Experimental code to compres HTML without losing the "structure", for a web scraping with ai use-case to reduce required token-usage.

HtmlCompressor.php

<?php

namespace App;

use Illuminate\Support\Str;
use Symfony\Component\DomCrawler\Crawler;

class HtmlCompressor
{
    protected bool $removeComments = true;

    protected bool $removeEmptyElements = true;

    protected array $elementsToRemove = [
        // Elements
        'script', 'style', 'link', 'head', 'noscript', 'template', 'footer', 'svg', 'br', 'hr',
        // Fontawesome icons
        'i.fa', 'i.fas', 'i.far', 'i.fal',
    ];

    protected array $attributesToRemove = [
        'alt',
        'for',
        'href',
        'onclick',
        'onerror',
        'onsubmit',
        'placeholder',
        'role',
        'src',
        'style',
        'tabindex',
    ];

    public function __construct(
        ?array $elementsToRemove = null,
        ?array $attributesToRemove = null,
        ?bool $removeComments = null,
        ?bool $removeEmptyElements = null,
    ) {
        $this->elementsToRemove = $elementsToRemove ?? $this->elementsToRemove;
        $this->attributesToRemove = $attributesToRemove ?? $this->attributesToRemove;
        $this->removeComments = $removeComments ?? $this->removeComments;
        $this->removeEmptyElements = $removeEmptyElements ?? $this->removeEmptyElements;
    }

    public static function compress(string $html): string
    {
        return (new self())->compressHtml($html);
    }

    public static function textOnly(string $html): string
    {
        $raw = self::compress($html);

        $dom = new Crawler(Str::of($raw)->replace('><', '> <')->toString());

        return Str::of($dom->text())->squish()->trim()->toString();

    }

    /** @noinspection PhpPossiblePolymorphicInvocationInspection */
    public function compressHtml(string $html): string
    {
        $crawler = new Crawler($html);

        // Remove elements we dont need
        foreach ($this->elementsToRemove as $element) {
            $crawler->filter($element)->each(function (Crawler $node) {
                $node->getNode(0)->parentNode->removeChild($node->getNode(0));
            });
        }

        if ($this->removeComments) {
            $crawler->filterXPath('//comment()')->each(function (Crawler $node) {
                $node->getNode(0)->parentNode->removeChild($node->getNode(0));
            });
        }

        if ($this->removeEmptyElements) {
            $crawler->filter('*')->each(function (Crawler $node) {
                if (in_array($node->nodeName(), ['div', 'p']) && $node->text() === '') {
                    $node->getNode(0)->parentNode->removeChild($node->getNode(0));
                }
            });
        }

        // Remove attributes
        $crawler->filter('*')->each(function (Crawler $node) {
            // Need to convert the iterator to an array,
            // because we are mutating the attributes as inside the loop.
            foreach (iterator_to_array($node->getNode(0)->attributes) as $attribute) {
                $name = $attribute->nodeName;

                if (in_array($name, $this->attributesToRemove)) {
                    $node->getNode(0)->removeAttribute($name);
                }

                // Remove aria and data attributes
                if (Str::startsWith($name, ['aria-', 'data-'])) {
                    $node->getNode(0)->removeAttribute($name);
                }
            }
        });

        $filteredHtml = $crawler->filter('body')->html();

        $compressedHtml = Str::of($filteredHtml)
            ->squish()
            ->replace('> <', '><') // Remove whitespace between closing and ending tags "</div> <div>"
            ->replace(' </', '</') // Remove whitespace before next 'whatever </tag>'
            ->replace(' ">', '">')
            ->replace(' " ', '"')
            ->replace('> ', '>')
            ->replace(' <', '<')
            ->remove('<span>×</span>')
            ->trim();

        return $compressedHtml;
    }

    public static function classes(bool|string $html)
    {

        $crawler = new Crawler($html);

        $classes = $crawler->filter('*')->each(function (Crawler $node) {
            return $node->attr('class');
        });

        return collect($classes)->filter()->unique()->values();
    }
}