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