src/Worker/DBClean.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Worker;
  23
  24 use Friendica\Core\Logger;
  25 use Friendica\Core\Worker;
  26 use Friendica\Database\DBA;
  27 use Friendica\DI;
  28
  29 /**
  30  * The script is called from time to time to clean the database entries and remove orphaned data.
  31  */
  32 class DBClean {
  33         public static function execute($stage = 0) {
  34
  35                 if (!DI::config()->get('system', 'dbclean', false)) {
  36                         return;
  37                 }
  38
  39                 if ($stage == 0) {
  40                         self::forkCleanProcess();
  41                 } else {
  42                         self::removeOrphans($stage);
  43                 }
  44         }
  45
  46         /**
  47          * Fork the different DBClean processes
  48          */
  49         private static function forkCleanProcess() {
  50                 // Get the expire days for step 8 and 9
  51                 $days = DI::config()->get('system', 'dbclean-expire-days', 0);
  52
  53                 for ($i = 1; $i <= 9; $i++) {
  54                         // Execute the background script for a step when it isn't finished.
  55                         // Execute step 8 and 9 only when $days is defined.
  56                         if (!DI::config()->get('system', 'finished-dbclean-'.$i, false) && (($i < 8) || ($i > 9) || ($days > 0))) {
  57                                 Worker::add(PRIORITY_LOW, 'DBClean', $i);
  58                         }
  59                 }
  60         }
  61
  62         /**
  63          * Remove orphaned database entries
  64          *
  65          * @param integer $stage What should be deleted?
  66          *
  67          * Values for $stage:
  68          * ------------------
  69          *  1:    Old global item entries from item table without user copy.
  70          *  2:    Items without parents.
  71          *  3:    Legacy functionality (removed)
  72          *  4:    Orphaned data from notify table.
  73          *  5:    Legacy functionality (removed)
  74          *  6:    Legacy functionality (removed)
  75          *  7:    Legacy functionality (removed)
  76          *  8:    Expired threads.
  77          *  9:    Old global item entries from expired threads.
  78          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
  79          */
  80         private static function removeOrphans($stage) {
  81                 // We split the deletion in many small tasks
  82                 $limit = DI::config()->get('system', 'dbclean-expire-limit', 1000);
  83
  84                 // Get the expire days for step 8 and 9
  85                 $days = DI::config()->get('system', 'dbclean-expire-days', 0);
  86                 $days_unclaimed = DI::config()->get('system', 'dbclean-expire-unclaimed', 90);
  87
  88                 if ($days_unclaimed == 0) {
  89                         $days_unclaimed = $days;
  90                 }
  91
  92                 if ($stage == 1) {
  93                         if ($days_unclaimed <= 0) {
  94                                 return;
  95                         }
  96
  97                         $last_id = DI::config()->get('system', 'dbclean-last-id-1', 0);
  98
  99                         Logger::log("Deleting old global item entries from item table without user copy. Last ID: ".$last_id);
 100                         $r = DBA::p("SELECT `id`, `guid` FROM `item` WHERE `uid` = 0 AND
 101                                                 NOT EXISTS (SELECT `guid` FROM `item` AS `i` WHERE `item`.`guid` = `i`.`guid` AND `i`.`uid` != 0) AND
 102                                                 `received` < UTC_TIMESTAMP() - INTERVAL ? DAY AND `id` >= ?
 103                                         ORDER BY `id` LIMIT ?", $days_unclaimed, $last_id, $limit);
 104                         $count = DBA::numRows($r);
 105                         if ($count > 0) {
 106                                 Logger::log("found global item orphans: ".$count);
 107                                 while ($orphan = DBA::fetch($r)) {
 108                                         $last_id = $orphan["id"];
 109                                         Logger::info('Delete global orphan item', ['id' => $orphan['id'], 'guid' => $orphan['guid']]);
 110                                         DBA::delete('item', ['id' => $orphan["id"]]);
 111                                 }
 112                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 1, $last_id);
 113                         } else {
 114                                 Logger::log("No global item orphans found");
 115                         }
 116                         DBA::close($r);
 117                         Logger::log("Done deleting ".$count." old global item entries from item table without user copy. Last ID: ".$last_id);
 118
 119                         DI::config()->set('system', 'dbclean-last-id-1', $last_id);
 120                 } elseif ($stage == 2) {
 121                         $last_id = DI::config()->get('system', 'dbclean-last-id-2', 0);
 122
 123                         Logger::log("Deleting items without parents. Last ID: ".$last_id);
 124                         $r = DBA::p("SELECT `id`, `guid` FROM `item`
 125                                         WHERE NOT EXISTS (SELECT `id` FROM `item` AS `i` WHERE `item`.`parent` = `i`.`id`)
 126                                         AND `id` >= ? ORDER BY `id` LIMIT ?", $last_id, $limit);
 127                         $count = DBA::numRows($r);
 128                         if ($count > 0) {
 129                                 Logger::log("found item orphans without parents: ".$count);
 130                                 while ($orphan = DBA::fetch($r)) {
 131                                         $last_id = $orphan["id"];
 132                                         Logger::info('Delete orphan item', ['id' => $orphan['id'], 'guid' => $orphan['guid']]);
 133                                         DBA::delete('item', ['id' => $orphan["id"]]);
 134                                 }
 135                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 2, $last_id);
 136                         } else {
 137                                 Logger::log("No item orphans without parents found");
 138                         }
 139                         DBA::close($r);
 140                         Logger::log("Done deleting ".$count." items without parents. Last ID: ".$last_id);
 141
 142                         DI::config()->set('system', 'dbclean-last-id-2', $last_id);
 143
 144                         if ($count < $limit) {
 145                                 DI::config()->set('system', 'finished-dbclean-2', true);
 146                         }
 147                 } elseif ($stage == 3) {
 148                         // The legacy functionality had been removed
 149                         DI::config()->set('system', 'finished-dbclean-3', true);
 150                 } elseif ($stage == 4) {
 151                         DBA::p("DELETE FROM `notify` WHERE NOT `type` IN (1, 2, 16, 32, 512) AND NOT `iid` IN (SELECT `id` FROM `item`)");
 152
 153                         Logger::notice("Deleted orphaned data from notify table.");
 154                         DI::config()->set('system', 'finished-dbclean-4', true);
 155                 } elseif ($stage == 5) {
 156                         // The legacy functionality had been removed
 157                         DI::config()->set('system', 'finished-dbclean-5', true);
 158                 } elseif ($stage == 6) {
 159                         // The legacy functionality had been removed
 160                         DI::config()->set('system', 'finished-dbclean-6', true);
 161                 } elseif ($stage == 7) {
 162                         // The legacy functionality had been removed
 163                         DI::config()->set('system', 'finished-dbclean-7', true);
 164                 } elseif ($stage == 8) {
 165                         if ($days <= 0) {
 166                                 return;
 167                         }
 168
 169                         $last_id = DI::config()->get('system', 'dbclean-last-id-8', 0);
 170
 171                         Logger::log("Deleting expired threads. Last ID: ".$last_id);
 172                         $r = DBA::p("SELECT `thread`.`iid` FROM `thread`
 173                                         INNER JOIN `contact` ON `thread`.`contact-id` = `contact`.`id` AND NOT `notify_new_posts`
 174                                         WHERE `thread`.`received` < UTC_TIMESTAMP() - INTERVAL ? DAY
 175                                                 AND NOT `thread`.`mention` AND NOT `thread`.`starred`
 176                                                 AND NOT `thread`.`wall` AND NOT `thread`.`origin`
 177                                                 AND `thread`.`uid` != 0 AND `thread`.`iid` >= ?
 178                                                 AND NOT `thread`.`iid` IN (SELECT `parent` FROM `item`
 179                                                                 WHERE (`item`.`starred` OR (`item`.`resource-id` != '')
 180                                                                         OR (`item`.`file` != '') OR (`item`.`event-id` != '')
 181                                                                         OR (`item`.`attach` != '') OR `item`.`wall` OR `item`.`origin`)
 182                                                                         AND `item`.`parent` = `thread`.`iid`)
 183                                         ORDER BY `thread`.`iid` LIMIT ?", $days, $last_id, $limit);
 184                         $count = DBA::numRows($r);
 185                         if ($count > 0) {
 186                                 Logger::log("found expired threads: ".$count);
 187                                 while ($thread = DBA::fetch($r)) {
 188                                         $last_id = $thread["iid"];
 189                                         DBA::delete('item', ['parent' => $thread["iid"]]);
 190                                 }
 191                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 8, $last_id);
 192                         } else {
 193                                 Logger::log("No expired threads found");
 194                         }
 195                         DBA::close($r);
 196                         Logger::log("Done deleting ".$count." expired threads. Last ID: ".$last_id);
 197
 198                         DI::config()->set('system', 'dbclean-last-id-8', $last_id);
 199                 } elseif ($stage == 9) {
 200                         if ($days <= 0) {
 201                                 return;
 202                         }
 203
 204                         $last_id = DI::config()->get('system', 'dbclean-last-id-9', 0);
 205                         $till_id = DI::config()->get('system', 'dbclean-last-id-8', 0);
 206
 207                         Logger::log("Deleting old global item entries from expired threads from ID ".$last_id." to ID ".$till_id);
 208                         $r = DBA::p("SELECT `id`, `guid` FROM `item` WHERE `uid` = 0 AND
 209                                                 NOT EXISTS (SELECT `guid` FROM `item` AS `i` WHERE `item`.`guid` = `i`.`guid` AND `i`.`uid` != 0) AND
 210                                                 `received` < UTC_TIMESTAMP() - INTERVAL 90 DAY AND `id` >= ? AND `id` <= ?
 211                                         ORDER BY `id` LIMIT ?", $last_id, $till_id, $limit);
 212                         $count = DBA::numRows($r);
 213                         if ($count > 0) {
 214                                 Logger::log("found global item entries from expired threads: ".$count);
 215                                 while ($orphan = DBA::fetch($r)) {
 216                                         $last_id = $orphan["id"];
 217                                         Logger::info('Delete expired thread item', ['id' => $orphan['id'], 'guid' => $orphan['guid']]);
 218                                         DBA::delete('item', ['id' => $orphan["id"]]);
 219                                 }
 220                                 Worker::add(PRIORITY_MEDIUM, 'DBClean', 9, $last_id);
 221                         } else {
 222                                 Logger::log("No global item entries from expired threads");
 223                         }
 224                         DBA::close($r);
 225                         Logger::log("Done deleting ".$count." old global item entries from expired threads. Last ID: ".$last_id);
 226
 227                         DI::config()->set('system', 'dbclean-last-id-9', $last_id);
 228                 }
 229         }
 230 }