, and tags, instead of the HTML
* that belongs within a tag.
*/
public function __construct( $content, $flags = false, $is_html_document = false ) {
if ( $is_html_document ) {
// The content is a full HTML document. Add the UTF-8 Content-Type meta tag to its
// , instead of wrapping the document in another document, which would result
// in the document's elements being output within its .
$content = $this->add_content_type_meta_tag( $content );
} else {
// Wrap content in , and tags with an UTF-8 Content-Type meta tag.
// Forcibly tell DOMDocument that this HTML uses the UTF-8 charset.
// isn't enough, as DOMDocument still interprets the HTML as ISO-8859, which breaks character encoding
// Use of mb_convert_encoding() with HTML-ENTITIES is deprecated in PHP 8.2, so we have to use this method.
// If we don't, special characters render incorrectly.
$content = '' . $content . '';
}
// Load the HTML into a DOMDocument.
libxml_use_internal_errors( true );
$this->html = new DOMDocument();
if ( $flags ) {
$this->html->loadHTML( $content, $flags );
} else {
$this->html->loadHTML( $content );
}
// Load DOMDocument into XPath.
$this->xpath = new DOMXPath( $this->html );
}
/**
* Adds a UTF-8 Content-Type meta tag to the given HTML document's .
*
* DOMDocument doesn't honor a document's tag, interpreting the
* HTML as ISO-8859 and therefore rendering special characters incorrectly. Use of
* mb_convert_encoding() with HTML-ENTITIES is deprecated in PHP 8.2, so we have to
* use this method.
*
* @since 3.4.1
*
* @param string $content HTML document.
* @return string HTML document, with a UTF-8 Content-Type meta tag.
*/
private function add_content_type_meta_tag( $content ) {
$meta_tag = '';
// Add the meta tag immediately after the opening tag.
$html = preg_replace( '/]*)?>/i', '$0' . $meta_tag, $content, 1, $count );
if ( $count ) {
return (string) $html;
}
// No tag exists; add one immediately after the opening tag.
$html = preg_replace( '/]*)?>/i', '$0' . $meta_tag . '', $content, 1, $count );
if ( $count ) {
return (string) $html;
}
// Neither a or tag exists; prepend the meta tag in its own tag.
return '' . $meta_tag . '' . $content;
}
/**
* Returns the HTML within the DOMDocument's tag as a string.
*
* @since 3.0.0
*
* @return string
*/
public function get_body_html() {
$body = $this->html->getElementsByTagName( 'body' )->item( 0 );
$html = '';
foreach ( $body->childNodes as $child ) { // phpcs:ignore WordPress.NamingConventions.ValidVariableName.UsedPropertyNotSnakeCase
$html .= $this->html->saveHTML( $child );
}
return $html;
}
}