with page-builder / block-editor classes ─────
if ( 'div' === $tag && $node->hasAttributes() ) {
$class = $node->getAttribute( 'class' );
if ( '' !== $class && $this->class_triggers_bailout( $class ) ) {
return array(
array(
'type' => 'BlockHtml',
'html' => $node->ownerDocument->saveHTML( $node ),
),
);
}
}
// ── Paragraphs & Headings (–
,
–) → BlockText ─────────
if ( 'p' === $tag || (bool) preg_match( '/^h[1-6]$/', $tag ) ) {
$inner_html = $this->get_inner_html( $node );
$text = trim( $node->textContent );
if ( '' === $text ) {
return array();
}
$style = ( 'p' === $tag ) ? 'paragraph' : $tag;
return array(
array(
'type' => 'BlockText',
'html_content' => trim( $inner_html ),
'text' => $text,
'style' => $style,
),
);
}
// ── Images (![]()
) → BlockImage ────────────────────────────────
if ( 'img' === $tag ) {
return $this->parse_img_element( $node );
}
// ── Lists (, ) → BlockList ─────────────────────────────
if ( 'ul' === $tag || 'ol' === $tag ) {
return $this->parse_list_element( $node );
}
// ── Blockquotes () → BlockQuote ────────────────────
if ( 'blockquote' === $tag ) {
return $this->parse_blockquote_element( $node );
}
// ── Default: recurse into children (div, section, figure, …) ───
$blocks = array();
foreach ( $node->childNodes as $child ) {
$blocks = array_merge( $blocks, $this->walk_node( $child ) );
}
return $blocks;
}
// ─── Element-Specific Parsers ───────────────────────────────────────
/**
* @param DOMElement $node An ![]()
element.
* @return array Zero or one BlockImage objects.
*/
private function parse_img_element( DOMElement $node ): array {
if ( ! $node->hasAttributes() ) {
return array();
}
$src = $node->getAttribute( 'src' );
if ( '' === $src ) {
return array();
}
$width = $node->hasAttribute( 'width' ) ? (int) $node->getAttribute( 'width' ) : null;
$height = $node->hasAttribute( 'height' ) ? (int) $node->getAttribute( 'height' ) : null;
$alt = $node->hasAttribute( 'alt' ) ? $node->getAttribute( 'alt' ) : '';
return array(
array(
'type' => 'BlockImage',
'url' => $src,
'width' => $width,
'height' => $height,
'alt' => $alt,
),
);
}
/**
* @param DOMElement $node A or element.
* @return array Zero or one BlockList objects.
*/
private function parse_list_element( DOMElement $node ): array {
$items = array();
foreach ( $node->childNodes as $child ) {
if ( XML_ELEMENT_NODE !== $child->nodeType ) {
continue;
}
if ( 'li' !== strtolower( $child->nodeName ) ) {
continue;
}
$inner = trim( $this->get_inner_html( $child ) );
if ( '' !== $inner ) {
$items[] = $inner;
}
}
if ( empty( $items ) ) {
return array();
}
return array(
array(
'type' => 'BlockList',
'items' => $items,
),
);
}
/**
* @param DOMElement $node A element.
* @return array Zero or one BlockQuote objects.
*/
private function parse_blockquote_element( DOMElement $node ): array {
$author = '';
$cites = $node->getElementsByTagName( 'cite' );
if ( $cites->length > 0 ) {
$cite_node = $cites->item( 0 );
$author = trim( $cite_node->textContent );
}
$full_text = trim( $node->textContent );
if ( '' !== $author ) {
$full_text = trim( str_replace( $author, '', $full_text ) );
}
if ( '' === $full_text ) {
return array();
}
return array(
array(
'type' => 'BlockQuote',
'text' => $full_text,
'author' => $author,
),
);
}
// ─── Helpers ────────────────────────────────────────────────────────
/**
* Returns true if a class string contains any bailout prefix.
*
* @param string $class_attr The element's class attribute value.
* @return bool
*/
private function class_triggers_bailout( string $class_attr ): bool {
foreach ( self::BAILOUT_CLASS_PREFIXES as $prefix ) {
if ( false !== strpos( $class_attr, $prefix ) ) {
return true;
}
}
return false;
}
/**
* Serialises all child nodes of an element back to an HTML string.
*
* This preserves inline formatting (, , , , )
* so the native app's rich-text renderer can interpret it.
*
* @param DOMNode $node Parent node whose children we want as HTML.
* @return string The concatenated inner HTML.
*/
private function get_inner_html( DOMNode $node ): string {
if ( ! $node->hasChildNodes() ) {
return '';
}
$html = '';
$doc = $node->ownerDocument;
foreach ( $node->childNodes as $child ) {
$html .= $doc->saveHTML( $child );
}
return $html;
}
}
- ,
- ) → BlockList ─────────────────────────────
if ( 'ul' === $tag || 'ol' === $tag ) {
return $this->parse_list_element( $node );
}
// ── Blockquotes (
) → BlockQuote ──────────────────── if ( 'blockquote' === $tag ) { return $this->parse_blockquote_element( $node ); } // ── Default: recurse into children (div, section, figure, …) ─── $blocks = array(); foreach ( $node->childNodes as $child ) { $blocks = array_merge( $blocks, $this->walk_node( $child ) ); } return $blocks; } // ─── Element-Specific Parsers ─────────────────────────────────────── /** * @param DOMElement $node Anelement. * @return array Zero or one BlockImage objects. */ private function parse_img_element( DOMElement $node ): array { if ( ! $node->hasAttributes() ) { return array(); } $src = $node->getAttribute( 'src' ); if ( '' === $src ) { return array(); } $width = $node->hasAttribute( 'width' ) ? (int) $node->getAttribute( 'width' ) : null; $height = $node->hasAttribute( 'height' ) ? (int) $node->getAttribute( 'height' ) : null; $alt = $node->hasAttribute( 'alt' ) ? $node->getAttribute( 'alt' ) : ''; return array( array( 'type' => 'BlockImage', 'url' => $src, 'width' => $width, 'height' => $height, 'alt' => $alt, ), ); } /** * @param DOMElement $node A
or
element. * @return array Zero or one BlockList objects. */ private function parse_list_element( DOMElement $node ): array { $items = array(); foreach ( $node->childNodes as $child ) { if ( XML_ELEMENT_NODE !== $child->nodeType ) { continue; } if ( 'li' !== strtolower( $child->nodeName ) ) { continue; } $inner = trim( $this->get_inner_html( $child ) ); if ( '' !== $inner ) { $items[] = $inner; } } if ( empty( $items ) ) { return array(); } return array( array( 'type' => 'BlockList', 'items' => $items, ), ); } /** * @param DOMElement $node A
element. * @return array Zero or one BlockQuote objects. */ private function parse_blockquote_element( DOMElement $node ): array { $author = ''; $cites = $node->getElementsByTagName( 'cite' ); if ( $cites->length > 0 ) { $cite_node = $cites->item( 0 ); $author = trim( $cite_node->textContent ); } $full_text = trim( $node->textContent ); if ( '' !== $author ) { $full_text = trim( str_replace( $author, '', $full_text ) ); } if ( '' === $full_text ) { return array(); } return array( array( 'type' => 'BlockQuote', 'text' => $full_text, 'author' => $author, ), ); } // ─── Helpers ──────────────────────────────────────────────────────── /** * Returns true if a class string contains any bailout prefix. * * @param string $class_attr The element's class attribute value. * @return bool */ private function class_triggers_bailout( string $class_attr ): bool { foreach ( self::BAILOUT_CLASS_PREFIXES as $prefix ) { if ( false !== strpos( $class_attr, $prefix ) ) { return true; } } return false; } /** * Serialises all child nodes of an element back to an HTML string. * * This preserves inline formatting (, , , , ) * so the native app's rich-text renderer can interpret it. * * @param DOMNode $node Parent node whose children we want as HTML. * @return string The concatenated inner HTML. */ private function get_inner_html( DOMNode $node ): string { if ( ! $node->hasChildNodes() ) { return ''; } $html = ''; $doc = $node->ownerDocument; foreach ( $node->childNodes as $child ) { $html .= $doc->saveHTML( $child ); } return $html; } }