Add FIX_DUPLICATE_STYLES option to modify() to remove duplicate <style> tags.

ivopetkov · ivopetkov · commit d26ef81739ee · 2023-02-05T14:42:33.000+02:00
diff --git a/src/HTML5DOMDocument.php b/src/HTML5DOMDocument.php
@@ -53,6 +53,11 @@ class HTML5DOMDocument extends \DOMDocument
      */
     const OPTIMIZE_HEAD = 32;
 
+    /**
+     * A modification (passed to modify()) that removes all but first styles with duplicate content.
+     */
+    const FIX_DUPLICATE_STYLES = 64;
+
     /**
      *
      * @var array
@@ -612,6 +617,7 @@ public function insertHTMLMulti(array $sources)
      *  - HTML5DOMDocument::FIX_MULTIPLE_HEADS - merges multiple head elements.
      *  - HTML5DOMDocument::FIX_MULTIPLE_BODIES - merges multiple body elements.
      *  - HTML5DOMDocument::OPTIMIZE_HEAD - moves charset metatag and title elements first.
+     *  - HTML5DOMDocument::FIX_DUPLICATE_STYLES - removes all but first styles with duplicate content.
      */
     public function modify($modifications = 0)
     {
@@ -621,6 +627,7 @@ public function modify($modifications = 0)
         $fixMultipleHeads = ($modifications & self::FIX_MULTIPLE_HEADS) !== 0;
         $fixMultipleBodies = ($modifications & self::FIX_MULTIPLE_BODIES) !== 0;
         $optimizeHead = ($modifications & self::OPTIMIZE_HEAD) !== 0;
+        $fixDuplicateStyles = ($modifications & self::FIX_DUPLICATE_STYLES) !== 0;
 
         /** @var \DOMNodeList<HTML5DOMElement> */
         $headElements = $this->getElementsByTagName('head');
@@ -696,6 +703,27 @@ public function modify($modifications = 0)
                 }
             }
 
+            if ($fixDuplicateStyles) {
+                $styles = $headElement->getElementsByTagName('style');
+                if ($styles->length > 0) {
+                    $stylesToRemove = [];
+                    $list = [];
+                    foreach ($styles as $style) {
+                        $innerHTML = trim($style->innerHTML);
+                        if (array_search($innerHTML, $list) === false) {
+                            $list[] = $innerHTML;
+                        } else {
+                            $stylesToRemove[] = $style;
+                        }
+                    }
+                    foreach ($stylesToRemove as $styleToRemove) {
+                        $styleToRemove->parentNode->removeChild($styleToRemove);
+                    }
+                    unset($list);
+                }
+                unset($styles);
+            }
+
             if ($optimizeHead) { // Moves charset metatag and title elements first.
                 $titleElement = $headElement->getElementsByTagName('title')->item(0);
                 $hasTitleElement = false;
diff --git a/tests/Test.php b/tests/Test.php
@@ -796,6 +796,25 @@ public function testDuplicateTags()
             . '<meta content="video.movie" property="og:type">'
             . '</head></html>';
         $this->assertEquals($expectedSource, $this->removeNewLines($dom->saveHTML()));
+
+        $dom = new HTML5DOMDocument();
+        $dom->loadHTML('<!DOCTYPE html><html><head>'
+            . '<style>body{color:red;}</style>'
+            . '<style>body{color:red;}</style>'
+            . '<style>div{color:blue;}</style>'
+            . '<style>span{color:green;}</style>'
+            . '<style>body{color:red;}</style>'
+            . '</head></html>');
+        $dom->insertHTML('<head>'
+            . '<style>div{color:blue;}</style>'
+            . '</head>');
+        $dom->modify(HTML5DOMDocument::FIX_DUPLICATE_STYLES);
+        $expectedSource = '<!DOCTYPE html><html><head>'
+            . '<style>body{color:red;}</style>'
+            . '<style>div{color:blue;}</style>'
+            . '<style>span{color:green;}</style>'
+            . '</head></html>';
+        $this->assertEquals($expectedSource, $this->removeNewLines($dom->saveHTML()));
     }
 
     /**
