Merge pull request #76 from zhaofengli/i75

Allow whitespaces in RCDATA end tags

2 files changed, 14 insertions, 1 deletions

diff --git a/src/HTML5/Parser/Tokenizer.php b/src/HTML5/Parser/Tokenizer.php
index 958ade8..e00b9a2 100644
--- a/src/HTML5/Parser/Tokenizer.php
+++ b/src/HTML5/Parser/Tokenizer.php
@@ -200,7 +200,7 @@ class Tokenizer
         if (is_null($this->untilTag)) {
             return $this->text();
         }
-        $sequence = '</' . $this->untilTag . '>';
+        $sequence = '</' . $this->untilTag;
         $txt = '';
         $tok = $this->scanner->current();
 
@@ -214,6 +214,13 @@ class Tokenizer
                 $tok = $this->scanner->next();
             }
         }
+        $len = strlen($sequence);
+        $this->scanner->consume($len);
+        $len += strlen($this->scanner->whitespace());
+        if ($this->scanner->current() !== '>') {
+            $this->parseError("Unclosed RCDATA end tag");
+        }
+        $this->scanner->unconsume($len);
         $this->events->text($txt);
         $this->setTextMode(0);
         return $this->endTag();
diff --git a/test/HTML5/Parser/TokenizerTest.php b/test/HTML5/Parser/TokenizerTest.php
index 8ff4d5c..3d834fd 100644
--- a/test/HTML5/Parser/TokenizerTest.php
+++ b/test/HTML5/Parser/TokenizerTest.php
@@ -884,6 +884,12 @@ class TokenizerTest extends \Masterminds\HTML5\Tests\TestCase
         $this->assertEventEquals('startTag', 'title', $events->get(0));
         $this->assertEventEquals('text', 'a test', $events->get(1));
         $this->assertEventEquals('endTag', 'title', $events->get(2));
+
+        // Testing end tags with whitespaces
+        $events = $this->parse('<title>Whitespaces are tasty</title >');
+        $this->assertEventEquals('startTag', 'title', $events->get(0));
+        $this->assertEventEquals('text', 'Whitespaces are tasty', $events->get(1));
+        $this->assertEventEquals('endTag', 'title', $events->get(2));
     }
 
     public function testRcdata()