Kirby bookmark

BookmarkPage.php

<?php

use Kirby\Http\Remote;
use Kirby\Cms\File;
use Kirby\Cms\Page;
use Kirby\Filesystem\F;
use Kirby\Filesystem\Mime;
use Kirby\Toolkit\Dom;
use Kirby\Toolkit\Str;
use Kirby\Toolkit\V;

/**
 * @method static Field note()
 * @method static Field description()
 * @method static Field dateCreated()
 * @method static Field dateFetched()
 * @method static Field lang()
 * @method static Field meta()
 * @method static Field text()
 * @method static Field html()
 */
class BookmarkPage extends Page
{
	public const jsonFlags = JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE;

	public const allowedTags = ['button', 'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'br', 'hr', 'ul', 'ol', 'li', 'p', 'a', 'pre', 'code', 'blockquote', 'video', 'img', 'picture', 'source'];

	public const allowedAttrs = ['src', 'href', 'alt', 'srcset', 'id'];

	public function storeRemoteData() {
		try {
			if ($this->content()->url()->isEmpty()) {
				throw new Exception('No url given');
			}

			$url = $this->content()->url()->toString();

			$remote = new Remote($url, [
				'agent' => option('remote.userAgent', 'Personal Bookmark Tool'),
			]);
			$remote = $remote->fetch();

			$html = $remote->content();
			$dom = $this->getDom($html);

			$canonicalUrl = $this->getCanonicalUrl($dom);
			$metaTags = $this->getMetaTags($dom);

			if ($metaTags['og:image'] ?? false) {
				$this->saveOgImage($metaTags['og:image'], $metaTags['og:image:alt'] ?? null);
			}

			$newContent = [
				'url' => $canonicalUrl ?? $url,
				'title' => $metaTags['og:title'] ?? $this->getTitle($dom),
				'description' => $metaTags['og:description'] ?? $this->getDescription($dom),
				'lang' => $this->getLang($dom),
				'meta' => json_encode($metaTags, self::jsonFlags),
				'html' => $html,
				'text' => $this->getText($dom),
				'dateFetched' => date('c'),
			];

			$updatedPage = $this->update($newContent);
			$updatedPage = $updatedPage->changeStatusToListed();

			if ($canonicalUrl !== null && $canonicalUrl !== $url) {
				$updatedPage->changeSlug(Str::slug($canonicalUrl));
			}
		} catch (Exception $exception) {
			// Delete page if anything went wrong
			$this->delete();
			throw $exception;
		}
	}

	public function getDom($html): Dom
	{
		$dom = new Dom($html);
		return $dom;
	}

	public function getCanonicalUrl($dom): ?string
	{
		/** @var ?DOMElement $element */
		$element = $dom->query('//link[@rel="canonical" and @href]')->item(0);
		return $element?->getAttribute('href');
	}


	public function getMetaTags(Dom $dom): array
	{
		$tags = $dom->query('//meta[@property and @content]');
		$tagsArray = [];
		foreach ($tags as $tag) {
			/** @var DOMElement $tag */
			$tagsArray[$tag->getAttribute('property')] = $tag->getAttribute('content');
		}
		return $tagsArray;
	}

	public function getLang(Dom $dom): ?string
	{
		/** @var ?DOMElement $element */
		$element = $dom->query('//html[@lang]')->item(0);
		return $element?->getAttribute('lang');
	}

	public function getTitle(Dom $dom): ?string
	{
		/** @var ?DOMElement $element */
		$element = $dom->query('//title')->item(0);
		return $element?->textContent;
	}

	public function getDescription(Dom $dom): ?string
	{
		/** @var ?DOMElement $element */
		$element = $dom->query('//meta[@property="description" and @content]')->item(0);
		return $element?->textContent;
	}

	/**
	 * Returns text of article, main or body
	 * element with very few semantic elements
	 * like, h1, h2, a, img, etc.
	 */
	public function getText(Dom $dom): ?string
	{
		$dom->sanitize([
			'allowedAttrs' => self::allowedAttrs,
		]);

		$tags = $dom->query('//*[self::article or self::main or self::body]');
		$main = $tags->item($tags->count() - 1);

		if ($main) {
			$htmlString = strip_tags($main->C14N(), self::allowedTags);
			$htmlString = preg_replace('/^\s+|\s+$/m', '', $htmlString);
			$htmlString = preg_replace('/\n{2,}/', "\n", $htmlString);
			return $htmlString;
		}
		return null;
	}

	public function saveOgImage(string $url, ?string $alt = null)
	{
		$file = null;

		if (V::url($url)) {
			$remote = new Remote($url);
			$remote = $remote->fetch();

			try {
				$mimeType = $this->getMimeTypeFromHeaders($remote->headers());
				$extension = Mime::toExtension($mimeType);
				if (F::extensionToType($extension) === 'image') {
					$filename = 'og-image.' . $extension;
					$file = $this->addFile($remote->content(), $filename, 'og-image', [
						'url' => $url,
						'alt' => $alt,
					]);
				}
			} catch(Exception $exception) {
				throw $exception;
			}
		}

		return $file;
	}

	public function getMimeTypeFromHeaders($headers): string
	{
		$contentType = $headers['Content-Type'] ?? $headers['content-type'] ?? null;
		$mimeType = preg_replace('/(\w+\/\w+).*/i', '${1}', $contentType);
		return $mimeType;
	}

	public function addFile(
		mixed $remoteContent,
		string $filename,
		string $template,
		array $content = [],
	): File
	{
		$tempFilePath = sys_get_temp_dir() . '/' . $filename;
		$tempFile = new Kirby\Filesystem\File([
			'root' => $tempFilePath,
		]);
		$tempFile->write($remoteContent);

		$file = $this->createFile([
			'parent' => $this,
			'filename' => $filename,
			'source' => $tempFilePath,
			'template' => $template,
			'content' => array_merge([
				'dateFetched' => date('c'),
			], $content),
		]);

		return $file;
	}

	public function ogImage(): ?File
	{
		$ogImage = $this->files()->template('og-image')->first();
		if ($ogImage) {
			return $ogImage;
		}
		return null;
	}

	public function tags(): array
	{
		$tags = $this->content()->tags()->split(',');
		sort($tags, SORT_NATURAL | SORT_FLAG_CASE);
		return $tags;
	}
}

hooks.php

<?php
return [
	'page.create:after' => function (Kirby\Cms\Page $page) {
		if ($page->intendedTemplate()->name() === 'bookmark') {
			/** @var BookmarkPage $page */
			$page = $page->update([
				'dateCreated' => date('c'),
			]);
			$page->storeRemoteData();
		}
	},
];

kirby-bookmarks.mp4