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('php',"include/queue.php");
  74
  75         // run the process to discover global contacts in the background
  76
  77         proc_run('php',"include/discover_poco.php");
  78
  79         // run the process to update locally stored global contacts in the background
  80
  81         proc_run('php',"include/discover_poco.php", "checkcontact");
  82
  83         // expire any expired accounts
  84
  85         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
  86                 AND `account_expires_on` != '0000-00-00 00:00:00'
  87                 AND `account_expires_on` < UTC_TIMESTAMP() ");
  88
  89         // delete user and contact records for recently removed accounts
  90
  91         $r = q("SELECT * FROM `user` WHERE `account_removed` = 1 AND `account_expires_on` < UTC_TIMESTAMP() - INTERVAL 3 DAY");
  92         if ($r) {
  93                 foreach($r as $user) {
  94                         q("DELETE FROM `contact` WHERE `uid` = %d", intval($user['uid']));
  95                         q("DELETE FROM `user` WHERE `uid` = %d", intval($user['uid']));
  96                 }
  97         }
  98
  99         $abandon_days = intval(get_config('system','account_abandon_days'));
 100         if($abandon_days < 1)
 101                 $abandon_days = 0;
 102
 103         // Check OStatus conversations
 104         // Check only conversations with mentions (for a longer time)
 105         ostatus::check_conversations(true);
 106
 107         // Check every conversation
 108         ostatus::check_conversations(false);
 109
 110         // Call possible post update functions
 111         // see include/post_update.php for more details
 112         post_update();
 113
 114         // update nodeinfo data
 115         nodeinfo_cron();
 116
 117         /// @TODO Regenerate usage statistics
 118         // q("ANALYZE TABLE `item`");
 119
 120         // once daily run birthday_updates and then expire in background
 121
 122         $d1 = get_config('system','last_expire_day');
 123         $d2 = intval(datetime_convert('UTC','UTC','now','d'));
 124
 125         if($d2 != intval($d1)) {
 126
 127                 update_contact_birthdays();
 128
 129                 proc_run('php',"include/discover_poco.php", "suggestions");
 130
 131                 set_config('system','last_expire_day',$d2);
 132
 133                 proc_run('php','include/expire.php');
 134         }
 135
 136         // Clear cache entries
 137         cron_clear_cache($a);
 138
 139         // Repair missing Diaspora values in contacts
 140         cron_repair_diaspora($a);
 141
 142         // Repair entries in the database
 143         cron_repair_database();
 144
 145         $manual_id  = 0;
 146         $generation = 0;
 147         $force      = false;
 148         $restart    = false;
 149
 150         if(($argc > 1) && ($argv[1] == 'force'))
 151                 $force = true;
 152
 153         if(($argc > 1) && ($argv[1] == 'restart')) {
 154                 $restart = true;
 155                 $generation = intval($argv[2]);
 156                 if(! $generation)
 157                         killme();
 158         }
 159
 160         if(($argc > 1) && intval($argv[1])) {
 161                 $manual_id = intval($argv[1]);
 162                 $force     = true;
 163         }
 164
 165         $interval = intval(get_config('system','poll_interval'));
 166         if(! $interval)
 167                 $interval = ((get_config('system','delivery_interval') === false) ? 3 : intval(get_config('system','delivery_interval')));
 168
 169         // If we are using the worker we don't need a delivery interval
 170         if (get_config("system", "worker"))
 171                 $interval = false;
 172
 173         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
 174
 175         reload_plugins();
 176
 177         $d = datetime_convert();
 178
 179         // Only poll from those with suitable relationships,
 180         // and which have a polling address and ignore Diaspora since
 181         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 182
 183         $abandon_sql = (($abandon_days)
 184                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
 185                 : ''
 186         );
 187
 188         $contacts = q("SELECT `contact`.`id` FROM `contact` INNER JOIN `user` ON `user`.`uid` = `contact`.`uid`
 189                 WHERE `rel` IN (%d, %d) AND `poll` != '' AND `network` IN ('%s', '%s', '%s', '%s', '%s', '%s')
 190                 $sql_extra
 191                 AND NOT `self` AND NOT `contact`.`blocked` AND NOT `contact`.`readonly` AND NOT `contact`.`archive`
 192                 AND NOT `user`.`account_expired` AND NOT `user`.`account_removed` $abandon_sql ORDER BY RAND()",
 193                 intval(CONTACT_IS_SHARING),
 194                 intval(CONTACT_IS_FRIEND),
 195                 dbesc(NETWORK_DFRN),
 196                 dbesc(NETWORK_ZOT),
 197                 dbesc(NETWORK_OSTATUS),
 198                 dbesc(NETWORK_FEED),
 199                 dbesc(NETWORK_MAIL),
 200                 dbesc(NETWORK_MAIL2)
 201         );
 202
 203         if(! count($contacts)) {
 204                 return;
 205         }
 206
 207         foreach($contacts as $c) {
 208
 209                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
 210                         intval($c['id'])
 211                 );
 212
 213                 if((! $res) || (! count($res)))
 214                         continue;
 215
 216                 foreach($res as $contact) {
 217
 218                         $xml = false;
 219
 220                         if($manual_id)
 221                                 $contact['last-update'] = '0000-00-00 00:00:00';
 222
 223                         if(in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS)))
 224                                 $contact['priority'] = 2;
 225
 226                         if($contact['subhub'] AND in_array($contact['network'], array(NETWORK_DFRN, NETWORK_ZOT, NETWORK_OSTATUS))) {
 227                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
 228                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 229                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
 230                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 231
 232                                 $poll_interval = get_config('system','pushpoll_frequency');
 233                                 $contact['priority'] = (($poll_interval !== false) ? intval($poll_interval) : 3);
 234                         }
 235
 236                         if($contact['priority'] AND !$force) {
 237
 238                                 $update     = false;
 239
 240                                 $t = $contact['last-update'];
 241
 242                                 /**
 243                                  * Based on $contact['priority'], should we poll this site now? Or later?
 244                                  */
 245
 246                                 switch ($contact['priority']) {
 247                                         case 5:
 248                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 249                                                         $update = true;
 250                                                 break;
 251                                         case 4:
 252                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 253                                                         $update = true;
 254                                                 break;
 255                                         case 3:
 256                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
 257                                                         $update = true;
 258                                                 break;
 259                                         case 2:
 260                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
 261                                                         $update = true;
 262                                                 break;
 263                                         case 1:
 264                                         default:
 265                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
 266                                                         $update = true;
 267                                                 break;
 268                                 }
 269                                 if(!$update)
 270                                         continue;
 271                         }
 272
 273                         logger("Polling ".$contact["network"]." ".$contact["id"]." ".$contact["nick"]." ".$contact["name"]);
 274
 275                         proc_run('php','include/onepoll.php',$contact['id']);
 276
 277                         if($interval)
 278                                 @time_sleep_until(microtime(true) + (float) $interval);
 279                 }
 280         }
 281
 282         logger('cron: end');
 283
 284         set_config('system','last_cron', time());
 285
 286         return;
 287 }
 288
 289 /**
 290  * @brief Clear cache entries
 291  *
 292  * @param App $a
 293  */
 294 function cron_clear_cache(&$a) {
 295
 296         $last = get_config('system','cache_last_cleared');
 297
 298         if($last) {
 299                 $next = $last + (3600); // Once per hour
 300                 $clear_cache = ($next <= time());
 301         } else
 302                 $clear_cache = true;
 303
 304         if (!$clear_cache)
 305                 return;
 306
 307         // clear old cache
 308         Cache::clear();
 309
 310         // clear old item cache files
 311         clear_cache();
 312
 313         // clear cache for photos
 314         clear_cache($a->get_basepath(), $a->get_basepath()."/photo");
 315
 316         // clear smarty cache
 317         clear_cache($a->get_basepath()."/view/smarty3/compiled", $a->get_basepath()."/view/smarty3/compiled");
 318
 319         // clear cache for image proxy
 320         if (!get_config("system", "proxy_disabled")) {
 321                 clear_cache($a->get_basepath(), $a->get_basepath()."/proxy");
 322
 323                 $cachetime = get_config('system','proxy_cache_time');
 324                 if (!$cachetime) $cachetime = PROXY_DEFAULT_TIME;
 325
 326                 q('DELETE FROM `photo` WHERE `uid` = 0 AND `resource-id` LIKE "pic:%%" AND `created` < NOW() - INTERVAL %d SECOND', $cachetime);
 327         }
 328
 329         // Delete the cached OEmbed entries that are older than one year
 330         q("DELETE FROM `oembed` WHERE `created` < NOW() - INTERVAL 1 YEAR");
 331
 332         // Delete the cached "parse_url" entries that are older than one year
 333         q("DELETE FROM `parsed_url` WHERE `created` < NOW() - INTERVAL 1 YEAR");
 334
 335         // Maximum table size in megabyte
 336         $max_tablesize = intval(get_config('system','optimize_max_tablesize')) * 1000000;
 337         if ($max_tablesize == 0)
 338                 $max_tablesize = 100 * 1000000; // Default are 100 MB
 339
 340         if ($max_tablesize > 0) {
 341                 // Minimum fragmentation level in percent
 342                 $fragmentation_level = intval(get_config('system','optimize_fragmentation')) / 100;
 343                 if ($fragmentation_level == 0)
 344                         $fragmentation_level = 0.3; // Default value is 30%
 345
 346                 // Optimize some tables that need to be optimized
 347                 $r = q("SHOW TABLE STATUS");
 348                 foreach($r as $table) {
 349
 350                         // Don't optimize tables that are too large
 351                         if ($table["Data_length"] > $max_tablesize)
 352                                 continue;
 353
 354                         // Don't optimize empty tables
 355                         if ($table["Data_length"] == 0)
 356                                 continue;
 357
 358                         // Calculate fragmentation
 359                         $fragmentation = $table["Data_free"] / ($table["Data_length"] + $table["Index_length"]);
 360
 361                         logger("Table ".$table["Name"]." - Fragmentation level: ".round($fragmentation * 100, 2), LOGGER_DEBUG);
 362
 363                         // Don't optimize tables that needn't to be optimized
 364                         if ($fragmentation < $fragmentation_level)
 365                                 continue;
 366
 367                         // So optimize it
 368                         logger("Optimize Table ".$table["Name"], LOGGER_DEBUG);
 369                         q("OPTIMIZE TABLE `%s`", dbesc($table["Name"]));
 370                 }
 371         }
 372
 373         set_config('system','cache_last_cleared', time());
 374 }
 375
 376 /**
 377  * @brief Repair missing values in Diaspora contacts
 378  *
 379  * @param App $a
 380  */
 381 function cron_repair_diaspora(&$a) {
 382         $r = q("SELECT `id`, `url` FROM `contact`
 383                 WHERE `network` = '%s' AND (`batch` = '' OR `notify` = '' OR `poll` = '' OR pubkey = '')
 384                         ORDER BY RAND() LIMIT 50", dbesc(NETWORK_DIASPORA));
 385         if ($r) {
 386                 foreach ($r AS $contact) {
 387                         if (poco_reachable($contact["url"])) {
 388                                 $data = probe_url($contact["url"]);
 389                                 if ($data["network"] == NETWORK_DIASPORA) {
 390                                         logger("Repair contact ".$contact["id"]." ".$contact["url"], LOGGER_DEBUG);
 391                                         q("UPDATE `contact` SET `batch` = '%s', `notify` = '%s', `poll` = '%s', pubkey = '%s' WHERE `id` = %d",
 392                                                 dbesc($data["batch"]), dbesc($data["notify"]), dbesc($data["poll"]), dbesc($data["pubkey"]),
 393                                                 intval($contact["id"]));
 394                                 }
 395                         }
 396                 }
 397         }
 398 }
 399
 400 /**
 401  * @brief Do some repairs in database entries
 402  *
 403  */
 404 function cron_repair_database() {
 405
 406         // Set the parent if it wasn't set. (Shouldn't happen - but does sometimes)
 407         // This call is very "cheap" so we can do it at any time without a problem
 408         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");
 409
 410         // There was an issue where the nick vanishes from the contact table
 411         q("UPDATE `contact` INNER JOIN `user` ON `contact`.`uid` = `user`.`uid` SET `nick` = `nickname` WHERE `self` AND `nick`=''");
 412
 413         /// @todo
 414         /// - remove thread entries without item
 415         /// - remove sign entries without item
 416         /// - remove children when parent got lost
 417         /// - set contact-id in item when not present
 418 }
 419
 420 if (array_search(__file__,get_included_files())===0){
 421         cron_run($_SERVER["argv"],$_SERVER["argc"]);
 422         killme();
 423 }