BogusCOmments. How cool.

2 files changed, 48 insertions, 23 deletions

diff --git a/src/HTML5/Parser/Tokenizer.php b/src/HTML5/Parser/Tokenizer.php
index a23e1f4..b84f9bf 100644
--- a/src/HTML5/Parser/Tokenizer.php
+++ b/src/HTML5/Parser/Tokenizer.php
@@ -235,8 +235,18 @@ class Tokenizer {
 
     // Comment:
     if ($tok == '-' && $this->scanner->peek() == '-') {
+      $this->scanner->next(); // Consume the other '-'
+      $this->scanner->next(); // Next char.
+      return $this->comment();
     }
+    elseif($tok == 'D') {
+    }
+    elseif($tok == '[') {
+    }
+
     // FINISH
+    $this->parseError("Expected <!--, <![CDATA[, or <!DOCTYPE. Got <!%s", $tok);
+    $this->bogusComment('<!');
     return TRUE;
   }
 
@@ -472,10 +482,11 @@ class Tokenizer {
     return TRUE;
   }
 
-  protected function commentStart() {
-  }
-  protected function commentStartDash() {
-  }
+  /**
+   * Read a comment.
+   *
+   * Expects the first tok to be inside of the comment.
+   */
   protected function comment() {
     $tok = $this->scanner->current();
     $comment = '';
@@ -504,17 +515,18 @@ class Tokenizer {
   }
 
   protected function isCommentEnd() {
+    // EOF
+    if($this->scanner->current() === FALSE) {
+      // Hit the end.
+      $this->parseError("Unexpected EOF in a comment.");
+      return TRUE;
+    }
+
     // If it doesn't start with -, not the end.
     if($this->scanner->current() != '-') {
       return FALSE;
     }
 
-    // EOF
-    if($this->scanner->Current() === FALSE) {
-      // Hit the end.
-      $this->events->parseError("Unexpected EOF in a comment.");
-      return TRUE;
-    }
 
     // Advance one, and test for '->'
     if ($this->scanner->next() == '-'
diff --git a/test/HTML5/Parser/TokenizerTest.php b/test/HTML5/Parser/TokenizerTest.php
index c057360..291bd0f 100644
--- a/test/HTML5/Parser/TokenizerTest.php
+++ b/test/HTML5/Parser/TokenizerTest.php
@@ -87,13 +87,18 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
   }
 
   public function testBogusComment() {
-    $str = '</+this is a bogus comment. +>';
-    $events = $this->parse($str . '   ');
-    $e0 = $events->get(0);
-    $this->assertEquals('error', $e0['name']);
-    $e1 = $events->get(1);
-    $this->assertEquals('comment', $e1['name']);
-    $this->assertEquals($str, $e1['data'][0]);
+    $bogus = array(
+      '</+this is a bogus comment. +>',
+      '<!+this is a bogus comment. !>',
+    );
+    foreach ($bogus as $str) {
+      $events = $this->parse($str . '   ');
+      $e0 = $events->get(0);
+      $this->assertEquals('error', $e0['name']);
+      $e1 = $events->get(1);
+      $this->assertEquals('comment', $e1['name']);
+      $this->assertEquals($str, $e1['data'][0]);
+    }
   }
 
   public function testEndTag() {
@@ -162,19 +167,27 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
       '<!-- --$i -->' => ' --$i ',
       '<!----$i-->' => '--$i',
       '<!-- 1 > 0 -->' => ' 1 > 0 ',
-      '<!--
-      Hello World.
-      -->' => "\nHello World\n",
+      "<!--\nHello World.\na-->" => "\nHello World.\na",
       '<!-- <!-- -->' => ' <!-- ',
     );
+    foreach ($good as $test => $expected) {
+      $events = $this->parse($test);
+      $e1 = $events->get(0);
+      $this->assertEquals('comment', $e1['name'], 'Expected a comment for ' . $test);
+      $this->assertEquals($expected, $e1['data'][0]);
+    }
+
     $fail = array(
       '<!-->' => '',
       '<!--Hello' => 'Hello',
+      "<!--\0Hello" => UTF8Utils::FFFD . 'Hello',
     );
+    foreach ($fail as $test => $expected) {
+      $events = $this->parse($test);
+      $e0 = $events->get(0);
+      $this->assertEquals('error', $e0['name'], 'Expected an error for ' . $test . print_r($events, TRUE));
 
-    foreach ($good as $test => $expected) {
-      $events = $this->parse($good);
-      $e1 = $events->get(0);
+      $e1 = $events->get(1);
       $this->assertEquals('comment', $e1['name'], 'Expected a comment for ' . $test);
       $this->assertEquals($expected, $e1['data'][0]);
     }