Ibochkarev · July 13, 2026 11:39
diff --git a/html_parser.php b/html_parser.php
 <?php

 defined('MODX_CORE_PATH') || exit;

 /**
 * Pull script/style/noscript/textarea out before DOMDocument so markup inside
 * JS/CSS strings is not rewritten (can inject literal </script> and break pages).
 *
 * @return array{0: string, 1: array<string, string>}
 */
 function imageoptimizer_extract_raw_blocks(string $html): array
 {
    $blocks = [];
    $replaced = preg_replace_callback(
        '#<(script|style|noscript|textarea)\b[^>]*>.*?</\1>#is',
        static function (array $match) use (&$blocks): string {
            $token = '<!--IMAGEOPTIMIZER_RAW_' . count($blocks) . '-->';
            $blocks[$token] = $match[0];

            return $token;
        },
        $html
    );

    return [$replaced ?? $html, $blocks];
 }

 /**
 * @param array<string, string> $blocks
 */
 function imageoptimizer_restore_raw_blocks(string $html, array $blocks): string
 {
    if ($blocks === []) {
        return $html;
    }

    return str_replace(array_keys($blocks), array_values($blocks), $html);
 }

 /**
 * Full page document (has doctype or html root) vs HTML fragment.
 */
 function imageoptimizer_is_full_html_document(string $html): bool
 {
    $sample = ltrim(substr($html, 0, 2048));
    if ($sample === '') {
        return false;
    }

    return (bool) preg_match('/^(?:<!DOCTYPE\b|<html\b)/i', $sample);
 }

 function imageoptimizer_load_html_document(string $html): ?DOMDocument
 {
    $doc = new DOMDocument();
    $flags = LIBXML_NOERROR | LIBXML_NOWARNING;
    if (defined('LIBXML_HTML_NODEFDTD')) {
        $flags |= LIBXML_HTML_NODEFDTD;
    }

    if (imageoptimizer_is_full_html_document($html)) {
        // Keep <html>/<head>/<body>; do not wrap in #imageoptimizer-root.
        $payload = '<?xml encoding="UTF-8">' . $html;
        if (!$doc->loadHTML($payload, $flags)) {
            return null;
        }

        return $doc;
    }

    if (defined('LIBXML_HTML_NOIMPLIED')) {
        $flags |= LIBXML_HTML_NOIMPLIED;
    }

    $wrapped = '<!DOCTYPE html><html><head><meta charset="UTF-8"></head><body><div id="imageoptimizer-root">'
        . $html . '</div></body></html>';
    if (!$doc->loadHTML($wrapped, $flags)) {
        return null;
    }

    return $doc;
 }

 /**
 * @return DOMElement[]
 */
 function imageoptimizer_extract_img_nodes(DOMDocument $doc): array
 {
    $images = [];
    $nodes = $doc->getElementsByTagName('img');
    foreach ($nodes as $node) {
        if ($node instanceof DOMElement) {
            $images[] = $node;
        }
    }

    return $images;
 }

 function imageoptimizer_serialize_document(DOMDocument $doc): string
 {
    $root = $doc->getElementById('imageoptimizer-root');
    if ($root) {
        $html = '';
        foreach ($root->childNodes as $child) {
            $html .= $doc->saveHTML($child);
        }

        return imageoptimizer_normalize_html_output($html);
    }

    $html = (string) $doc->saveHTML();
    // loadHTML UTF-8 hint may serialize as PI or as HTML comment after doctype.
    $html = preg_replace('/<\?xml[^?]*\?>\s*/i', '', $html) ?? $html;
    $html = preg_replace('/<!--\?xml[^>]*-->\s*/i', '', $html) ?? $html;

    return imageoptimizer_normalize_html_output($html);
 }

 function imageoptimizer_normalize_html_output(string $html): string
 {
    $voidTags = ['area', 'base', 'br', 'col', 'embed', 'hr', 'img', 'input', 'link', 'meta', 'param', 'source', 'track', 'wbr'];
    foreach ($voidTags as $tag) {
        $html = preg_replace('#(<' . $tag . '\b[^>]*>)\s*</' . $tag . '>#i', '$1', $html) ?? $html;
    }
    $html = preg_replace('#</source>\s*</picture>#i', '</picture>', $html) ?? $html;

    return $html;
 }
	<?php

	defined('MODX_CORE_PATH') \|\| exit;

	/**
	* Pull script/style/noscript/textarea out before DOMDocument so markup inside
	* JS/CSS strings is not rewritten (can inject literal </script> and break pages).
	*
	* @return array{0: string, 1: array<string, string>}
	*/
	function imageoptimizer_extract_raw_blocks(string $html): array
	{
	$blocks = [];
	$replaced = preg_replace_callback(
	'#<(script\|style\|noscript\|textarea)\b[^>]>.?</\1>#is',
	static function (array $match) use (&$blocks): string {
	$token = '<!--IMAGEOPTIMIZER_RAW_' . count($blocks) . '-->';
	$blocks[$token] = $match[0];

	return $token;
	},
	$html
	);

	return [$replaced ?? $html, $blocks];
	}

	/**
	* @param array<string, string> $blocks
	*/
	function imageoptimizer_restore_raw_blocks(string $html, array $blocks): string
	{
	if ($blocks === []) {
	return $html;
	}

	return str_replace(array_keys($blocks), array_values($blocks), $html);
	}

	/**
	* Full page document (has doctype or html root) vs HTML fragment.
	*/
	function imageoptimizer_is_full_html_document(string $html): bool
	{
	$sample = ltrim(substr($html, 0, 2048));
	if ($sample === '') {
	return false;
	}

	return (bool) preg_match('/^(?:<!DOCTYPE\b\|<html\b)/i', $sample);
	}

	function imageoptimizer_load_html_document(string $html): ?DOMDocument
	{
	$doc = new DOMDocument();
	$flags = LIBXML_NOERROR \| LIBXML_NOWARNING;
	if (defined('LIBXML_HTML_NODEFDTD')) {
	$flags \|= LIBXML_HTML_NODEFDTD;
	}

	if (imageoptimizer_is_full_html_document($html)) {
	// Keep <html>/<head>/<body>; do not wrap in #imageoptimizer-root.
	$payload = '<?xml encoding="UTF-8">' . $html;
	if (!$doc->loadHTML($payload, $flags)) {
	return null;
	}

	return $doc;
	}

	if (defined('LIBXML_HTML_NOIMPLIED')) {
	$flags \|= LIBXML_HTML_NOIMPLIED;
	}

	$wrapped = '<!DOCTYPE html><html><head><meta charset="UTF-8"></head><body><div id="imageoptimizer-root">'
	. $html . '</div></body></html>';
	if (!$doc->loadHTML($wrapped, $flags)) {
	return null;
	}

	return $doc;
	}

	/**
	* @return DOMElement[]
	*/
	function imageoptimizer_extract_img_nodes(DOMDocument $doc): array
	{
	$images = [];
	$nodes = $doc->getElementsByTagName('img');
	foreach ($nodes as $node) {
	if ($node instanceof DOMElement) {
	$images[] = $node;
	}
	}

	return $images;
	}

	function imageoptimizer_serialize_document(DOMDocument $doc): string
	{
	$root = $doc->getElementById('imageoptimizer-root');
	if ($root) {
	$html = '';
	foreach ($root->childNodes as $child) {
	$html .= $doc->saveHTML($child);
	}

	return imageoptimizer_normalize_html_output($html);
	}

	$html = (string) $doc->saveHTML();
	// loadHTML UTF-8 hint may serialize as PI or as HTML comment after doctype.
	$html = preg_replace('/<\?xml[^?]\?>\s/i', '', $html) ?? $html;
	$html = preg_replace('/<!--\?xml[^>]-->\s/i', '', $html) ?? $html;

	return imageoptimizer_normalize_html_output($html);
	}

	function imageoptimizer_normalize_html_output(string $html): string
	{
	$voidTags = ['area', 'base', 'br', 'col', 'embed', 'hr', 'img', 'input', 'link', 'meta', 'param', 'source', 'track', 'wbr'];
	foreach ($voidTags as $tag) {
	$html = preg_replace('#(<' . $tag . '\b[^>]>)\s</' . $tag . '>#i', '$1', $html) ?? $html;
	}
	$html = preg_replace('#</source>\s*</picture>#i', '</picture>', $html) ?? $html;

	return $html;
	}
No results found