Working on closing tag and bogus comments.

2 files changed, 51 insertions, 9 deletions

diff --git a/test/HTML5/Parser/EventStack.php b/test/HTML5/Parser/EventStack.php
index 36e2f29..4d82629 100644
--- a/test/HTML5/Parser/EventStack.php
+++ b/test/HTML5/Parser/EventStack.php
@@ -64,7 +64,7 @@ class EventStack implements EventHandler {
   public function parseError($msg, $line, $col) {
     //throw new EventStackParseError(sprintf("%s (line %d, col %d)", $msg, $line, $col));
     //$this->store(sprintf("%s (line %d, col %d)", $msg, $line, $col));
-    $this->store('comment', func_get_args());
+    $this->store('error', func_get_args());
   }
 
 
diff --git a/test/HTML5/Parser/TokenizerTest.php b/test/HTML5/Parser/TokenizerTest.php
index e006fb9..c31d74b 100644
--- a/test/HTML5/Parser/TokenizerTest.php
+++ b/test/HTML5/Parser/TokenizerTest.php
@@ -4,18 +4,21 @@ require __DIR__ . '/../TestCase.php';
 require 'EventStack.php';
 
 class TokenizerTest extends \HTML5\Tests\TestCase {
-  protected function createTokenizer($string) {
+  protected function createTokenizer($string, $debug = FALSE) {
     $eventHandler = new EventStack();
     $stream = new StringInputStream($string);
     $scanner = new Scanner($stream);
+
+    $scanner->debug = $debug;
+
     return array(
       new Tokenizer($scanner, $eventHandler),
       $eventHandler,
     );
   }
 
-  public function parse($string) {
-    list($tok, $events) = $this->createTokenizer($string);
+  public function parse($string, $debug = FALSE) {
+    list($tok, $events) = $this->createTokenizer($string, $debug);
     $tok->parse();
 
     return $events;
@@ -75,20 +78,59 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
 
   }
 
-  /**
-   * @expectedException \HTML5\Parser\EventStackParseError
-   */
   public function testBrokenCharacterReference() {
     // Test with broken charref
     $str = '&foo';
     $events = $this->parse($str);
+    $e1 = $events->get(0);
+    $this->assertEquals('error', $e1['name']);
   }
 
   public function testBogusComment() {
     $str = '</+this is a bogus comment. +>';
-    $events = $this->parse($str);
-    $e1 = $events->get(0);
+    $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() {
+    $succeed = array(
+      '</a>' => 'a',
+      '</test>' => 'test',
+      '</test 
+      >' => 'test',
+      // See 8.2.4.10, which requires this and does not say error.
+      '</a<b>' => 'a<b', 
+    );
+    foreach ($succeed as $test => $result) {
+      $events = $this->parse($test);
+      $this->assertEquals(2, $events->depth());
+      $e1 = $events->get(0);
+      $this->assertEquals('endTag', $e1['name'], "Parsing $test expects $result.");
+      $this->assertEquals($result, $e1['data'][0], "Parse end tag " . $test);
+    }
+
+
+    $fail = array(
+      '</a class="monkey">' => 'a',
+      '</a <b>' => 'a',
+      '</ a>' => 'a',
+      '</>' => '',
+      '</ >' => '',
+    );
+    foreach ($fail as $test => $result) {
+      $events = $this->parse($test);
+      // Should have triggered an error.
+      $e0 = $events->get(0);
+      $this->assertEquals('error', $e0['name'], "Parsing $test expects a leading error." . print_r($events, TRUE));
+      // Should have tried to parse anyway.
+      $e1 = $events->get(1);
+      $this->assertEquals('endTag', $e1['name'], "Parsing $test expects resolution to $result." . print_r($events, TRUE));
+      $this->assertEquals($result, $e1['data'][0], "Parse end tag " . $test);
+      $this->assertEquals(3, $events->depth());
+    }
+  }
 }