[PHP] Compress HTML output from TYPO3

Google Pagespeed also complains from time to time: In addition to the recommended GZIP compression, you can also remove unnecessary characters (e.g. spaces, breaks and comments) from the rendered HTML of a TYPO3 page. Although there is already a ViewHelper (f:spaceless) that helps you to compress the code a bit in the Fluid, it usually only takes care of the body area of your website. If you really want to compress everything, the use of PSR-15 middleware is recommended.

PHP package wyrihaximus/html-compress helps with compressing. Just add it to your composer.json:

{
  "name": "vendor/sitepackage",
  "description": "Sitepackage extension",
  "type": "typo3-cms-extension",
  "homepage": "https://www.in2code.de",
  "require": {
    "typo3/cms-core": "^10.4",
    "wyrihaximus/html-compress": "^1"
  },
  "autoload": {
    "psr-4": {
      "Vendor\\Sitepackage\\": "Classes/"
    }
  }
}

EXT:sitepackage/Configuration/RequestMiddlewares.php:

<?php
return [
    'frontend' => [
        'sitepackage-compress-html' => [
            'target' => \Vendor\Sitepackage\Middleware\HtmlCompress::class,
            'before' => [
                'typo3/cms-frontend/output-compression'
            ],
            'after' => [
                'typo3/cms-adminpanel/renderer'
            ]
        ]
    ]
];

EXT:sitepackage/Classes/Middleware/HtmlCompress.php:

<?php
declare(strict_types=1);
namespace Vendor\Sitepackage\Middleware;

use Psr\Http\Message\ResponseInterface;
use Psr\Http\Message\ServerRequestInterface;
use Psr\Http\Server\MiddlewareInterface;
use Psr\Http\Server\RequestHandlerInterface;
use TYPO3\CMS\Core\Http\StreamFactory;
use WyriHaximus\HtmlCompress\Factory;

/**
 * Class HtmlCompress
 */
class HtmlCompress implements MiddlewareInterface
{
    /**
     * @param ServerRequestInterface $request
     * @param RequestHandlerInterface $handler
     * @return ResponseInterface
     */
    public function process(ServerRequestInterface $request, RequestHandlerInterface $handler): ResponseInterface
    {
        $response = $handler->handle($request);
        if ($this->isTypeNumSet($request) === false) {
            $stream = $response->getBody();
            $stream->rewind();
            $content = $stream->getContents();
            $newBody = (new StreamFactory())->createStream($this->compressHtml($content));
            $response = $response->withBody($newBody);
        }
        return $response;
    }

    /**
     * @param string $html
     * @return string
     */
    protected function compressHtml(string $html): string
    {
        $parser = Factory::construct();
        $html = $parser->compress($html);
        $html = $this->removeComments($html);
        return $html;
    }

    /**
     * Remove all html comments but not "<!--TYPO3SEARCH_begin-->" and "<!--TYPO3SEARCH-end-->"
     * @param string $html
     * @return string
     */
    protected function removeComments(string $html): string
    {
        return preg_replace('/<!--((?!TYPO3SEARCH)[\s\S])*?-->/', '', $html);
    }

    /**
     * @param ServerRequestInterface $request
     * @return bool
     */
    protected function isTypeNumSet(ServerRequestInterface $request): bool
    {
        return $request->getAttribute('routing')->getPageType() > 0;
    }
}

Note: If you remove HTML comments, please note that you need the comments "TYPO3SEARCH _..." for Solr or the index search! The code above already takes this into account.