML5/Elements.php
@@ -66,6 +66,11 @@ class Elements
      */
     const BLOCK_TAG = 64;
 
+    /**
+     * Indicates that the tag allows only inline elements as child nodes.
+     */
+    const BLOCK_ONLY_INLINE = 128;
+
     /**
      * The HTML5 elements as defined in http://dev.w3.org/html5/markup/elements.html.
      *
@@ -120,7 +125,7 @@ class Elements
         "head" => 1,
         "header" => 81, // NORMAL | AUTOCLOSE_P | BLOCK_TAG
         "hgroup" => 81, // NORMAL | AUTOCLOSE_P | BLOCK_TAG
-        "hr" => 73, // NORMAL | VOID_TAG | BLOCK_TAG
+        "hr" => 73, // NORMAL | VOID_TAG
         "html" => 1,
         "i" => 1,
         "iframe" => 3, // NORMAL | TEXT_RAW
@@ -145,7 +150,7 @@ class Elements
         "optgroup" => 1,
         "option" => 1,
         "output" => 65, // NORMAL | BLOCK_TAG
-        "p" => 81, // NORMAL | AUTOCLOSE_P | BLOCK_TAG
+        "p" => 209, // NORMAL | AUTOCLOSE_P | BLOCK_TAG | BLOCK_ONLY_INLINE
         "param" => 9, // NORMAL | VOID_TAG
         "pre" => 81, // NORMAL | AUTOCLOSE_P | BLOCK_TAG
         "progress" => 1,

diff --git a/src/HTML5/Parser/DOMTreeBuilder.php b/src/HTML5/Parser/DOMTreeBuilder.php
@@ -138,6 +138,12 @@ class DOMTreeBuilder implements EventHandler
 
     protected $insertMode = 0;
 
+    /**
+     * Track if we are in an element that allows only inline child nodes
+     * @var string|null
+     */
+    protected $onlyInline;
+
     /**
      * Quirks mode is enabled by default.
      * Any document that is missing the
@@ -320,6 +326,11 @@ public function startTag($name, $attributes = array(), $selfClosing = false)
             }
         }
 
+        if ($this->onlyInline && Elements::isA($lname, Elements::BLOCK_TAG)) {
+        	$this->autoclose($this->onlyInline);
+        	$this->onlyInline = null;
+        }
+
         try {
             $prefix = ($pos = strpos($lname, ':')) ? substr($lname, 0, $pos) : '';
 
@@ -346,6 +357,10 @@ public function startTag($name, $attributes = array(), $selfClosing = false)
             $ele = $this->doc->createElement('invalid');
         }
 
+        if (Elements::isA($lname, Elements::BLOCK_ONLY_INLINE)) {
+        	$this->onlyInline = $lname;
+        }
+
         // When we add some namespacess, we have to track them. Later, when "endElement" is invoked, we have to remove them.
         // When we are on a void tag, we do not need to care about namesapce nesting.
         if ($pushes > 0 && !Elements::isA($name, Elements::VOID_TAG)) {

diff --git a/test/HTML5/Parser/DOMTreeBuilderTest.php b/test/HTML5/Parser/DOMTreeBuilderTest.php
@@ -204,6 +204,15 @@ public function testXmlNamespaceNesting()
         $this->assertEquals("http://www.prefixed.com/bar5_x", $prefixed->namespaceURI);
     }
 
+    public function testMoveNonInlineElements()
+    {
+    	$doc = $this->parse('<p>line1<br/><hr/>line2</p>');
+		$this->assertEquals('<html xmlns="http://www.w3.org/1999/xhtml"><p>line1<br/></p><hr/>line2</html>', $doc->saveXML($doc->documentElement), 'Move non-inline elements outside of inline containers.');
+
+		$doc = $this->parse('<p>line1<div>line2</div></p>');
+		$this->assertEquals('<html xmlns="http://www.w3.org/1999/xhtml"><p>line1</p><div>line2</div></html>', $doc->saveXML($doc->documentElement), 'Move non-inline elements outside of inline containers.');
+    }
+
     public function testAttributes()
     {
         $html = "<!DOCTYPE html>