include/poller.php

   1 <?php
   2
   3 require_once("boot.php");
   4
   5
   6 function poller_run($argv, $argc){
   7         global $a, $db;
   8
   9         if(is_null($a)) {
  10                 $a = new App;
  11         }
  12
  13         if(is_null($db)) {
  14             @include(".htconfig.php");
  15         require_once("dba.php");
  16             $db = new dba($db_host, $db_user, $db_pass, $db_data);
  17         unset($db_host, $db_user, $db_pass, $db_data);
  18         };
  19
  20
  21         require_once('include/session.php');
  22         require_once('include/datetime.php');
  23         require_once('library/simplepie/simplepie.inc');
  24         require_once('include/items.php');
  25         require_once('include/Contact.php');
  26         require_once('include/email.php');
  27
  28         load_config('config');
  29         load_config('system');
  30
  31         $a->set_baseurl(get_config('system','url'));
  32
  33         load_hooks();
  34
  35         logger('poller: start');
  36
  37         // run queue delivery process in the background
  38
  39         proc_run('php',"include/queue.php");
  40
  41         // expire any expired accounts
  42
  43         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
  44                 AND `account_expires_on` != '0000-00-00 00:00:00'
  45                 AND `account_expires_on` < UTC_TIMESTAMP() ");
  46
  47         $abandon_days = intval(get_config('system','account_abandon_days'));
  48         if($abandon_days < 1)
  49                 $abandon_days = 0;
  50
  51
  52
  53         // once daily run expire in background
  54
  55         $d1 = get_config('system','last_expire_day');
  56         $d2 = intval(datetime_convert('UTC','UTC','now','d'));
  57
  58         if($d2 != intval($d1)) {
  59                 set_config('system','last_expire_day',$d2);
  60                 proc_run('php','include/expire.php');
  61         }
  62
  63         // clear old cache
  64         q("DELETE FROM `cache` WHERE `updated` < '%s'",
  65                 dbesc(datetime_convert('UTC','UTC',"now - 30 days")));
  66
  67         $manual_id  = 0;
  68         $generation = 0;
  69         $hub_update = false;
  70         $force      = false;
  71         $restart    = false;
  72
  73         if(($argc > 1) && ($argv[1] == 'force'))
  74                 $force = true;
  75
  76         if(($argc > 1) && ($argv[1] == 'restart')) {
  77                 $restart = true;
  78                 $generation = intval($argv[2]);
  79                 if(! $generation)
  80                         killme();
  81         }
  82
  83         if(($argc > 1) && intval($argv[1])) {
  84                 $manual_id = intval($argv[1]);
  85                 $force     = true;
  86         }
  87
  88         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
  89
  90         reload_plugins();
  91
  92         $d = datetime_convert();
  93
  94         if(! $restart)
  95                 proc_run('php','include/cronhooks.php');
  96
  97         // Only poll from those with suitable relationships,
  98         // and which have a polling address and ignore Diaspora since
  99         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 100
 101         $abandon_sql = (($abandon_days)
 102                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
 103                 : ''
 104         );
 105
 106         $contacts = q("SELECT `contact`.`id` FROM `contact` LEFT JOIN `user` ON `user`.`uid` = `contact`.`uid`
 107                 WHERE ( `rel` = %d OR `rel` = %d ) AND `poll` != ''
 108                 AND `network` != '%s'
 109                 $sql_extra
 110                 AND `self` = 0 AND `contact`.`blocked` = 0 AND `contact`.`readonly` = 0
 111                 AND `user`.`account_expired` = 0 $abandon_sql ORDER BY RAND()",
 112                 intval(CONTACT_IS_SHARING),
 113                 intval(CONTACT_IS_FRIEND),
 114                 dbesc(NETWORK_DIASPORA)
 115         );
 116
 117         if(! count($contacts)) {
 118                 return;
 119         }
 120
 121         foreach($contacts as $c) {
 122
 123                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
 124                         intval($c['id'])
 125                 );
 126
 127                 if((! $res) || (! count($res)))
 128                         continue;
 129
 130                 foreach($res as $contact) {
 131
 132                         $xml = false;
 133
 134                         if($manual_id)
 135                                 $contact['last-update'] = '0000-00-00 00:00:00';
 136
 137                         if($contact['priority'] || $contact['subhub']) {
 138
 139                                 $hub_update = true;
 140                                 $update     = false;
 141
 142                                 $t = $contact['last-update'];
 143
 144                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
 145                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 146                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
 147                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 148
 149
 150                                 if($contact['subhub']) {
 151                                         $interval = get_config('system','pushpoll_frequency');
 152                                         $contact['priority'] = (($interval !== false) ? intval($interval) : 3);
 153                                         $hub_update = false;
 154
 155                                         if((datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day")) || $force)
 156                                                         $hub_update = true;
 157                                 }
 158                                 else
 159                                         $hub_update = false;
 160
 161                                 /**
 162                                  * Based on $contact['priority'], should we poll this site now? Or later?
 163                                  */
 164
 165                                 switch ($contact['priority']) {
 166                                         case 5:
 167                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 168                                                         $update = true;
 169                                                 break;
 170                                         case 4:
 171                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 172                                                         $update = true;
 173                                                 break;
 174                                         case 3:
 175                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
 176                                                         $update = true;
 177                                                 break;
 178                                         case 2:
 179                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
 180                                                         $update = true;
 181                                                 break;
 182                                         case 1:
 183                                         default:
 184                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
 185                                                         $update = true;
 186                                                 break;
 187                                 }
 188                                 if((! $update) && (! $force))
 189                                         continue;
 190                         }
 191
 192                         // Check to see if we are running out of memory - if so spawn a new process and kill this one
 193
 194                         $avail_memory = return_bytes(ini_get('memory_limit'));
 195                         $memused = memory_get_peak_usage(true);
 196                         if(intval($avail_memory)) {
 197                                 if(($memused / $avail_memory) > 0.95) {
 198                                         if($generation + 1 > 10) {
 199                                                 logger('poller: maximum number of spawns exceeded. Terminating.');
 200                                                 killme();
 201                                         }
 202                                         logger('poller: memory exceeded. ' . $memused . ' bytes used. Spawning new poll.');
 203                                         proc_run('php', 'include/poller.php', 'restart', (string) $generation + 1);
 204                                         killme();
 205                                 }
 206                         }
 207
 208                         $importer_uid = $contact['uid'];
 209
 210                         $r = q("SELECT * FROM `contact` WHERE `uid` = %d AND `self` = 1 LIMIT 1",
 211                                 intval($importer_uid)
 212                         );
 213                         if(! count($r))
 214                                 continue;
 215
 216                         $importer = $r[0];
 217
 218                         logger("poller: poll: IMPORTER: {$importer['name']}, CONTACT: {$contact['name']}");
 219
 220                         $last_update = (($contact['last-update'] === '0000-00-00 00:00:00')
 221                                 ? datetime_convert('UTC','UTC','now - 30 days', ATOM_TIME)
 222                                 : datetime_convert('UTC','UTC',$contact['last-update'], ATOM_TIME)
 223                         );
 224
 225                         if($contact['network'] === NETWORK_DFRN) {
 226
 227                                 $idtosend = $orig_id = (($contact['dfrn-id']) ? $contact['dfrn-id'] : $contact['issued-id']);
 228
 229                                 if(intval($contact['duplex']) && $contact['dfrn-id'])
 230                                         $idtosend = '0:' . $orig_id;
 231                                 if(intval($contact['duplex']) && $contact['issued-id'])
 232                                         $idtosend = '1:' . $orig_id;
 233
 234                                 // they have permission to write to us. We already filtered this in the contact query.
 235                                 $perm = 'rw';
 236
 237                                 $url = $contact['poll'] . '?dfrn_id=' . $idtosend
 238                                         . '&dfrn_version=' . DFRN_PROTOCOL_VERSION
 239                                         . '&type=data&last_update=' . $last_update
 240                                         . '&perm=' . $perm ;
 241
 242                                 $handshake_xml = fetch_url($url);
 243
 244                                 logger('poller: handshake with url ' . $url . ' returns xml: ' . $handshake_xml, LOGGER_DATA);
 245
 246
 247                                 if(! $handshake_xml) {
 248                                         logger("poller: $url appears to be dead - marking for death ");
 249                                         // dead connection - might be a transient event, or this might
 250                                         // mean the software was uninstalled or the domain expired.
 251                                         // Will keep trying for one month.
 252                                         mark_for_death($contact);
 253
 254                                         // set the last-update so we don't keep polling
 255
 256                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 257                                                 dbesc(datetime_convert()),
 258                                                 intval($contact['id'])
 259                                         );
 260
 261                                         continue;
 262                                 }
 263
 264                                 if(! strstr($handshake_xml,'<?xml')) {
 265                                         logger('poller: response from ' . $url . ' did not contain XML.');
 266                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 267                                                 dbesc(datetime_convert()),
 268                                                 intval($contact['id'])
 269                                         );
 270                                         continue;
 271                                 }
 272
 273
 274                                 $res = parse_xml_string($handshake_xml);
 275
 276                                 if(intval($res->status) == 1) {
 277                                         logger("poller: $url replied status 1 - marking for death ");
 278
 279                                         // we may not be friends anymore. Will keep trying for one month.
 280                                         // set the last-update so we don't keep polling
 281
 282                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 283                                                 dbesc(datetime_convert()),
 284                                                 intval($contact['id'])
 285                                         );
 286
 287                                         mark_for_death($contact);
 288                                 }
 289                                 else {
 290                                         if($contact['term-date'] != '0000-00-00 00:00:00') {
 291                                                 logger("poller: $url back from the dead - removing mark for death");
 292                                                 unmark_for_death($contact);
 293                                         }
 294                                 }
 295
 296                                 if((intval($res->status) != 0) || (! strlen($res->challenge)) || (! strlen($res->dfrn_id)))
 297                                         continue;
 298
 299                                 $postvars = array();
 300
 301                                 $sent_dfrn_id = hex2bin((string) $res->dfrn_id);
 302                                 $challenge    = hex2bin((string) $res->challenge);
 303
 304                                 $final_dfrn_id = '';
 305
 306                                 if(($contact['duplex']) && strlen($contact['prvkey'])) {
 307                                         openssl_private_decrypt($sent_dfrn_id,$final_dfrn_id,$contact['prvkey']);
 308                                         openssl_private_decrypt($challenge,$postvars['challenge'],$contact['prvkey']);
 309                                 }
 310                                 else {
 311                                         openssl_public_decrypt($sent_dfrn_id,$final_dfrn_id,$contact['pubkey']);
 312                                         openssl_public_decrypt($challenge,$postvars['challenge'],$contact['pubkey']);
 313                                 }
 314
 315                                 $final_dfrn_id = substr($final_dfrn_id, 0, strpos($final_dfrn_id, '.'));
 316
 317                                 if(strpos($final_dfrn_id,':') == 1)
 318                                         $final_dfrn_id = substr($final_dfrn_id,2);
 319
 320                                 if($final_dfrn_id != $orig_id) {
 321                                         logger('poller: ID did not decode: ' . $contact['id'] . ' orig: ' . $orig_id . ' final: ' . $final_dfrn_id);
 322                                         // did not decode properly - cannot trust this site
 323                                         continue;
 324                                 }
 325
 326                                 $postvars['dfrn_id'] = $idtosend;
 327                                 $postvars['dfrn_version'] = DFRN_PROTOCOL_VERSION;
 328                                 $postvars['perm'] = 'rw';
 329
 330                                 $xml = post_url($contact['poll'],$postvars);
 331                         }
 332                         elseif(($contact['network'] === NETWORK_OSTATUS)
 333                                 || ($contact['network'] === NETWORK_DIASPORA)
 334                                 || ($contact['network'] === NETWORK_FEED) ) {
 335
 336                                 // Upgrading DB fields from an older Friendika version
 337                                 // Will only do this once per notify-enabled OStatus contact
 338                                 // or if relationship changes
 339
 340                                 $stat_writeable = ((($contact['notify']) && ($contact['rel'] == CONTACT_IS_FOLLOWER || $contact['rel'] == CONTACT_IS_FRIEND)) ? 1 : 0);
 341
 342                                 if($stat_writeable != $contact['writable']) {
 343                                         q("UPDATE `contact` SET `writable` = %d WHERE `id` = %d LIMIT 1",
 344                                                 intval($stat_writeable),
 345                                                 intval($contact['id'])
 346                                         );
 347                                 }
 348
 349                                 // Are we allowed to import from this person?
 350
 351                                 if($contact['rel'] == CONTACT_IS_FOLLOWER || $contact['blocked'] || $contact['readonly'])
 352                                         continue;
 353
 354                                 $xml = fetch_url($contact['poll']);
 355                         }
 356                         elseif($contact['network'] === NETWORK_MAIL) {
 357
 358                                 $mail_disabled = ((function_exists('imap_open') && (! get_config('system','imap_disabled'))) ? 0 : 1);
 359                                 if($mail_disabled)
 360                                         continue;
 361
 362                                 $mbox = null;
 363                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 364                                         intval($importer_uid)
 365                                 );
 366                                 $mailconf = q("SELECT * FROM `mailacct` WHERE `server` != '' AND `uid` = %d LIMIT 1",
 367                                         intval($importer_uid)
 368                                 );
 369                                 if(count($x) && count($mailconf)) {
 370                                     $mailbox = construct_mailbox_name($mailconf[0]);
 371                                         $password = '';
 372                                         openssl_private_decrypt(hex2bin($mailconf[0]['pass']),$password,$x[0]['prvkey']);
 373                                         $mbox = email_connect($mailbox,$mailconf[0]['user'],$password);
 374                                         unset($password);
 375                                         if($mbox) {
 376                                                 q("UPDATE `mailacct` SET `last_check` = '%s' WHERE `id` = %d AND `uid` = %d LIMIT 1",
 377                                                         dbesc(datetime_convert()),
 378                                                         intval($mailconf[0]['id']),
 379                                                         intval($importer_uid)
 380                                                 );
 381                                         }
 382                                 }
 383                                 if($mbox) {
 384
 385                                         $msgs = email_poll($mbox,$contact['addr']);
 386
 387                                         if(count($msgs)) {
 388                                                 foreach($msgs as $msg_uid) {
 389                                                         $datarray = array();
 390                                                         $meta = email_msg_meta($mbox,$msg_uid);
 391                                                         $headers = email_msg_headers($mbox,$msg_uid);
 392
 393                                                         // look for a 'references' header and try and match with a parent item we have locally.
 394
 395                                                         $raw_refs = ((x($headers,'references')) ? str_replace("\t",'',$headers['references']) : '');
 396                                                         $datarray['uri'] = trim($meta->message_id,'<>');
 397
 398                                                         if($raw_refs) {
 399                                                                 $refs_arr = explode(' ', $raw_refs);
 400                                                                 if(count($refs_arr)) {
 401                                                                         for($x = 0; $x < count($refs_arr); $x ++)
 402                                                                                 $refs_arr[$x] = "'" . str_replace(array('<','>',' '),array('','',''),dbesc($refs_arr[$x])) . "'";
 403                                                                 }
 404                                                                 $qstr = implode(',',$refs_arr);
 405                                                                 $r = q("SELECT `uri` , `parent-uri` FROM `item` WHERE `uri` IN ( $qstr ) AND `uid` = %d LIMIT 1",
 406                                                                         intval($importer_uid)
 407                                                                 );
 408                                                                 if(count($r))
 409                                                                         $datarray['parent-uri'] = $r[0]['uri'];
 410                                                         }
 411
 412
 413                                                         if(! x($datarray,'parent-uri'))
 414                                                                 $datarray['parent-uri'] = $datarray['uri'];
 415
 416                                                         // Have we seen it before?
 417                                                         $r = q("SELECT * FROM `item` WHERE `uid` = %d AND `uri` = '%s' LIMIT 1",
 418                                                                 intval($importer_uid),
 419                                                                 dbesc($datarray['uri'])
 420                                                         );
 421
 422                                                         if(count($r)) {
 423                                                                 if($meta->deleted && ! $r[0]['deleted']) {
 424                                                                         q("UPDATE `item` SET `deleted` = 1, `changed` = '%s' WHERE `id` = %d LIMIT 1",
 425                                                                                 dbesc(datetime_convert()),
 426                                                                                 intval($r[0]['id'])
 427                                                                         );
 428                                                                 }
 429                                                                 continue;
 430                                                         }
 431                                                         $datarray['title'] = notags(trim($meta->subject));
 432                                                         $datarray['created'] = datetime_convert('UTC','UTC',$meta->date);
 433
 434                                                         $r = email_get_msg($mbox,$msg_uid);
 435                                                         if(! $r)
 436                                                                 continue;
 437                                                         $datarray['body'] = escape_tags($r['body']);
 438
 439                                                         // some mailing lists have the original author as 'from' - add this sender info to msg body.
 440                                                         // todo: adding a gravatar for the original author would be cool
 441
 442                                                         if(! stristr($meta->from,$contact['addr']))
 443                                                                 $datarray['body'] = t('From: ') . escape_tags($meta->from) . "\n\n" . $datarray['body'];
 444
 445                                                         $datarray['uid'] = $importer_uid;
 446                                                         $datarray['contact-id'] = $contact['id'];
 447                                                         if($datarray['parent-uri'] === $datarray['uri'])
 448                                                                 $datarray['private'] = 1;
 449                                                         if(! get_pconfig($importer_uid,'system','allow_public_email_replies')) {
 450                                                                 $datarray['private'] = 1;
 451                                                                 $datarray['allow_cid'] = '<' . $contact['id'] . '>';
 452                                                         }
 453                                                         $datarray['author-name'] = $contact['name'];
 454                                                         $datarray['author-link'] = 'mailbox';
 455                                                         $datarray['author-avatar'] = $contact['photo'];
 456
 457                                                         $stored_item = item_store($datarray);
 458                                                         q("UPDATE `item` SET `last-child` = 0 WHERE `parent-uri` = '%s' AND `uid` = %d",
 459                                                                 dbesc($datarray['parent-uri']),
 460                                                                 intval($importer_uid)
 461                                                         );
 462                                                         q("UPDATE `item` SET `last-child` = 1 WHERE `id` = %d LIMIT 1",
 463                                                                 intval($stored_item)
 464                                                         );
 465                                                 }
 466                                         }
 467
 468                                         imap_close($mbox);
 469                                 }
 470                         }
 471                         elseif($contact['network'] === NETWORK_FACEBOOK) {
 472                                 // This is picked up by the Facebook plugin on a cron hook.
 473                                 // Ignored here.
 474                         }
 475
 476                         if($xml) {
 477                                 logger('poller: received xml : ' . $xml, LOGGER_DATA);
 478
 479                                 if(! strstr($xml,'<?xml')) {
 480                                         logger('poller: post_handshake: response from ' . $url . ' did not contain XML.');
 481                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 482                                                 dbesc(datetime_convert()),
 483                                                 intval($contact['id'])
 484                                         );
 485                                         continue;
 486                                 }
 487
 488
 489                                 consume_feed($xml,$importer,$contact,$hub,1, true);
 490
 491                                 // do it twice. Ensures that children of parents which may be later in the stream aren't tossed
 492
 493                                 consume_feed($xml,$importer,$contact,$hub,1);
 494
 495                                 $hubmode = 'subscribe';
 496                                 if($contact['network'] === NETWORK_DFRN || $contact['blocked'] || $contact['readonly'])
 497                                         $hubmode = 'unsubscribe';
 498
 499                                 if((strlen($hub)) && ($hub_update) && ($contact['rel'] != CONTACT_IS_FOLLOWER)) {
 500                                         logger('poller: hub ' . $hubmode . ' : ' . $hub . ' contact name : ' . $contact['name'] . ' local user : ' . $importer['name']);
 501                                         $hubs = explode(',', $hub);
 502                                         if(count($hubs)) {
 503                                                 foreach($hubs as $h) {
 504                                                         $h = trim($h);
 505                                                         if(! strlen($h))
 506                                                                 continue;
 507                                                         subscribe_to_hub($h,$importer,$contact,$hubmode);
 508                                                 }
 509                                         }
 510                                 }
 511                         }
 512
 513
 514                         $updated = datetime_convert();
 515
 516                         $r = q("UPDATE `contact` SET `last-update` = '%s', `success_update` = '%s' WHERE `id` = %d LIMIT 1",
 517                                 dbesc($updated),
 518                                 dbesc($updated),
 519                                 intval($contact['id'])
 520                         );
 521
 522                         // loop - next contact
 523                 }
 524         }
 525
 526
 527         return;
 528 }
 529
 530 if (array_search(__file__,get_included_files())===0){
 531   poller_run($argv,$argc);
 532   killme();
 533 }