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