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