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