Avoid fetching posts in a loop

diff --git a/src/Protocol/ActivityPub/Processor.php b/src/Protocol/ActivityPub/Processor.php
index 33cbfc98a4e291594ffe1ffb7f00ac0f6379027f..672589a64f4dc2f93fc524e3355590e0a3797c21 100644 (file)
--- a/src/Protocol/ActivityPub/Processor.php
+++ b/src/Protocol/ActivityPub/Processor.php
@@ -563,6 +563,11 @@ class Processor
                        return '';
                }
 
+               if (in_array($activity['reply-to-id'], $activity['children'] ?? [])) {
+                       Logger::notice('reply-to-id is already in the list of children', ['id' => $activity['reply-to-id'], 'children' => $activity['children'], 'depth' => count($activity['children'])]);
+                       return '';
+               } 
+
                self::addActivityId($activity['reply-to-id']);
 
                $completion = $activity['completion-mode'] ?? Receiver::COMPLETION_NONE;
@@ -1632,6 +1637,11 @@ class Processor
                        return null;
                }
 
+               if (!empty($child['children']) && in_array($url, $child['children'])) {
+                       Logger::notice('id is already in the list of children', ['depth' => count($child['children']), 'children' => $child['children'], 'id' => $url]);
+                       return null;
+               }
+
                try {
                        $curlResult = HTTPSignature::fetchRaw($url, $uid);
                } catch (\Exception $exception) {
@@ -1725,6 +1735,16 @@ class Processor
                }
 
                $ldactivity['recursion-depth'] = !empty($child['recursion-depth']) ? $child['recursion-depth'] + 1 : 0;
+               $ldactivity['children']        = $child['children'] ?? [];
+               // This check ist mostly superflous, since there are similar checks before. This covers the case, when the fetched id doesn't match the url
+               if (in_array($activity['id'], $ldactivity['children'])) {                       
+                       Logger::notice('Fetched id is already in the list of children. It will not be processed.', ['id' => $activity['id'], 'children' => $ldactivity['children'], 'depth' => count($ldactivity['children'])]);
+                       return null;
+               }
+               if (!empty($child['id'])) {
+                       $ldactivity['children'][] = $child['id'];
+               }
+
 
                if ($object_actor != $actor) {
                        Contact::updateByUrlIfNeeded($object_actor);
@@ -1782,6 +1802,10 @@ class Processor
                                if (Processor::alreadyKnown($id, $child['id'] ?? '')) {
                                        continue;
                                }
+                               if (!empty($child['children']) && in_array($id, $child['children'])) {
+                                       Logger::debug('Replies id is already in the list of children', ['depth' => count($child['children']), 'children' => $child['children'], 'id' => $id]);
+                                       continue;
+                               }
                                if (parse_url($id, PHP_URL_HOST) == parse_url($url, PHP_URL_HOST)) {
                                        Logger::debug('Incluced activity will be processed', ['replies' => $url, 'id' => $id]);
                                        self::processActivity($reply, $id, $child, '', Receiver::COMPLETION_REPLIES);

diff --git a/src/Protocol/ActivityPub/Queue.php b/src/Protocol/ActivityPub/Queue.php
index 53761585ffb8fbf16ce526088185243c58d8ad90..b2e8cb301b0594e1b81470fd0fa03e758174c6df 100644 (file)
--- a/src/Protocol/ActivityPub/Queue.php
+++ b/src/Protocol/ActivityPub/Queue.php
@@ -318,7 +318,7 @@ class Queue
                                self::retrial($id);
                                return false;
                        }
-                       if (!Post::exists(['uri' => $entry['in-reply-to-id']])) {
+                       if (!Processor::alreadyKnown($entry['in-reply-to-id'], '')) {
                                // This entry belongs to some other entry that need to be fetched first
                                if (Fetch::hasWorker($entry['in-reply-to-id'])) {
                                        Logger::debug('Fetching of the activity is already queued', ['id' => $entry['activity-id'], 'reply-to-id' => $entry['in-reply-to-id']]);
@@ -327,6 +327,11 @@ class Queue
                                }
                                Fetch::add($entry['in-reply-to-id']);
                                $activity = json_decode($entry['activity'], true);
+                               if (in_array($entry['in-reply-to-id'], $activity['children'] ?? [])) {
+                                       Logger::notice('reply-to-id is already in the list of children', ['id' => $entry['in-reply-to-id'], 'children' => $activity['children'], 'depth' => count($activity['children'])]);
+                                       self::retrial($id);
+                                       return false;
+                               }
                                $activity['recursion-depth'] = 0;
                                $wid = Worker::add(Worker::PRIORITY_HIGH, 'FetchMissingActivity', $entry['in-reply-to-id'], $activity, '', Receiver::COMPLETION_ASYNC);
                                Fetch::setWorkerId($entry['in-reply-to-id'], $wid);

diff --git a/src/Protocol/ActivityPub/Receiver.php b/src/Protocol/ActivityPub/Receiver.php
index bab386979ae309188e53647689c0ed5b51eb9c2c..9660827cd3251bc1e9f1220239f4d0f04d554bb3 100644 (file)
--- a/src/Protocol/ActivityPub/Receiver.php
+++ b/src/Protocol/ActivityPub/Receiver.php
@@ -757,6 +757,8 @@ class Receiver
                        $object_data['recursion-depth'] = $activity['recursion-depth'];
                }
 
+               $object_data['children'] = $activity['children'] ?? [];
+
                if (!self::routeActivities($object_data, $type, $push, true, $uid)) {
                        self::storeUnhandledActivity(true, $type, $object_data, $activity, $body, $uid, $trust_source, $push, $signer);
                        Queue::remove($object_data);