#26: Updated the case handling for tags to allow for uppercase tags and normalizing tag names to lowercase (per 8.2.4.9) except for SVG foreign tags that are case sensitive.

4 files changed, 17 insertions, 6 deletions

diff --git a/src/HTML5/Parser/DOMTreeBuilder.php b/src/HTML5/Parser/DOMTreeBuilder.php
index 6dbd25e..13ae3bc 100644
--- a/src/HTML5/Parser/DOMTreeBuilder.php
+++ b/src/HTML5/Parser/DOMTreeBuilder.php
@@ -289,6 +289,11 @@ class DOMTreeBuilder implements EventHandler {
       return;
     }
 
+    // Special case handling for SVG.
+    if ($this->insertMode == static::IM_IN_SVG) {
+      $lname = Elements::normalizeSvgElement($lname);
+    }
+
     // XXX: Not sure whether we need this anymore.
     // if ($name != $lname) {
     //  return $this->quirksTreeResolver($lname);
@@ -301,8 +306,8 @@ class DOMTreeBuilder implements EventHandler {
     }
 
     //$this->current = $this->current->parentNode;
-    if (!$this->autoclose($name)) {
-      $this->parseError('Could not find closing tag for ' . $name);
+    if (!$this->autoclose($lname)) {
+      $this->parseError('Could not find closing tag for ' . $lname);
     }
 
     //switch ($this->insertMode) {
diff --git a/src/HTML5/Parser/Tokenizer.php b/src/HTML5/Parser/Tokenizer.php
index f81f069..04baa10 100644
--- a/src/HTML5/Parser/Tokenizer.php
+++ b/src/HTML5/Parser/Tokenizer.php
@@ -295,7 +295,7 @@ class Tokenizer {
       return $this->bogusComment('</');
     }
 
-    $name = $this->scanner->charsUntil("\n\f \t>");
+    $name = strtolower($this->scanner->charsUntil("\n\f \t>"));
     // Trash whitespace.
     $this->scanner->whitespace();
 
@@ -831,7 +831,7 @@ class Tokenizer {
       $buffer .= $this->scanner->charsUntil($first);
 
       // Stop as soon as we hit the stopping condition.
-      if ($this->sequenceMatches($sequence)) {
+      if ($this->sequenceMatches($sequence) || $this->sequenceMatches(strtoupper($sequence))) {
         return $buffer;
       }
       $buffer .= $this->scanner->current();
diff --git a/test/HTML5/Html5Test.php b/test/HTML5/Html5Test.php
index 4e7c300..dcb51cd 100644
--- a/test/HTML5/Html5Test.php
+++ b/test/HTML5/Html5Test.php
@@ -134,7 +134,7 @@ class Html5Test extends TestCase {
         </body>
       </html>');
 
-    $this->assertEmpty($dom->errors);
+    $this->assertEmpty($dom->errors, print_r($dom->errors, TRUE));
 
     // Test a mixed case attribute.
     $list = $dom->getElementsByTagName('svg');
diff --git a/test/HTML5/Parser/TokenizerTest.php b/test/HTML5/Parser/TokenizerTest.php
index edc427c..9f335b0 100644
--- a/test/HTML5/Parser/TokenizerTest.php
+++ b/test/HTML5/Parser/TokenizerTest.php
@@ -142,7 +142,7 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
       '</test 
       >' => 'test',
       '</thisIsTheTagThatDoesntEndItJustGoesOnAndOnMyFriend>' =>
-        'thisIsTheTagThatDoesntEndItJustGoesOnAndOnMyFriend',
+        'thisisthetagthatdoesntenditjustgoesonandonmyfriend',
       // See 8.2.4.10, which requires this and does not say error.
       '</a<b>' => 'a<b', 
     );
@@ -419,6 +419,12 @@ class TokenizerTest extends \HTML5\Tests\TestCase {
       $this->assertEventEquals('text', $expects, $events->get(2));
     }
 
+    // Testing case sensitivity
+    $events = $this->parse('<TITLE>a test</TITLE>');
+    $this->assertEventEquals('startTag', 'title', $events->get(0));
+    $this->assertEventEquals('text', 'a test', $events->get(1));
+    $this->assertEventEquals('endTag', 'title', $events->get(2));
+
   }
 
   public function testText() {