From 256a9682fa26dee67c80807ec7cdc13f38fb1b6e Mon Sep 17 00:00:00 2001
From: Zankaria <zankaria.auxa@mailu.io>
Date: Mon, 17 Mar 2025 12:47:42 +0100
Subject: [PATCH] EmbedService.php: download and handle thumbnails

---
 inc/Services/Embed/EmbedService.php | 140 ++++++++++++++++++++--------
 1 file changed, 103 insertions(+), 37 deletions(-)

diff --git a/inc/Services/Embed/EmbedService.php b/inc/Services/Embed/EmbedService.php
index eb3ba21b..9db84d4a 100644
--- a/inc/Services/Embed/EmbedService.php
+++ b/inc/Services/Embed/EmbedService.php
@@ -1,64 +1,130 @@
 <?php
 namespace Vichan\Services\Embed;
 
-use RuntimeException;
-use Vichan\Context;
+use Vichan\Data\Driver\HttpDriver;
 use Vichan\Data\Driver\LogDriver;
 
-class EmbedService {
-	private array $tuples;
-	private LogDriver $log;
 
-	public function __construct(LogDriver $log) {
+class EmbedService {
+	private const TMP_FILE_PREFIX = 'oembed-thumb-';
+
+	/**
+	 * The input embed text isn't a valid URL.
+	 */
+	public const MATCH_EMBED_ERR_NOT_AN_URL = 0;
+	/**
+	 * The embed url doesn't match any supported url.
+	 */
+	public const MATCH_EMBED_ERR_NO_MATCH = 1;
+	/**
+	 * The embed url doesn't have a thumbnail.
+	 */
+	public const MATCH_EMBED_ERR_NO_THUMBNAIL = 2;
+
+	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, array $embed_entries, int $thumb_download_timeout) {
 		$this->log = $log;
+		$this->oembed_extractor = $oembed_extractor;
+		$this->embed_entries = $embed_entries;
+		$this->thumb_download_timeout = $thumb_download_timeout;
+	}
+
+	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;
 	}
 
 	/**
-	 * Undocumented function
+	 * Downloads the thumbnail into a temporary file.
 	 *
-	 * @param Context $ctx
-	 * @param string $rawText
-	 * @return void
+	 * @return ?string The path to the temporary file, null if the file was too large.
 	 */
-	public function matchEmbed(Context $ctx, string $rawText) {
-		if (\filter_var($rawText, \FILTER_VALIDATE_URL) === false) {
-			return null;
+	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");
 		}
-		$rawText = \trim($rawText);
 
-		foreach ($this->tuples as $cfg) {
-			if (!isset($cfg['match_regex'])) {
-				throw new \RuntimeException('Missing \'match_regex\' field');
-			}
-			$match_regex = $cfg['match_regex'];
+		$ret = $this->http->requestGetInto($thumbnail_url, null, $fd, $this->thumb_download_timeout);
+		return $ret ? $tmp_file : null;
+	}
+
+	/**
+	 * Matches an alleged embed url and returns the path to the thumbnail, if any.
+	 *
+	 * @param string $rawText
+	 * @return ?array Returns the url to the thumbnail and the path to the fallback if any embedding matches, null otherwise.
+	 */
+	private function matchAndExtract(string $rawText) {
+		foreach ($this->embed_entries as $embed_entry) {
+			$match_regex = $embed_entry['match_regex'];
 
 			if (\preg_match($match_regex, $rawText, $matches)) {
-				if (!isset($cfg['type'])) {
-					throw new \RuntimeException('Missing \'type\' field');
-				}
-				$type = $cfg['type'];
+				$type = $embed_entry['type'];
 
 				if ($type === 'oembed') {
-					if (!isset($cfg['provider'])) {
-						throw new \RuntimeException('Missing \'provider\' field');
-					}
-					$provider = $cfg['provider'];
-
-					$extractor = $ctx->get(OembedExtractor::class);
-					$oembed_resp = $extractor->fetch($provider, $rawText);
-
+					$thumbnail_url_fallback = $embed_entry['thumbnail_url_fallback'] ?? null;
+					$provider = $embed_entry['provider'];
+					$oembed_resp = $this->oembed_extractor->fetch($provider, $rawText);
 
+					return [ $oembed_resp->thumbnail_url, $thumbnail_url_fallback ];
 				} elseif ($type === 'regex') {
-					if (!isset($cfg['thumbnail_url'])) {
-						throw new \RuntimeException('Missing \'thumbnail_url\' field');
-					}
-					$thumbnail_url_regex = $cfg['thumbnail_url'];
+					$thumbnail_url_regex = $embed_entry['thumbnail_url'];
 					// Plz somebody review this.
-					$thumbnail_url = \preg_replace($match_regex, $thumbnail_url_regex, $rawText);
+					return [ \preg_replace($match_regex, $thumbnail_url_regex, $rawText), null ];
 				} else {
-					$this->log->log(LogDriver::ERROR, "Unknown embed type '$type', ignoring");
+					$this->log->log(LogDriver::ERROR, "Unknown embed type '$type', ignoring the embed entry");
 				}
 			}
 		}
+
+		return null;
+	}
+
+	/**
+	 * @return array|int Returns the url to the thumbnail and if it should be moved if any embedding matches,
+	 *                   otherwise it returns a MATCH_EMBED_ERR_* constant.
+	 */
+	public function matchEmbed(string $rawText) {
+		$rawText = \trim($rawText);
+		if (\filter_var($rawText, \FILTER_VALIDATE_URL) === false) {
+			return self::MATCH_EMBED_ERR_NOT_AN_URL;
+		}
+
+		$ret = $this->matchAndExtract($rawText);
+		if ($ret === null) {
+			return self::MATCH_EMBED_ERR_NO_MATCH;
+		}
+		list($thumbnail_url, $thumbnail_url_fallback) = $ret;
+		if (!isset($thumbnail_url, $thumbnail_url_fallback)) {
+			return self::MATCH_EMBED_ERR_NO_THUMBNAIL;
+		}
+
+		if (\filter_var($thumbnail_url, \FILTER_VALIDATE_URL) === false) {
+			$this->log->log(LogDriver::ERROR, "Thumbnail URL '$thumbnail_url' 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 self::MATCH_EMBED_ERR_NO_THUMBNAIL;
+		}
+		return [ $thumbnail_url_fallback, false ];
 	}
 }