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