leftypol/inc/Service/Embed/EmbedService.php

<?php
namespace Vichan\Service\Embed;

use Vichan\Service\Embed\OembedExtractor;
use Vichan\Data\Driver\{HttpDriver, LogDriver};


class EmbedService {
	private const TMP_FILE_PREFIX = 'oembed-thumb-';

	private LogDriver $log;
	private OembedExtractor $oembed_extractor;
	private HttpDriver $http;
	private string $tmp_dir;
	private array $embed_entries;
	private int $thumb_download_timeout;


	public function __construct(
		LogDriver $log,
		OembedExtractor $oembed_extractor,
		HttpDriver $http,
		array $embed_entries,
		int $thumb_download_timeout,
		string $tmp_dir,
	) {
		$this->log = $log;
		$this->oembed_extractor = $oembed_extractor;
		$this->http = $http;
		$this->embed_entries = $embed_entries;
		$this->thumb_download_timeout = $thumb_download_timeout;
		$this->tmp_dir = $tmp_dir;
	}

	private function make_tmp_file(): string {
		$ret = \tempnam($this->tmp_dir, self::TMP_FILE_PREFIX);
		if ($ret === false) {
			throw new \RuntimeException("Could not create temporary file in {$this->tmp_dir}");
		}
		\register_shutdown_function(fn() => @unlink($ret));
		return $ret;
	}

	/**
	 * Downloads the thumbnail into a temporary file.
	 *
	 * @return ?string The path to the temporary file, null if the file was too large.
	 */
	private function fetchThumbnail(string $thumbnail_url): ?string {
		$tmp_file = $this->make_tmp_file();
		$fd = \fopen($tmp_file, 'w+b');
		if ($fd === false) {
			throw new \RuntimeException("Could not open temporary file $tmp_file for read/write");
		}

		$ret = $this->http->requestGetInto($thumbnail_url, null, $fd, $this->thumb_download_timeout);
		return $ret ? $tmp_file : null;
	}

	/**
	 * Returns the path to the thumbnail from a matched url, if any.
	 *
	 * @param string $url The url to embed.
	 * @param int $entry_index The index of the embedding entry.
	 * @return ?array Returns the url to the thumbnail and the path to the fallback.
	 */
	private function extractThumb(string $url, int $entry_index) {
		$embed_entry = $this->embed_entries[$entry_index];
		$match_regex = $embed_entry['match_regex'];
		$type = $embed_entry['type'];

		if ($type === 'oembed') {
			$thumbnail_url_fallback = $embed_entry['thumbnail_url_fallback'] ?? null;
			$provider = $embed_entry['provider_url'];
			$oembed_resp = $this->oembed_extractor->fetch($provider, $url);

			return [ $oembed_resp->thumbnail_url, $thumbnail_url_fallback ];
		} elseif ($type === 'regex') {
			$thumbnail_url_regex = $embed_entry['thumbnail_url'];
			// Plz somebody review this.
			return [ \preg_replace($match_regex, $thumbnail_url_regex, $url), null ];
		} else {
			$this->log->log(LogDriver::ERROR, "Unknown embed type '$type' in embed entry $entry_index, ignoring the entry");
			return [ null, null ];
		}
	}

	/**
	 * Find the embed entry matching with the url, if any.
	 *
	 * @param string $url Url to embed. MUST BE ALREADY VALIDATED.
	 * @return int The index of the matched embed entry or null.
	 */
	public function matchEmbed(string $url): ?int {
		for ($i = 0; $i < \count($this->embed_entries); $i++) {
			$match_regex = $this->embed_entries[$i]['match_regex'];
			if (\preg_match($match_regex, $url, $matches)) {
				return $i;
			}
		}

		return null;
	}

	/**
	 * Get the embed's thumbnail if possible. May download it from the network into a temporary file, or use a static file.
	 *
	 * @param string $url Url to embed. MUST BE ALREADY VALIDATED.
	 * @param int The index of the matched embed entry.
	 * @return ?array Null if no thumbnail can be selected, otherwise an array with the local file path to the thumbnail
	 *                and if the the file is a temporary or a static one.
	 */
	public function getEmbedThumb(string $url, int $entry_index): ?array {
		$ret = $this->extractThumb($url, $entry_index);
		list($thumbnail_url, $thumbnail_url_fallback) = $ret;
		if (!isset($thumbnail_url, $thumbnail_url_fallback)) {
			return null;
		}

		if (\filter_var($thumbnail_url, \FILTER_VALIDATE_URL) === false) {
			$this->log->log(LogDriver::ERROR, "Thumbnail URL '$thumbnail_url' from embed entry $entry_index is not a valid URL, trying fallback");
		} else {
			$tmp_file = $this->fetchThumbnail($thumbnail_url);
			if ($tmp_file !== null) {
				return [ $tmp_file, true ];
			}
			$this->log->log(LogDriver::NOTICE, "Thumbnail at '$thumbnail_url' was too large, trying fallback");
		}

		if ($thumbnail_url_fallback === null) {
			return null;
		}
		return [ $thumbnail_url_fallback, false ];
	}

	public function renderEmbed(string $url, int $entry_index, string $thumbnail_path): string {
		$embed_entry = $this->embed_entries[$entry_index];
		$match_regex = $embed_entry['match_regex'];
		$html = $embed_entry['html'];

		$ret = \preg_replace($match_regex, $html, $url);
		if (!\is_string($ret)) {
			throw new \RuntimeException("Error while applying regex replacement for embed entry $entry_index");
		}

		\str_replace('%%embed_url%%', $url, $ret);
		\str_replace('%%thumbnail_path%%', $thumbnail_path, $ret);
		return $ret;
	}
}