Remove unused contacts

diff --git a/src/Protocol/ActivityPub/Transmitter.php b/src/Protocol/ActivityPub/Transmitter.php
index 20ff0e6f3b47e1841cfc8f68b60f64cf6380a0c7..2792eab2d61479ac902ea8e7abb117ea25805a17 100644 (file)
--- a/src/Protocol/ActivityPub/Transmitter.php
+++ b/src/Protocol/ActivityPub/Transmitter.php
@@ -928,7 +928,7 @@ class Transmitter
         *
         * @param array $item
         * @param boolean $blindcopy
-        * @return void
+        * @return array
         */
        public static function getReceiversForUriId(int $uri_id, bool $blindcopy): array
        {

diff --git a/src/Worker/RemoveUnusedContacts.php b/src/Worker/RemoveUnusedContacts.php
index f1a746eb5c165892a08d75fd24329ac7ce3cc0cd..6f9d2124e1247f9bc5cf81b72cdeacc7a9ab59e8 100644 (file)
--- a/src/Worker/RemoveUnusedContacts.php
+++ b/src/Worker/RemoveUnusedContacts.php
@@ -24,8 +24,10 @@ namespace Friendica\Worker;
 use Friendica\Contact\Avatar;
 use Friendica\Core\Logger;
 use Friendica\Core\Protocol;
+use Friendica\Core\Worker;
 use Friendica\Database\DBA;
 use Friendica\Database\DBStructure;
+use Friendica\DI;
 use Friendica\Model\Contact;
 use Friendica\Model\Photo;
 use Friendica\Util\DateTimeFormat;
@@ -37,28 +39,53 @@ class RemoveUnusedContacts
 {
        public static function execute()
        {
-               $condition = ["`id` != ? AND `uid` = ? AND NOT `self` AND NOT `nurl` IN (SELECT `nurl` FROM `contact` WHERE `uid` != ?)
-                       AND (NOT `network` IN (?, ?, ?, ?, ?, ?) OR (`archive` AND `success_update` < ?))
-                       AND NOT `id` IN (SELECT `author-id` FROM `post-user` WHERE `author-id` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `owner-id` FROM `post-user` WHERE `owner-id` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `causer-id` FROM `post-user` WHERE `causer-id` IS NOT NULL AND `causer-id` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `cid` FROM `post-tag` WHERE `cid` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `contact-id` FROM `post-user` WHERE `contact-id` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `cid` FROM `user-contact` WHERE `cid` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `cid` FROM `event` WHERE `cid` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `cid` FROM `group` WHERE `cid` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `cid` FROM `delivery-queue` WHERE `cid` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `author-id` FROM `mail` WHERE `author-id` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `contact-id` FROM `mail` WHERE `contact-id` = `contact`.`id`)
-                       AND NOT `id` IN (SELECT `contact-id` FROM `group_member` WHERE `contact-id` = `contact`.`id`)
-                       AND `created` < ?",
-                       0, 0, 0, Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS, Protocol::FEED, Protocol::MAIL, Protocol::ACTIVITYPUB, DateTimeFormat::utc('now - 365 days'), DateTimeFormat::utc('now - 30 days')];
+               $loop = 0;
+               while (self::removeContacts(++$loop)) {
+                       Logger::info('In removal', ['loop' => $loop]);
+               }
+
+               Logger::notice('Remove apcontact entries with no related contact');
+               DBA::delete('apcontact', ["`uri-id` NOT IN (SELECT `uri-id` FROM `contact`) AND `updated` < ?", DateTimeFormat::utc('now - 30 days')]);
+               Logger::notice('Removed apcontact entries with no related contact', ['count' => DBA::affectedRows()]);
+
+               Logger::notice('Remove diaspora-contact entries with no related contact');
+               DBA::delete('diaspora-contact', ["`uri-id` NOT IN (SELECT `uri-id` FROM `contact`) AND `updated` < ?", DateTimeFormat::utc('now - 30 days')]);
+               Logger::notice('Removed diaspora-contact entries with no related contact', ['count' => DBA::affectedRows()]);
+       }
+
+       public static function removeContacts(int $loop): bool
+       {
+               Logger::notice('Starting removal', ['loop' => $loop]);
 
-               $total = DBA::count('contact', $condition);
-               Logger::notice('Starting removal', ['total' => $total]);
+               $condition = [
+                       "`id` != ? AND `uid` = ? AND NOT `self` AND NOT `uri-id` IN (SELECT `uri-id` FROM `contact` WHERE `uid` != ?)
+                       AND NOT EXISTS(SELECT `author-id` FROM `post-user` WHERE `author-id` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `owner-id` FROM `post-user` WHERE `owner-id` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `causer-id` FROM `post-user` WHERE `causer-id` IS NOT NULL AND `causer-id` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `cid` FROM `post-tag` WHERE `cid` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `contact-id` FROM `post-user` WHERE `contact-id` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `cid` FROM `user-contact` WHERE `cid` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `cid` FROM `event` WHERE `cid` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `cid` FROM `group` WHERE `cid` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `author-id` FROM `mail` WHERE `author-id` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `contact-id` FROM `mail` WHERE `contact-id` = `contact`.`id`)
+                       AND NOT EXISTS(SELECT `contact-id` FROM `group_member` WHERE `contact-id` = `contact`.`id`)
+                       AND `created` < ?", 0, 0, 0, DateTimeFormat::utc('now - 7 days')
+               ];
+
+               if (!DI::config()->get('remove_all_unused_contacts')) {
+                       $condition2 = [
+                               "(NOT `network` IN (?, ?, ?, ?, ?, ?) OR `archive`)",
+                               Protocol::DFRN, Protocol::DIASPORA, Protocol::OSTATUS, Protocol::FEED, Protocol::MAIL, Protocol::ACTIVITYPUB
+                       ];
+                       
+                       $condition = DBA::mergeConditions($condition2, $condition);
+               }
+
+               $contacts = DBA::select('contact', ['id', 'uid', 'photo', 'thumb', 'micro'], $condition, ['limit' => 1000]);
                $count = 0;
-               $contacts = DBA::select('contact', ['id', 'uid', 'photo', 'thumb', 'micro'], $condition);
                while ($contact = DBA::fetch($contacts)) {
+                       ++$count;
                        Photo::delete(['uid' => $contact['uid'], 'contact-id' => $contact['id']]);
                        Avatar::deleteCache($contact);
 
@@ -87,19 +114,9 @@ class RemoveUnusedContacts
                        DBA::delete('post-thread-user', ['causer-id' => $contact['id']]);
 
                        Contact::deleteById($contact['id']);
-                       if ((++$count % 1000) == 0) {
-                               Logger::info('In removal', ['count' => $count, 'total' => $total]);
-                       }
                }
                DBA::close($contacts);
-               Logger::notice('Removal done', ['count' => $count, 'total' => $total]);
-               
-               Logger::notice('Remove apcontact entries with no related contact');
-               DBA::delete('apcontact', ["`uri-id` NOT IN (SELECT `uri-id` FROM `contact`) AND `updated` < ?", DateTimeFormat::utc('now - 30 days')]);
-               Logger::notice('Removed apcontact entries with no related contact', ['count' => DBA::affectedRows()]);
-
-               Logger::notice('Remove diaspora-contact entries with no related contact');
-               DBA::delete('diaspora-contact', ["`uri-id` NOT IN (SELECT `uri-id` FROM `contact`) AND `updated` < ?", DateTimeFormat::utc('now - 30 days')]);
-               Logger::notice('Removed diaspora-contact entries with no related contact', ['count' => DBA::affectedRows()]);
+               Logger::notice('Removal done', ['count' => $count]);
+               return ($count == 1000 && Worker::isInMaintenanceWindow());
        }
 }

diff --git a/static/defaults.config.php b/static/defaults.config.php
index 6bdc9c53b55ce1039f98955cc96010d8efcac7e1..49f8d6e5fd714518d28cdcb596f85dcbf2f10536 100644 (file)
--- a/static/defaults.config.php
+++ b/static/defaults.config.php
@@ -514,6 +514,12 @@ return [
                // Redistribute incoming activities via ActivityPub
                'redistribute_activities' => true,
 
+               // remove_all_unused_contacts (Boolean)
+               // Remove all unused contacts.
+               // Per default only archived contacts are removed from federated services.
+               // Unused contacts from connector networks will be removed in any case.
+               'remove_all_unused_contacts' => false,
+
                // session_handler (database|cache|native)
                // Whether to use Cache to store session data or to use PHP native session storage.
                'session_handler' => 'database',