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