From ffcfa507b081cf132db5b90c26bfad66d79a4eb4 Mon Sep 17 00:00:00 2001
From: Technosophos <technosophos@gmail.com>
Date: Fri, 12 Apr 2013 17:53:01 -0500
Subject: DOCTYPE bogus comments handled.

---
 src/HTML5/Parser/Tokenizer.php      | 11 +++++++++++
 test/HTML5/Parser/TokenizerTest.php |  1 +
 2 files changed, 12 insertions(+)

diff --git a/src/HTML5/Parser/Tokenizer.php b/src/HTML5/Parser/Tokenizer.php
index b84f9bf..899e908 100644
--- a/src/HTML5/Parser/Tokenizer.php
+++ b/src/HTML5/Parser/Tokenizer.php
@@ -539,6 +539,17 @@ class Tokenizer {
     return FALSE;
   }
   protected function doctype() {
+    if ($this->scanner->current() != 'D') {
+      return FALSE;
+    }
+    // Check that string is DOCTYPE
+    $chars = $this->scanner->charsWhile("DOCTYPE");
+    if ($chars != 'DOCTYPE') {
+      $this->parseError('Expected DOCTYPE, got %s', $chars);
+      return $this->bogusComment('<!' .  $chars);
+    }
+
+    // Now we need to parse the DOCTYPE.
   }
   protected function beforeDoctype() {
   }
diff --git a/test/HTML5/Parser/TokenizerTest.php b/test/HTML5/Parser/TokenizerTest.php
index 291bd0f..ead02d0 100644
--- a/test/HTML5/Parser/TokenizerTest.php
+++ b/test/HTML5/Parser/TokenizerTest.php
@@ -90,6 +90,7 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
     $bogus = array(
       '</+this is a bogus comment. +>',
       '<!+this is a bogus comment. !>',
+      '<!DOCTYEP foo bar>',
     );
     foreach ($bogus as $str) {
       $events = $this->parse($str . '   ');
-- 
cgit v1.2.3