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         require_once('include/socgraph.php');
  28
  29         load_config('config');
  30         load_config('system');
  31
  32         $a->set_baseurl(get_config('system','url'));
  33
  34         load_hooks();
  35
  36         logger('poller: start');
  37
  38         // run queue delivery process in the background
  39
  40         proc_run('php',"include/queue.php");
  41
  42         // expire any expired accounts
  43
  44         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
  45                 AND `account_expires_on` != '0000-00-00 00:00:00'
  46                 AND `account_expires_on` < UTC_TIMESTAMP() ");
  47
  48         $abandon_days = intval(get_config('system','account_abandon_days'));
  49         if($abandon_days < 1)
  50                 $abandon_days = 0;
  51
  52
  53
  54         // once daily run birthday_updates and then expire in background
  55
  56         $d1 = get_config('system','last_expire_day');
  57         $d2 = intval(datetime_convert('UTC','UTC','now','d'));
  58
  59         if($d2 != intval($d1)) {
  60
  61                 update_contact_birthdays();
  62
  63                 set_config('system','last_expire_day',$d2);
  64                 proc_run('php','include/expire.php');
  65         }
  66
  67         // clear old cache
  68         Cache::clear();
  69
  70         $manual_id  = 0;
  71         $generation = 0;
  72         $hub_update = false;
  73         $force      = false;
  74         $restart    = false;
  75
  76         if(($argc > 1) && ($argv[1] == 'force'))
  77                 $force = true;
  78
  79         if(($argc > 1) && ($argv[1] == 'restart')) {
  80                 $restart = true;
  81                 $generation = intval($argv[2]);
  82                 if(! $generation)
  83                         killme();
  84         }
  85
  86         if(($argc > 1) && intval($argv[1])) {
  87                 $manual_id = intval($argv[1]);
  88                 $force     = true;
  89         }
  90
  91         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
  92
  93         reload_plugins();
  94
  95         $d = datetime_convert();
  96
  97         if(! $restart)
  98                 proc_run('php','include/cronhooks.php');
  99
 100         // Only poll from those with suitable relationships,
 101         // and which have a polling address and ignore Diaspora since
 102         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 103
 104         $abandon_sql = (($abandon_days)
 105                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
 106                 : ''
 107         );
 108
 109         $contacts = q("SELECT `contact`.`id` FROM `contact` LEFT JOIN `user` ON `user`.`uid` = `contact`.`uid`
 110                 WHERE ( `rel` = %d OR `rel` = %d ) AND `poll` != ''
 111                 AND `network` != '%s'
 112                 $sql_extra
 113                 AND `self` = 0 AND `contact`.`blocked` = 0 AND `contact`.`readonly` = 0
 114                 AND `user`.`account_expired` = 0 $abandon_sql ORDER BY RAND()",
 115                 intval(CONTACT_IS_SHARING),
 116                 intval(CONTACT_IS_FRIEND),
 117                 dbesc(NETWORK_DIASPORA)
 118         );
 119
 120         if(! count($contacts)) {
 121                 return;
 122         }
 123
 124         foreach($contacts as $c) {
 125
 126                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
 127                         intval($c['id'])
 128                 );
 129
 130                 if((! $res) || (! count($res)))
 131                         continue;
 132
 133                 foreach($res as $contact) {
 134
 135                         $xml = false;
 136
 137                         if($manual_id)
 138                                 $contact['last-update'] = '0000-00-00 00:00:00';
 139
 140                         if($contact['priority'] || $contact['subhub']) {
 141
 142                                 $hub_update = true;
 143                                 $update     = false;
 144
 145                                 $t = $contact['last-update'];
 146
 147                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
 148                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 149                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
 150                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 151
 152
 153                                 if($contact['subhub']) {
 154                                         $interval = get_config('system','pushpoll_frequency');
 155                                         $contact['priority'] = (($interval !== false) ? intval($interval) : 3);
 156                                         $hub_update = false;
 157
 158                                         if((datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day")) || $force)
 159                                                         $hub_update = true;
 160                                 }
 161                                 else
 162                                         $hub_update = false;
 163
 164                                 /**
 165                                  * Based on $contact['priority'], should we poll this site now? Or later?
 166                                  */
 167
 168                                 switch ($contact['priority']) {
 169                                         case 5:
 170                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 171                                                         $update = true;
 172                                                 break;
 173                                         case 4:
 174                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 175                                                         $update = true;
 176                                                 break;
 177                                         case 3:
 178                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
 179                                                         $update = true;
 180                                                 break;
 181                                         case 2:
 182                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
 183                                                         $update = true;
 184                                                 break;
 185                                         case 1:
 186                                         default:
 187                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
 188                                                         $update = true;
 189                                                 break;
 190                                 }
 191                                 if((! $update) && (! $force))
 192                                         continue;
 193                         }
 194
 195                         // Check to see if we are running out of memory - if so spawn a new process and kill this one
 196
 197                         $avail_memory = return_bytes(ini_get('memory_limit'));
 198                         $memused = memory_get_peak_usage(true);
 199                         if(intval($avail_memory)) {
 200                                 if(($memused / $avail_memory) > 0.95) {
 201                                         if($generation + 1 > 10) {
 202                                                 logger('poller: maximum number of spawns exceeded. Terminating.');
 203                                                 killme();
 204                                         }
 205                                         logger('poller: memory exceeded. ' . $memused . ' bytes used. Spawning new poll.');
 206                                         proc_run('php', 'include/poller.php', 'restart', (string) $generation + 1);
 207                                         killme();
 208                                 }
 209                         }
 210
 211                         $importer_uid = $contact['uid'];
 212
 213                         $r = q("SELECT * FROM `contact` WHERE `uid` = %d AND `self` = 1 LIMIT 1",
 214                                 intval($importer_uid)
 215                         );
 216                         if(! count($r))
 217                                 continue;
 218
 219                         $importer = $r[0];
 220
 221                         logger("poller: poll: IMPORTER: {$importer['name']}, CONTACT: {$contact['name']}");
 222
 223                         $last_update = (($contact['last-update'] === '0000-00-00 00:00:00')
 224                                 ? datetime_convert('UTC','UTC','now - 30 days', ATOM_TIME)
 225                                 : datetime_convert('UTC','UTC',$contact['last-update'], ATOM_TIME)
 226                         );
 227
 228                         if($contact['network'] === NETWORK_DFRN) {
 229
 230                                 $idtosend = $orig_id = (($contact['dfrn-id']) ? $contact['dfrn-id'] : $contact['issued-id']);
 231
 232                                 if(intval($contact['duplex']) && $contact['dfrn-id'])
 233                                         $idtosend = '0:' . $orig_id;
 234                                 if(intval($contact['duplex']) && $contact['issued-id'])
 235                                         $idtosend = '1:' . $orig_id;
 236
 237                                 // they have permission to write to us. We already filtered this in the contact query.
 238                                 $perm = 'rw';
 239
 240                                 $url = $contact['poll'] . '?dfrn_id=' . $idtosend
 241                                         . '&dfrn_version=' . DFRN_PROTOCOL_VERSION
 242                                         . '&type=data&last_update=' . $last_update
 243                                         . '&perm=' . $perm ;
 244
 245                                 $handshake_xml = fetch_url($url);
 246
 247                                 logger('poller: handshake with url ' . $url . ' returns xml: ' . $handshake_xml, LOGGER_DATA);
 248
 249
 250                                 if(! $handshake_xml) {
 251                                         logger("poller: $url appears to be dead - marking for death ");
 252                                         // dead connection - might be a transient event, or this might
 253                                         // mean the software was uninstalled or the domain expired.
 254                                         // Will keep trying for one month.
 255                                         mark_for_death($contact);
 256
 257                                         // set the last-update so we don't keep polling
 258
 259                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 260                                                 dbesc(datetime_convert()),
 261                                                 intval($contact['id'])
 262                                         );
 263
 264                                         continue;
 265                                 }
 266
 267                                 if(! strstr($handshake_xml,'<?xml')) {
 268                                         logger('poller: response from ' . $url . ' did not contain XML.');
 269                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 270                                                 dbesc(datetime_convert()),
 271                                                 intval($contact['id'])
 272                                         );
 273                                         continue;
 274                                 }
 275
 276
 277                                 $res = parse_xml_string($handshake_xml);
 278
 279                                 if(intval($res->status) == 1) {
 280                                         logger("poller: $url replied status 1 - marking for death ");
 281
 282                                         // we may not be friends anymore. Will keep trying for one month.
 283                                         // set the last-update so we don't keep polling
 284
 285                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 286                                                 dbesc(datetime_convert()),
 287                                                 intval($contact['id'])
 288                                         );
 289
 290                                         mark_for_death($contact);
 291                                 }
 292                                 else {
 293                                         if($contact['term-date'] != '0000-00-00 00:00:00') {
 294                                                 logger("poller: $url back from the dead - removing mark for death");
 295                                                 unmark_for_death($contact);
 296                                         }
 297                                 }
 298
 299                                 if((intval($res->status) != 0) || (! strlen($res->challenge)) || (! strlen($res->dfrn_id)))
 300                                         continue;
 301
 302                                 if(((float) $res->dfrn_version > 2.21) && ($contact['poco'] == '')) {
 303                                         q("update contact set poco = '%s' where id = %d limit 1",
 304                                                 dbesc(str_replace('/profile/','/poco/', $contact['url'])),
 305                                                 intval($contact['id'])
 306                                         );
 307                                 }
 308
 309                                 $postvars = array();
 310
 311                                 $sent_dfrn_id = hex2bin((string) $res->dfrn_id);
 312                                 $challenge    = hex2bin((string) $res->challenge);
 313
 314                                 $final_dfrn_id = '';
 315
 316                                 if(($contact['duplex']) && strlen($contact['prvkey'])) {
 317                                         openssl_private_decrypt($sent_dfrn_id,$final_dfrn_id,$contact['prvkey']);
 318                                         openssl_private_decrypt($challenge,$postvars['challenge'],$contact['prvkey']);
 319                                 }
 320                                 else {
 321                                         openssl_public_decrypt($sent_dfrn_id,$final_dfrn_id,$contact['pubkey']);
 322                                         openssl_public_decrypt($challenge,$postvars['challenge'],$contact['pubkey']);
 323                                 }
 324
 325                                 $final_dfrn_id = substr($final_dfrn_id, 0, strpos($final_dfrn_id, '.'));
 326
 327                                 if(strpos($final_dfrn_id,':') == 1)
 328                                         $final_dfrn_id = substr($final_dfrn_id,2);
 329
 330                                 if($final_dfrn_id != $orig_id) {
 331                                         logger('poller: ID did not decode: ' . $contact['id'] . ' orig: ' . $orig_id . ' final: ' . $final_dfrn_id);
 332                                         // did not decode properly - cannot trust this site
 333                                         continue;
 334                                 }
 335
 336                                 $postvars['dfrn_id'] = $idtosend;
 337                                 $postvars['dfrn_version'] = DFRN_PROTOCOL_VERSION;
 338                                 $postvars['perm'] = 'rw';
 339
 340                                 $xml = post_url($contact['poll'],$postvars);
 341                         }
 342                         elseif(($contact['network'] === NETWORK_OSTATUS)
 343                                 || ($contact['network'] === NETWORK_DIASPORA)
 344                                 || ($contact['network'] === NETWORK_FEED) ) {
 345
 346                                 // Upgrading DB fields from an older Friendika version
 347                                 // Will only do this once per notify-enabled OStatus contact
 348                                 // or if relationship changes
 349
 350                                 $stat_writeable = ((($contact['notify']) && ($contact['rel'] == CONTACT_IS_FOLLOWER || $contact['rel'] == CONTACT_IS_FRIEND)) ? 1 : 0);
 351
 352                                 if($stat_writeable != $contact['writable']) {
 353                                         q("UPDATE `contact` SET `writable` = %d WHERE `id` = %d LIMIT 1",
 354                                                 intval($stat_writeable),
 355                                                 intval($contact['id'])
 356                                         );
 357                                 }
 358
 359                                 // Are we allowed to import from this person?
 360
 361                                 if($contact['rel'] == CONTACT_IS_FOLLOWER || $contact['blocked'] || $contact['readonly'])
 362                                         continue;
 363
 364                                 $xml = fetch_url($contact['poll']);
 365                         }
 366                         elseif($contact['network'] === NETWORK_MAIL) {
 367
 368                                 $mail_disabled = ((function_exists('imap_open') && (! get_config('system','imap_disabled'))) ? 0 : 1);
 369                                 if($mail_disabled)
 370                                         continue;
 371
 372                                 $mbox = null;
 373                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 374                                         intval($importer_uid)
 375                                 );
 376                                 $mailconf = q("SELECT * FROM `mailacct` WHERE `server` != '' AND `uid` = %d LIMIT 1",
 377                                         intval($importer_uid)
 378                                 );
 379                                 if(count($x) && count($mailconf)) {
 380                                     $mailbox = construct_mailbox_name($mailconf[0]);
 381                                         $password = '';
 382                                         openssl_private_decrypt(hex2bin($mailconf[0]['pass']),$password,$x[0]['prvkey']);
 383                                         $mbox = email_connect($mailbox,$mailconf[0]['user'],$password);
 384                                         unset($password);
 385                                         if($mbox) {
 386                                                 q("UPDATE `mailacct` SET `last_check` = '%s' WHERE `id` = %d AND `uid` = %d LIMIT 1",
 387                                                         dbesc(datetime_convert()),
 388                                                         intval($mailconf[0]['id']),
 389                                                         intval($importer_uid)
 390                                                 );
 391                                         }
 392                                 }
 393                                 if($mbox) {
 394
 395                                         $msgs = email_poll($mbox,$contact['addr']);
 396
 397                                         if(count($msgs)) {
 398                                                 foreach($msgs as $msg_uid) {
 399                                                         $datarray = array();
 400                                                         $meta = email_msg_meta($mbox,$msg_uid);
 401                                                         $headers = email_msg_headers($mbox,$msg_uid);
 402
 403                                                         // look for a 'references' header and try and match with a parent item we have locally.
 404
 405                                                         $raw_refs = ((x($headers,'references')) ? str_replace("\t",'',$headers['references']) : '');
 406                                                         $datarray['uri'] = trim($meta->message_id,'<>');
 407
 408                                                         if($raw_refs) {
 409                                                                 $refs_arr = explode(' ', $raw_refs);
 410                                                                 if(count($refs_arr)) {
 411                                                                         for($x = 0; $x < count($refs_arr); $x ++)
 412                                                                                 $refs_arr[$x] = "'" . str_replace(array('<','>',' '),array('','',''),dbesc($refs_arr[$x])) . "'";
 413                                                                 }
 414                                                                 $qstr = implode(',',$refs_arr);
 415                                                                 $r = q("SELECT `uri` , `parent-uri` FROM `item` WHERE `uri` IN ( $qstr ) AND `uid` = %d LIMIT 1",
 416                                                                         intval($importer_uid)
 417                                                                 );
 418                                                                 if(count($r))
 419                                                                         $datarray['parent-uri'] = $r[0]['uri'];
 420                                                         }
 421
 422
 423                                                         if(! x($datarray,'parent-uri'))
 424                                                                 $datarray['parent-uri'] = $datarray['uri'];
 425
 426                                                         // Have we seen it before?
 427                                                         $r = q("SELECT * FROM `item` WHERE `uid` = %d AND `uri` = '%s' LIMIT 1",
 428                                                                 intval($importer_uid),
 429                                                                 dbesc($datarray['uri'])
 430                                                         );
 431
 432                                                         if(count($r)) {
 433                                                                 if($meta->deleted && ! $r[0]['deleted']) {
 434                                                                         q("UPDATE `item` SET `deleted` = 1, `changed` = '%s' WHERE `id` = %d LIMIT 1",
 435                                                                                 dbesc(datetime_convert()),
 436                                                                                 intval($r[0]['id'])
 437                                                                         );
 438                                                                 }
 439                                                                 continue;
 440                                                         }
 441                                                         $datarray['title'] = notags(trim($meta->subject));
 442                                                         $datarray['created'] = datetime_convert('UTC','UTC',$meta->date);
 443
 444                                                         $r = email_get_msg($mbox,$msg_uid);
 445                                                         if(! $r)
 446                                                                 continue;
 447                                                         $datarray['body'] = escape_tags($r['body']);
 448
 449                                                         // some mailing lists have the original author as 'from' - add this sender info to msg body.
 450                                                         // todo: adding a gravatar for the original author would be cool
 451
 452                                                         if(! stristr($meta->from,$contact['addr']))
 453                                                                 $datarray['body'] = t('From: ') . escape_tags($meta->from) . "\n\n" . $datarray['body'];
 454
 455                                                         $datarray['uid'] = $importer_uid;
 456                                                         $datarray['contact-id'] = $contact['id'];
 457                                                         if($datarray['parent-uri'] === $datarray['uri'])
 458                                                                 $datarray['private'] = 1;
 459                                                         if(! get_pconfig($importer_uid,'system','allow_public_email_replies')) {
 460                                                                 $datarray['private'] = 1;
 461                                                                 $datarray['allow_cid'] = '<' . $contact['id'] . '>';
 462                                                         }
 463                                                         $datarray['author-name'] = $contact['name'];
 464                                                         $datarray['author-link'] = 'mailbox';
 465                                                         $datarray['author-avatar'] = $contact['photo'];
 466
 467                                                         $stored_item = item_store($datarray);
 468                                                         q("UPDATE `item` SET `last-child` = 0 WHERE `parent-uri` = '%s' AND `uid` = %d",
 469                                                                 dbesc($datarray['parent-uri']),
 470                                                                 intval($importer_uid)
 471                                                         );
 472                                                         q("UPDATE `item` SET `last-child` = 1 WHERE `id` = %d LIMIT 1",
 473                                                                 intval($stored_item)
 474                                                         );
 475                                                 }
 476                                         }
 477
 478                                         imap_close($mbox);
 479                                 }
 480                         }
 481                         elseif($contact['network'] === NETWORK_FACEBOOK) {
 482                                 // This is picked up by the Facebook plugin on a cron hook.
 483                                 // Ignored here.
 484                         }
 485
 486                         if($xml) {
 487                                 logger('poller: received xml : ' . $xml, LOGGER_DATA);
 488
 489                                 if(! strstr($xml,'<?xml')) {
 490                                         logger('poller: post_handshake: response from ' . $url . ' did not contain XML.');
 491                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 492                                                 dbesc(datetime_convert()),
 493                                                 intval($contact['id'])
 494                                         );
 495                                         continue;
 496                                 }
 497
 498
 499                                 consume_feed($xml,$importer,$contact,$hub,1,1);
 500
 501                                 // do it twice. Ensures that children of parents which may be later in the stream aren't tossed
 502
 503                                 consume_feed($xml,$importer,$contact,$hub,1,2);
 504
 505                                 $hubmode = 'subscribe';
 506                                 if($contact['network'] === NETWORK_DFRN || $contact['blocked'] || $contact['readonly'])
 507                                         $hubmode = 'unsubscribe';
 508
 509                                 if((strlen($hub)) && ($hub_update) && ($contact['rel'] != CONTACT_IS_FOLLOWER)) {
 510                                         logger('poller: hub ' . $hubmode . ' : ' . $hub . ' contact name : ' . $contact['name'] . ' local user : ' . $importer['name']);
 511                                         $hubs = explode(',', $hub);
 512                                         if(count($hubs)) {
 513                                                 foreach($hubs as $h) {
 514                                                         $h = trim($h);
 515                                                         if(! strlen($h))
 516                                                                 continue;
 517                                                         subscribe_to_hub($h,$importer,$contact,$hubmode);
 518                                                 }
 519                                         }
 520                                 }
 521                         }
 522
 523
 524                         $updated = datetime_convert();
 525
 526                         $r = q("UPDATE `contact` SET `last-update` = '%s', `success_update` = '%s' WHERE `id` = %d LIMIT 1",
 527                                 dbesc($updated),
 528                                 dbesc($updated),
 529                                 intval($contact['id'])
 530                         );
 531
 532
 533                         // load current friends if possible.
 534
 535                         if($contact['poco']) {
 536                                 $r = q("SELECT count(*) as total from glink
 537                                         where `cid` = %d and updated > UTC_TIMESTAMP() - INTERVAL 1 DAY",
 538                                         intval($contact['id'])
 539                                 );
 540                         }
 541                         if(count($r)) {
 542                                 if(! $r[0]['total'])
 543                                         poco_load($contact['id'],$importer_uid,$contact['poco']);
 544                         }
 545
 546                         // loop - next contact
 547                 }
 548         }
 549
 550
 551         return;
 552 }
 553
 554 if (array_search(__file__,get_included_files())===0){
 555   poller_run($argv,$argc);
 556   killme();
 557 }