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
  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('php',"include/queue.php");
  73
  74         // run the process to discover global contacts in the background
  75
  76         proc_run('php',"include/discover_poco.php");
  77
  78         // run the process to update locally stored global contacts in the background
  79
  80         proc_run('php',"include/discover_poco.php", "checkcontact");
  81
  82         // expire any expired accounts
  83
  84         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
  85                 AND `account_expires_on` != '0000-00-00 00:00:00'
  86                 AND `account_expires_on` < UTC_TIMESTAMP() ");
  87
  88         // delete user and contact records for recently removed accounts
  89
  90         $r = q("SELECT * FROM `user` WHERE `account_removed` = 1 AND `account_expires_on` < UTC_TIMESTAMP() - INTERVAL 3 DAY");
  91         if ($r) {
  92                 foreach($r as $user) {
  93                         q("DELETE FROM `contact` WHERE `uid` = %d", intval($user['uid']));
  94                         q("DELETE FROM `user` WHERE `uid` = %d", intval($user['uid']));
  95                 }
  96         }
  97
  98         $abandon_days = intval(get_config('system','account_abandon_days'));
  99         if($abandon_days < 1)
 100                 $abandon_days = 0;
 101
 102         // Check OStatus conversations
 103         // Check only conversations with mentions (for a longer time)
 104         ostatus::check_conversations(true);
 105
 106         // Check every conversation
 107         ostatus::check_conversations(false);
 108
 109         // Set the gcontact-id in the item table if missing
 110         item_set_gcontact();
 111
 112         // update nodeinfo data
 113         nodeinfo_cron();
 114
 115         /// @TODO Regenerate usage statistics
 116         // q("ANALYZE TABLE `item`");
 117
 118         // once daily run birthday_updates and then expire in background
 119
 120         $d1 = get_config('system','last_expire_day');
 121         $d2 = intval(datetime_convert('UTC','UTC','now','d'));
 122
 123         if($d2 != intval($d1)) {
 124
 125                 update_contact_birthdays();
 126
 127                 proc_run('php',"include/discover_poco.php", "suggestions");
 128
 129                 set_config('system','last_expire_day',$d2);
 130
 131                 proc_run('php','include/expire.php');
 132         }
 133
 134         // Clear cache entries
 135         cron_clear_cache($a);
 136
 137         // Repair missing Diaspora values in contacts
 138         cron_repair_diaspora($a);
 139
 140         // Repair entries in the database
 141         cron_repair_database();
 142
 143         $manual_id  = 0;
 144         $generation = 0;
 145         $force      = false;
 146         $restart    = false;
 147
 148         if(($argc > 1) && ($argv[1] == 'force'))
 149                 $force = true;
 150
 151         if(($argc > 1) && ($argv[1] == 'restart')) {
 152                 $restart = true;
 153                 $generation = intval($argv[2]);
 154                 if(! $generation)
 155                         killme();
 156         }
 157
 158         if(($argc > 1) && intval($argv[1])) {
 159                 $manual_id = intval($argv[1]);
 160                 $force     = true;
 161         }
 162
 163         $interval = intval(get_config('system','poll_interval'));
 164         if(! $interval)
 165                 $interval = ((get_config('system','delivery_interval') === false) ? 3 : intval(get_config('system','delivery_interval')));
 166
 167         // If we are using the worker we don't need a delivery interval
 168         if (get_config("system", "worker"))
 169                 $interval = false;
 170
 171         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
 172
 173         reload_plugins();
 174
 175         $d = datetime_convert();
 176
 177         // Only poll from those with suitable relationships,
 178         // and which have a polling address and ignore Diaspora since
 179         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 180
 181         $abandon_sql = (($abandon_days)
 182                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
 183                 : ''
 184         );
 185
 186         $contacts = q("SELECT `contact`.`id` FROM `contact` INNER JOIN `user` ON `user`.`uid` = `contact`.`uid`
 187                 WHERE `rel` IN (%d, %d) AND `poll` != '' AND `network` IN ('%s', '%s', '%s', '%s', '%s', '%s')
 188                 $sql_extra
 189                 AND NOT `self` AND NOT `contact`.`blocked` AND NOT `contact`.`readonly` AND NOT `contact`.`archive`
 190                 AND NOT `user`.`account_expired` AND NOT `user`.`account_removed` $abandon_sql ORDER BY RAND()",
 191                 intval(CONTACT_IS_SHARING),
 192                 intval(CONTACT_IS_FRIEND),
 193                 dbesc(NETWORK_DFRN),
 194                 dbesc(NETWORK_ZOT),
 195                 dbesc(NETWORK_OSTATUS),
 196                 dbesc(NETWORK_FEED),
 197                 dbesc(NETWORK_MAIL),
 198                 dbesc(NETWORK_MAIL2)
 199         );
 200
 201         if(! count($contacts)) {
 202                 return;
 203         }
 204
 205         foreach($contacts as $c) {
 206
 207                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
 208                         intval($c['id'])
 209                 );
 210
 211                 if((! $res) || (! count($res)))
 212                         continue;
 213
 214                 foreach($res as $contact) {
 215
 216                         $xml = false;
 217
 218                         if($manual_id)
 219                                 $contact['last-update'] = '0000-00-00 00:00:00';
 220
 221                         if(in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS)))
 222                                 $contact['priority'] = 2;
 223
 224                         if($contact['subhub'] AND in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS))) {
 225                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
 226                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 227                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
 228                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 229
 230                                 $poll_interval = get_config('system','pushpoll_frequency');
 231                                 $contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
 232                         }
 233
 234                         if($contact['priority'] AND !$force) {
 235
 236                                 $update     = false;
 237
 238                                 $t = $contact['last-update'];
 239
 240                                 /**
 241                                  * Based on $contact['priority'], should we poll this site now? Or later?
 242                                  */
 243
 244                                 switch ($contact['priority']) {
 245                                         case 5:
 246                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 247                                                         $update = true;
 248                                                 break;
 249                                         case 4:
 250                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 251                                                         $update = true;
 252                                                 break;
 253                                         case 3:
 254                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
 255                                                         $update = true;
 256                                                 break;
 257                                         case 2:
 258                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
 259                                                         $update = true;
 260                                                 break;
 261                                         case 1:
 262                                         default:
 263                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
 264                                                         $update = true;
 265                                                 break;
 266                                 }
 267                                 if(!$update)
 268                                         continue;
 269                         }
 270
 271                         logger("Polling ".$contact["network"]." ".$contact["id"]." ".$contact["nick"]." ".$contact["name"]);
 272
 273                         proc_run('php','include/onepoll.php',$contact['id']);
 274
 275                         if($interval)
 276                                 @time_sleep_until(microtime(true) + (float) $interval);
 277                 }
 278         }
 279
 280         logger('cron: end');
 281
 282         set_config('system','last_cron', time());
 283
 284         return;
 285 }
 286
 287 /**
 288  * @brief Clear cache entries
 289  *
 290  * @param App $a
 291  */
 292 function cron_clear_cache(&$a) {
 293
 294         $last = get_config('system','cache_last_cleared');
 295
 296         if($last) {
 297                 $next = $last + (3600); // Once per hour
 298                 $clear_cache = ($next <= time());
 299         } else
 300                 $clear_cache = true;
 301
 302         if (!$clear_cache)
 303                 return;
 304
 305         // clear old cache
 306         Cache::clear();
 307
 308         // clear old item cache files
 309         clear_cache();
 310
 311         // clear cache for photos
 312         clear_cache($a->get_basepath(), $a->get_basepath()."/photo");
 313
 314         // clear smarty cache
 315         clear_cache($a->get_basepath()."/view/smarty3/compiled", $a->get_basepath()."/view/smarty3/compiled");
 316
 317         // clear cache for image proxy
 318         if (!get_config("system", "proxy_disabled")) {
 319                 clear_cache($a->get_basepath(), $a->get_basepath()."/proxy");
 320
 321                 $cachetime = get_config('system','proxy_cache_time');
 322                 if (!$cachetime) $cachetime = PROXY_DEFAULT_TIME;
 323
 324                 q('DELETE FROM `photo` WHERE `uid` = 0 AND `resource-id` LIKE "pic:%%" AND `created` < NOW() - INTERVAL %d SECOND', $cachetime);
 325         }
 326
 327         // Delete the cached OEmbed entries that are older than one year
 328         q("DELETE FROM `oembed` WHERE `created` < NOW() - INTERVAL 1 YEAR");
 329
 330         // Delete the cached "parse_url" entries that are older than one year
 331         q("DELETE FROM `parsed_url` WHERE `created` < NOW() - INTERVAL 1 YEAR");
 332
 333         // Maximum table size in megabyte
 334         $max_tablesize = intval(get_config('system','optimize_max_tablesize')) * 1000000;
 335         if ($max_tablesize == 0)
 336                 $max_tablesize = 100 * 1000000; // Default are 100 MB
 337
 338         if ($max_tablesize > 0) {
 339                 // Minimum fragmentation level in percent
 340                 $fragmentation_level = intval(get_config('system','optimize_fragmentation')) / 100;
 341                 if ($fragmentation_level == 0)
 342                         $fragmentation_level = 0.3; // Default value is 30%
 343
 344                 // Optimize some tables that need to be optimized
 345                 $r = q("SHOW TABLE STATUS");
 346                 foreach($r as $table) {
 347
 348                         // Don't optimize tables that are too large
 349                         if ($table["Data_length"] > $max_tablesize)
 350                                 continue;
 351
 352                         // Don't optimize empty tables
 353                         if ($table["Data_length"] == 0)
 354                                 continue;
 355
 356                         // Calculate fragmentation
 357                         $fragmentation = $table["Data_free"] / ($table["Data_length"] + $table["Index_length"]);
 358
 359                         logger("Table ".$table["Name"]." - Fragmentation level: ".round($fragmentation * 100, 2), LOGGER_DEBUG);
 360
 361                         // Don't optimize tables that needn't to be optimized
 362                         if ($fragmentation < $fragmentation_level)
 363                                 continue;
 364
 365                         // So optimize it
 366                         logger("Optimize Table ".$table["Name"], LOGGER_DEBUG);
 367                         q("OPTIMIZE TABLE `%s`", dbesc($table["Name"]));
 368                 }
 369         }
 370
 371         set_config('system','cache_last_cleared', time());
 372 }
 373
 374 /**
 375  * @brief Repair missing values in Diaspora contacts
 376  *
 377  * @param App $a
 378  */
 379 function cron_repair_diaspora(&$a) {
 380         $r = q("SELECT `id`, `url` FROM `contact`
 381                 WHERE `network` = '%s' AND (`batch` = '' OR `notify` = '' OR `poll` = '' OR pubkey = '')
 382                         ORDER BY RAND() LIMIT 50", dbesc(NETWORK_DIASPORA));
 383         if ($r) {
 384                 foreach ($r AS $contact) {
 385                         if (poco_reachable($contact["url"])) {
 386                                 $data = probe_url($contact["url"]);
 387                                 if ($data["network"] == NETWORK_DIASPORA) {
 388                                         logger("Repair contact ".$contact["id"]." ".$contact["url"], LOGGER_DEBUG);
 389                                         q("UPDATE `contact` SET `batch` = '%s', `notify` = '%s', `poll` = '%s', pubkey = '%s' WHERE `id` = %d",
 390                                                 dbesc($data["batch"]), dbesc($data["notify"]), dbesc($data["poll"]), dbesc($data["pubkey"]),
 391                                                 intval($contact["id"]));
 392                                 }
 393                         }
 394                 }
 395         }
 396 }
 397
 398 /**
 399  * @brief Do some repairs in database entries
 400  *
 401  */
 402 function cron_repair_database() {
 403
 404         // Set the parent if it wasn't set. (Shouldn't happen - but does sometimes)
 405         // This call is very "cheap" so we can do it at any time without a problem
 406         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");
 407
 408         // There was an issue where the nick vanishes from the contact table
 409         q("UPDATE `contact` INNER JOIN `user` ON `contact`.`uid` = `user`.`uid` SET `nick` = `nickname` WHERE `self` AND `nick`=''");
 410
 411         /// @todo
 412         /// - remove thread entries without item
 413         /// - remove sign entries without item
 414         /// - remove children when parent got lost
 415         /// - set contact-id in item when not present
 416 }
 417
 418 if (array_search(__file__,get_included_files())===0){
 419         cron_run($_SERVER["argv"],$_SERVER["argc"]);
 420         killme();
 421 }