The post-media table now works with the "attach" field as well

mod/item.php

@@ -35,7 +35,6 @@ use Friendica\Content\Text\BBCode;
 use Friendica\Core\Hook;
 use Friendica\Core\Logger;
 use Friendica\Core\Protocol;
-use Friendica\Core\Renderer;
 use Friendica\Core\Session;
 use Friendica\Core\System;
 use Friendica\Core\Worker;
@@ -48,6 +47,7 @@ use Friendica\Model\FileTag;
 use Friendica\Model\Item;
 use Friendica\Model\Notify\Type;
 use Friendica\Model\Photo;
+use Friendica\Model\Post;
 use Friendica\Model\Tag;
 use Friendica\Network\HTTPException;
 use Friendica\Object\EMail\ItemCCEMail;
@@ -55,7 +55,6 @@ use Friendica\Protocol\Activity;
 use Friendica\Protocol\Diaspora;
 use Friendica\Util\DateTimeFormat;
 use Friendica\Security\Security;
-use Friendica\Util\Strings;
 use Friendica\Worker\Delivery;
 
 function item_post(App $a) {
@@ -532,9 +531,8 @@ function item_post(App $a) {
 				if (strlen($attachments)) {
 					$attachments .= ',';
 				}
-				$attachments .= '[attach]href="' . DI::baseUrl() . '/attach/' . $attachment['id'] .
+				$attachments .= Post\Media::getAttachElement(DI::baseUrl() . '/attach/' . $attachment['id'],
-						'" length="' . $attachment['filesize'] . '" type="' . $attachment['filetype'] .
+					$attachment['filesize'], $attachment['filetype'], $attachment['filename'] ?? '');
-						'" title="' . ($attachment['filename'] ? $attachment['filename'] : '') . '"[/attach]';
 			}
 			$body = str_replace($match[1],'',$body);
 		}

src/Model/Item.php

@@ -968,6 +968,14 @@ class Item
 
 		while ($item = DBA::fetch($items)) {
 			if (empty($content_fields['verb']) || !in_array($content_fields['verb'], self::ACTIVITIES)) {
+				if (!empty($content_fields['body'])) {
+					$content_fields['raw-body'] = trim($content_fields['raw-body'] ?? $content_fields['body']);
+		
+					// Remove all media attachments from the body and store them in the post-media table
+					$content_fields['raw-body'] = Post\Media::insertFromBody($item['uri-id'], $content_fields['raw-body']);
+					$content_fields['raw-body'] = self::setHashtags($content_fields['raw-body']);
+				}
+		
 				self::updateContent($content_fields, ['uri-id' => $item['uri-id']]);
 
 				if (empty($item['icid'])) {
@@ -994,6 +1002,10 @@ class Item
 				}
 			}
 
+			if (!empty($fields['attach'])) {
+				Post\Media::insertFromAttachment($item['uri-id'], $fields['attach']);
+			}
+
 			Post\DeliveryData::update($item['uri-id'], $delivery_data);
 
 			self::updateThread($item['id']);
@@ -1826,6 +1838,10 @@ class Item
 		// Check for hashtags in the body and repair or add hashtag links
 		$item['body'] = self::setHashtags($item['body']);
 
+		if (!empty($item['attach'])) {
+			Post\Media::insertFromAttachment($item['uri-id'], $item['attach']);
+		}
+
 		// Fill the cache field
 		self::putInCache($item);
 

src/Model/Post/Media.php

@@ -24,6 +24,7 @@ namespace Friendica\Model\Post;
 use Friendica\Core\Logger;
 use Friendica\Core\System;
 use Friendica\Database\DBA;
+use Friendica\DI;
 use Friendica\Util\Images;
 
 /**
@@ -34,11 +35,12 @@ use Friendica\Util\Images;
  */
 class Media
 {
-	const UNKNOWN = 0;
+	const UNKNOWN  = 0;
-	const IMAGE   = 1;
+	const IMAGE    = 1;
-	const VIDEO   = 2;
+	const VIDEO    = 2;
-	const AUDIO   = 3;
+	const AUDIO    = 3;
-	const TORRENT = 16;
+	const TORRENT  = 16;
+	const DOCUMENT = 128;
 
 	/**
 	 * Insert a post-media record
@@ -46,25 +48,90 @@ class Media
 	 * @param array $media
 	 * @return void
 	 */
-	public static function insert(array $media)
+	public static function insert(array $media, bool $force = false)
 	{
-		if (empty($media['url']) || empty($media['uri-id'])) {
+		if (empty($media['url']) || empty($media['uri-id']) || empty($media['type'])) {
+			Logger::warning('Incomplete media data', ['media' => $media]);
 			return;
 		}
 
-		if (DBA::exists('post-media', ['uri-id' => $media['uri-id'], 'url' => $media['url']])) {
+		// "document" has got the lowest priority. So when the same file is both attached as document
+		// and embedded as picture then we only store the picture or replace the document 
+		$found = DBA::selectFirst('post-media', ['type'], ['uri-id' => $media['uri-id'], 'url' => $media['url']]);
+		if (!$force && !empty($found) && (($found['type'] != self::DOCUMENT) || ($media['type'] == self::DOCUMENT))) {
 			Logger::info('Media already exists', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'callstack' => System::callstack()]);
 			return;
 		}
 
-		$fields = ['type', 'mimetype', 'height', 'width', 'size', 'preview', 'preview-height', 'preview-width', 'description'];
+		$fields = ['mimetype', 'height', 'width', 'size', 'preview', 'preview-height', 'preview-width', 'description'];
 		foreach ($fields as $field) {
 			if (empty($media[$field])) {
 				unset($media[$field]);
 			}
 		}
 
-		if ($media['type'] == self::IMAGE) {
+		// We are storing as fast as possible to avoid duplicated network requests
+		// when fetching additional information for pictures and other content.
+		$result = DBA::insert('post-media', $media, true);
+		Logger::info('Stored media', ['result' => $result, 'media' => $media, 'callstack' => System::callstack()]);
+		$stored = $media;
+
+		$media = self::fetchAdditionalData($media);
+
+		if (array_diff_assoc($media, $stored)) {
+			$result = DBA::insert('post-media', $media, true);
+			Logger::info('Updated media', ['result' => $result, 'media' => $media]);
+		} else {
+			Logger::info('Norhing to update', ['media' => $media]);
+		}
+	}
+
+	/**
+	 * Creates the "[attach]" element from the given attributes
+	 *
+	 * @param string $href
+	 * @param integer $length
+	 * @param string $type
+	 * @param string $title
+	 * @return string "[attach]" element
+	 */
+	public static function getAttachElement(string $href, int $length, string $type, string $title = '')
+	{
+		$media = self::fetchAdditionalData(['type' => self::DOCUMENT, 'url' => $href,
+			'size' => $length, 'mimetype' => $type, 'description' => $title]);
+
+		return '[attach]href="' . $media['url'] . '" length="' . $media['size'] .
+			'" type="' . $media['mimetype'] . '" title="' . $media['description'] . '"[/attach]';
+	}
+
+	/**
+	 * Fetch additional data for the provided media array
+	 *
+	 * @param array $media
+	 * @return array media array with additional data
+	 */
+	public static function fetchAdditionalData(array $media)
+	{
+		// Fetch the mimetype or size if missing.
+		// We don't do it for torrent links since they need special treatment.
+		// We don't do this for images, since we are fetching their details some lines later anyway.
+		if (!in_array($media['type'], [self::TORRENT, self::IMAGE]) && (empty($media['mimetype']) || empty($media['size']))) {
+			$timeout = DI::config()->get('system', 'xrd_timeout');
+			$curlResult = DI::httpRequest()->head($media['url'], ['timeout' => $timeout]);
+			if ($curlResult->isSuccess()) {
+				$header = $curlResult->getHeaderArray();
+				if (empty($media['mimetype']) && !empty($header['content-type'])) {
+					$media['mimetype'] = $header['content-type'];
+				}
+				if (empty($media['size']) && !empty($header['content-length'])) {
+					$media['size'] = $header['content-length'];
+				}
+			}
+		}
+
+		$filetype = !empty($media['mimetype']) ? strtolower(substr($media['mimetype'], 0, strpos($media['mimetype'], '/'))) : '';
+
+		if (($media['type'] == self::IMAGE) || ($filetype == 'image')) {
 			$imagedata = Images::getInfoFromURLCached($media['url']);
 			if (!empty($imagedata)) {
 				$media['mimetype'] = $imagedata['mime'];
@@ -80,9 +147,7 @@ class Media
 				}
 			}
 		}
-
+		return $media;
-		$result = DBA::insert('post-media', $media, true);
-		Logger::info('Stored media', ['result' => $result, 'media' => $media, 'callstack' => System::callstack()]);
 	}
 
 	/**
@@ -168,4 +233,29 @@ class Media
 
 		return trim($body);
 	}
+
+	/**
+	 * Add media links from the attach field
+	 *
+	 * @param integer $uriid
+	 * @param string $attach
+	 * @return void
+	 */
+	public static function insertFromAttachment(int $uriid, string $attach)
+	{
+		if (!preg_match_all('|\[attach\]href=\"(.*?)\" length=\"(.*?)\" type=\"(.*?)\"(?: title=\"(.*?)\")?|', $attach, $matches, PREG_SET_ORDER)) {
+			return;
+		}
+
+		foreach ($matches as $attachment) {
+			$media['type'] = self::DOCUMENT;
+			$media['uri-id'] = $uriid;
+			$media['url'] = $attachment[1];
+			$media['size'] = $attachment[2];
+			$media['mimetype'] = $attachment[3];
+			$media['description'] = $attachment[4] ?? '';
+
+			self::insert($media);
+		}
+	}
 }

src/Protocol/ActivityPub/Processor.php

@@ -196,7 +196,8 @@ class Processor
 							$item['attach'] = '';
 						}
 
-						$item['attach'] .= '[attach]href="' . $attach['url'] . '" length="' . ($attach['length'] ?? '0') . '" type="' . $attach['mediaType'] . '" title="' . ($attach['name'] ?? '') . '"[/attach]';
+						$item['attach'] .= Post\Media::getAttachElement($attach['url'],
+							$attach['length'] ?? 0, $attach['mediaType'], $attach['name'] ?? '');
 					}
 			}
 		}

src/Protocol/DFRN.php

@@ -39,6 +39,7 @@ use Friendica\Model\ItemURI;
 use Friendica\Model\Mail;
 use Friendica\Model\Notify\Type;
 use Friendica\Model\PermissionSet;
+use Friendica\Model\Post;
 use Friendica\Model\Post\Category;
 use Friendica\Model\Profile;
 use Friendica\Model\Tag;
@@ -2176,7 +2177,7 @@ class DFRN
 							$item["attach"] = "";
 						}
 
-						$item["attach"] .= '[attach]href="' . $href . '" length="' . $length . '" type="' . $type . '" title="' . $title . '"[/attach]';
+						$item["attach"] .= Post\Media::getAttachElement($href, $length, $type, $title);
 						break;
 				}
 			}

src/Protocol/Feed.php

@@ -33,6 +33,7 @@ use Friendica\Database\DBA;
 use Friendica\DI;
 use Friendica\Model\Contact;
 use Friendica\Model\Item;
+use Friendica\Model\Post;
 use Friendica\Model\Tag;
 use Friendica\Model\User;
 use Friendica\Util\DateTimeFormat;
@@ -457,7 +458,7 @@ class Feed
 
 				$attachments[] = ["link" => $href, "type" => $type, "length" => $length];
 
-				$item["attach"] .= '[attach]href="' . $href . '" length="' . $length . '" type="' . $type . '"[/attach]';
+				$item["attach"] .= Post\Media::getAttachElement($href, $length, $type);
 			}
 
 			$taglist = [];

src/Protocol/OStatus.php

@@ -36,6 +36,7 @@ use Friendica\Model\Contact;
 use Friendica\Model\Conversation;
 use Friendica\Model\Item;
 use Friendica\Model\ItemURI;
+use Friendica\Model\Post;
 use Friendica\Model\Tag;
 use Friendica\Model\User;
 use Friendica\Network\Probe;
@@ -1126,7 +1127,8 @@ class OStatus
 							if (!isset($attribute['length'])) {
 								$attribute['length'] = "0";
 							}
-							$item["attach"] .= '[attach]href="'.$attribute['href'].'" length="'.$attribute['length'].'" type="'.$attribute['type'].'" title="'.($attribute['title'] ?? '') .'"[/attach]';
+							$item["attach"] .= Post\Media::getAttachElement($attribute['href'],
+								$attribute['length'], $attribute['type'], $attribute['title'] ?? '');
 						}
 						break;
 					case "related":