b2in/app/Services/CmsFluxEditorHtmlTransformer.php

<?php

declare(strict_types=1);

namespace App\Services;

use DOMDocument;
use DOMElement;
use DOMNode;
use DOMXPath;

/**
 * Der Flux-Editor (TipTap) nutzt für Hervorhebungen das Element `mark` (Highlight-Extension).
 * Im Frontend wird derselbe optische Zweck oft mit `span.text-secondary` umgesetzt.
 * Diese Klasse wandelt beim Laden in den Editor und beim Speichern zwischen beiden Formen um.
 */
final class CmsFluxEditorHtmlTransformer
{
    /**
     * JSON-Felder, die im CMS-Modal mit flux:editor (Rich-Text) bearbeitet werden.
     *
     * @var list<string>
     */
    public const RICH_TEXT_JSON_FIELDS = [
        'description',
        'text',
        'content',
        'help',
        'answer',
        'quote',
    ];

    public static function toEditor(string $html): string
    {
        if ($html === '' || ! str_contains($html, 'text-secondary')) {
            return $html;
        }

        return self::spansWithClassToMarks($html);
    }

    public static function fromEditor(string $html): string
    {
        if ($html === '' || ! str_contains($html, '<mark')) {
            return $html;
        }

        return self::marksToSecondarySpans($html);
    }

    /**
     * Der Editor packt jeden Text in Absätze – auch eingefügten Klartext (12.10.2026). Für Felder,
     * die die Website innerhalb einer Zeile oder eines eigenen Absatzes ausgibt, wird daraus
     * Inline-HTML: ein Absatz ohne `<p>`, mehrere durch `<br><br>` getrennt, reiner Text als
     * Klartext. Enthält der Text Blockelemente (Überschriften, Listen …), bleibt er unverändert.
     */
    public static function unwrapParagraphs(string $html): string
    {
        $html = trim((string) preg_replace('#<p[^>]*>(\s|&nbsp;|<br\s*/?>)*</p>#i', '', $html));

        if ($html === '' || ! str_contains($html, '<p')) {
            return $html;
        }

        if (preg_match('#<(?!/?(?:p|strong|b|em|i|mark|span|br|a)\b)[a-z]#i', $html) === 1) {
            return $html;
        }

        $html = (string) preg_replace('#</p>\s*<p[^>]*>#i', '<br><br>', $html);
        $html = trim((string) preg_replace('#^<p[^>]*>|</p>$#i', '', $html));

        return str_contains($html, '<') ? $html : html_entity_decode($html, ENT_QUOTES | ENT_HTML5, 'UTF-8');
    }

    /**
     * @param  array<int, array<string, mixed>|array{_value: string}>  $items
     * @return array<int, array<string, mixed>|array{_value: string}>
     */
    public static function toEditorJsonItems(array $items, bool $isStringArray): array
    {
        if ($isStringArray) {
            return $items;
        }

        return array_map(function ($item) {
            if (! is_array($item)) {
                return $item;
            }
            $out = [];
            foreach ($item as $key => $value) {
                if (in_array($key, self::RICH_TEXT_JSON_FIELDS, true) && is_string($value)) {
                    $out[$key] = self::toEditor($value);
                } else {
                    $out[$key] = $value;
                }
            }

            return $out;
        }, $items);
    }

    /**
     * @param  array<int, array<string, mixed>|array{_value: string}>  $items
     * @return array<int, array<string, mixed>|array{_value: string}>
     */
    public static function fromEditorJsonItems(array $items, bool $isStringArray): array
    {
        if ($isStringArray) {
            return $items;
        }

        return array_map(function ($item) {
            if (! is_array($item)) {
                return $item;
            }
            $out = [];
            foreach ($item as $key => $value) {
                if (in_array($key, self::RICH_TEXT_JSON_FIELDS, true) && is_string($value)) {
                    $out[$key] = self::unwrapParagraphs(self::fromEditor($value));
                } else {
                    $out[$key] = $value;
                }
            }

            return $out;
        }, $items);
    }

    private static function spansWithClassToMarks(string $html): string
    {
        libxml_use_internal_errors(true);
        $dom = new DOMDocument;
        $wrapped = '<?xml encoding="utf-8"?><div id="cms-flux-root">'.$html.'</div>';
        $dom->loadHTML($wrapped, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
        libxml_clear_errors();

        $root = $dom->getElementById('cms-flux-root');
        if (! $root instanceof DOMElement) {
            return $html;
        }

        $xpath = new DOMXPath($dom);
        $expression = '//span[contains(concat(" ", normalize-space(@class), " "), " text-secondary ")]';
        $nodes = [];
        foreach ($xpath->query($expression) ?? [] as $node) {
            $nodes[] = $node;
        }

        usort($nodes, fn ($a, $b) => self::nodeDepth($b) <=> self::nodeDepth($a));

        foreach ($nodes as $span) {
            if (! $span instanceof DOMElement || $span->tagName !== 'span') {
                continue;
            }
            $mark = $dom->createElement('mark');
            while ($span->firstChild) {
                $mark->appendChild($span->firstChild);
            }
            $span->parentNode?->replaceChild($mark, $span);
        }

        return self::extractInnerHtml($dom, $root);
    }

    private static function marksToSecondarySpans(string $html): string
    {
        libxml_use_internal_errors(true);
        $dom = new DOMDocument;
        $wrapped = '<?xml encoding="utf-8"?><div id="cms-flux-root">'.$html.'</div>';
        $dom->loadHTML($wrapped, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
        libxml_clear_errors();

        $root = $dom->getElementById('cms-flux-root');
        if (! $root instanceof DOMElement) {
            return $html;
        }

        $xpath = new DOMXPath($dom);
        $nodes = [];
        foreach ($xpath->query('//mark') ?? [] as $node) {
            $nodes[] = $node;
        }

        usort($nodes, fn ($a, $b) => self::nodeDepth($b) <=> self::nodeDepth($a));

        foreach ($nodes as $mark) {
            if (! $mark instanceof DOMElement) {
                continue;
            }
            $span = $dom->createElement('span');
            $span->setAttribute('class', 'text-secondary');
            while ($mark->firstChild) {
                $span->appendChild($mark->firstChild);
            }
            $mark->parentNode?->replaceChild($span, $mark);
        }

        return self::extractInnerHtml($dom, $root);
    }

    private static function nodeDepth(DOMNode $node): int
    {
        $d = 0;
        while ($node->parentNode) {
            $d++;
            $node = $node->parentNode;
        }

        return $d;
    }

    private static function extractInnerHtml(DOMDocument $dom, DOMElement $root): string
    {
        $html = '';
        foreach ($root->childNodes as $child) {
            $html .= $dom->saveHTML($child);
        }

        return $html;
    }
}