From d6c97dbde23c3aa71b8501888a8f8ffa3f5d5b86 Mon Sep 17 00:00:00 2001
From: Michael <heluecht@pirati.ca>
Date: Wed, 8 Dec 2021 13:32:20 +0000
Subject: [PATCH] Fix attachment handling

---
 src/Protocol/ActivityPub/Transmitter.php | 48 +-----------------------
 src/Protocol/Diaspora.php                |  2 +-
 src/Protocol/Feed.php                    | 11 +++++-
 src/Util/ParseUrl.php                    | 18 ++++++++-
 4 files changed, 29 insertions(+), 50 deletions(-)

diff --git a/src/Protocol/ActivityPub/Transmitter.php b/src/Protocol/ActivityPub/Transmitter.php
index 9b5d052138..74c1893d9d 100644
--- a/src/Protocol/ActivityPub/Transmitter.php
+++ b/src/Protocol/ActivityPub/Transmitter.php
@@ -1293,7 +1293,7 @@ class Transmitter
 
 		$urls = [];
 		foreach ($uriids as $uriid) {
-			foreach (Post\Media::getByURIId($uriid, [Post\Media::DOCUMENT, Post\Media::TORRENT]) as $attachment) {
+			foreach (Post\Media::getByURIId($uriid, [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO, Post\Media::DOCUMENT, Post\Media::TORRENT]) as $attachment) {
 				if (in_array($attachment['url'], $urls)) {
 					continue;
 				}
@@ -1320,52 +1320,6 @@ class Transmitter
 			}
 		}
 
-		if ($type != 'Note') {
-			return $attachments;
-		}
-
-		foreach ($uriids as $uriid) {
-			foreach (Post\Media::getByURIId($uriid, [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO]) as $attachment) {
-				if (in_array($attachment['url'], $urls)) {
-					continue;
-				}
-				$urls[] = $attachment['url'];
-
-				$attach = ['type' => 'Document',
-					'mediaType' => $attachment['mimetype'],
-					'url' => $attachment['url'],
-					'name' => $attachment['description']];
-
-				if (!empty($attachment['height'])) {
-					$attach['height'] = $attachment['height'];
-				}
-
-				if (!empty($attachment['width'])) {
-					$attach['width'] = $attachment['width'];
-				}
-
-				if (!empty($attachment['preview'])) {
-					$attach['image'] = $attachment['preview'];
-				}
-
-				$attachments[] = $attach;
-			}
-			// Currently deactivated, since it creates side effects on Mastodon and Pleroma.
-			// It will be activated, once this cleared.
-			/*
-			foreach (Post\Media::getByURIId($uriid, [Post\Media::HTML]) as $attachment) {
-				if (in_array($attachment['url'], $urls)) {
-					continue;
-				}
-				$urls[] = $attachment['url'];
-
-				$attachments[] = ['type' => 'Page',
-					'mediaType' => $attachment['mimetype'],
-					'url' => $attachment['url'],
-					'name' => $attachment['description']];
-			}*/
-		}
-
 		return $attachments;
 	}
 
diff --git a/src/Protocol/Diaspora.php b/src/Protocol/Diaspora.php
index 57f3502cc9..5caa6c7a15 100644
--- a/src/Protocol/Diaspora.php
+++ b/src/Protocol/Diaspora.php
@@ -3411,7 +3411,7 @@ class Diaspora
 
 			$attachments = Post\Media::getByURIId($item['uri-id'], [Post\Media::DOCUMENT, Post\Media::TORRENT, Post\Media::UNKNOWN]);
 			if (!empty($attachments)) {
-				$body .= "\n".DI::l10n()->t("Attachments:")."\n";
+				$body .= "\n[hr]\n";
 				foreach ($attachments as $attachment) {
 					$body .= "[" . $attachment['description'] . "](" . $attachment['url'] . ")\n";
 				}
diff --git a/src/Protocol/Feed.php b/src/Protocol/Feed.php
index cc65efc143..8cb08f65df 100644
--- a/src/Protocol/Feed.php
+++ b/src/Protocol/Feed.php
@@ -401,7 +401,16 @@ class Feed
 				}
 
 				if (!empty($href)) {
-					$attachments[] = ['type' => Post\Media::DOCUMENT, 'url' => $href, 'mimetype' => $type, 'size' => $length];
+					$attachment = ['type' => Post\Media::UNKNOWN, 'url' => $href, 'mimetype' => $type, 'size' => $length];
+
+					$attachment = Post\Media::fetchAdditionalData($attachment);
+
+					// By now we separate the visible media types (audio, video, image) from the rest
+					// In the future we should try to avoid the DOCUMENT type and only use the real one - but not in the RC phase.
+					if (!in_array($attachment['type'], [Post\Media::AUDIO, Post\Media::IMAGE, Post\Media::VIDEO])) {
+						$attachment['type'] = Post\Media::DOCUMENT;
+					}
+					$attachments[] = $attachment;
 				}
 			}
 
diff --git a/src/Util/ParseUrl.php b/src/Util/ParseUrl.php
index 03cfd7d761..d65ec0bb55 100644
--- a/src/Util/ParseUrl.php
+++ b/src/Util/ParseUrl.php
@@ -451,6 +451,12 @@ class ParseUrl
 					case 'og:site_name':
 						$siteinfo['publisher_name'] = trim($meta_tag['content']);
 						break;
+					case 'og:locale':
+						$siteinfo['language'] = trim($meta_tag['content']);
+						break;
+					case 'og:type':
+						$siteinfo['type'] = trim($meta_tag['content']);
+						break;
 					case 'twitter:description':
 						$siteinfo['text'] = trim($meta_tag['content']);
 						break;
@@ -521,7 +527,7 @@ class ParseUrl
 	 *
 	 * @param string $page_url
 	 * @param array $siteinfo
-	 * @return void
+	 * @return array
 	 */
 	private static function checkMedia(string $page_url, array $siteinfo) : array
 	{
@@ -967,6 +973,16 @@ class ParseUrl
 			}
 		}
 
+		$content = JsonLD::fetchElement($jsonld, 'datePublished');
+		if (!empty($content) && is_string($content)) {
+			$jsonldinfo['published'] = DateTimeFormat::utc($content);
+		}
+
+		$content = JsonLD::fetchElement($jsonld, 'dateModified');
+		if (!empty($content) && is_string($content)) {
+			$jsonldinfo['modified'] = DateTimeFormat::utc($content);
+		}
+
 		$jsonldinfo = self::parseJsonLdAuthor($jsonldinfo, $jsonld);
 
 		Logger::info('Fetched article information', ['url' => $siteinfo['url'], 'fetched' => $jsonldinfo]);