Reduce the number of HTTP requests in the media handling

diff --git a/database.sql b/database.sql
index 8aa5c1d5b36672316093c2b11325f6c42d7c10f4..f7c093a4c948197532e6af9c0e712d3fd36829c8 100644 (file)
--- a/database.sql
+++ b/database.sql
@@ -2110,6 +2110,7 @@ CREATE VIEW `post-engagement-user-view` AS SELECT
        `post-thread-user`.`received` AS `received`,
        `post-thread-user`.`created` AS `created`,
        `post-thread-user`.`network` AS `network`,
+       `post-user`.`protocol` AS `protocol`,
        `post-engagement`.`language` AS `restricted`,
        0 AS `comments`,
        0 AS `activities`
@@ -2236,6 +2237,7 @@ CREATE VIEW `post-searchindex-user-view` AS SELECT
        `post-thread-user`.`received` AS `received`,
        `post-thread-user`.`created` AS `created`,
        `post-thread-user`.`network` AS `network`,
+       `post-user`.`protocol` AS `protocol`,
        `post-searchindex`.`language` AS `restricted`,
        0 AS `comments`,
        0 AS `activities`
@@ -2494,6 +2496,7 @@ CREATE VIEW `post-thread-origin-view` AS SELECT
        `post-user`.`global` AS `global`,
        EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post-thread-user`.`uri-id`) AS `featured`,
        `post-thread-user`.`network` AS `network`,
+       `post-user`.`protocol` AS `protocol`,
        `post-origin`.`vid` AS `vid`,
        `post-thread-user`.`psid` AS `psid`,
        IF (`post-origin`.`vid` IS NULL, '', `verb`.`name`) AS `verb`,
@@ -2881,6 +2884,7 @@ CREATE VIEW `post-thread-user-view` AS SELECT
        `post-user`.`global` AS `global`,
        EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post-thread-user`.`uri-id`) AS `featured`,
        `post-thread-user`.`network` AS `network`,
+       `post-user`.`protocol` AS `protocol`,
        `post-user`.`vid` AS `vid`,
        `post-thread-user`.`psid` AS `psid`,
        IF (`post-user`.`vid` IS NULL, '', `verb`.`name`) AS `verb`,
@@ -3061,6 +3065,7 @@ CREATE VIEW `post-view` AS SELECT
        `post`.`global` AS `global`,
        EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post`.`uri-id`) AS `featured`,
        `post`.`network` AS `network`,
+       255 AS `protocol`,
        `post`.`vid` AS `vid`,
        IF (`post`.`vid` IS NULL, '', `verb`.`name`) AS `verb`,
        `post-content`.`title` AS `title`,
@@ -3215,6 +3220,7 @@ CREATE VIEW `post-thread-view` AS SELECT
        `post`.`global` AS `global`,
        EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post-thread`.`uri-id`) AS `featured`,
        `post-thread`.`network` AS `network`,
+       255 AS `protocol`,
        `post`.`vid` AS `vid`,
        IF (`post`.`vid` IS NULL, '', `verb`.`name`) AS `verb`,
        `post-content`.`title` AS `title`,
@@ -3418,6 +3424,7 @@ CREATE VIEW `network-thread-view` AS SELECT
        `post-thread-user`.`starred` AS `starred`,
        `post-thread-user`.`mention` AS `mention`,
        `post-thread-user`.`network` AS `network`,
+       `post-user`.`protocol` AS `protocol`,
        `post-thread-user`.`contact-id` AS `contact-id`,
        `ownercontact`.`contact-type` AS `contact-type`
        FROM `post-thread-user`
@@ -3446,6 +3453,7 @@ CREATE VIEW `network-thread-circle-view` AS SELECT
        `post-thread-user`.`starred` AS `starred`,
        `post-thread-user`.`mention` AS `mention`,
        `post-thread-user`.`network` AS `network`,
+       `post-user`.`protocol` AS `protocol`,
        `post-thread-user`.`contact-id` AS `contact-id`,
        `ownercontact`.`contact-type` AS `contact-type`
        FROM `post-thread-user`
@@ -3818,6 +3826,7 @@ CREATE VIEW `tag-search-view` AS SELECT
        `post-user`.`gravity` AS `gravity`,
        `post-user`.`received` AS `received`,
        `post-user`.`network` AS `network`,
+       `post-user`.`protocol` AS `protocol`,
        `post-user`.`author-id` AS `author-id`,
        `tag`.`name` AS `name`
        FROM `post-tag`

diff --git a/src/Model/Item.php b/src/Model/Item.php
index a9b1435664a00a9ee2b210d52b42ce015bc56483..93de0500e5b18b4e1dabcfe7aab4abd6edace095 100644 (file)
--- a/src/Model/Item.php
+++ b/src/Model/Item.php
@@ -86,7 +86,7 @@ class Item
 
        // Field list that is used to display the items
        const DISPLAY_FIELDLIST = [
-               'uid', 'id', 'parent', 'guid', 'network', 'gravity',
+               'uid', 'id', 'parent', 'guid', 'network', 'protocol', 'gravity',
                'uri-id', 'uri', 'thr-parent-id', 'thr-parent', 'parent-uri-id', 'parent-uri', 'conversation',
                'commented', 'created', 'edited', 'received', 'verb', 'object-type', 'postopts', 'plink',
                'wall', 'private', 'starred', 'origin', 'parent-origin', 'title', 'body', 'language', 'sensitive',
@@ -4169,10 +4169,11 @@ class Item
         * @param string $uri
         * @param int    $uid
         * @param int    $completion
+        * @param string $mimetype
         *
         * @return integer item id
         */
-       public static function fetchByLink(string $uri, int $uid = 0, int $completion = ActivityPub\Receiver::COMPLETION_MANUAL): int
+       public static function fetchByLink(string $uri, int $uid = 0, int $completion = ActivityPub\Receiver::COMPLETION_MANUAL, string $mimetype = ''): int
        {
                Logger::info('Trying to fetch link', ['uid' => $uid, 'uri' => $uri]);
                $item_id = self::searchByLink($uri, $uid);
@@ -4194,35 +4195,49 @@ class Item
                Hook::callAll('item_by_link', $hookData);
 
                if (isset($hookData['item_id'])) {
+                       Logger::info('Hook link fetched', ['uid' => $uid, 'uri' => $uri, 'id' => $hookData['item_id']]);
                        return is_numeric($hookData['item_id']) ? $hookData['item_id'] : 0;
                }
 
-               try {
-                       $curlResult = DI::httpClient()->head($uri, [HttpClientOptions::ACCEPT_CONTENT => HttpClientAccept::JSON_AS, HttpClientOptions::REQUEST => HttpClientRequest::ACTIVITYPUB]);
-                       if (!HTTPSignature::isValidContentType($curlResult->getContentType(), $uri) && (current(explode(';', $curlResult->getContentType())) == 'application/json')) {
+               if (!$mimetype) {
+                       try {
+                               $curlResult = DI::httpClient()->head($uri, [HttpClientOptions::ACCEPT_CONTENT => HttpClientAccept::JSON_AS, HttpClientOptions::REQUEST => HttpClientRequest::ACTIVITYPUB]);
+                               $mimetype = empty($curlResult) ? '' : $curlResult->getContentType();
+                       } catch (\Throwable $th) {
+                               Logger::info('Error while fetching HTTP link via HEAD', ['uid' => $uid, 'uri' => $uri, 'code' => $th->getCode(), 'message' => $th->getMessage()]);
+                               return 0;
+                       }
+               }
+
+               if (!HTTPSignature::isValidContentType($mimetype, $uri) && (current(explode(';', $mimetype)) == 'application/json')) {
+                       try {
                                // Issue 14126: Workaround for Mastodon servers that return "application/json" on a "head" request.
                                $curlResult = HTTPSignature::fetchRaw($uri, $uid);
+                               $mimetype = empty($curlResult) ? '' : $curlResult->getContentType();
+                       } catch (\Throwable $th) {
+                               Logger::info('Error while fetching HTTP link via signed GET', ['uid' => $uid, 'uri' => $uri, 'code' => $th->getCode(), 'message' => $th->getMessage()]);
+                               return 0;
                        }
-                       if (HTTPSignature::isValidContentType($curlResult->getContentType(), $uri)) {
-                               $fetched_uri = ActivityPub\Processor::fetchMissingActivity($uri, [], '', $completion, $uid);
-                       }
-               } catch (\Throwable $th) {
-                       Logger::info('Invalid link', ['uid' => $uid, 'uri' => $uri, 'code' => $th->getCode(), 'message' => $th->getMessage()]);
-                       return 0;
                }
 
-               if (!empty($fetched_uri)) {
-                       $item_id = self::searchByLink($fetched_uri, $uid);
-               } else {
-                       $item_id = Diaspora::fetchByURL($uri);
+               if (HTTPSignature::isValidContentType($mimetype, $uri)) {
+                       $fetched_uri = ActivityPub\Processor::fetchMissingActivity($uri, [], '', $completion, $uid);
+                       if (!empty($fetched_uri)) {
+                               $item_id = self::searchByLink($fetched_uri, $uid);
+                               if ($item_id) {
+                                       Logger::info('ActivityPub link fetched', ['uid' => $uid, 'uri' => $uri, 'id' => $item_id]);
+                                       return $item_id;
+                               }
+                       }
                }
 
-               if (!empty($item_id)) {
-                       Logger::info('Link fetched', ['uid' => $uid, 'uri' => $uri, 'id' => $item_id]);
+               $item_id = Diaspora::fetchByURL($uri);
+               if ($item_id) {
+                       Logger::info('Diaspora link fetched', ['uid' => $uid, 'uri' => $uri, 'id' => $item_id]);
                        return $item_id;
                }
 
-               Logger::info('Link not found', ['uid' => $uid, 'uri' => $uri]);
+               Logger::info('This is not an item link', ['uid' => $uid, 'uri' => $uri]);
                return 0;
        }
 

diff --git a/src/Model/Post/Media.php b/src/Model/Post/Media.php
index e707c79cedb08c2783b2e73f281b90fc72ad299c..fb363dd57a55e26fb383a7be695b730accd301ec 100644 (file)
--- a/src/Model/Post/Media.php
+++ b/src/Model/Post/Media.php
@@ -51,6 +51,8 @@ class Media
        const ACTIVITY    = 20;
        const ACCOUNT     = 21;
        const HLS         = 22;
+       const JSON        = 23;
+       const LD          = 24;
        const DOCUMENT    = 128;
 
        /**
@@ -180,14 +182,14 @@ class Media
                }
 
                // Fetch the mimetype or size if missing.
-               if (Network::isValidHttpUrl($media['url']) && (empty($media['mimetype']) || empty($media['size']))) {
+               if (Network::isValidHttpUrl($media['url']) && empty($media['mimetype']) && !in_array($media['type'], [self::IMAGE, self::HLS])) {
                        $timeout = DI::config()->get('system', 'xrd_timeout');
                        try {
-                               $curlResult = DI::httpClient()->head($media['url'], [HttpClientOptions::TIMEOUT => $timeout, HttpClientOptions::REQUEST => HttpClientRequest::CONTENTTYPE]);
+                               $curlResult = DI::httpClient()->head($media['url'], [HttpClientOptions::ACCEPT_CONTENT => HttpClientAccept::AS_DEFAULT, HttpClientOptions::TIMEOUT => $timeout, HttpClientOptions::REQUEST => HttpClientRequest::CONTENTTYPE]);
 
                                // Workaround for systems that can't handle a HEAD request
-                               if (!$curlResult->isSuccess() && ($curlResult->getReturnCode() == 405)) {
-                                       $curlResult = DI::httpClient()->get($media['url'], HttpClientAccept::DEFAULT, [HttpClientOptions::TIMEOUT => $timeout]);
+                               if (!$curlResult->isSuccess() && in_array($curlResult->getReturnCode(), [400, 403, 405])) {
+                                       $curlResult = DI::httpClient()->get($media['url'], HttpClientAccept::AS_DEFAULT, [HttpClientOptions::TIMEOUT => $timeout]);
                                }
                                if ($curlResult->isSuccess()) {
                                        if (empty($media['mimetype'])) {
@@ -197,16 +199,20 @@ class Media
                                                $media['size'] = (int)($curlResult->getHeader('Content-Length')[0] ?? strlen($curlResult->getBodyString() ?? ''));
                                        }
                                } else {
-                                       Logger::notice('Could not fetch head', ['media' => $media]);
+                                       Logger::notice('Could not fetch head', ['media' => $media, 'code' => $curlResult->getReturnCode()]);
                                }
                        } catch (\Throwable $th) {
                                Logger::notice('Got exception', ['code' => $th->getCode(), 'message' => $th->getMessage()]);
                        }
                }
 
-               $filetype = !empty($media['mimetype']) ? strtolower(current(explode('/', $media['mimetype']))) : '';
+               if (($media['type'] != self::DOCUMENT) && !empty($media['mimetype'])) {
+                       $media = self::addType($media);
+               }
+
+               Logger::debug('Got type for url', ['type' => $media['type'], 'mimetype' => $media['mimetype'] ?? '', 'url' => $media['url']]);
 
-               if (($media['type'] == self::IMAGE) || ($filetype == 'image')) {
+               if ($media['type'] == self::IMAGE) {
                        $imagedata = Images::getInfoFromURLCached($media['url'], empty($media['description']));
                        if ($imagedata) {
                                $media['mimetype'] = $imagedata['mime'];
@@ -223,23 +229,19 @@ class Media
                        }
                }
 
-               if ($media['type'] != self::DOCUMENT) {
-                       $media = self::addType($media);
-               }
-
                if (!empty($media['preview'])) {
                        $media = self::addPreviewData($media);
                }
 
-               if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
-                       $media = self::addActivity($media);
+               if (in_array($media['type'], [self::TEXT, self::ACTIVITY, self::LD, self::JSON, self::HTML, self::XML, self::PLAIN])) {
+                       $media = self::addAccount($media);
                }
 
-               if (in_array($media['type'], [self::TEXT, self::APPLICATION, self::HTML, self::XML, self::PLAIN])) {
-                       $media = self::addAccount($media);
+               if (in_array($media['type'], [self::ACTIVITY, self::LD, self::JSON])) {
+                       $media = self::addActivity($media);
                }
 
-               if ($media['type'] == self::HTML) {
+               if (in_array($media['type'], [self::HTML, self::LD, self::JSON])) {
                        $media = self::addPage($media);
                }
 
@@ -254,6 +256,16 @@ class Media
 
                $imagedata = Images::getInfoFromURLCached($media['preview']);
                if ($imagedata) {
+                       $media['blurhash'] = $imagedata['blurhash'] ?? null;
+
+                       // When the original picture is potentially animated but the preview isn't, we override the preview
+                       if (in_array($media['mimetype'] ?? '', ['image/gif', 'image/png']) && !in_array($imagedata['mime'], ['image/gif', 'image/png'])) {
+                               $media['preview'] = $media['url'];
+                               $media['preview-width'] = $media['width'];
+                               $media['preview-height'] = $media['height'];
+                               return $media;
+                       }
+
                        $media['preview-width'] = $imagedata[0];
                        $media['preview-height'] = $imagedata[1];
                }
@@ -269,19 +281,22 @@ class Media
         */
        private static function addActivity(array $media): array
        {
-               $id = Item::fetchByLink($media['url'], 0, ActivityPub\Receiver::COMPLETION_ASYNC);
+               $id = Item::fetchByLink($media['url'], 0, ActivityPub\Receiver::COMPLETION_ASYNC, $media['mimetype'] ?? '');
                if (empty($id)) {
+                       $media['type'] = $media['type'] == self::ACTIVITY ? self::JSON : $media['type'];
                        return $media;
                }
 
                $item = Post::selectFirst([], ['id' => $id, 'network' => Protocol::FEDERATED]);
                if (empty($item['id'])) {
                        Logger::debug('Not a federated activity', ['id' => $id, 'uri-id' => $media['uri-id'], 'url' => $media['url']]);
+                       $media['type'] = $media['type'] == self::ACTIVITY ? self::JSON : $media['type'];
                        return $media;
                }
 
                if ($item['uri-id'] == $media['uri-id']) {
                        Logger::info('Media-Uri-Id is identical to Uri-Id', ['uri-id' => $media['uri-id']]);
+                       $media['type'] = $media['type'] == self::ACTIVITY ? self::JSON : $media['type'];
                        return $media;
                }
 
@@ -290,6 +305,7 @@ class Media
                        parse_url($item['plink'], PHP_URL_HOST) != parse_url($item['uri'], PHP_URL_HOST)
                ) {
                        Logger::debug('Not a link to an activity', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'plink' => $item['plink'], 'uri' => $item['uri']]);
+                       $media['type'] = $media['type'] == self::ACTIVITY ? self::JSON : $media['type'];
                        return $media;
                }
 
@@ -375,14 +391,23 @@ class Media
         */
        private static function addPage(array $media): array
        {
-               $data = ParseUrl::getSiteinfoCached($media['url']);
+               $data = ParseUrl::getSiteinfoCached($media['url'], $media['mimetype'] ?? '');
+               if (empty($data['images'][0]['src']) && empty($data['text']) && empty($data['title'])) {
+                       if (!empty($media['preview'])) {
+                               $media = self::addPreviewData($media);
+                               Logger::debug('Detected site data is empty, use suggested media data instead', ['uri-id' => $media['uri-id'], 'url' => $media['url'], 'type' => $data['type']]);
+                       }
+               } else {
+                       $media['preview'] = $data['images'][0]['src'] ?? null;
+                       $media['preview-height'] = $data['images'][0]['height'] ?? null;
+                       $media['preview-width'] = $data['images'][0]['width'] ?? null;
+                       $media['blurhash'] = $data['images'][0]['blurhash'] ?? null;
+                       $media['description'] = $data['text'] ?? null;
+                       $media['name'] = $data['title'] ?? null;        
+               }
+               
+               $media['type'] = self::HTML;
                $media['size'] = $data['size'] ?? null;
-               $media['preview'] = $data['images'][0]['src'] ?? null;
-               $media['preview-height'] = $data['images'][0]['height'] ?? null;
-               $media['preview-width'] = $data['images'][0]['width'] ?? null;
-               $media['blurhash'] = $data['images'][0]['blurhash'] ?? null;
-               $media['description'] = $data['text'] ?? null;
-               $media['name'] = $data['title'] ?? null;
                $media['author-url'] = $data['author_url'] ?? null;
                $media['author-name'] = $data['author_name'] ?? null;
                $media['author-image'] = $data['author_img'] ?? null;
@@ -481,6 +506,12 @@ class Media
                        $type = self::TORRENT;
                } elseif (($filetype == 'application') && ($subtype == 'vnd.apple.mpegurl')) {
                        $type = self::HLS;
+               } elseif (($filetype == 'application') && ($subtype == 'activity+json')) {
+                       $type = self::ACTIVITY;
+               } elseif (($filetype == 'application') && ($subtype == 'ld+json')) {
+                       $type = self::LD;
+               } elseif (($filetype == 'application') && ($subtype == 'json')) {
+                       $type = self::JSON;
                } elseif ($filetype == 'application') {
                        $type = self::APPLICATION;
                } else {

diff --git a/src/Module/Stats.php b/src/Module/Stats.php
index d675aa80ee1e6cbdfc64af41cc2bc65217cdda1e..4bc9d957119bc53cf76f48eeafec74f4bc7a3177 100644 (file)
--- a/src/Module/Stats.php
+++ b/src/Module/Stats.php
@@ -94,6 +94,13 @@ class Stats extends BaseModule
                                'deferred'      => [],
                                'total'         => [],
                        ],
+                       'jetstream' => [
+                               'drift'     => intval($this->keyValue->get('jetstream_drift')),
+                               'did_count' => intval($this->keyValue->get('jetstream_did_count')),
+                               'did_limit' => intval($this->keyValue->get('jetstream_did_limit')),
+                               'messages'  => intval($this->keyValue->get('jetstream_messages')),
+                               'timestamp' => intval($this->keyValue->get('jetstream_timestamp')),
+                       ],
                        'users' => [
                                'total'          => intval($this->keyValue->get('nodeinfo_total_users')),
                                'activeWeek'     => intval($this->keyValue->get('nodeinfo_active_users_weekly')),

diff --git a/src/Network/HTTPClient/Client/HttpClientAccept.php b/src/Network/HTTPClient/Client/HttpClientAccept.php
index 601c2d732e7d5da8e14e78b3481c27d556fce926..24326330322879bac39aec8195caa2476042d544 100644 (file)
--- a/src/Network/HTTPClient/Client/HttpClientAccept.php
+++ b/src/Network/HTTPClient/Client/HttpClientAccept.php
@@ -15,6 +15,9 @@ class HttpClientAccept
        /** @var string Default value for "Accept" header */
        public const DEFAULT = '*/*';
 
+       /** @var string Accept all types with a preferences of ActivityStream content */
+       public const AS_DEFAULT = 'application/activity+json,application/ld+json; profile="https://www.w3.org/ns/activitystreams",*/*;q=0.9';
+
        public const ATOM_XML  = 'application/atom+xml,text/xml;q=0.9,*/*;q=0.8';
        public const FEED_XML  = 'application/atom+xml,application/rss+xml;q=0.9,application/rdf+xml;q=0.8,text/xml;q=0.7,*/*;q=0.6';
        public const HTML      = 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8';

diff --git a/src/Object/Post.php b/src/Object/Post.php
index 450af95c84373b8746db16551a44f2e2e4be849f..a88ea8ecf9824851fd4d5109e5b6ecd1c3a97cd4 100644 (file)
--- a/src/Object/Post.php
+++ b/src/Object/Post.php
@@ -16,6 +16,7 @@ use Friendica\Core\Protocol;
 use Friendica\Core\Renderer;
 use Friendica\DI;
 use Friendica\Model\Contact;
+use Friendica\Model\Conversation;
 use Friendica\Model\Item;
 use Friendica\Model\Post as PostModel;
 use Friendica\Model\Tag;
@@ -193,7 +194,7 @@ class Post
 
                $privacy   = $this->fetchPrivacy($item);
                $lock      = ($item['private'] == Item::PRIVATE) ? $privacy : false;
-               $connector = !in_array($item['network'], Protocol::NATIVE_SUPPORT) ? DI::l10n()->t('Connector Message') : false;
+               $connector = !in_array($item['network'], Protocol::NATIVE_SUPPORT) && ($item['protocol'] != Conversation::PARCEL_JETSTREAM) ? DI::l10n()->t('Connector Message') : false;
 
                $shareable    = in_array($conv->getProfileOwner(), [0, DI::userSession()->getLocalUserId()]) && $item['private'] != Item::PRIVATE;
                $announceable = $shareable && in_array($item['network'], [Protocol::ACTIVITYPUB, Protocol::DFRN, Protocol::DIASPORA, Protocol::TWITTER, Protocol::TUMBLR, Protocol::BLUESKY]);

diff --git a/src/Protocol/ATProtocol.php b/src/Protocol/ATProtocol.php
index 139453224f40880b0ada68e3095ea90c69d5b50b..72383755fd274f883eb60eed59a520bc51c22122 100644 (file)
--- a/src/Protocol/ATProtocol.php
+++ b/src/Protocol/ATProtocol.php
@@ -147,7 +147,7 @@ final class ATProtocol
                return $data;
        }
 
-       private function post(int $uid, string $url, string $params, array $headers): ?stdClass
+       public function post(int $uid, string $url, string $params, array $headers): ?stdClass
        {
                $pds = $this->getUserPds($uid);
                if (empty($pds)) {
@@ -172,8 +172,11 @@ final class ATProtocol
                        $data->code = $curlResult->getReturnCode();
                }
 
-               $this->pConfig->set($uid, 'bluesky', 'status', self::STATUS_SUCCESS);
-               Item::incrementOutbound(Protocol::BLUESKY);
+               if (!empty($data->code) && ($data->code >= 200) && ($data->code < 400)) {
+                       $this->pConfig->set($uid, 'bluesky', 'status', self::STATUS_SUCCESS);
+               } else {
+                       $this->pConfig->set($uid, 'bluesky', 'status', self::STATUS_API_FAIL);
+               }
                return $data;
        }
 
@@ -230,7 +233,7 @@ final class ATProtocol
                return $did;
        }
 
-       private function getDid(string $handle): string
+       public function getDid(string $handle): string
        {
                if ($handle == '') {
                        return '';
@@ -373,7 +376,7 @@ final class ATProtocol
                return in_array('at://' . $handle, $data->alsoKnownAs);
        }
 
-       private function getUserToken(int $uid): string
+       public function getUserToken(int $uid): string
        {
                $token   = $this->pConfig->get($uid, 'bluesky', 'access_token');
                $created = $this->pConfig->get($uid, 'bluesky', 'token_created');
@@ -393,6 +396,11 @@ final class ATProtocol
 
                $data = $this->post($uid, '/xrpc/com.atproto.server.refreshSession', '', ['Authorization' => ['Bearer ' . $token]]);
                if (empty($data) || empty($data->accessJwt)) {
+                       $this->logger->debug('Refresh failed', ['return' => $data]);
+                       $password = $this->pConfig->get($uid, 'bluesky', 'password');
+                       if (!empty($password)) {
+                               return $this->createUserToken($uid, $password);
+                       }
                        $this->pConfig->set($uid, 'bluesky', 'status', self::STATUS_TOKEN_FAIL);
                        return '';
                }

diff --git a/src/Protocol/ATProtocol/Jetstream.php b/src/Protocol/ATProtocol/Jetstream.php
index 834d6ac5c6e932d46aa32e3bf8c1b74cb05f9efc..bf53aa46747c6be488ae7c882e79578debb83410 100755 (executable)
--- a/src/Protocol/ATProtocol/Jetstream.php
+++ b/src/Protocol/ATProtocol/Jetstream.php
@@ -38,9 +38,10 @@ use stdClass;
  */
 class Jetstream
 {
-       private $uids   = [];
-       private $self   = [];
-       private $capped = false;
+       private $uids      = [];
+       private $self      = [];
+       private $capped    = false;
+       private $next_stat = 0;
 
        /** @var LoggerInterface */
        private $logger;
@@ -108,6 +109,7 @@ class Jetstream
                                                $timestamp = $data->time_us;
                                                $this->route($data);
                                                $this->keyValue->set('jetstream_timestamp', $timestamp);
+                                               $this->incrementMessages();
                                        } else {
                                                $this->logger->warning('Unexpected return value', ['data' => $data]);
                                                break;
@@ -135,6 +137,15 @@ class Jetstream
                }
        }
 
+       private function incrementMessages()
+       {
+               $packets = (int)($this->keyValue->get('jetstream_messages') ?? 0);
+               if ($packets >= PHP_INT_MAX) {
+                       $packets = 0;
+               }
+               $this->keyValue->set('jetstream_messages', $packets + 1);
+       }
+
        private function syncContacts()
        {
                $active_uids = $this->atprotocol->getUids();
@@ -184,10 +195,14 @@ class Jetstream
                }
 
                if (!$this->capped && count($dids) < $did_limit) {
-                       $contacts = Contact::selectToArray(['url'], ['uid' => 0, 'network' => Protocol::BLUESKY], ['order' => ['last-item' => true], 'limit' => $did_limit]);
+                       $condition = ["`uid` = ? AND `network` = ? AND EXISTS(SELECT `author-id` FROM `post-user` WHERE `author-id` = `contact`.`id` AND `post-user`.`uid` != ?)", 0, Protocol::BLUESKY, 0];
+                       $contacts = Contact::selectToArray(['url'], $condition, ['order' => ['last-item' => true], 'limit' => $did_limit]);
                        $dids     = $this->addDids($contacts, $uids, $did_limit, $dids);
                }
 
+               $this->keyValue->set('jetstream_did_count', count($dids));
+               $this->keyValue->set('jetstream_did_limit', $did_limit);
+
                $this->logger->debug('Selected DIDs', ['uids' => $active_uids, 'count' => count($dids), 'capped' => $this->capped]);
                $update = [
                        'type'    => 'options_update',
@@ -241,17 +256,7 @@ class Jetstream
 
        private function routeCommits(stdClass $data)
        {
-               $drift = max(0, round(time() - $data->time_us / 1000000));
-               if ($drift > 60 && !$this->capped) {
-                       $this->capped = true;
-                       $this->setOptions();
-                       $this->logger->notice('Drift is too high, dids will be capped');
-               } elseif ($drift == 0 && $this->capped) {
-                       $this->capped = false;
-                       $this->setOptions();
-                       $this->logger->notice('Drift is low enough, dids will be uncapped');
-               }
-
+               $drift = $this->getDrift($data);
                $this->logger->notice('Received commit', ['time' => date(DateTimeFormat::ATOM, $data->time_us / 1000000), 'drift' => $drift, 'capped' => $this->capped, 'did' => $data->did, 'operation' => $data->commit->operation, 'collection' => $data->commit->collection, 'timestamp' => $data->time_us]);
                $timestamp = microtime(true);
 
@@ -299,6 +304,23 @@ class Jetstream
                }
        }
 
+       private function getDrift(stdClass $data): int
+       {
+               $drift = max(0, round(time() - $data->time_us / 1000000));
+               $this->keyValue->set('jetstream_drift', $drift);
+
+               if ($drift > 60 && !$this->capped) {
+                       $this->capped = true;
+                       $this->setOptions();
+                       $this->logger->notice('Drift is too high, dids will be capped');
+               } elseif ($drift == 0 && $this->capped) {
+                       $this->capped = false;
+                       $this->setOptions();
+                       $this->logger->notice('Drift is low enough, dids will be uncapped');
+               }
+               return $drift;
+       }
+
        private function routePost(stdClass $data, int $drift)
        {
                switch ($data->commit->operation) {

diff --git a/src/Protocol/ATProtocol/Processor.php b/src/Protocol/ATProtocol/Processor.php
index 9416ef7a2f2e1213794e1b9a16c3705a72599928..737112ec4c040f45f309f7b1af1e7a2fa37aa400 100755 (executable)
--- a/src/Protocol/ATProtocol/Processor.php
+++ b/src/Protocol/ATProtocol/Processor.php
@@ -13,6 +13,7 @@ namespace Friendica\Protocol\ATProtocol;
 use Friendica\App\BaseURL;
 use Friendica\Core\Protocol;
 use Friendica\Database\Database;
+use Friendica\Database\DBA;
 use Friendica\Model\Contact;
 use Friendica\Model\Conversation;
 use Friendica\Model\Item;
@@ -128,12 +129,12 @@ class Processor
                if (!empty($data->commit->record->reply)) {
                        $root   = $this->getUri($data->commit->record->reply->root);
                        $parent = $this->getUri($data->commit->record->reply->parent);
-                       $uids   = $this->getPostUids($root);
+                       $uids   = $this->getPostUids($root, true);
                        if (!$uids) {
                                $this->logger->debug('Comment is not imported since the root post is not found.', ['root' => $root, 'parent' => $parent]);
                                return;
                        }
-                       if ($dont_fetch && !$this->getPostUids($parent)) {
+                       if ($dont_fetch && !$this->getPostUids($parent, false)) {
                                $this->logger->debug('Comment is not imported since the parent post is not found.', ['root' => $root, 'parent' => $parent]);
                                return;
                        }
@@ -166,7 +167,8 @@ class Processor
                                                return;
                                        }
                                }
-                               $item = $this->addMedia($post->thread->post->embed, $item, 0, 0, 0);
+                               $item['source'] = json_encode($post);
+                               $item = $this->addMedia($post->thread->post->embed, $item, 0);
                        }
 
                        $id = Item::insert($item);
@@ -183,7 +185,7 @@ class Processor
 
        public function createRepost(stdClass $data, array $uids, bool $dont_fetch)
        {
-               if ($dont_fetch && !$this->getPostUids($this->getUri($data->commit->record->subject))) {
+               if ($dont_fetch && !$this->getPostUids($this->getUri($data->commit->record->subject), true)) {
                        $this->logger->debug('Repost is not imported since the subject is not found.', ['subject' => $this->getUri($data->commit->record->subject)]);
                        return;
                }
@@ -213,7 +215,7 @@ class Processor
 
        public function createLike(stdClass $data)
        {
-               $uids = $this->getPostUids($this->getUri($data->commit->record->subject));
+               $uids = $this->getPostUids($this->getUri($data->commit->record->subject), false);
                if (!$uids) {
                        $this->logger->debug('Like is not imported since the subject is not found.', ['subject' => $this->getUri($data->commit->record->subject)]);
                        return;
@@ -270,7 +272,7 @@ class Processor
                return true;
        }
 
-       private function processPost(stdClass $post, int $uid, int $post_reason, int $causer, int $level, int $protocol): int
+       public function processPost(stdClass $post, int $uid, int $post_reason, int $causer, int $level, int $protocol): int
        {
                $uri = $this->getUri($post);
 
@@ -295,7 +297,7 @@ class Processor
                }
 
                if (!empty($post->embed)) {
-                       $item = $this->addMedia($post->embed, $item, $uid, $level);
+                       $item = $this->addMedia($post->embed, $item, $level);
                }
 
                $item['restrictions'] = $this->getRestrictionsForUser($post, $item, $post_reason);
@@ -378,7 +380,7 @@ class Processor
                return $item;
        }
 
-       private function getHeaderFromPost(stdClass $post, string $uri, int $uid, int $protocol): array
+       public function getHeaderFromPost(stdClass $post, string $uri, int $uid, int $protocol): array
        {
                $parts = $this->getUriParts($uri);
                if (empty($post->author) || empty($post->cid) || empty($parts->rkey)) {
@@ -538,6 +540,8 @@ class Processor
                                                'url'         => $image->fullsize,
                                                'preview'     => $image->thumb,
                                                'description' => $image->alt,
+                                               'height'      => $image->aspectRatio->height ?? null,
+                                               'width'       => $image->aspectRatio->width ?? null,
                                        ];
                                        Post\Media::insert($media);
                                }
@@ -561,6 +565,7 @@ class Processor
                                        'uri-id'      => $item['uri-id'],
                                        'type'        => Post\Media::HTML,
                                        'url'         => $embed->external->uri,
+                                       'preview'     => $embed->external->thumb ?? null,
                                        'name'        => $embed->external->title,
                                        'description' => $embed->external->description,
                                ];
@@ -686,7 +691,7 @@ class Processor
                return $restrict ? Item::CANT_REPLY : null;
        }
 
-       private function fetchMissingPost(string $uri, int $uid, int $post_reason, int $causer, int $level, string $fallback = '', bool $always_fetch = false, int $Protocol = Conversation::PARCEL_JETSTREAM): string
+       public function fetchMissingPost(string $uri, int $uid, int $post_reason, int $causer, int $level, string $fallback = '', bool $always_fetch = false, int $Protocol = Conversation::PARCEL_JETSTREAM): string
        {
                $timestamp = microtime(true);
                $stamp     = Strings::getRandomHex(30);
@@ -794,7 +799,7 @@ class Processor
                return $uri;
        }
 
-       private function getUriParts(string $uri): ?stdClass
+       public function getUriParts(string $uri): ?stdClass
        {
                $class = $this->getUriClass($uri);
                if (empty($class)) {
@@ -812,7 +817,7 @@ class Processor
                return $class;
        }
 
-       private function getUriClass(string $uri): ?stdClass
+       public function getUriClass(string $uri): ?stdClass
        {
                if (empty($uri)) {
                        return null;
@@ -837,7 +842,7 @@ class Processor
                return $class;
        }
 
-       private function fetchUriId(string $uri, int $uid): string
+       public function fetchUriId(string $uri, int $uid): string
        {
                $reply = Post::selectFirst(['uri-id'], ['uri' => $uri, 'uid' => [$uid, 0]]);
                if (!empty($reply['uri-id'])) {
@@ -852,16 +857,18 @@ class Processor
                return 0;
        }
 
-       private function getPostUids(string $uri): array
+       private function getPostUids(string $uri, bool $with_public_user): array
        {
+               $condition = $with_public_user ? [] : ["`uid` != ?", 0];
+
                $uids  = [];
-               $posts = Post::select(['uid'], ['uri' => $uri]);
+               $posts = Post::select(['uid'], DBA::mergeConditions(['uri' => $uri], $condition));
                while ($post = Post::fetch($posts)) {
                        $uids[] = $post['uid'];
                }
                $this->db->close($posts);
 
-               $posts = Post::select(['uid'], ['extid' => $uri]);
+               $posts = Post::select(['uid'], DBA::mergeConditions(['extid' => $uri], $condition));
                while ($post = Post::fetch($posts)) {
                        $uids[] = $post['uid'];
                }
@@ -878,7 +885,7 @@ class Processor
                return Post::exists(['extid' => $uri, 'uid' => $uids]);
        }
 
-       private function getUri(stdClass $post): string
+       public function getUri(stdClass $post): string
        {
                if (empty($post->cid)) {
                        $this->logger->info('Invalid URI', ['post' => $post]);
@@ -887,7 +894,7 @@ class Processor
                return $post->uri . ':' . $post->cid;
        }
 
-       private function getPostUri(string $uri, int $uid): string
+       public function getPostUri(string $uri, int $uid): string
        {
                if (Post::exists(['uri' => $uri, 'uid' => [$uid, 0]])) {
                        $this->logger->debug('Post exists', ['uri' => $uri]);

diff --git a/src/Util/ParseUrl.php b/src/Util/ParseUrl.php
index 4238499bab954833c43ebf8512e504d18e4d6cf0..9d2e807f24a4e38a1df8d802ed459683a07747c4 100644 (file)
--- a/src/Util/ParseUrl.php
+++ b/src/Util/ParseUrl.php
@@ -89,7 +89,8 @@ class ParseUrl
        /**
         * Search for cached embeddable data of an url otherwise fetch it
         *
-        * @param string $url         The url of the page which should be scraped
+        * @param string $url      The url of the page which should be scraped
+        * @param string $mimetype Optional mimetype that had already been detected for this page
         *
         * @return array which contains needed data for embedding
         *    string 'url'      => The url of the parsed page
@@ -104,7 +105,7 @@ class ParseUrl
         * @see   ParseUrl::getSiteinfo() for more information about scraping
         * embeddable content
         */
-       public static function getSiteinfoCached(string $url): array
+       public static function getSiteinfoCached(string $url, string $mimetype = ''): array
        {
                if (empty($url)) {
                        return [
@@ -123,7 +124,7 @@ class ParseUrl
                        return $data;
                }
 
-               $data = self::getSiteinfo($url);
+               $data = self::getSiteinfo($url, $mimetype);
 
                $expires = $data['expires'];
 
@@ -155,8 +156,9 @@ class ParseUrl
         * like \<title\>Awesome Title\</title\> or
         * \<meta name="description" content="An awesome description"\>
         *
-        * @param string $url         The url of the page which should be scraped
-        * @param int    $count       Internal counter to avoid endless loops
+        * @param string $url      The url of the page which should be scraped
+        * @param string $mimetype Optional mimetype that had already been detected for this page
+        * @param int    $count    Internal counter to avoid endless loops
         *
         * @return array which contains needed data for embedding
         *    string 'url'      => The url of the parsed page
@@ -181,7 +183,7 @@ class ParseUrl
         * </body>
         * @endverbatim
         */
-       public static function getSiteinfo(string $url, int $count = 1): array
+       public static function getSiteinfo(string $url, string $mimetype = '', int $count = 1): array
        {
                if (empty($url)) {
                        return [
@@ -212,7 +214,11 @@ class ParseUrl
                        return $siteinfo;
                }
 
-               $type = self::getContentType($url);
+               if (!empty($mimetype)) {
+                       $type = explode('/', current(explode(';', $mimetype)));
+               } else {
+                       $type = self::getContentType($url);
+               }
                Logger::info('Got content-type', ['content-type' => $type, 'url' => $url]);
                if (!empty($type) && in_array($type[0], ['image', 'video', 'audio'])) {
                        $siteinfo['type'] = $type[0];
@@ -309,7 +315,7 @@ class ParseUrl
                                        }
                                }
                                if ($content != '') {
-                                       $siteinfo = self::getSiteinfo($content, ++$count);
+                                       $siteinfo = self::getSiteinfo($content, $mimetype, ++$count);
                                        return $siteinfo;
                                }
                        }

diff --git a/static/dbview.config.php b/static/dbview.config.php
index b1c91d594223147ee0d61371f81e0cded6f42720..f3f81938bced881488d5ab2163f1a86ade2b1f8f 100644 (file)
--- a/static/dbview.config.php
+++ b/static/dbview.config.php
@@ -101,6 +101,7 @@ return [
                        "received" => ["post-thread-user", "received"],
                        "created" => ["post-thread-user", "created"],
                        "network" => ["post-thread-user", "network"],
+                       "protocol" => ["post-user", "protocol"],
                        "restricted" => ["post-engagement", "language"],
                        "comments" => "0",
                        "activities" => "0",
@@ -221,6 +222,7 @@ return [
                        "received" => ["post-thread-user", "received"],
                        "created" => ["post-thread-user", "created"],
                        "network" => ["post-thread-user", "network"],
+                       "protocol" => ["post-user", "protocol"],
                        "restricted" => ["post-searchindex", "language"],
                        "comments" => "0",
                        "activities" => "0",
@@ -475,6 +477,7 @@ return [
                        "global" => ["post-user", "global"],
                        "featured" => "EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post-thread-user`.`uri-id`)",
                        "network" => ["post-thread-user", "network"],
+                       "protocol" => ["post-user", "protocol"],
                        "vid" => ["post-origin", "vid"],
                        "psid" => ["post-thread-user", "psid"],
                        "verb" => "IF (`post-origin`.`vid` IS NULL, '', `verb`.`name`)",
@@ -858,6 +861,7 @@ return [
                        "global" => ["post-user", "global"],
                        "featured" => "EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post-thread-user`.`uri-id`)",
                        "network" => ["post-thread-user", "network"],
+                       "protocol" => ["post-user", "protocol"],
                        "vid" => ["post-user", "vid"],
                        "psid" => ["post-thread-user", "psid"],
                        "verb" => "IF (`post-user`.`vid` IS NULL, '', `verb`.`name`)",
@@ -1036,6 +1040,7 @@ return [
                        "global" => ["post", "global"],
                        "featured" => "EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post`.`uri-id`)",
                        "network" => ["post", "network"],
+                       "protocol" => "255",
                        "vid" => ["post", "vid"],
                        "verb" => "IF (`post`.`vid` IS NULL, '', `verb`.`name`)",
                        "title" => ["post-content", "title"],
@@ -1188,6 +1193,7 @@ return [
                        "global" => ["post", "global"],
                        "featured" => "EXISTS(SELECT `type` FROM `post-collection` WHERE `type` = 0 AND `uri-id` = `post-thread`.`uri-id`)",
                        "network" => ["post-thread", "network"],
+                       "protocol" => "255",
                        "vid" => ["post", "vid"],
                        "verb" => "IF (`post`.`vid` IS NULL, '', `verb`.`name`)",
                        "title" => ["post-content", "title"],
@@ -1381,6 +1387,7 @@ return [
                        "starred" => ["post-thread-user", "starred"],
                        "mention" => ["post-thread-user", "mention"],
                        "network" => ["post-thread-user", "network"],
+                       "protocol" => ["post-user", "protocol"],
                        "contact-id" => ["post-thread-user", "contact-id"],
                        "contact-type" => ["ownercontact", "contact-type"],
                ],
@@ -1407,6 +1414,7 @@ return [
                        "starred" => ["post-thread-user", "starred"],
                        "mention" => ["post-thread-user", "mention"],
                        "network" => ["post-thread-user", "network"],
+                       "protocol" => ["post-user", "protocol"],
                        "contact-id" => ["post-thread-user", "contact-id"],
                        "contact-type" => ["ownercontact", "contact-type"],
                ],
@@ -1769,6 +1777,7 @@ return [
                        "gravity" => ["post-user", "gravity"],
                        "received" => ["post-user", "received"],
                        "network" => ["post-user", "network"],
+                       "protocol" => ["post-user", "protocol"],
                        "author-id" => ["post-user", "author-id"],
                        "name" => ["tag", "name"],
                ],