include/poller.php

   1 <?php
   2
   3 require_once("boot.php");
   4
   5
   6 function poller_run($argv, $argc){
   7         global $a, $db;
   8
   9         if(is_null($a)) {
  10                 $a = new App;
  11         }
  12
  13         if(is_null($db)) {
  14             @include(".htconfig.php");
  15         require_once("dba.php");
  16             $db = new dba($db_host, $db_user, $db_pass, $db_data);
  17         unset($db_host, $db_user, $db_pass, $db_data);
  18         };
  19
  20
  21         require_once('include/session.php');
  22         require_once('include/datetime.php');
  23         require_once('library/simplepie/simplepie.inc');
  24         require_once('include/items.php');
  25         require_once('include/Contact.php');
  26         require_once('include/email.php');
  27
  28         load_config('config');
  29         load_config('system');
  30
  31         $a->set_baseurl(get_config('system','url'));
  32
  33         load_hooks();
  34
  35         logger('poller: start');
  36
  37         // run queue delivery process in the background
  38
  39         proc_run('php',"include/queue.php");
  40
  41         // expire any expired accounts
  42
  43         q("UPDATE user SET `account_expired` = 1 where `account_expired` = 0
  44                 AND `account_expires_on` != '0000-00-00 00:00:00'
  45                 AND `account_expires_on` < UTC_TIMESTAMP() ");
  46
  47         $abandon_days = intval(get_config('system','account_abandon_days'));
  48         if($abandon_days < 1)
  49                 $abandon_days = 0;
  50
  51
  52
  53         // once daily run birthday_updates and then expire in background
  54
  55         $d1 = get_config('system','last_expire_day');
  56         $d2 = intval(datetime_convert('UTC','UTC','now','d'));
  57
  58         if($d2 != intval($d1)) {
  59
  60                 update_contact_birthdays();
  61
  62                 set_config('system','last_expire_day',$d2);
  63                 proc_run('php','include/expire.php');
  64         }
  65
  66         // clear old cache
  67         Cache::clear();
  68
  69         $manual_id  = 0;
  70         $generation = 0;
  71         $hub_update = false;
  72         $force      = false;
  73         $restart    = false;
  74
  75         if(($argc > 1) && ($argv[1] == 'force'))
  76                 $force = true;
  77
  78         if(($argc > 1) && ($argv[1] == 'restart')) {
  79                 $restart = true;
  80                 $generation = intval($argv[2]);
  81                 if(! $generation)
  82                         killme();
  83         }
  84
  85         if(($argc > 1) && intval($argv[1])) {
  86                 $manual_id = intval($argv[1]);
  87                 $force     = true;
  88         }
  89
  90         $sql_extra = (($manual_id) ? " AND `id` = $manual_id " : "");
  91
  92         reload_plugins();
  93
  94         $d = datetime_convert();
  95
  96         if(! $restart)
  97                 proc_run('php','include/cronhooks.php');
  98
  99         // Only poll from those with suitable relationships,
 100         // and which have a polling address and ignore Diaspora since
 101         // we are unable to match those posts with a Diaspora GUID and prevent duplicates.
 102
 103         $abandon_sql = (($abandon_days)
 104                 ? sprintf(" AND `user`.`login_date` > UTC_TIMESTAMP() - INTERVAL %d DAY ", intval($abandon_days))
 105                 : ''
 106         );
 107
 108         $contacts = q("SELECT `contact`.`id` FROM `contact` LEFT JOIN `user` ON `user`.`uid` = `contact`.`uid`
 109                 WHERE ( `rel` = %d OR `rel` = %d ) AND `poll` != ''
 110                 AND `network` != '%s'
 111                 $sql_extra
 112                 AND `self` = 0 AND `contact`.`blocked` = 0 AND `contact`.`readonly` = 0
 113                 AND `user`.`account_expired` = 0 $abandon_sql ORDER BY RAND()",
 114                 intval(CONTACT_IS_SHARING),
 115                 intval(CONTACT_IS_FRIEND),
 116                 dbesc(NETWORK_DIASPORA)
 117         );
 118
 119         if(! count($contacts)) {
 120                 return;
 121         }
 122
 123         foreach($contacts as $c) {
 124
 125                 $res = q("SELECT * FROM `contact` WHERE `id` = %d LIMIT 1",
 126                         intval($c['id'])
 127                 );
 128
 129                 if((! $res) || (! count($res)))
 130                         continue;
 131
 132                 foreach($res as $contact) {
 133
 134                         $xml = false;
 135
 136                         if($manual_id)
 137                                 $contact['last-update'] = '0000-00-00 00:00:00';
 138
 139                         if($contact['priority'] || $contact['subhub']) {
 140
 141                                 $hub_update = true;
 142                                 $update     = false;
 143
 144                                 $t = $contact['last-update'];
 145
 146                                 // We should be getting everything via a hub. But just to be sure, let's check once a day.
 147                                 // (You can make this more or less frequent if desired by setting 'pushpoll_frequency' appropriately)
 148                                 // This also lets us update our subscription to the hub, and add or replace hubs in case it
 149                                 // changed. We will only update hubs once a day, regardless of 'pushpoll_frequency'.
 150
 151
 152                                 if($contact['subhub']) {
 153                                         $interval = get_config('system','pushpoll_frequency');
 154                                         $contact['priority'] = (($interval !== false) ? intval($interval) : 3);
 155                                         $hub_update = false;
 156
 157                                         if((datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day")) || $force)
 158                                                         $hub_update = true;
 159                                 }
 160                                 else
 161                                         $hub_update = false;
 162
 163                                 /**
 164                                  * Based on $contact['priority'], should we poll this site now? Or later?
 165                                  */
 166
 167                                 switch ($contact['priority']) {
 168                                         case 5:
 169                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 month"))
 170                                                         $update = true;
 171                                                 break;
 172                                         case 4:
 173                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 week"))
 174                                                         $update = true;
 175                                                 break;
 176                                         case 3:
 177                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 day"))
 178                                                         $update = true;
 179                                                 break;
 180                                         case 2:
 181                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 12 hour"))
 182                                                         $update = true;
 183                                                 break;
 184                                         case 1:
 185                                         default:
 186                                                 if(datetime_convert('UTC','UTC', 'now') > datetime_convert('UTC','UTC', $t . " + 1 hour"))
 187                                                         $update = true;
 188                                                 break;
 189                                 }
 190                                 if((! $update) && (! $force))
 191                                         continue;
 192                         }
 193
 194                         // Check to see if we are running out of memory - if so spawn a new process and kill this one
 195
 196                         $avail_memory = return_bytes(ini_get('memory_limit'));
 197                         $memused = memory_get_peak_usage(true);
 198                         if(intval($avail_memory)) {
 199                                 if(($memused / $avail_memory) > 0.95) {
 200                                         if($generation + 1 > 10) {
 201                                                 logger('poller: maximum number of spawns exceeded. Terminating.');
 202                                                 killme();
 203                                         }
 204                                         logger('poller: memory exceeded. ' . $memused . ' bytes used. Spawning new poll.');
 205                                         proc_run('php', 'include/poller.php', 'restart', (string) $generation + 1);
 206                                         killme();
 207                                 }
 208                         }
 209
 210                         $importer_uid = $contact['uid'];
 211
 212                         $r = q("SELECT * FROM `contact` WHERE `uid` = %d AND `self` = 1 LIMIT 1",
 213                                 intval($importer_uid)
 214                         );
 215                         if(! count($r))
 216                                 continue;
 217
 218                         $importer = $r[0];
 219
 220                         logger("poller: poll: IMPORTER: {$importer['name']}, CONTACT: {$contact['name']}");
 221
 222                         $last_update = (($contact['last-update'] === '0000-00-00 00:00:00')
 223                                 ? datetime_convert('UTC','UTC','now - 30 days', ATOM_TIME)
 224                                 : datetime_convert('UTC','UTC',$contact['last-update'], ATOM_TIME)
 225                         );
 226
 227                         if($contact['network'] === NETWORK_DFRN) {
 228
 229                                 $idtosend = $orig_id = (($contact['dfrn-id']) ? $contact['dfrn-id'] : $contact['issued-id']);
 230
 231                                 if(intval($contact['duplex']) && $contact['dfrn-id'])
 232                                         $idtosend = '0:' . $orig_id;
 233                                 if(intval($contact['duplex']) && $contact['issued-id'])
 234                                         $idtosend = '1:' . $orig_id;
 235
 236                                 // they have permission to write to us. We already filtered this in the contact query.
 237                                 $perm = 'rw';
 238
 239                                 $url = $contact['poll'] . '?dfrn_id=' . $idtosend
 240                                         . '&dfrn_version=' . DFRN_PROTOCOL_VERSION
 241                                         . '&type=data&last_update=' . $last_update
 242                                         . '&perm=' . $perm ;
 243
 244                                 $handshake_xml = fetch_url($url);
 245
 246                                 logger('poller: handshake with url ' . $url . ' returns xml: ' . $handshake_xml, LOGGER_DATA);
 247
 248
 249                                 if(! $handshake_xml) {
 250                                         logger("poller: $url appears to be dead - marking for death ");
 251                                         // dead connection - might be a transient event, or this might
 252                                         // mean the software was uninstalled or the domain expired.
 253                                         // Will keep trying for one month.
 254                                         mark_for_death($contact);
 255
 256                                         // set the last-update so we don't keep polling
 257
 258                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 259                                                 dbesc(datetime_convert()),
 260                                                 intval($contact['id'])
 261                                         );
 262
 263                                         continue;
 264                                 }
 265
 266                                 if(! strstr($handshake_xml,'<?xml')) {
 267                                         logger('poller: response from ' . $url . ' did not contain XML.');
 268                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 269                                                 dbesc(datetime_convert()),
 270                                                 intval($contact['id'])
 271                                         );
 272                                         continue;
 273                                 }
 274
 275
 276                                 $res = parse_xml_string($handshake_xml);
 277
 278                                 if(intval($res->status) == 1) {
 279                                         logger("poller: $url replied status 1 - marking for death ");
 280
 281                                         // we may not be friends anymore. Will keep trying for one month.
 282                                         // set the last-update so we don't keep polling
 283
 284                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 285                                                 dbesc(datetime_convert()),
 286                                                 intval($contact['id'])
 287                                         );
 288
 289                                         mark_for_death($contact);
 290                                 }
 291                                 else {
 292                                         if($contact['term-date'] != '0000-00-00 00:00:00') {
 293                                                 logger("poller: $url back from the dead - removing mark for death");
 294                                                 unmark_for_death($contact);
 295                                         }
 296                                 }
 297
 298                                 if((intval($res->status) != 0) || (! strlen($res->challenge)) || (! strlen($res->dfrn_id)))
 299                                         continue;
 300
 301                                 if(((float) $res->dfrn_version > 2.21) && ($contact['poco'] == '')) {
 302                                         q("update contact set poco = '%s' where id = %d limit 1",
 303                                                 dbesc(str_replace('/profile/','/poco/', $contact['url'])),
 304                                                 intval($contact['id'])
 305                                         );
 306                                 }
 307
 308                                 $postvars = array();
 309
 310                                 $sent_dfrn_id = hex2bin((string) $res->dfrn_id);
 311                                 $challenge    = hex2bin((string) $res->challenge);
 312
 313                                 $final_dfrn_id = '';
 314
 315                                 if(($contact['duplex']) && strlen($contact['prvkey'])) {
 316                                         openssl_private_decrypt($sent_dfrn_id,$final_dfrn_id,$contact['prvkey']);
 317                                         openssl_private_decrypt($challenge,$postvars['challenge'],$contact['prvkey']);
 318                                 }
 319                                 else {
 320                                         openssl_public_decrypt($sent_dfrn_id,$final_dfrn_id,$contact['pubkey']);
 321                                         openssl_public_decrypt($challenge,$postvars['challenge'],$contact['pubkey']);
 322                                 }
 323
 324                                 $final_dfrn_id = substr($final_dfrn_id, 0, strpos($final_dfrn_id, '.'));
 325
 326                                 if(strpos($final_dfrn_id,':') == 1)
 327                                         $final_dfrn_id = substr($final_dfrn_id,2);
 328
 329                                 if($final_dfrn_id != $orig_id) {
 330                                         logger('poller: ID did not decode: ' . $contact['id'] . ' orig: ' . $orig_id . ' final: ' . $final_dfrn_id);
 331                                         // did not decode properly - cannot trust this site
 332                                         continue;
 333                                 }
 334
 335                                 $postvars['dfrn_id'] = $idtosend;
 336                                 $postvars['dfrn_version'] = DFRN_PROTOCOL_VERSION;
 337                                 $postvars['perm'] = 'rw';
 338
 339                                 $xml = post_url($contact['poll'],$postvars);
 340                         }
 341                         elseif(($contact['network'] === NETWORK_OSTATUS)
 342                                 || ($contact['network'] === NETWORK_DIASPORA)
 343                                 || ($contact['network'] === NETWORK_FEED) ) {
 344
 345                                 // Upgrading DB fields from an older Friendika version
 346                                 // Will only do this once per notify-enabled OStatus contact
 347                                 // or if relationship changes
 348
 349                                 $stat_writeable = ((($contact['notify']) && ($contact['rel'] == CONTACT_IS_FOLLOWER || $contact['rel'] == CONTACT_IS_FRIEND)) ? 1 : 0);
 350
 351                                 if($stat_writeable != $contact['writable']) {
 352                                         q("UPDATE `contact` SET `writable` = %d WHERE `id` = %d LIMIT 1",
 353                                                 intval($stat_writeable),
 354                                                 intval($contact['id'])
 355                                         );
 356                                 }
 357
 358                                 // Are we allowed to import from this person?
 359
 360                                 if($contact['rel'] == CONTACT_IS_FOLLOWER || $contact['blocked'] || $contact['readonly'])
 361                                         continue;
 362
 363                                 $xml = fetch_url($contact['poll']);
 364                         }
 365                         elseif($contact['network'] === NETWORK_MAIL) {
 366
 367                                 $mail_disabled = ((function_exists('imap_open') && (! get_config('system','imap_disabled'))) ? 0 : 1);
 368                                 if($mail_disabled)
 369                                         continue;
 370
 371                                 $mbox = null;
 372                                 $x = q("SELECT `prvkey` FROM `user` WHERE `uid` = %d LIMIT 1",
 373                                         intval($importer_uid)
 374                                 );
 375                                 $mailconf = q("SELECT * FROM `mailacct` WHERE `server` != '' AND `uid` = %d LIMIT 1",
 376                                         intval($importer_uid)
 377                                 );
 378                                 if(count($x) && count($mailconf)) {
 379                                     $mailbox = construct_mailbox_name($mailconf[0]);
 380                                         $password = '';
 381                                         openssl_private_decrypt(hex2bin($mailconf[0]['pass']),$password,$x[0]['prvkey']);
 382                                         $mbox = email_connect($mailbox,$mailconf[0]['user'],$password);
 383                                         unset($password);
 384                                         if($mbox) {
 385                                                 q("UPDATE `mailacct` SET `last_check` = '%s' WHERE `id` = %d AND `uid` = %d LIMIT 1",
 386                                                         dbesc(datetime_convert()),
 387                                                         intval($mailconf[0]['id']),
 388                                                         intval($importer_uid)
 389                                                 );
 390                                         }
 391                                 }
 392                                 if($mbox) {
 393
 394                                         $msgs = email_poll($mbox,$contact['addr']);
 395
 396                                         if(count($msgs)) {
 397                                                 foreach($msgs as $msg_uid) {
 398                                                         $datarray = array();
 399                                                         $meta = email_msg_meta($mbox,$msg_uid);
 400                                                         $headers = email_msg_headers($mbox,$msg_uid);
 401
 402                                                         // look for a 'references' header and try and match with a parent item we have locally.
 403
 404                                                         $raw_refs = ((x($headers,'references')) ? str_replace("\t",'',$headers['references']) : '');
 405                                                         $datarray['uri'] = trim($meta->message_id,'<>');
 406
 407                                                         if($raw_refs) {
 408                                                                 $refs_arr = explode(' ', $raw_refs);
 409                                                                 if(count($refs_arr)) {
 410                                                                         for($x = 0; $x < count($refs_arr); $x ++)
 411                                                                                 $refs_arr[$x] = "'" . str_replace(array('<','>',' '),array('','',''),dbesc($refs_arr[$x])) . "'";
 412                                                                 }
 413                                                                 $qstr = implode(',',$refs_arr);
 414                                                                 $r = q("SELECT `uri` , `parent-uri` FROM `item` WHERE `uri` IN ( $qstr ) AND `uid` = %d LIMIT 1",
 415                                                                         intval($importer_uid)
 416                                                                 );
 417                                                                 if(count($r))
 418                                                                         $datarray['parent-uri'] = $r[0]['uri'];
 419                                                         }
 420
 421
 422                                                         if(! x($datarray,'parent-uri'))
 423                                                                 $datarray['parent-uri'] = $datarray['uri'];
 424
 425                                                         // Have we seen it before?
 426                                                         $r = q("SELECT * FROM `item` WHERE `uid` = %d AND `uri` = '%s' LIMIT 1",
 427                                                                 intval($importer_uid),
 428                                                                 dbesc($datarray['uri'])
 429                                                         );
 430
 431                                                         if(count($r)) {
 432                                                                 if($meta->deleted && ! $r[0]['deleted']) {
 433                                                                         q("UPDATE `item` SET `deleted` = 1, `changed` = '%s' WHERE `id` = %d LIMIT 1",
 434                                                                                 dbesc(datetime_convert()),
 435                                                                                 intval($r[0]['id'])
 436                                                                         );
 437                                                                 }
 438                                                                 continue;
 439                                                         }
 440                                                         $datarray['title'] = notags(trim($meta->subject));
 441                                                         $datarray['created'] = datetime_convert('UTC','UTC',$meta->date);
 442
 443                                                         $r = email_get_msg($mbox,$msg_uid);
 444                                                         if(! $r)
 445                                                                 continue;
 446                                                         $datarray['body'] = escape_tags($r['body']);
 447
 448                                                         // some mailing lists have the original author as 'from' - add this sender info to msg body.
 449                                                         // todo: adding a gravatar for the original author would be cool
 450
 451                                                         if(! stristr($meta->from,$contact['addr']))
 452                                                                 $datarray['body'] = t('From: ') . escape_tags($meta->from) . "\n\n" . $datarray['body'];
 453
 454                                                         $datarray['uid'] = $importer_uid;
 455                                                         $datarray['contact-id'] = $contact['id'];
 456                                                         if($datarray['parent-uri'] === $datarray['uri'])
 457                                                                 $datarray['private'] = 1;
 458                                                         if(! get_pconfig($importer_uid,'system','allow_public_email_replies')) {
 459                                                                 $datarray['private'] = 1;
 460                                                                 $datarray['allow_cid'] = '<' . $contact['id'] . '>';
 461                                                         }
 462                                                         $datarray['author-name'] = $contact['name'];
 463                                                         $datarray['author-link'] = 'mailbox';
 464                                                         $datarray['author-avatar'] = $contact['photo'];
 465
 466                                                         $stored_item = item_store($datarray);
 467                                                         q("UPDATE `item` SET `last-child` = 0 WHERE `parent-uri` = '%s' AND `uid` = %d",
 468                                                                 dbesc($datarray['parent-uri']),
 469                                                                 intval($importer_uid)
 470                                                         );
 471                                                         q("UPDATE `item` SET `last-child` = 1 WHERE `id` = %d LIMIT 1",
 472                                                                 intval($stored_item)
 473                                                         );
 474                                                 }
 475                                         }
 476
 477                                         imap_close($mbox);
 478                                 }
 479                         }
 480                         elseif($contact['network'] === NETWORK_FACEBOOK) {
 481                                 // This is picked up by the Facebook plugin on a cron hook.
 482                                 // Ignored here.
 483                         }
 484
 485                         if($xml) {
 486                                 logger('poller: received xml : ' . $xml, LOGGER_DATA);
 487
 488                                 if(! strstr($xml,'<?xml')) {
 489                                         logger('poller: post_handshake: response from ' . $url . ' did not contain XML.');
 490                                         $r = q("UPDATE `contact` SET `last-update` = '%s' WHERE `id` = %d LIMIT 1",
 491                                                 dbesc(datetime_convert()),
 492                                                 intval($contact['id'])
 493                                         );
 494                                         continue;
 495                                 }
 496
 497
 498                                 consume_feed($xml,$importer,$contact,$hub,1,1);
 499
 500                                 // do it twice. Ensures that children of parents which may be later in the stream aren't tossed
 501
 502                                 consume_feed($xml,$importer,$contact,$hub,1,2);
 503
 504                                 $hubmode = 'subscribe';
 505                                 if($contact['network'] === NETWORK_DFRN || $contact['blocked'] || $contact['readonly'])
 506                                         $hubmode = 'unsubscribe';
 507
 508                                 if((strlen($hub)) && ($hub_update) && ($contact['rel'] != CONTACT_IS_FOLLOWER)) {
 509                                         logger('poller: hub ' . $hubmode . ' : ' . $hub . ' contact name : ' . $contact['name'] . ' local user : ' . $importer['name']);
 510                                         $hubs = explode(',', $hub);
 511                                         if(count($hubs)) {
 512                                                 foreach($hubs as $h) {
 513                                                         $h = trim($h);
 514                                                         if(! strlen($h))
 515                                                                 continue;
 516                                                         subscribe_to_hub($h,$importer,$contact,$hubmode);
 517                                                 }
 518                                         }
 519                                 }
 520                         }
 521
 522
 523                         $updated = datetime_convert();
 524
 525                         $r = q("UPDATE `contact` SET `last-update` = '%s', `success_update` = '%s' WHERE `id` = %d LIMIT 1",
 526                                 dbesc($updated),
 527                                 dbesc($updated),
 528                                 intval($contact['id'])
 529                         );
 530
 531                         // loop - next contact
 532                 }
 533         }
 534
 535
 536         return;
 537 }
 538
 539 if (array_search(__file__,get_included_files())===0){
 540   poller_run($argv,$argc);
 541   killme();
 542 }