include/cron.php

   1 <?php
   2 if (!file_exists("boot.php") AND (sizeof($_SERVER["argv"]) != 0)) {
   3         $directory = dirname($_SERVER["argv"][0]);
   4
   5         if (substr($directory, 0, 1) != "/")
   6                 $directory = $_SERVER["PWD"]."/".$directory;
   7
   8         $directory = realpath($directory."/..");
   9
  10         chdir($directory);
  11 }
  12
  13 require_once("boot.php");
  14 require_once("include/photos.php");
  15 require_once("include/user.php");
  16
  17
  18 function cron_run(&$argv, &$argc){
  19         global $a, $db;
  20
  21         if(is_null($a)) {
  22                 $a = new App;
  23         }
  24
  25         if(is_null($db)) {
  26                 @include(".htconfig.php");
  27                 require_once("include/dba.php");
  28                 $db = new dba($db_host, $db_user, $db_pass, $db_data);
  29                 unset($db_host, $db_user, $db_pass, $db_data);
  30         };
  31
  32         require_once('include/session.php');
  33         require_once('include/datetime.php');
  34         require_once('include/items.php');
  35         require_once('include/Contact.php');
  36         require_once('include/email.php');
  37         require_once('include/socgraph.php');
  38         require_once('mod/nodeinfo.php');
  39         require_once('include/post_update.php');
  40
  41         load_config('config');
  42         load_config('system');
  43
  44         // Don't check this stuff if the function is called by the poller
  45         if (App::callstack() != "poller_run") {
  46                 if ($a->maxload_reached())
  47                         return;
  48                 if (App::is_already_running('cron', 'include/cron.php', 540))
  49                         return;
  50         }
  51
  52         $last = get_config('system','last_cron');
  53
  54         $poll_interval = intval(get_config('system','cron_interval'));
  55         if(! $poll_interval)
  56                 $poll_interval = 10;
  57
  58         if($last) {
  59                 $next = $last + ($poll_interval * 60);
  60                 if($next > time()) {
  61                         logger('cron intervall not reached');
  62                         return;
  63                 }
  64         }
  65
  66         $a->set_baseurl(get_config('system','url'));
  67
  68         load_hooks();
  69
  70         logger('cron: start');
  71
  72         // run queue delivery process in the background
  73
  74         proc_run(PRIORITY_NEGLIGIBLE, "include/queue.php");
  75
  76         // run the process to discover global contacts in the background
  77
  78         proc_run(PRIORITY_LOW, "include/discover_poco.php");
  79
  80         // run the process to update locally stored global contacts in the background
  81
  82         proc_run(PRIORITY_LOW, "include/discover_poco.php", "checkcontact");
  83
  84         // Expire and remove user entries
  85         cron_expire_and_remove_users();
  86
  87         // If the worker is active, split the jobs in several sub processes
  88         if (get_config("system", "worker")) {
  89                 // Check OStatus conversations
  90                 proc_run(PRIORITY_MEDIUM, "include/cronjobs.php", "ostatus_mentions");
  91
  92                 // Check every conversation
  93                 proc_run(PRIORITY_MEDIUM, "include/cronjobs.php", "ostatus_conversations");
  94
  95                 // Call possible post update functions
  96                 proc_run(PRIORITY_LOW, "include/cronjobs.php", "post_update");
  97
  98                 // update nodeinfo data
  99                 proc_run(PRIORITY_LOW, "include/cronjobs.php", "nodeinfo");
 100         } else {
 101                 // Check OStatus conversations
 102                 // Check only conversations with mentions (for a longer time)
 103                 ostatus::check_conversations(true);
 104
 105                 // Check every conversation
 106                 ostatus::check_conversations(false);
 107
 108                 // Call possible post update functions
 109                 // see include/post_update.php for more details
 110                 post_update();
 111
 112                 // update nodeinfo data
 113                 nodeinfo_cron();
 114         }
 115
 116         // once daily run birthday_updates and then expire in background
 117
 118         $d1 = get_config('system','last_expire_day');
 119         $d2 = intval(datetime_convert('UTC','UTC','now','d'));
 120
 121         if($d2 != intval($d1)) {
 122
 123                 update_contact_birthdays();
 124
 125                 proc_run(PRIORITY_LOW, "include/discover_poco.php", "suggestions");
 126
 127                 set_config('system','last_expire_day',$d2);
 128
 129                 proc_run(PRIORITY_LOW, 'include/expire.php');
 130
 131                 proc_run(PRIORITY_MEDIUM, 'include/dbclean.php');
 132
 133                 cron_update_photo_albums();
 134         }
 135
 136         // Clear cache entries
 137         cron_clear_cache($a);
 138
 139         // Repair missing Diaspora values in contacts
 140         cron_repair_diaspora($a);
 141
 142         // Repair entries in the database
 143         cron_repair_database();
 144
 145         // Poll contacts
 146         cron_poll_contacts($argc, $argv);
 147
 148         logger('cron: end');
 149
 150         set_config('system','last_cron', time());
 151
 152         return;
 153 }
 154
 155 /**
 156  * @brief Update the cached values for the number of photo albums per user
 157  */
 158 function cron_update_photo_albums() {
 159         $r = q("SELECT `uid` FROM `user` WHERE NOT `account_expired` AND NOT `account_removed`");
 160         if (!dbm::is_result($r)) {
 161                 return;
 162         }
 163
 164         foreach ($r AS $user) {
 165                 photo_albums($user['uid'], true);
 166         }
 167 }
 168
 169 /**
 170  * @brief Expire and remove user entries
 171  */
 172 function cron_expire_and_remove_users() {
 173         // expire any expired accounts
 174         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
 175                 AND `account_expires_on` != '0000-00-00 00:00:00'
 176                 AND `account_expires_on` < UTC_TIMESTAMP() ");
 177
 178         // delete user and contact records for recently removed accounts
 179         $r = q("SELECT * FROM `user` WHERE `account_removed` AND `account_expires_on` < UTC_TIMESTAMP() - INTERVAL 3 DAY");
 180         if ($r) {
 181                 foreach($r as $user) {
 182                         q("DELETE FROM `contact` WHERE `uid` = %d", intval($user['uid']));
 183                         q("DELETE FROM `user` WHERE `uid` = %d", intval($user['uid']));
 184                 }
 185         }
 186 }
 187
 188 /**
 189  * @brief Poll contacts for unreceived messages
 190  *
 191  * @param Integer $argc Number of command line arguments
 192  * @param Array $argv Array of command line arguments
 193  */
 194 function cron_poll_contacts($argc, $argv) {
 195         $manual_id  = 0;
 196         $generation = 0;
 197         $force      = false;
 198         $restart    = false;
 199
 200         if (($argc > 1) && ($argv[1] == 'force'))
 201                 $force = true;
 202
 203         if (($argc > 1) && ($argv[1] == 'restart')) {
 204                 $restart = true;
 205                 $generation = intval($argv[2]);
 206                 if (!$generation)
 207                         killme();
 208         }
 209
 210         if (($argc > 1) && intval($argv[1])) {
 211                 $manual_id = intval($argv[1]);
 212                 $force     = true;
 213         }
 214
 215         $interval = intval(get_config('system','poll_interval'));
 216         if (!$interval)
 217                 $interval = ((get_config('system','delivery_interval') === false) ? 3 : intval(get_config('system','delivery_interval')));
 218
 219         // If we are using the worker we don't need a delivery interval
 220         if (get_config("system", "worker"))
 221                 $interval = false;
 222
 223         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
 224
 225         reload_plugins();
 226
 227         $d = datetime_convert();
 228
 229         // Only poll from those with suitable relationships,
 230         // and which have a polling address and ignore Diaspora since
 231         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 232
 233         $abandon_days = intval(get_config('system','account_abandon_days'));
 234         if($abandon_days < 1)
 235                 $abandon_days = 0;
 236
 237         $abandon_sql = (($abandon_days)
 238                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
 239                 : ''
 240         );
 241
 242         $contacts = q("SELECT `contact`.`id` FROM `user`
 243                         STRAIGHT_JOIN `contact`
 244                         ON `contact`.`uid` = `user`.`uid` AND `contact`.`rel` IN (%d, %d) AND `contact`.`poll` != ''
 245                                 AND `contact`.`network` IN ('%s', '%s', '%s', '%s', '%s', '%s') $sql_extra
 246                                 AND NOT `contact`.`self` AND NOT `contact`.`blocked` AND NOT `contact`.`readonly`
 247                                 AND NOT `contact`.`archive`
 248                         WHERE NOT `user`.`account_expired` AND NOT `user`.`account_removed` $abandon_sql ORDER BY RAND()",
 249                 intval(CONTACT_IS_SHARING),
 250                 intval(CONTACT_IS_FRIEND),
 251                 dbesc(NETWORK_DFRN),
 252                 dbesc(NETWORK_ZOT),
 253                 dbesc(NETWORK_OSTATUS),
 254                 dbesc(NETWORK_FEED),
 255                 dbesc(NETWORK_MAIL),
 256                 dbesc(NETWORK_MAIL2)
 257         );
 258
 259         if (!count($contacts)) {
 260                 return;
 261         }
 262
 263         foreach ($contacts as $c) {
 264
 265                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
 266                         intval($c['id'])
 267                 );
 268
 269                 if (!dbm::is_result($res)) {
 270                         continue;
 271                 }
 272
 273                 foreach($res as $contact) {
 274
 275                         $xml = false;
 276
 277                         if($manual_id)
 278                                 $contact['last-update'] = '0000-00-00 00:00:00';
 279
 280                         if(in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS)))
 281                                 $contact['priority'] = 2;
 282
 283                         if($contact['subhub'] AND in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS))) {
 284                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
 285                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 286                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
 287                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 288
 289                                 $poll_interval = get_config('system','pushpoll_frequency');
 290                                 $contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
 291                         }
 292
 293                         if($contact['priority'] AND !$force) {
 294
 295                                 $update     = false;
 296
 297                                 $t = $contact['last-update'];
 298
 299                                 /**
 300                                  * Based on $contact['priority'], should we poll this site now? Or later?
 301                                  */
 302
 303                                 switch ($contact['priority']) {
 304                                         case 5:
 305                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 306                                                         $update = true;
 307                                                 break;
 308                                         case 4:
 309                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 310                                                         $update = true;
 311                                                 break;
 312                                         case 3:
 313                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
 314                                                         $update = true;
 315                                                 break;
 316                                         case 2:
 317                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
 318                                                         $update = true;
 319                                                 break;
 320                                         case 1:
 321                                         default:
 322                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
 323                                                         $update = true;
 324                                                 break;
 325                                 }
 326                                 if (!$update)
 327                                         continue;
 328                         }
 329
 330                         logger("Polling ".$contact["network"]." ".$contact["id"]." ".$contact["nick"]." ".$contact["name"]);
 331
 332                         if (($contact['network'] == NETWORK_FEED) AND ($contact['priority'] <= 3)) {
 333                                 proc_run(PRIORITY_MEDIUM, 'include/onepoll.php', $contact['id']);
 334                         } else {
 335                                 proc_run(PRIORITY_LOW, 'include/onepoll.php', $contact['id']);
 336                         }
 337
 338                         if($interval)
 339                                 @time_sleep_until(microtime(true) + (float) $interval);
 340                 }
 341         }
 342 }
 343
 344 /**
 345  * @brief Clear cache entries
 346  *
 347  * @param App $a
 348  */
 349 function cron_clear_cache(App $a) {
 350
 351         $last = get_config('system','cache_last_cleared');
 352
 353         if($last) {
 354                 $next = $last + (3600); // Once per hour
 355                 $clear_cache = ($next <= time());
 356         } else
 357                 $clear_cache = true;
 358
 359         if (!$clear_cache)
 360                 return;
 361
 362         // clear old cache
 363         Cache::clear();
 364
 365         // clear old item cache files
 366         clear_cache();
 367
 368         // clear cache for photos
 369         clear_cache($a->get_basepath(), $a->get_basepath()."/photo");
 370
 371         // clear smarty cache
 372         clear_cache($a->get_basepath()."/view/smarty3/compiled", $a->get_basepath()."/view/smarty3/compiled");
 373
 374         // clear cache for image proxy
 375         if (!get_config("system", "proxy_disabled")) {
 376                 clear_cache($a->get_basepath(), $a->get_basepath()."/proxy");
 377
 378                 $cachetime = get_config('system','proxy_cache_time');
 379                 if (!$cachetime) $cachetime = PROXY_DEFAULT_TIME;
 380
 381                 q('DELETE FROM `photo` WHERE `uid` = 0 AND `resource-id` LIKE "pic:%%" AND `created` < NOW() - INTERVAL %d SECOND', $cachetime);
 382         }
 383
 384         // Delete the cached OEmbed entries that are older than one year
 385         q("DELETE FROM `oembed` WHERE `created` < NOW() - INTERVAL 3 MONTH");
 386
 387         // Delete the cached "parse_url" entries that are older than one year
 388         q("DELETE FROM `parsed_url` WHERE `created` < NOW() - INTERVAL 3 MONTH");
 389
 390         // Maximum table size in megabyte
 391         $max_tablesize = intval(get_config('system','optimize_max_tablesize')) * 1000000;
 392         if ($max_tablesize == 0)
 393                 $max_tablesize = 100 * 1000000; // Default are 100 MB
 394
 395         if ($max_tablesize > 0) {
 396                 // Minimum fragmentation level in percent
 397                 $fragmentation_level = intval(get_config('system','optimize_fragmentation')) / 100;
 398                 if ($fragmentation_level == 0)
 399                         $fragmentation_level = 0.3; // Default value is 30%
 400
 401                 // Optimize some tables that need to be optimized
 402                 $r = q("SHOW TABLE STATUS");
 403                 foreach($r as $table) {
 404
 405                         // Don't optimize tables that are too large
 406                         if ($table["Data_length"] > $max_tablesize)
 407                                 continue;
 408
 409                         // Don't optimize empty tables
 410                         if ($table["Data_length"] == 0)
 411                                 continue;
 412
 413                         // Calculate fragmentation
 414                         $fragmentation = $table["Data_free"] / ($table["Data_length"] + $table["Index_length"]);
 415
 416                         logger("Table ".$table["Name"]." - Fragmentation level: ".round($fragmentation * 100, 2), LOGGER_DEBUG);
 417
 418                         // Don't optimize tables that needn't to be optimized
 419                         if ($fragmentation < $fragmentation_level)
 420                                 continue;
 421
 422                         // So optimize it
 423                         logger("Optimize Table ".$table["Name"], LOGGER_DEBUG);
 424                         q("OPTIMIZE TABLE `%s`", dbesc($table["Name"]));
 425                 }
 426         }
 427
 428         set_config('system','cache_last_cleared', time());
 429 }
 430
 431 /**
 432  * @brief Repair missing values in Diaspora contacts
 433  *
 434  * @param App $a
 435  */
 436 function cron_repair_diaspora(App $a) {
 437         $r = q("SELECT `id`, `url` FROM `contact`
 438                 WHERE `network` = '%s' AND (`batch` = '' OR `notify` = '' OR `poll` = '' OR pubkey = '')
 439                         ORDER BY RAND() LIMIT 50", dbesc(NETWORK_DIASPORA));
 440         if (dbm::is_result($r)) {
 441                 foreach ($r AS $contact) {
 442                         if (poco_reachable($contact["url"])) {
 443                                 $data = probe_url($contact["url"]);
 444                                 if ($data["network"] == NETWORK_DIASPORA) {
 445                                         logger("Repair contact ".$contact["id"]." ".$contact["url"], LOGGER_DEBUG);
 446                                         q("UPDATE `contact` SET `batch` = '%s', `notify` = '%s', `poll` = '%s', pubkey = '%s' WHERE `id` = %d",
 447                                                 dbesc($data["batch"]), dbesc($data["notify"]), dbesc($data["poll"]), dbesc($data["pubkey"]),
 448                                                 intval($contact["id"]));
 449                                 }
 450                         }
 451                 }
 452         }
 453 }
 454
 455 /**
 456  * @brief Do some repairs in database entries
 457  *
 458  */
 459 function cron_repair_database() {
 460
 461         // Sometimes there seem to be issues where the "self" contact vanishes.
 462         // We haven't found the origin of the problem by now.
 463         $r = q("SELECT `uid` FROM `user` WHERE NOT EXISTS (SELECT `uid` FROM `contact` WHERE `contact`.`uid` = `user`.`uid` AND `contact`.`self`)");
 464         if (dbm::is_result($r)) {
 465                 foreach ($r AS $user) {
 466                         logger('Create missing self contact for user '.$user['uid']);
 467                         user_create_self_contact($user['uid']);
 468                 }
 469         }
 470
 471         // Set the parent if it wasn't set. (Shouldn't happen - but does sometimes)
 472         // This call is very "cheap" so we can do it at any time without a problem
 473         q("UPDATE `item` INNER JOIN `item` AS `parent` ON `parent`.`uri` = `item`.`parent-uri` AND `parent`.`uid` = `item`.`uid` SET `item`.`parent` = `parent`.`id` WHERE `item`.`parent` = 0");
 474
 475         // There was an issue where the nick vanishes from the contact table
 476         q("UPDATE `contact` INNER JOIN `user` ON `contact`.`uid` = `user`.`uid` SET `nick` = `nickname` WHERE `self` AND `nick`=''");
 477
 478         // Update the global contacts for local users
 479         $r = q("SELECT `uid` FROM `user` WHERE `verified` AND NOT `blocked` AND NOT `account_removed` AND NOT `account_expired`");
 480         if (dbm::is_result($r))
 481                 foreach ($r AS $user)
 482                         update_gcontact_for_user($user["uid"]);
 483
 484         /// @todo
 485         /// - remove thread entries without item
 486         /// - remove sign entries without item
 487         /// - remove children when parent got lost
 488         /// - set contact-id in item when not present
 489 }
 490
 491 if (array_search(__file__,get_included_files())===0){
 492         cron_run($_SERVER["argv"],$_SERVER["argc"]);
 493         killme();
 494 }