DOMDocument::loadHTML() UTF-8 issue

tijsverkoyen · Mar 30, 2016 · a942e3a · mlocati · May 10, 2016 · techi602
1 parent bb535b0
commit a942e3a
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 0 deletions.
diff --git a/src/CssToInlineStyles.php b/src/CssToInlineStyles.php
@@ -100,6 +100,12 @@ public function getInlineStyles(\DOMElement $element)
      */
     protected function createDomDocumentFromHtml($html)
     {
+        $html = trim($html);
+        if (strstr('<?xml', $html) !== 0) {
+            $xmlHeader = '<?xml encoding="utf-8" ?>';
+            $html = $xmlHeader . $html;
+        }
+
         $document = new \DOMDocument('1.0', 'UTF-8');
         $internalErrors = libxml_use_internal_errors(true);
         $document->loadHTML($html);

diff --git a/tests/CssToInlineStylesTest.php b/tests/CssToInlineStylesTest.php
@@ -198,6 +198,14 @@ public function testInvalidSelector()
         $this->assertCorrectConversion($expected, $html, $css);
     }
 
+    public function testHtmlEncoding()
+    {
+        $text = 'Žluťoučký kůň pije pivo nebo jak to je dál';
+        $expectedText = '&#x17D;lu&#x165;ou&#x10D;k&#xFD; k&#x16F;&#x148; pije pivo nebo jak to je d&#xE1;l';
+
+        $this->assertEquals($expectedText, trim(strip_tags($this->cssToInlineStyles->convert($text, ''))));
+    }
+
     private function assertCorrectConversion($expected, $html, $css = null)
     {
         $this->assertEquals(