Save as html instead of xml.

This solves the issues with self-closing tags (<br> going to <br></br>) and assists with utf-8 encoding.
tijsverkoyen · Aug 30, 2016 · 4e1620a · 4e1620a
1 parent b081b74
commit 4e1620a
Show file tree

Hide file tree

Showing 2 changed files with 53 additions and 20 deletions.
diff --git a/src/CssToInlineStyles.php b/src/CssToInlineStyles.php
@@ -100,15 +100,9 @@ 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);
+        $document->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
         libxml_use_internal_errors($internalErrors);
         $document->formatOutput = true;
 
@@ -121,15 +115,23 @@ protected function createDomDocumentFromHtml($html)
      */
     protected function getHtmlFromDocument(\DOMDocument $document)
     {
-        $xml = $document->saveXML(null, LIBXML_NOEMPTYTAG);
+        // retrieve the document element
+        // we do it this way to preserve the utf-8 encoding
+        $htmlElement = $document->documentElement;
+        $html = $document->saveHTML($htmlElement);
+        $html = trim($html);
 
-        $html = preg_replace(
-            '|<\?xml (.*)\?>|',
-            '',
-            $xml
-        );
+        // retrieve the doctype
+        $document->removeChild($htmlElement);
+        $doctype = $document->saveHTML();
+        $doctype = trim($doctype);
+
+        // if it is the html5 doctype convert it to lowercase
+        if ($doctype === '<!DOCTYPE html>') {
+            $doctype = strtolower($doctype);
+        }
 
-        return ltrim($html);
+        return $doctype."\n".$html;
     }
 
     /**

diff --git a/tests/CssToInlineStylesTest.php b/tests/CssToInlineStylesTest.php
@@ -84,11 +84,17 @@ public function testApplyBasicStylesOnElementWithInlineStyles()
 
     public function testBasicRealHTMLExample()
     {
-        $html = '<!DOCTYPE html><html><body><p>foo</p></body></html>';
+        $html = '<!doctype html><html><head><style>body{color:blue}</style></head><body><p>foo</p></body></html>';
         $css = 'p { color: red; }';
-        $expected = '<p style="color: red;">foo</p>';
+        $expected = <<<EOF
+<!doctype html>
+<html>
+<head><style>body{color:blue}</style></head>
+<body style="color: blue;"><p style="color: red;">foo</p></body>
+</html>
+EOF;
 
-        $this->assertCorrectConversion($expected, $html, $css);
+        $this->assertEquals($expected, $this->cssToInlineStyles->convert($html, $css));
     }
 
     public function testSimpleElementSelector()
@@ -175,7 +181,7 @@ public function testSpecificity()
 EOF;
         $expected = <<<EOF
 <a class="one" id="ONE" style="padding: 100px; border: 1px solid red; margin: 10px; width: 20px !important;">
-  <img class="two" id="TWO" style="border: none;"></img></a>
+  <img class="two" id="TWO" style="border: none;"></a>
 EOF;
         $this->assertCorrectConversion($expected, $html, $css);
     }
@@ -201,9 +207,34 @@ public function testInvalidSelector()
     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';
+        $expected = $text;
 
-        $this->assertEquals($expectedText, trim(strip_tags($this->cssToInlineStyles->convert($text, ''))));
+        $this->assertEquals($expected, trim(strip_tags($this->cssToInlineStyles->convert($text))));
+    }
+
+    public function testSpecialCharacters()
+    {
+        $text = '1 &lt; 2';
+        $expected = $text;
+
+        $this->assertEquals($expected, trim(strip_tags($this->cssToInlineStyles->convert($text))));
+    }
+
+    public function testSpecialCharactersExplicit()
+    {
+        $text = '&amp;lt;script&amp;&gt;';
+        $expected = $text;
+
+        $this->assertEquals($expected, trim(strip_tags($this->cssToInlineStyles->convert($text))));
+    }
+
+    public function testSelfClosingTags()
+    {
+        $html = '<br>';
+        $css = '';
+        $expected = $html;
+
+        $this->assertCorrectConversion($expected, $html, $css);
     }
 
     private function assertCorrectConversion($expected, $html, $css = null)