tijsverkoyen · bytestream · May 17, 2021 · May 17, 2021
diff --git a/composer.json b/composer.json
@@ -18,7 +18,11 @@
     "symfony/css-selector": "^2.7 || ^3.0 || ^4.0 || ^5.0"
   },
   "require-dev": {
-    "phpunit/phpunit": "^4.8.35 || ^5.7 || ^6.0 || ^7.5"
+    "phpunit/phpunit": "^4.8.35 || ^5.7 || ^6.0 || ^7.5",
+    "masterminds/html5": "^2.7"
+  },
+  "suggest": {
+    "masterminds/html5": "To use a HTML5 parser instead of native DOM parser."
   },
   "autoload": {
     "psr-4": {

diff --git a/src/CssToInlineStyles.php b/src/CssToInlineStyles.php
@@ -2,6 +2,8 @@
 
 namespace TijsVerkoyen\CssToInlineStyles;
 
+use LogicException;
+use Masterminds\HTML5;
 use Symfony\Component\CssSelector\CssSelector;
 use Symfony\Component\CssSelector\CssSelectorConverter;
 use Symfony\Component\CssSelector\Exception\ExceptionInterface;
@@ -13,11 +15,28 @@ class CssToInlineStyles
 {
     private $cssConverter;
 
-    public function __construct()
+    /** @var HTML5|null */
+    private $html5Parser;
+
+    /** @var bool */
+    private $isHtml5Document = false;
+
+    /**
+     * @param bool|null $useHtml5Parser Whether to use a HTML5 parser or the native DOM parser
+     */
+    public function __construct($useHtml5Parser = null)
     {
         if (class_exists('Symfony\Component\CssSelector\CssSelectorConverter')) {
             $this->cssConverter = new CssSelectorConverter();
         }
+
+        if ($useHtml5Parser) {
+            if (! class_exists(HTML5::class)) {
+                throw new LogicException('Using the HTML5 parser requires the html5-php library. Try running "composer require masterminds/html5".');
+            }
+
+            $this->html5Parser = new HTML5(['disable_html_ns' => true]);
+        }
     }
 
     /**
@@ -110,27 +129,95 @@ public function getInlineStyles(\DOMElement $element)
      * @return \DOMDocument
      */
     protected function createDomDocumentFromHtml($html)
+    {
+        $this->isHtml5Document = false;
+
+        if ($this->canParseHtml5String($html)) {
+            return $this->parseHtml5($html);
+        }
+
+        return $this->parseXhtml($html);
+    }
+
+    /**
+     * @param string $html
+     * @return \DOMDocument
+     */
+    protected function parseHtml5($html)
+    {
+        $this->isHtml5Document = true;
+
+        return $this->html5Parser->parse($this->convertToHtmlEntities($html));
+    }
+
+    /**
+     * @param string $html
+     * @return \DOMDocument
+     */
+    protected function parseXhtml($html)
     {
         $document = new \DOMDocument('1.0', 'UTF-8');
         $internalErrors = libxml_use_internal_errors(true);
-        $document->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
+        $document->loadHTML($this->convertToHtmlEntities($html));
         libxml_use_internal_errors($internalErrors);
         $document->formatOutput = true;
 
         return $document;
     }
 
+    /**
+     * @param string $content
+     * @return bool
+     */
+    protected function canParseHtml5String($content)
+    {
+        if (null === $this->html5Parser) {
+            return false;
+        }
+
+        if (false === ($pos = stripos($content, '<!doctype html>'))) {
+            return false;
+        }
+
+        $header = substr($content, 0, $pos);
+
+        return '' === $header || $this->isValidHtml5Heading($header);
+    }
+
+    /**
+     * @param string $heading
+     * @return bool
+     */
+    protected function isValidHtml5Heading($heading)
+    {
+        return 1 === preg_match('/^\x{FEFF}?\s*(<!--[^>]*?-->\s*)*$/u', $heading);
+    }
+
+    /**
+     * @param string $html
+     * @return array|false|string
+     */
+    protected function convertToHtmlEntities($html)
+    {
+        return mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8');
+    }
+
     /**
      * @param \DOMDocument $document
      *
      * @return string
      */
     protected function getHtmlFromDocument(\DOMDocument $document)
     {
+        $parser = $document;
+        if (null !== $this->html5Parser && $this->isHtml5Document) {
+            $parser = $this->html5Parser;
+        }
+
         // retrieve the document element
         // we do it this way to preserve the utf-8 encoding
         $htmlElement = $document->documentElement;
-        $html = $document->saveHTML($htmlElement);
+        $html = $parser->saveHTML($htmlElement);
         $html = trim($html);
 
         // retrieve the doctype

diff --git a/tests/HTML5ParserTest.php b/tests/HTML5ParserTest.php
@@ -0,0 +1,66 @@
+<?php
+
+namespace TijsVerkoyen\CssToInlineStyles\tests;
+
+use PHPUnit\Framework\TestCase;
+use TijsVerkoyen\CssToInlineStyles\CssToInlineStyles;
+
+class HTML5ParserTest extends TestCase
+{
+    /**
+     * @var CssToInlineStyles
+     */
+    protected $cssToInlineStyles;
+
+    /**
+     * @before
+     */
+    protected function prepare()
+    {
+        $this->cssToInlineStyles = new CssToInlineStyles(true);
+    }
+
+    /**
+     * @after
+     */
+    protected function clear()
+    {
+        $this->cssToInlineStyles = null;
+    }
+
+    public function testBasicHtml()
+    {
+        $html = '<!doctype html><html><head><style>body{color:blue}</style></head><body><p>foo</p></body></html>';
+        $css = 'p { color: red; }';
+        $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;
+
+        $this->assertEquals($expected, $this->cssToInlineStyles->convert($html, $css));
+    }
+
+    public function testSwitchingParser()
+    {
+        // HTML4
+        $html = '<html><head><style>body{color:blue}</style></head><body><p>foo</p></body></html>';
+        $css = 'p { color: red; }';
+        $expected = <<<EOF
+<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">
+<html>
+<head><style>body{color:blue}</style></head>
+<body style="color: blue;"><p style="color: red;">foo</p></body>
+</html>
+EOF;
+
+        $this->assertEquals($expected, $this->cssToInlineStyles->convert($html, $css));
+
+        // HTML5
+        $html = '<!doctype html>' . $html;
+        $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;
+        $this->assertEquals($expected, $this->cssToInlineStyles->convert($html, $css));
+    }
+}