Working on entity resolution.

2 files changed, 43 insertions, 3 deletions

diff --git a/test/HTML5/Parser/EventStack.php b/test/HTML5/Parser/EventStack.php
index 213d893..a75402e 100644
--- a/test/HTML5/Parser/EventStack.php
+++ b/test/HTML5/Parser/EventStack.php
@@ -1,6 +1,10 @@
 <?php
 namespace HTML5\Parser;
 
+/**
+ * This testing class gathers events from a parser and builds a stack of events.
+ * It is useful for checking the output of a tokenizer.
+ */
 class EventStack implements EventHandler {
   protected $stack;
 
@@ -19,8 +23,12 @@ class EventStack implements EventHandler {
     return count($this->stack);
   }
 
+  public function get($index) {
+    return $this->stack[$index];
+  }
+
   protected function store($event, $data = NULL) {
-    $stack[] = array(
+    $this->stack[] = array(
       'name' => $event,
       'data' => $data,
     );
@@ -45,7 +53,8 @@ class EventStack implements EventHandler {
   }
 
   public function text($cdata) {
-    $this->store('character', array($cdata));
+    fprintf(STDOUT, "Received TEXT event with: " . $cdata);
+    $this->store('text', array($cdata));
   }
 
   public function eof() {
diff --git a/test/HTML5/Parser/TokenizerTest.php b/test/HTML5/Parser/TokenizerTest.php
index d10c683..0e93bb5 100644
--- a/test/HTML5/Parser/TokenizerTest.php
+++ b/test/HTML5/Parser/TokenizerTest.php
@@ -19,6 +19,37 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
 
     $tok->parse();
 
-    $this->assertEquals(0, $events->Depth());
+    $this->assertEquals(1, $events->Depth());
+    $this->assertEquals('eof', $e1['name']);
+  }
+
+  public function testWhitespace() {
+    $spaces = '    ';
+    list($tok, $events) = $this->createTokenizer($spaces);
+
+    $tok->parse();
+
+    $this->assertEquals(2, $events->depth());
+
+    $e1 = $events->get(0);
+
+    $this->assertEquals('text', $e1['name']);
+    $this->assertEquals($spaces, $e1['data'][0]);
+  }
+
+  public function testCharacterReference() {
+    $str = '&amp;';
+    list($tok, $events) = $this->createTokenizer($str);
+
+    $tok->parse();
+    $this->assertEquals(2, $events->depth());
+    $e1 = $events->get(0);
+
+    $this->assertEquals('&', $e1['data'][0]);
+
+    // Test with hex charref
+    // Test with decimal charref
+    // Test with broken charref
+    // Test with stand-alone ampersand
   }
 }