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                 proc_run(PRIORITY_LOW,'include/dbclean.php');
 130         }
 131
 132         // Clear cache entries
 133         cron_clear_cache($a);
 134
 135         // Repair missing Diaspora values in contacts
 136         cron_repair_diaspora($a);
 137
 138         // Repair entries in the database
 139         cron_repair_database();
 140
 141         // Poll contacts
 142         cron_poll_contacts($argc, $argv);
 143
 144         logger('cron: end');
 145
 146         set_config('system','last_cron', time());
 147
 148         return;
 149 }
 150
 151 /**
 152  * @brief Expire and remove user entries
 153  */
 154 function cron_expire_and_remove_users() {
 155         // expire any expired accounts
 156         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
 157                 AND `account_expires_on` != '0000-00-00 00:00:00'
 158                 AND `account_expires_on` < UTC_TIMESTAMP() ");
 159
 160         // delete user and contact records for recently removed accounts
 161         $r = q("SELECT * FROM `user` WHERE `account_removed` AND `account_expires_on` < UTC_TIMESTAMP() - INTERVAL 3 DAY");
 162         if ($r) {
 163                 foreach($r as $user) {
 164                         q("DELETE FROM `contact` WHERE `uid` = %d", intval($user['uid']));
 165                         q("DELETE FROM `user` WHERE `uid` = %d", intval($user['uid']));
 166                 }
 167         }
 168 }
 169
 170 /**
 171  * @brief Poll contacts for unreceived messages
 172  *
 173  * @param Integer $argc Number of command line arguments
 174  * @param Array $argv Array of command line arguments
 175  */
 176 function cron_poll_contacts($argc, $argv) {
 177         $manual_id  = 0;
 178         $generation = 0;
 179         $force      = false;
 180         $restart    = false;
 181
 182         if (($argc > 1) && ($argv[1] == 'force'))
 183                 $force = true;
 184
 185         if (($argc > 1) && ($argv[1] == 'restart')) {
 186                 $restart = true;
 187                 $generation = intval($argv[2]);
 188                 if (!$generation)
 189                         killme();
 190         }
 191
 192         if (($argc > 1) && intval($argv[1])) {
 193                 $manual_id = intval($argv[1]);
 194                 $force     = true;
 195         }
 196
 197         $interval = intval(get_config('system','poll_interval'));
 198         if (!$interval)
 199                 $interval = ((get_config('system','delivery_interval') === false) ? 3 : intval(get_config('system','delivery_interval')));
 200
 201         // If we are using the worker we don't need a delivery interval
 202         if (get_config("system", "worker"))
 203                 $interval = false;
 204
 205         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
 206
 207         reload_plugins();
 208
 209         $d = datetime_convert();
 210
 211         // Only poll from those with suitable relationships,
 212         // and which have a polling address and ignore Diaspora since
 213         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 214
 215         $abandon_days = intval(get_config('system','account_abandon_days'));
 216         if($abandon_days < 1)
 217                 $abandon_days = 0;
 218
 219         $abandon_sql = (($abandon_days)
 220                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
 221                 : ''
 222         );
 223
 224         $contacts = q("SELECT `contact`.`id` FROM `contact` INNER JOIN `user` ON `user`.`uid` = `contact`.`uid`
 225                 WHERE `rel` IN (%d, %d) AND `poll` != '' AND `network` IN ('%s', '%s', '%s', '%s', '%s', '%s')
 226                 $sql_extra
 227                 AND NOT `self` AND NOT `contact`.`blocked` AND NOT `contact`.`readonly` AND NOT `contact`.`archive`
 228                 AND NOT `user`.`account_expired` AND NOT `user`.`account_removed` $abandon_sql ORDER BY RAND()",
 229                 intval(CONTACT_IS_SHARING),
 230                 intval(CONTACT_IS_FRIEND),
 231                 dbesc(NETWORK_DFRN),
 232                 dbesc(NETWORK_ZOT),
 233                 dbesc(NETWORK_OSTATUS),
 234                 dbesc(NETWORK_FEED),
 235                 dbesc(NETWORK_MAIL),
 236                 dbesc(NETWORK_MAIL2)
 237         );
 238
 239         if (!count($contacts)) {
 240                 return;
 241         }
 242
 243         foreach ($contacts as $c) {
 244
 245                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
 246                         intval($c['id'])
 247                 );
 248
 249                 if((! $res) || (! count($res)))
 250                         continue;
 251
 252                 foreach($res as $contact) {
 253
 254                         $xml = false;
 255
 256                         if($manual_id)
 257                                 $contact['last-update'] = '0000-00-00 00:00:00';
 258
 259                         if(in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS)))
 260                                 $contact['priority'] = 2;
 261
 262                         if($contact['subhub'] AND in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS))) {
 263                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
 264                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 265                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
 266                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 267
 268                                 $poll_interval = get_config('system','pushpoll_frequency');
 269                                 $contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
 270                         }
 271
 272                         if($contact['priority'] AND !$force) {
 273
 274                                 $update     = false;
 275
 276                                 $t = $contact['last-update'];
 277
 278                                 /**
 279                                  * Based on $contact['priority'], should we poll this site now? Or later?
 280                                  */
 281
 282                                 switch ($contact['priority']) {
 283                                         case 5:
 284                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 285                                                         $update = true;
 286                                                 break;
 287                                         case 4:
 288                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 289                                                         $update = true;
 290                                                 break;
 291                                         case 3:
 292                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
 293                                                         $update = true;
 294                                                 break;
 295                                         case 2:
 296                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
 297                                                         $update = true;
 298                                                 break;
 299                                         case 1:
 300                                         default:
 301                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
 302                                                         $update = true;
 303                                                 break;
 304                                 }
 305                                 if (!$update)
 306                                         continue;
 307                         }
 308
 309                         logger("Polling ".$contact["network"]." ".$contact["id"]." ".$contact["nick"]." ".$contact["name"]);
 310
 311                         proc_run(PRIORITY_MEDIUM,'include/onepoll.php',$contact['id']);
 312
 313                         if($interval)
 314                                 @time_sleep_until(microtime(true) + (float) $interval);
 315                 }
 316         }
 317 }
 318
 319 /**
 320  * @brief Clear cache entries
 321  *
 322  * @param App $a
 323  */
 324 function cron_clear_cache(&$a) {
 325
 326         $last = get_config('system','cache_last_cleared');
 327
 328         if($last) {
 329                 $next = $last + (3600); // Once per hour
 330                 $clear_cache = ($next <= time());
 331         } else
 332                 $clear_cache = true;
 333
 334         if (!$clear_cache)
 335                 return;
 336
 337         // clear old cache
 338         Cache::clear();
 339
 340         // clear old item cache files
 341         clear_cache();
 342
 343         // clear cache for photos
 344         clear_cache($a->get_basepath(), $a->get_basepath()."/photo");
 345
 346         // clear smarty cache
 347         clear_cache($a->get_basepath()."/view/smarty3/compiled", $a->get_basepath()."/view/smarty3/compiled");
 348
 349         // clear cache for image proxy
 350         if (!get_config("system", "proxy_disabled")) {
 351                 clear_cache($a->get_basepath(), $a->get_basepath()."/proxy");
 352
 353                 $cachetime = get_config('system','proxy_cache_time');
 354                 if (!$cachetime) $cachetime = PROXY_DEFAULT_TIME;
 355
 356                 q('DELETE FROM `photo` WHERE `uid` = 0 AND `resource-id` LIKE "pic:%%" AND `created` < NOW() - INTERVAL %d SECOND', $cachetime);
 357         }
 358
 359         // Delete the cached OEmbed entries that are older than one year
 360         q("DELETE FROM `oembed` WHERE `created` < NOW() - INTERVAL 3 MONTH");
 361
 362         // Delete the cached "parse_url" entries that are older than one year
 363         q("DELETE FROM `parsed_url` WHERE `created` < NOW() - INTERVAL 3 MONTH");
 364
 365         // Maximum table size in megabyte
 366         $max_tablesize = intval(get_config('system','optimize_max_tablesize')) * 1000000;
 367         if ($max_tablesize == 0)
 368                 $max_tablesize = 100 * 1000000; // Default are 100 MB
 369
 370         if ($max_tablesize > 0) {
 371                 // Minimum fragmentation level in percent
 372                 $fragmentation_level = intval(get_config('system','optimize_fragmentation')) / 100;
 373                 if ($fragmentation_level == 0)
 374                         $fragmentation_level = 0.3; // Default value is 30%
 375
 376                 // Optimize some tables that need to be optimized
 377                 $r = q("SHOW TABLE STATUS");
 378                 foreach($r as $table) {
 379
 380                         // Don't optimize tables that are too large
 381                         if ($table["Data_length"] > $max_tablesize)
 382                                 continue;
 383
 384                         // Don't optimize empty tables
 385                         if ($table["Data_length"] == 0)
 386                                 continue;
 387
 388                         // Calculate fragmentation
 389                         $fragmentation = $table["Data_free"] / ($table["Data_length"] + $table["Index_length"]);
 390
 391                         logger("Table ".$table["Name"]." - Fragmentation level: ".round($fragmentation * 100, 2), LOGGER_DEBUG);
 392
 393                         // Don't optimize tables that needn't to be optimized
 394                         if ($fragmentation < $fragmentation_level)
 395                                 continue;
 396
 397                         // So optimize it
 398                         logger("Optimize Table ".$table["Name"], LOGGER_DEBUG);
 399                         q("OPTIMIZE TABLE `%s`", dbesc($table["Name"]));
 400                 }
 401         }
 402
 403         set_config('system','cache_last_cleared', time());
 404 }
 405
 406 /**
 407  * @brief Repair missing values in Diaspora contacts
 408  *
 409  * @param App $a
 410  */
 411 function cron_repair_diaspora(&$a) {
 412         $r = q("SELECT `id`, `url` FROM `contact`
 413                 WHERE `network` = '%s' AND (`batch` = '' OR `notify` = '' OR `poll` = '' OR pubkey = '')
 414                         ORDER BY RAND() LIMIT 50", dbesc(NETWORK_DIASPORA));
 415         if ($r) {
 416                 foreach ($r AS $contact) {
 417                         if (poco_reachable($contact["url"])) {
 418                                 $data = probe_url($contact["url"]);
 419                                 if ($data["network"] == NETWORK_DIASPORA) {
 420                                         logger("Repair contact ".$contact["id"]." ".$contact["url"], LOGGER_DEBUG);
 421                                         q("UPDATE `contact` SET `batch` = '%s', `notify` = '%s', `poll` = '%s', pubkey = '%s' WHERE `id` = %d",
 422                                                 dbesc($data["batch"]), dbesc($data["notify"]), dbesc($data["poll"]), dbesc($data["pubkey"]),
 423                                                 intval($contact["id"]));
 424                                 }
 425                         }
 426                 }
 427         }
 428 }
 429
 430 /**
 431  * @brief Do some repairs in database entries
 432  *
 433  */
 434 function cron_repair_database() {
 435
 436         // Set the parent if it wasn't set. (Shouldn't happen - but does sometimes)
 437         // This call is very "cheap" so we can do it at any time without a problem
 438         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");
 439
 440         // There was an issue where the nick vanishes from the contact table
 441         q("UPDATE `contact` INNER JOIN `user` ON `contact`.`uid` = `user`.`uid` SET `nick` = `nickname` WHERE `self` AND `nick`=''");
 442
 443         // Update the global contacts for local users
 444         $r = q("SELECT `uid` FROM `user` WHERE `verified` AND NOT `blocked` AND NOT `account_removed` AND NOT `account_expired`");
 445         if ($r)
 446                 foreach ($r AS $user)
 447                         update_gcontact_for_user($user["uid"]);
 448
 449         /// @todo
 450         /// - remove thread entries without item
 451         /// - remove sign entries without item
 452         /// - remove children when parent got lost
 453         /// - set contact-id in item when not present
 454 }
 455
 456 if (array_search(__file__,get_included_files())===0){
 457         cron_run($_SERVER["argv"],$_SERVER["argc"]);
 458         killme();
 459 }