diff --git a/src/Dom/Document.php b/src/Dom/Document.php
index 5bad55dddb0..f798bf5cb2d 100644
--- a/src/Dom/Document.php
+++ b/src/Dom/Document.php
@@ -340,6 +340,8 @@ public function loadHTML( $source, $options = 0 ) {
libxml_use_internal_errors( $libxml_previous_state );
if ( $success ) {
+ $this->normalize_html_attributes();
+
// Remove http-equiv charset again.
$meta = $this->head->firstChild;
if (
@@ -1021,6 +1023,33 @@ private function restore_doctype_node( $html ) {
return preg_replace( self::HTML_RESTORE_DOCTYPE_PATTERN, '\1!\3\4>', $html, 1 );
}
+ /**
+ * Normalizes HTML attributes to be HTML5 compatible.
+ *
+ * Conditionally removes html[xmlns], and converts html[xml:lang] to html[lang].
+ */
+ private function normalize_html_attributes() {
+ $html = $this->documentElement;
+ if ( ! $html->hasAttributes() ) {
+ return;
+ }
+
+ $xmlns = $html->attributes->getNamedItem( 'xmlns' );
+ if ( $xmlns && 'http://www.w3.org/1999/xhtml' === $xmlns->nodeValue ) {
+ $html->removeAttributeNode( $xmlns );
+ }
+
+ $xml_lang = $html->attributes->getNamedItem( 'xml:lang' );
+ if ( $xml_lang ) {
+ $lang_node = $html->attributes->getNamedItem( 'lang' );
+ if ( ( ! $lang_node || ! $lang_node->nodeValue ) && $xml_lang->nodeValue ) {
+ // Move the html[xml:lang] value to html[lang].
+ $html->setAttribute( 'lang', $xml_lang->nodeValue );
+ }
+ $html->removeAttributeNode( $xml_lang );
+ }
+ }
+
/**
* Deduplicate a given tag.
*
diff --git a/tests/php/test-class-amp-dom-document.php b/tests/php/test-class-amp-dom-document.php
index 199dce78f84..a067df34f31 100644
--- a/tests/php/test-class-amp-dom-document.php
+++ b/tests/php/test-class-amp-dom-document.php
@@ -88,6 +88,31 @@ public function data_dom_document() {
'' . $head . 'Text
',
'' . $head . 'Text
',
],
+ 'html_with_xmlns_and_xml_lang' => [
+ 'utf-8',
+ '' . $head . '',
+ '' . $head . '',
+ ],
+ 'html_with_xmlns_value_that_should_remain' => [
+ 'utf-8',
+ '' . $head . '',
+ '' . $head . '',
+ ],
+ 'html_with_lang_and_xml_lang' => [
+ 'utf-8',
+ '' . $head . '',
+ '' . $head . '',
+ ],
+ 'html_with_empty_xml_lang' => [
+ 'utf-8',
+ '' . $head . '',
+ '' . $head . '',
+ ],
+ 'html_with_empty_lang' => [
+ 'utf-8',
+ '' . $head . '',
+ '' . $head . '',
+ ],
'slashes_on_closing_tags' => [
'utf-8',
'Text
',