src/Model/GServer.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Model;
  23
  24 use DOMDocument;
  25 use DOMXPath;
  26 use Friendica\Core\Logger;
  27 use Friendica\Core\Protocol;
  28 use Friendica\Core\System;
  29 use Friendica\Core\Worker;
  30 use Friendica\Database\Database;
  31 use Friendica\Database\DBA;
  32 use Friendica\DI;
  33 use Friendica\Module\Register;
  34 use Friendica\Network\CurlResult;
  35 use Friendica\Protocol\Relay;
  36 use Friendica\Util\DateTimeFormat;
  37 use Friendica\Util\Network;
  38 use Friendica\Util\Strings;
  39 use Friendica\Util\XML;
  40
  41 /**
  42  * This class handles GServer related functions
  43  */
  44 class GServer
  45 {
  46         // Directory types
  47         const DT_NONE = 0;
  48         const DT_POCO = 1;
  49         const DT_MASTODON = 2;
  50
  51         // Methods to detect server types
  52
  53         // Non endpoint specific methods
  54         const DETECT_MANUAL = 0;
  55         const DETECT_HEADER = 1;
  56         const DETECT_BODY = 2;
  57
  58         // Implementation specific endpoints
  59         const DETECT_FRIENDIKA = 10;
  60         const DETECT_FRIENDICA = 11;
  61         const DETECT_STATUSNET = 12;
  62         const DETECT_GNUSOCIAL = 13;
  63         const DETECT_CONFIG_JSON = 14; // Statusnet, GNU Social, Older Hubzilla/Redmatrix
  64         const DETECT_SITEINFO_JSON = 15; // Newer Hubzilla
  65         const DETECT_MASTODON_API = 16;
  66         const DETECT_STATUS_PHP = 17; // Nextcloud
  67
  68         // Standardized endpoints
  69         const DETECT_STATISTICS_JSON = 100;
  70         const DETECT_NODEINFO_1 = 101;
  71         const DETECT_NODEINFO_2 = 102;
  72
  73         /**
  74          * Get the ID for the given server URL
  75          *
  76          * @param string $url
  77          * @param boolean $no_check Don't check if the server hadn't been found
  78          * @return int gserver id
  79          */
  80         public static function getID(string $url, bool $no_check = false)
  81         {
  82                 if (empty($url)) {
  83                         return null;
  84                 }
  85
  86                 $url = self::cleanURL($url);
  87
  88                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
  89                 if (DBA::isResult($gserver)) {
  90                         Logger::info('Got ID for URL', ['id' => $gserver['id'], 'url' => $url, 'callstack' => System::callstack(20)]);
  91                         return $gserver['id'];
  92                 }
  93
  94                 if ($no_check || !self::check($url)) {
  95                         return null;
  96                 }
  97
  98                 return self::getID($url, true);
  99         }
 100
 101         /**
 102          * Checks if the given server is reachable
 103          *
 104          * @param string  $profile URL of the given profile
 105          * @param string  $server  URL of the given server (If empty, taken from profile)
 106          * @param string  $network Network value that is used, when detection failed
 107          * @param boolean $force   Force an update.
 108          *
 109          * @return boolean 'true' if server seems vital
 110          */
 111         public static function reachable(string $profile, string $server = '', string $network = '', bool $force = false)
 112         {
 113                 if ($server == '') {
 114                         $contact = Contact::getByURL($profile, null, ['baseurl']);
 115                         if (!empty($contact['baseurl'])) {
 116                                 $server = $contact['baseurl'];
 117                         }
 118                 }
 119
 120                 if ($server == '') {
 121                         return true;
 122                 }
 123
 124                 return self::check($server, $network, $force);
 125         }
 126
 127         /**
 128          * Decides if a server needs to be updated, based upon several date fields
 129          *
 130          * @param date $created      Creation date of that server entry
 131          * @param date $updated      When had the server entry be updated
 132          * @param date $last_failure Last failure when contacting that server
 133          * @param date $last_contact Last time the server had been contacted
 134          *
 135          * @return boolean Does the server record needs an update?
 136          */
 137         public static function updateNeeded($created, $updated, $last_failure, $last_contact)
 138         {
 139                 $now = strtotime(DateTimeFormat::utcNow());
 140
 141                 if ($updated > $last_contact) {
 142                         $contact_time = strtotime($updated);
 143                 } else {
 144                         $contact_time = strtotime($last_contact);
 145                 }
 146
 147                 $failure_time = strtotime($last_failure);
 148                 $created_time = strtotime($created);
 149
 150                 // If there is no "created" time then use the current time
 151                 if ($created_time <= 0) {
 152                         $created_time = $now;
 153                 }
 154
 155                 // If the last contact was less than 24 hours then don't update
 156                 if (($now - $contact_time) < (60 * 60 * 24)) {
 157                         return false;
 158                 }
 159
 160                 // If the last failure was less than 24 hours then don't update
 161                 if (($now - $failure_time) < (60 * 60 * 24)) {
 162                         return false;
 163                 }
 164
 165                 // If the last contact was less than a week ago and the last failure is older than a week then don't update
 166                 //if ((($now - $contact_time) < (60 * 60 * 24 * 7)) && ($contact_time > $failure_time))
 167                 //      return false;
 168
 169                 // If the last contact time was more than a week ago and the contact was created more than a week ago, then only try once a week
 170                 if ((($now - $contact_time) > (60 * 60 * 24 * 7)) && (($now - $created_time) > (60 * 60 * 24 * 7)) && (($now - $failure_time) < (60 * 60 * 24 * 7))) {
 171                         return false;
 172                 }
 173
 174                 // If the last contact time was more than a month ago and the contact was created more than a month ago, then only try once a month
 175                 if ((($now - $contact_time) > (60 * 60 * 24 * 30)) && (($now - $created_time) > (60 * 60 * 24 * 30)) && (($now - $failure_time) < (60 * 60 * 24 * 30))) {
 176                         return false;
 177                 }
 178
 179                 return true;
 180         }
 181
 182         /**
 183          * Checks the state of the given server.
 184          *
 185          * @param string  $server_url    URL of the given server
 186          * @param string  $network       Network value that is used, when detection failed
 187          * @param boolean $force         Force an update.
 188          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 189          *
 190          * @return boolean 'true' if server seems vital
 191          */
 192         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false)
 193         {
 194                 $server_url = self::cleanURL($server_url);
 195
 196                 if ($server_url == '') {
 197                         return false;
 198                 }
 199
 200                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
 201                 if (DBA::isResult($gserver)) {
 202                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
 203                                 $fields = ['created' => DateTimeFormat::utcNow()];
 204                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
 205                                 DBA::update('gserver', $fields, $condition);
 206                         }
 207
 208                         $last_contact = $gserver['last_contact'];
 209                         $last_failure = $gserver['last_failure'];
 210
 211                         // See discussion under https://forum.friendi.ca/display/0b6b25a8135aabc37a5a0f5684081633
 212                         // It can happen that a zero date is in the database, but storing it again is forbidden.
 213                         if ($last_contact < DBA::NULL_DATETIME) {
 214                                 $last_contact = DBA::NULL_DATETIME;
 215                         }
 216
 217                         if ($last_failure < DBA::NULL_DATETIME) {
 218                                 $last_failure = DBA::NULL_DATETIME;
 219                         }
 220
 221                         if (!$force && !self::updateNeeded($gserver['created'], '', $last_failure, $last_contact)) {
 222                                 Logger::info('No update needed', ['server' => $server_url]);
 223                                 return ($last_contact >= $last_failure);
 224                         }
 225                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force, 'Created' => $gserver['created'], 'Failure' => $last_failure, 'Contact' => $last_contact]);
 226                 } else {
 227                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
 228                 }
 229
 230                 return self::detect($server_url, $network, $only_nodeinfo);
 231         }
 232
 233         /**
 234          * Set failed server status
 235          *
 236          * @param string $url
 237          */
 238         private static function setFailure(string $url)
 239         {
 240                 if (DBA::exists('gserver', ['nurl' => Strings::normaliseLink($url)])) {
 241                         DBA::update('gserver', ['failed' => true, 'last_failure' => DateTimeFormat::utcNow(), 'detection-method' => null],
 242                         ['nurl' => Strings::normaliseLink($url)]);
 243                         Logger::info('Set failed status for existing server', ['url' => $url]);
 244                         return;
 245                 }
 246                 DBA::insert('gserver', ['url' => $url, 'nurl' => Strings::normaliseLink($url),
 247                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
 248                         'failed' => true, 'last_failure' => DateTimeFormat::utcNow()]);
 249                 Logger::info('Set failed status for new server', ['url' => $url]);
 250         }
 251
 252         /**
 253          * Remove unwanted content from the given URL
 254          *
 255          * @param string $url
 256          * @return string cleaned URL
 257          */
 258         public static function cleanURL(string $url)
 259         {
 260                 $url = trim($url, '/');
 261                 $url = str_replace('/index.php', '', $url);
 262
 263                 $urlparts = parse_url($url);
 264                 unset($urlparts['user']);
 265                 unset($urlparts['pass']);
 266                 unset($urlparts['query']);
 267                 unset($urlparts['fragment']);
 268                 return Network::unparseURL($urlparts);
 269         }
 270
 271         /**
 272          * Return the base URL
 273          *
 274          * @param string $url
 275          * @return string base URL
 276          */
 277         private static function getBaseURL(string $url)
 278         {
 279                 $urlparts = parse_url(self::cleanURL($url));
 280                 unset($urlparts['path']);
 281                 return Network::unparseURL($urlparts);
 282         }
 283
 284         /**
 285          * Detect server data (type, protocol, version number, ...)
 286          * The detected data is then updated or inserted in the gserver table.
 287          *
 288          * @param string  $url           URL of the given server
 289          * @param string  $network       Network value that is used, when detection failed
 290          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 291          *
 292          * @return boolean 'true' if server could be detected
 293          */
 294         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false)
 295         {
 296                 Logger::info('Detect server type', ['server' => $url]);
 297                 $serverdata = ['detection-method' => self::DETECT_MANUAL];
 298
 299                 $original_url = $url;
 300
 301                 // Remove URL content that is not supposed to exist for a server url
 302                 $url = self::cleanURL($url);
 303
 304                 // Get base URL
 305                 $baseurl = self::getBaseURL($url);
 306
 307                 // If the URL missmatches, then we mark the old entry as failure
 308                 if ($url != $original_url) {
 309                         DBA::update('gserver', ['failed' => true, 'last_failure' => DateTimeFormat::utcNow()],
 310                                 ['nurl' => Strings::normaliseLink($original_url)]);
 311                 }
 312
 313                 // When a nodeinfo is present, we don't need to dig further
 314                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
 315                 $curlResult = DI::httpRequest()->get($url . '/.well-known/nodeinfo', ['timeout' => $xrd_timeout]);
 316                 if ($curlResult->isTimeout()) {
 317                         self::setFailure($url);
 318                         return false;
 319                 }
 320
 321                 $nodeinfo = self::fetchNodeinfo($url, $curlResult);
 322                 if ($only_nodeinfo && empty($nodeinfo)) {
 323                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 324                         self::setFailure($url);
 325                         return false;
 326                 }
 327
 328                 // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 329                 if (empty($nodeinfo)) {
 330                         $nodeinfo = self::fetchStatistics($url);
 331                 }
 332
 333                 // If that didn't work out well, we use some protocol specific endpoints
 334                 // For Friendica and Zot based networks we have to dive deeper to reveal more details
 335                 if (empty($nodeinfo['network']) || in_array($nodeinfo['network'], [Protocol::DFRN, Protocol::ZOT])) {
 336                         if (!empty($nodeinfo['detection-method'])) {
 337                                 $serverdata['detection-method'] = $nodeinfo['detection-method'];
 338                         }
 339
 340                         // Fetch the landing page, possibly it reveals some data
 341                         if (empty($nodeinfo['network'])) {
 342                                 if ($baseurl == $url) {
 343                                         $basedata = $serverdata;
 344                                 } else {
 345                                         $basedata = ['detection-method' => self::DETECT_MANUAL];
 346                                 }
 347
 348                                 $curlResult = DI::httpRequest()->get($baseurl, ['timeout' => $xrd_timeout]);
 349                                 if ($curlResult->isSuccess()) {
 350                                         $basedata = self::analyseRootHeader($curlResult, $basedata);
 351                                         $basedata = self::analyseRootBody($curlResult, $basedata, $baseurl);
 352                                 }
 353
 354                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody()) || self::invalidBody($curlResult->getBody())) {
 355                                         self::setFailure($url);
 356                                         return false;
 357                                 }
 358
 359                                 if ($baseurl == $url) {
 360                                         $serverdata = $basedata;
 361                                 } else {
 362                                         // When the base path doesn't seem to contain a social network we try the complete path.
 363                                         // Most detectable system have to be installed in the root directory.
 364                                         // We checked the base to avoid false positives.
 365                                         $curlResult = DI::httpRequest()->get($url, ['timeout' => $xrd_timeout]);
 366                                         if ($curlResult->isSuccess()) {
 367                                                 $urldata = self::analyseRootHeader($curlResult, $serverdata);
 368                                                 $urldata = self::analyseRootBody($curlResult, $urldata, $url);
 369
 370                                                 $comparebase = $basedata;
 371                                                 unset($comparebase['info']);
 372                                                 unset($comparebase['site_name']);
 373                                                 $compareurl = $urldata;
 374                                                 unset($compareurl['info']);
 375                                                 unset($compareurl['site_name']);
 376
 377                                                 // We assume that no one will install the identical system in the root and a subfolder
 378                                                 if (!empty(array_diff($comparebase, $compareurl))) {
 379                                                         $serverdata = $urldata;
 380                                                 }
 381                                         }
 382                                 }
 383                         }
 384
 385                         if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ACTIVITYPUB)) {
 386                                 $serverdata = self::detectMastodonAlikes($url, $serverdata);
 387                         }
 388
 389                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 390                         // With this check we don't have to waste time and ressources for dead systems.
 391                         // Also this hopefully prevents us from receiving abuse messages.
 392                         if (empty($serverdata['network']) && !self::validHostMeta($url)) {
 393                                 self::setFailure($url);
 394                                 return false;
 395                         }
 396
 397                         if (empty($serverdata['network']) || in_array($serverdata['network'], [Protocol::DFRN, Protocol::ACTIVITYPUB])) {
 398                                 $serverdata = self::detectFriendica($url, $serverdata);
 399                         }
 400
 401                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 402                         if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ZOT)) {
 403                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 404                         }
 405
 406                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations
 407                         if (empty($serverdata['network'])) {
 408                                 $serverdata = self::detectHubzilla($url, $serverdata);
 409                         }
 410
 411                         if (empty($serverdata['network'])) {
 412                                 $serverdata = self::detectNextcloud($url, $serverdata);
 413                         }
 414
 415                         if (empty($serverdata['network'])) {
 416                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 417                         }
 418
 419                         $serverdata = array_merge($nodeinfo, $serverdata);
 420                 } else {
 421                         $serverdata = $nodeinfo;
 422                 }
 423
 424                 // Detect the directory type
 425                 $serverdata['directory-type'] = self::DT_NONE;
 426                 $serverdata = self::checkPoCo($url, $serverdata);
 427                 $serverdata = self::checkMastodonDirectory($url, $serverdata);
 428
 429                 // We can't detect the network type. Possibly it is some system that we don't know yet
 430                 if (empty($serverdata['network'])) {
 431                         $serverdata['network'] = Protocol::PHANTOM;
 432                 }
 433
 434                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 435                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 436                         $serverdata['network'] = $network;
 437                 }
 438
 439                 $serverdata['url'] = $url;
 440                 $serverdata['nurl'] = Strings::normaliseLink($url);
 441
 442                 // We take the highest number that we do find
 443                 $registeredUsers = $serverdata['registered-users'] ?? 0;
 444
 445                 // On an active server there has to be at least a single user
 446                 if (($serverdata['network'] != Protocol::PHANTOM) && ($registeredUsers == 0)) {
 447                         $registeredUsers = 1;
 448                 }
 449
 450                 if ($serverdata['network'] == Protocol::PHANTOM) {
 451                         $serverdata['registered-users'] = $registeredUsers;
 452                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 453                 }
 454
 455                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 456                 $serverdata['failed'] = false;
 457
 458                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 459                 if (!DBA::isResult($gserver)) {
 460                         $serverdata['created'] = DateTimeFormat::utcNow();
 461                         $ret = DBA::insert('gserver', $serverdata);
 462                         $id = DBA::lastInsertId();
 463                 } else {
 464                         // Don't override the network with 'unknown' when there had been a valid entry before
 465                         if (($serverdata['network'] == Protocol::PHANTOM) && !empty($gserver['network'])) {
 466                                 unset($serverdata['network']);
 467                         }
 468
 469                         $ret = DBA::update('gserver', $serverdata, ['nurl' => $serverdata['nurl']]);
 470                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 471                         if (DBA::isResult($gserver)) {
 472                                 $id = $gserver['id'];
 473                         }
 474                 }
 475
 476                 if (!empty($serverdata['network']) && !empty($id) && ($serverdata['network'] != Protocol::PHANTOM)) {
 477                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 478                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id]);
 479                         $max_users = max($apcontacts, $contacts, $registeredUsers);
 480                         if ($max_users > $registeredUsers) {
 481                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 482                                 DBA::update('gserver', ['registered-users' => $max_users], ['id' => $id]);
 483                         }
 484                 }
 485
 486                 if (!empty($serverdata['network']) && in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 487                         self::discoverRelay($url);
 488                 }
 489
 490                 return $ret;
 491         }
 492
 493         /**
 494          * Fetch relay data from a given server url
 495          *
 496          * @param string $server_url address of the server
 497          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 498          */
 499         private static function discoverRelay(string $server_url)
 500         {
 501                 Logger::info('Discover relay data', ['server' => $server_url]);
 502
 503                 $curlResult = DI::httpRequest()->get($server_url . '/.well-known/x-social-relay');
 504                 if (!$curlResult->isSuccess()) {
 505                         return;
 506                 }
 507
 508                 $data = json_decode($curlResult->getBody(), true);
 509                 if (!is_array($data)) {
 510                         return;
 511                 }
 512
 513                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 514                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
 515
 516                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 517                         $data['scope'] = '';
 518                         $data['subscribe'] = false;
 519                         $data['tags'] = [];
 520                 }
 521
 522                 $gserver = DBA::selectFirst('gserver', ['id', 'url', 'network', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 523                 if (!DBA::isResult($gserver)) {
 524                         return;
 525                 }
 526
 527                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 528                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 529                         DBA::update('gserver', $fields, ['id' => $gserver['id']]);
 530                 }
 531
 532                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 533
 534                 if ($data['scope'] == 'tags') {
 535                         // Avoid duplicates
 536                         $tags = [];
 537                         foreach ($data['tags'] as $tag) {
 538                                 $tag = mb_strtolower($tag);
 539                                 if (strlen($tag) < 100) {
 540                                         $tags[$tag] = $tag;
 541                                 }
 542                         }
 543
 544                         foreach ($tags as $tag) {
 545                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], Database::INSERT_IGNORE);
 546                         }
 547                 }
 548
 549                 // Create or update the relay contact
 550                 $fields = [];
 551                 if (isset($data['protocols'])) {
 552                         if (isset($data['protocols']['diaspora'])) {
 553                                 $fields['network'] = Protocol::DIASPORA;
 554
 555                                 if (isset($data['protocols']['diaspora']['receive'])) {
 556                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 557                                 } elseif (is_string($data['protocols']['diaspora'])) {
 558                                         $fields['batch'] = $data['protocols']['diaspora'];
 559                                 }
 560                         }
 561
 562                         if (isset($data['protocols']['dfrn'])) {
 563                                 $fields['network'] = Protocol::DFRN;
 564
 565                                 if (isset($data['protocols']['dfrn']['receive'])) {
 566                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 567                                 } elseif (is_string($data['protocols']['dfrn'])) {
 568                                         $fields['batch'] = $data['protocols']['dfrn'];
 569                                 }
 570                         }
 571
 572                         if (isset($data['protocols']['activitypub'])) {
 573                                 $fields['network'] = Protocol::ACTIVITYPUB;
 574
 575                                 if (!empty($data['protocols']['activitypub']['actor'])) {
 576                                         $fields['url'] = $data['protocols']['activitypub']['actor'];
 577                                 }
 578                                 if (!empty($data['protocols']['activitypub']['receive'])) {
 579                                         $fields['batch'] = $data['protocols']['activitypub']['receive'];
 580                                 }
 581                         }
 582                 }
 583
 584                 Logger::info('Discovery ended', ['server' => $server_url, 'data' => $fields]);
 585
 586                 Relay::updateContact($gserver, $fields);
 587         }
 588
 589         /**
 590          * Fetch server data from '/statistics.json' on the given server
 591          *
 592          * @param string $url URL of the given server
 593          *
 594          * @return array server data
 595          */
 596         private static function fetchStatistics(string $url)
 597         {
 598                 $curlResult = DI::httpRequest()->get($url . '/statistics.json');
 599                 if (!$curlResult->isSuccess()) {
 600                         return [];
 601                 }
 602
 603                 $data = json_decode($curlResult->getBody(), true);
 604                 if (empty($data)) {
 605                         return [];
 606                 }
 607
 608                 $serverdata = ['detection-method' => self::DETECT_STATISTICS_JSON];
 609
 610                 if (!empty($data['version'])) {
 611                         $serverdata['version'] = $data['version'];
 612                         // Version numbers on statistics.json are presented with additional info, e.g.:
 613                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 614                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 615                 }
 616
 617                 if (!empty($data['name'])) {
 618                         $serverdata['site_name'] = $data['name'];
 619                 }
 620
 621                 if (!empty($data['network'])) {
 622                         $serverdata['platform'] = strtolower($data['network']);
 623
 624                         if ($serverdata['platform'] == 'diaspora') {
 625                                 $serverdata['network'] = Protocol::DIASPORA;
 626                         } elseif ($serverdata['platform'] == 'friendica') {
 627                                 $serverdata['network'] = Protocol::DFRN;
 628                         } elseif ($serverdata['platform'] == 'hubzilla') {
 629                                 $serverdata['network'] = Protocol::ZOT;
 630                         } elseif ($serverdata['platform'] == 'redmatrix') {
 631                                 $serverdata['network'] = Protocol::ZOT;
 632                         }
 633                 }
 634
 635
 636                 if (!empty($data['registrations_open'])) {
 637                         $serverdata['register_policy'] = Register::OPEN;
 638                 } else {
 639                         $serverdata['register_policy'] = Register::CLOSED;
 640                 }
 641
 642                 return $serverdata;
 643         }
 644
 645         /**
 646          * Detect server type by using the nodeinfo data
 647          *
 648          * @param string     $url        address of the server
 649          * @param CurlResult $curlResult
 650          * @return array Server data
 651          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 652          */
 653         private static function fetchNodeinfo(string $url, CurlResult $curlResult)
 654         {
 655                 if (!$curlResult->isSuccess()) {
 656                         return [];
 657                 }
 658
 659                 $nodeinfo = json_decode($curlResult->getBody(), true);
 660
 661                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
 662                         return [];
 663                 }
 664
 665                 $nodeinfo1_url = '';
 666                 $nodeinfo2_url = '';
 667
 668                 foreach ($nodeinfo['links'] as $link) {
 669                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
 670                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
 671                                 continue;
 672                         }
 673                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 674                                 $nodeinfo1_url = $link['href'];
 675                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 676                                 $nodeinfo2_url = $link['href'];
 677                         }
 678                 }
 679
 680                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 681                         return [];
 682                 }
 683
 684                 $server = [];
 685
 686                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 687                 if (!empty($nodeinfo2_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo2_url, PHP_URL_HOST))) {
 688                         $server = self::parseNodeinfo2($nodeinfo2_url);
 689                 }
 690
 691                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 692                 if (empty($server) && !empty($nodeinfo1_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo1_url, PHP_URL_HOST))) {
 693                         $server = self::parseNodeinfo1($nodeinfo1_url);
 694                 }
 695
 696                 return $server;
 697         }
 698
 699         /**
 700          * Parses Nodeinfo 1
 701          *
 702          * @param string $nodeinfo_url address of the nodeinfo path
 703          * @return array Server data
 704          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 705          */
 706         private static function parseNodeinfo1(string $nodeinfo_url)
 707         {
 708                 $curlResult = DI::httpRequest()->get($nodeinfo_url);
 709
 710                 if (!$curlResult->isSuccess()) {
 711                         return [];
 712                 }
 713
 714                 $nodeinfo = json_decode($curlResult->getBody(), true);
 715
 716                 if (!is_array($nodeinfo)) {
 717                         return [];
 718                 }
 719
 720                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
 721                         'register_policy' => Register::CLOSED];
 722
 723                 if (!empty($nodeinfo['openRegistrations'])) {
 724                         $server['register_policy'] = Register::OPEN;
 725                 }
 726
 727                 if (is_array($nodeinfo['software'])) {
 728                         if (!empty($nodeinfo['software']['name'])) {
 729                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 730                         }
 731
 732                         if (!empty($nodeinfo['software']['version'])) {
 733                                 $server['version'] = $nodeinfo['software']['version'];
 734                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 735                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 736                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 737                         }
 738                 }
 739
 740                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 741                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 742                 }
 743
 744                 if (!empty($nodeinfo['usage']['users']['total'])) {
 745                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 746                 }
 747
 748                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
 749                         $protocols = [];
 750                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
 751                                 $protocols[$protocol] = true;
 752                         }
 753
 754                         if (!empty($protocols['friendica'])) {
 755                                 $server['network'] = Protocol::DFRN;
 756                         } elseif (!empty($protocols['activitypub'])) {
 757                                 $server['network'] = Protocol::ACTIVITYPUB;
 758                         } elseif (!empty($protocols['diaspora'])) {
 759                                 $server['network'] = Protocol::DIASPORA;
 760                         } elseif (!empty($protocols['ostatus'])) {
 761                                 $server['network'] = Protocol::OSTATUS;
 762                         } elseif (!empty($protocols['gnusocial'])) {
 763                                 $server['network'] = Protocol::OSTATUS;
 764                         } elseif (!empty($protocols['zot'])) {
 765                                 $server['network'] = Protocol::ZOT;
 766                         }
 767                 }
 768
 769                 if (empty($server)) {
 770                         return [];
 771                 }
 772
 773                 return $server;
 774         }
 775
 776         /**
 777          * Parses Nodeinfo 2
 778          *
 779          * @param string $nodeinfo_url address of the nodeinfo path
 780          * @return array Server data
 781          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 782          */
 783         private static function parseNodeinfo2(string $nodeinfo_url)
 784         {
 785                 $curlResult = DI::httpRequest()->get($nodeinfo_url);
 786                 if (!$curlResult->isSuccess()) {
 787                         return [];
 788                 }
 789
 790                 $nodeinfo = json_decode($curlResult->getBody(), true);
 791
 792                 if (!is_array($nodeinfo)) {
 793                         return [];
 794                 }
 795
 796                 $server = ['detection-method' => self::DETECT_NODEINFO_2,
 797                         'register_policy' => Register::CLOSED];
 798
 799                 if (!empty($nodeinfo['openRegistrations'])) {
 800                         $server['register_policy'] = Register::OPEN;
 801                 }
 802
 803                 if (is_array($nodeinfo['software'])) {
 804                         if (!empty($nodeinfo['software']['name'])) {
 805                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 806                         }
 807
 808                         if (!empty($nodeinfo['software']['version'])) {
 809                                 $server['version'] = $nodeinfo['software']['version'];
 810                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 811                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 812                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 813                         }
 814                 }
 815
 816                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 817                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 818                 }
 819
 820                 if (!empty($nodeinfo['usage']['users']['total'])) {
 821                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 822                 }
 823
 824                 if (!empty($nodeinfo['protocols'])) {
 825                         $protocols = [];
 826                         foreach ($nodeinfo['protocols'] as $protocol) {
 827                                 $protocols[$protocol] = true;
 828                         }
 829
 830                         if (!empty($protocols['dfrn'])) {
 831                                 $server['network'] = Protocol::DFRN;
 832                         } elseif (!empty($protocols['activitypub'])) {
 833                                 $server['network'] = Protocol::ACTIVITYPUB;
 834                         } elseif (!empty($protocols['diaspora'])) {
 835                                 $server['network'] = Protocol::DIASPORA;
 836                         } elseif (!empty($protocols['ostatus'])) {
 837                                 $server['network'] = Protocol::OSTATUS;
 838                         } elseif (!empty($protocols['gnusocial'])) {
 839                                 $server['network'] = Protocol::OSTATUS;
 840                         } elseif (!empty($protocols['zot'])) {
 841                                 $server['network'] = Protocol::ZOT;
 842                         }
 843                 }
 844
 845                 if (empty($server)) {
 846                         return [];
 847                 }
 848
 849                 return $server;
 850         }
 851
 852         /**
 853          * Fetch server information from a 'siteinfo.json' file on the given server
 854          *
 855          * @param string $url        URL of the given server
 856          * @param array  $serverdata array with server data
 857          *
 858          * @return array server data
 859          */
 860         private static function fetchSiteinfo(string $url, array $serverdata)
 861         {
 862                 $curlResult = DI::httpRequest()->get($url . '/siteinfo.json');
 863                 if (!$curlResult->isSuccess()) {
 864                         return $serverdata;
 865                 }
 866
 867                 $data = json_decode($curlResult->getBody(), true);
 868                 if (empty($data)) {
 869                         return $serverdata;
 870                 }
 871
 872                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
 873                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
 874                 }
 875
 876                 if (!empty($data['url'])) {
 877                         $serverdata['platform'] = strtolower($data['platform']);
 878                         $serverdata['version'] = $data['version'];
 879                 }
 880
 881                 if (!empty($data['plugins'])) {
 882                         if (in_array('pubcrawl', $data['plugins'])) {
 883                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
 884                         } elseif (in_array('diaspora', $data['plugins'])) {
 885                                 $serverdata['network'] = Protocol::DIASPORA;
 886                         } elseif (in_array('gnusoc', $data['plugins'])) {
 887                                 $serverdata['network'] = Protocol::OSTATUS;
 888                         } else {
 889                                 $serverdata['network'] = Protocol::ZOT;
 890                         }
 891                 }
 892
 893                 if (!empty($data['site_name'])) {
 894                         $serverdata['site_name'] = $data['site_name'];
 895                 }
 896
 897                 if (!empty($data['channels_total'])) {
 898                         $serverdata['registered-users'] = $data['channels_total'];
 899                 }
 900
 901                 if (!empty($data['register_policy'])) {
 902                         switch ($data['register_policy']) {
 903                                 case 'REGISTER_OPEN':
 904                                         $serverdata['register_policy'] = Register::OPEN;
 905                                         break;
 906
 907                                 case 'REGISTER_APPROVE':
 908                                         $serverdata['register_policy'] = Register::APPROVE;
 909                                         break;
 910
 911                                 case 'REGISTER_CLOSED':
 912                                 default:
 913                                         $serverdata['register_policy'] = Register::CLOSED;
 914                                         break;
 915                         }
 916                 }
 917
 918                 return $serverdata;
 919         }
 920
 921         /**
 922          * Checks if the server contains a valid host meta file
 923          *
 924          * @param string $url URL of the given server
 925          *
 926          * @return boolean 'true' if the server seems to be vital
 927          */
 928         private static function validHostMeta(string $url)
 929         {
 930                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
 931                 $curlResult = DI::httpRequest()->get($url . '/.well-known/host-meta', ['timeout' => $xrd_timeout]);
 932                 if (!$curlResult->isSuccess()) {
 933                         return false;
 934                 }
 935
 936                 $xrd = XML::parseString($curlResult->getBody());
 937                 if (!is_object($xrd)) {
 938                         return false;
 939                 }
 940
 941                 $elements = XML::elementToArray($xrd);
 942                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
 943                         return false;
 944                 }
 945
 946                 $valid = false;
 947                 foreach ($elements['xrd']['link'] as $link) {
 948                         // When there is more than a single "link" element, the array looks slightly different
 949                         if (!empty($link['@attributes'])) {
 950                                 $link = $link['@attributes'];
 951                         }
 952
 953                         if (empty($link['rel']) || empty($link['template'])) {
 954                                 continue;
 955                         }
 956
 957                         if ($link['rel'] == 'lrdd') {
 958                                 // When the webfinger host is the same like the system host, it should be ok.
 959                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
 960                         }
 961                 }
 962
 963                 return $valid;
 964         }
 965
 966         /**
 967          * Detect the network of the given server via their known contacts
 968          *
 969          * @param string $url        URL of the given server
 970          * @param array  $serverdata array with server data
 971          *
 972          * @return array server data
 973          */
 974         private static function detectNetworkViaContacts(string $url, array $serverdata)
 975         {
 976                 $contacts = [];
 977
 978                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $serverdata['nurl']]]);
 979                 while ($apcontact = DBA::fetch($apcontacts)) {
 980                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
 981                 }
 982                 DBA::close($apcontacts);
 983
 984                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $serverdata['nurl']]]);
 985                 while ($pcontact = DBA::fetch($pcontacts)) {
 986                         $contacts[$pcontact['nurl']] = $pcontact['url'];
 987                 }
 988                 DBA::close($pcontacts);
 989
 990                 if (empty($contacts)) {
 991                         return $serverdata;
 992                 }
 993
 994                 foreach ($contacts as $contact) {
 995                         $probed = Contact::getByURL($contact);
 996                         if (!empty($probed) && in_array($probed['network'], Protocol::FEDERATED)) {
 997                                 $serverdata['network'] = $probed['network'];
 998                                 break;
 999                         }
1000                 }
1001
1002                 $serverdata['registered-users'] = max($serverdata['registered-users'], count($contacts));
1003
1004                 return $serverdata;
1005         }
1006
1007         /**
1008          * Checks if the given server does have a '/poco' endpoint.
1009          * This is used for the 'PortableContact' functionality,
1010          * which is used by both Friendica and Hubzilla.
1011          *
1012          * @param string $url        URL of the given server
1013          * @param array  $serverdata array with server data
1014          *
1015          * @return array server data
1016          */
1017         private static function checkPoCo(string $url, array $serverdata)
1018         {
1019                 $serverdata['poco'] = '';
1020
1021                 $curlResult = DI::httpRequest()->get($url . '/poco');
1022                 if (!$curlResult->isSuccess()) {
1023                         return $serverdata;
1024                 }
1025
1026                 $data = json_decode($curlResult->getBody(), true);
1027                 if (empty($data)) {
1028                         return $serverdata;
1029                 }
1030
1031                 if (!empty($data['totalResults'])) {
1032                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1033                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers);
1034                         $serverdata['directory-type'] = self::DT_POCO;
1035                         $serverdata['poco'] = $url . '/poco';
1036                 }
1037
1038                 return $serverdata;
1039         }
1040
1041         /**
1042          * Checks if the given server does have a Mastodon style directory endpoint.
1043          *
1044          * @param string $url        URL of the given server
1045          * @param array  $serverdata array with server data
1046          *
1047          * @return array server data
1048          */
1049         public static function checkMastodonDirectory(string $url, array $serverdata)
1050         {
1051                 $curlResult = DI::httpRequest()->get($url . '/api/v1/directory?limit=1');
1052                 if (!$curlResult->isSuccess()) {
1053                         return $serverdata;
1054                 }
1055
1056                 $data = json_decode($curlResult->getBody(), true);
1057                 if (empty($data)) {
1058                         return $serverdata;
1059                 }
1060
1061                 if (count($data) == 1) {
1062                         $serverdata['directory-type'] = self::DT_MASTODON;
1063                 }
1064
1065                 return $serverdata;
1066         }
1067
1068         /**
1069          * Detects the version number of a given server when it was a NextCloud installation
1070          *
1071          * @param string $url        URL of the given server
1072          * @param array  $serverdata array with server data
1073          *
1074          * @return array server data
1075          */
1076         private static function detectNextcloud(string $url, array $serverdata)
1077         {
1078                 $curlResult = DI::httpRequest()->get($url . '/status.php');
1079
1080                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1081                         return $serverdata;
1082                 }
1083
1084                 $data = json_decode($curlResult->getBody(), true);
1085                 if (empty($data)) {
1086                         return $serverdata;
1087                 }
1088
1089                 if (!empty($data['version'])) {
1090                         $serverdata['platform'] = 'nextcloud';
1091                         $serverdata['version'] = $data['version'];
1092                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1093
1094                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1095                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1096                         }
1097                 }
1098
1099                 return $serverdata;
1100         }
1101
1102         /**
1103          * Detects data from a given server url if it was a mastodon alike system
1104          *
1105          * @param string $url        URL of the given server
1106          * @param array  $serverdata array with server data
1107          *
1108          * @return array server data
1109          */
1110         private static function detectMastodonAlikes(string $url, array $serverdata)
1111         {
1112                 $curlResult = DI::httpRequest()->get($url . '/api/v1/instance');
1113
1114                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1115                         return $serverdata;
1116                 }
1117
1118                 $data = json_decode($curlResult->getBody(), true);
1119                 if (empty($data)) {
1120                         return $serverdata;
1121                 }
1122
1123                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1124                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1125                 }
1126
1127                 if (!empty($data['version'])) {
1128                         $serverdata['platform'] = 'mastodon';
1129                         $serverdata['version'] = $data['version'] ?? '';
1130                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1131                 }
1132
1133                 if (!empty($data['title'])) {
1134                         $serverdata['site_name'] = $data['title'];
1135                 }
1136
1137                 if (!empty($data['title']) && empty($serverdata['platform']) && empty($serverdata['network'])) {
1138                         $serverdata['platform'] = 'mastodon';
1139                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1140                 }
1141
1142                 if (!empty($data['description'])) {
1143                         $serverdata['info'] = trim($data['description']);
1144                 }
1145
1146                 if (!empty($data['stats']['user_count'])) {
1147                         $serverdata['registered-users'] = $data['stats']['user_count'];
1148                 }
1149
1150                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1151                         $serverdata['platform'] = strtolower($matches[1]);
1152                         $serverdata['version'] = $matches[2];
1153                 }
1154
1155                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1156                         $serverdata['platform'] = 'pleroma';
1157                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1158                 }
1159
1160                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1161                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1162                         $serverdata['platform'] = 'pleroma';
1163                 }
1164
1165                 return $serverdata;
1166         }
1167
1168         /**
1169          * Detects data from typical Hubzilla endpoints
1170          *
1171          * @param string $url        URL of the given server
1172          * @param array  $serverdata array with server data
1173          *
1174          * @return array server data
1175          */
1176         private static function detectHubzilla(string $url, array $serverdata)
1177         {
1178                 $curlResult = DI::httpRequest()->get($url . '/api/statusnet/config.json');
1179                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1180                         return $serverdata;
1181                 }
1182
1183                 $data = json_decode($curlResult->getBody(), true);
1184                 if (empty($data) || empty($data['site'])) {
1185                         return $serverdata;
1186                 }
1187
1188                 if (!empty($data['site']['name'])) {
1189                         $serverdata['site_name'] = $data['site']['name'];
1190                 }
1191
1192                 if (!empty($data['site']['platform'])) {
1193                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1194                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1195                         $serverdata['network'] = Protocol::ZOT;
1196                 }
1197
1198                 if (!empty($data['site']['hubzilla'])) {
1199                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1200                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1201                         $serverdata['network'] = Protocol::ZOT;
1202                 }
1203
1204                 if (!empty($data['site']['redmatrix'])) {
1205                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1206                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1207                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1208                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1209                         }
1210
1211                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1212                         $serverdata['network'] = Protocol::ZOT;
1213                 }
1214
1215                 $private = false;
1216                 $inviteonly = false;
1217                 $closed = false;
1218
1219                 if (!empty($data['site']['closed'])) {
1220                         $closed = self::toBoolean($data['site']['closed']);
1221                 }
1222
1223                 if (!empty($data['site']['private'])) {
1224                         $private = self::toBoolean($data['site']['private']);
1225                 }
1226
1227                 if (!empty($data['site']['inviteonly'])) {
1228                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1229                 }
1230
1231                 if (!$closed && !$private and $inviteonly) {
1232                         $serverdata['register_policy'] = Register::APPROVE;
1233                 } elseif (!$closed && !$private) {
1234                         $serverdata['register_policy'] = Register::OPEN;
1235                 } else {
1236                         $serverdata['register_policy'] = Register::CLOSED;
1237                 }
1238
1239                 if (!empty($serverdata['network']) && in_array($serverdata['detection-method'],
1240                         [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1241                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1242                 }
1243
1244                 return $serverdata;
1245         }
1246
1247         /**
1248          * Converts input value to a boolean value
1249          *
1250          * @param string|integer $val
1251          *
1252          * @return boolean
1253          */
1254         private static function toBoolean($val)
1255         {
1256                 if (($val == 'true') || ($val == 1)) {
1257                         return true;
1258                 } elseif (($val == 'false') || ($val == 0)) {
1259                         return false;
1260                 }
1261
1262                 return $val;
1263         }
1264
1265         /**
1266          * Detect if the URL belongs to a GNU Social server
1267          *
1268          * @param string $url        URL of the given server
1269          * @param array  $serverdata array with server data
1270          *
1271          * @return array server data
1272          */
1273         private static function detectGNUSocial(string $url, array $serverdata)
1274         {
1275                 // Test for GNU Social
1276                 $curlResult = DI::httpRequest()->get($url . '/api/gnusocial/version.json');
1277                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1278                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1279                         $serverdata['platform'] = 'gnusocial';
1280                         // Remove junk that some GNU Social servers return
1281                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1282                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1283                         $serverdata['version'] = trim($serverdata['version'], '"');
1284                         $serverdata['network'] = Protocol::OSTATUS;
1285
1286                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1287                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1288                         }
1289
1290                         return $serverdata;
1291                 }
1292
1293                 // Test for Statusnet
1294                 $curlResult = DI::httpRequest()->get($url . '/api/statusnet/version.json');
1295                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1296                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1297
1298                         // Remove junk that some GNU Social servers return
1299                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1300                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1301                         $serverdata['version'] = trim($serverdata['version'], '"');
1302
1303                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1304                                 $serverdata['platform'] = 'pleroma';
1305                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1306                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1307                         } else {
1308                                 $serverdata['platform'] = 'statusnet';
1309                                 $serverdata['network'] = Protocol::OSTATUS;
1310                         }
1311
1312                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1313                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1314                         }
1315                 }
1316
1317                 return $serverdata;
1318         }
1319
1320         /**
1321          * Detect if the URL belongs to a Friendica server
1322          *
1323          * @param string $url        URL of the given server
1324          * @param array  $serverdata array with server data
1325          *
1326          * @return array server data
1327          */
1328         private static function detectFriendica(string $url, array $serverdata)
1329         {
1330                 $curlResult = DI::httpRequest()->get($url . '/friendica/json');
1331                 if (!$curlResult->isSuccess()) {
1332                         $curlResult = DI::httpRequest()->get($url . '/friendika/json');
1333                         $friendika = true;
1334                         $platform = 'Friendika';
1335                 } else {
1336                         $friendika = false;
1337                         $platform = 'Friendica';
1338                 }
1339
1340                 if (!$curlResult->isSuccess()) {
1341                         return $serverdata;
1342                 }
1343
1344                 $data = json_decode($curlResult->getBody(), true);
1345                 if (empty($data) || empty($data['version'])) {
1346                         return $serverdata;
1347                 }
1348
1349                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1350                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1351                 }
1352
1353                 $serverdata['network'] = Protocol::DFRN;
1354                 $serverdata['version'] = $data['version'];
1355
1356                 if (!empty($data['no_scrape_url'])) {
1357                         $serverdata['noscrape'] = $data['no_scrape_url'];
1358                 }
1359
1360                 if (!empty($data['site_name'])) {
1361                         $serverdata['site_name'] = $data['site_name'];
1362                 }
1363
1364                 if (!empty($data['info'])) {
1365                         $serverdata['info'] = trim($data['info']);
1366                 }
1367
1368                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1369                 switch ($register_policy) {
1370                         case 'REGISTER_OPEN':
1371                                 $serverdata['register_policy'] = Register::OPEN;
1372                                 break;
1373
1374                         case 'REGISTER_APPROVE':
1375                                 $serverdata['register_policy'] = Register::APPROVE;
1376                                 break;
1377
1378                         case 'REGISTER_CLOSED':
1379                         case 'REGISTER_INVITATION':
1380                                 $serverdata['register_policy'] = Register::CLOSED;
1381                                 break;
1382                         default:
1383                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1384                                 $serverdata['register_policy'] = Register::CLOSED;
1385                                 break;
1386                 }
1387
1388                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1389
1390                 return $serverdata;
1391         }
1392
1393         /**
1394          * Analyses the landing page of a given server for hints about type and system of that server
1395          *
1396          * @param object $curlResult result of curl execution
1397          * @param array  $serverdata array with server data
1398          * @param string $url        Server URL
1399          *
1400          * @return array server data
1401          */
1402         private static function analyseRootBody($curlResult, array $serverdata, string $url)
1403         {
1404                 $doc = new DOMDocument();
1405                 @$doc->loadHTML($curlResult->getBody());
1406                 $xpath = new DOMXPath($doc);
1407
1408                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1409                 if (!empty($title)) {
1410                         $serverdata['site_name'] = $title;
1411                 }
1412
1413                 $list = $xpath->query('//meta[@name]');
1414
1415                 foreach ($list as $node) {
1416                         $attr = [];
1417                         if ($node->attributes->length) {
1418                                 foreach ($node->attributes as $attribute) {
1419                                         $value = trim($attribute->value);
1420                                         if (empty($value)) {
1421                                                 continue;
1422                                         }
1423
1424                                         $attr[$attribute->name] = $value;
1425                                 }
1426
1427                                 if (empty($attr['name']) || empty($attr['content'])) {
1428                                         continue;
1429                                 }
1430                         }
1431
1432                         if ($attr['name'] == 'description') {
1433                                 $serverdata['info'] = $attr['content'];
1434                         }
1435
1436                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1437                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad'])) {
1438                                 $serverdata['platform'] = strtolower($attr['content']);
1439                                 if (in_array($attr['content'], ['Misskey', 'Write.as'])) {
1440                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1441                                 }
1442                         }
1443                         if (($attr['name'] == 'generator') && (empty($serverdata['platform']) || (substr(strtolower($attr['content']), 0, 9) == 'wordpress'))) {
1444                                 $serverdata['platform'] = strtolower($attr['content']);
1445                                 $version_part = explode(' ', $attr['content']);
1446
1447                                 if (count($version_part) == 2) {
1448                                         if (in_array($version_part[0], ['WordPress'])) {
1449                                                 $serverdata['platform'] = strtolower($version_part[0]);
1450                                                 $serverdata['version'] = $version_part[1];
1451
1452                                                 // We still do need a reliable test if some AP plugin is activated
1453                                                 if (DBA::exists('apcontact', ['baseurl' => $url])) {
1454                                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1455                                                 } else {
1456                                                         $serverdata['network'] = Protocol::FEED;
1457                                                 }
1458
1459                                                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1460                                                         $serverdata['detection-method'] = self::DETECT_BODY;
1461                                                 }
1462                                         }
1463                                         if (in_array($version_part[0], ['Friendika', 'Friendica'])) {
1464                                                 $serverdata['platform'] = strtolower($version_part[0]);
1465                                                 $serverdata['version'] = $version_part[1];
1466                                                 $serverdata['network'] = Protocol::DFRN;
1467                                         }
1468                                 }
1469                         }
1470                 }
1471
1472                 $list = $xpath->query('//meta[@property]');
1473
1474                 foreach ($list as $node) {
1475                         $attr = [];
1476                         if ($node->attributes->length) {
1477                                 foreach ($node->attributes as $attribute) {
1478                                         $value = trim($attribute->value);
1479                                         if (empty($value)) {
1480                                                 continue;
1481                                         }
1482
1483                                         $attr[$attribute->name] = $value;
1484                                 }
1485
1486                                 if (empty($attr['property']) || empty($attr['content'])) {
1487                                         continue;
1488                                 }
1489                         }
1490
1491                         if ($attr['property'] == 'og:site_name') {
1492                                 $serverdata['site_name'] = $attr['content'];
1493                         }
1494
1495                         if ($attr['property'] == 'og:description') {
1496                                 $serverdata['info'] = $attr['content'];
1497                         }
1498
1499                         if ($attr['property'] == 'og:platform') {
1500                                 $serverdata['platform'] = strtolower($attr['content']);
1501
1502                                 if (in_array($attr['content'], ['PeerTube'])) {
1503                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1504                                 }
1505                         }
1506
1507                         if ($attr['property'] == 'generator') {
1508                                 $serverdata['platform'] = strtolower($attr['content']);
1509
1510                                 if (in_array($attr['content'], ['hubzilla'])) {
1511                                         // We later check which compatible protocol modules are loaded.
1512                                         $serverdata['network'] = Protocol::ZOT;
1513                                 }
1514                         }
1515                 }
1516
1517                 if (!empty($serverdata['network']) && ($serverdata['detection-method'] == self::DETECT_MANUAL)) {
1518                         $serverdata['detection-method'] = self::DETECT_BODY;
1519                 }
1520
1521                 return $serverdata;
1522         }
1523
1524         /**
1525          * Analyses the header data of a given server for hints about type and system of that server
1526          *
1527          * @param object $curlResult result of curl execution
1528          * @param array  $serverdata array with server data
1529          *
1530          * @return array server data
1531          */
1532         private static function analyseRootHeader($curlResult, array $serverdata)
1533         {
1534                 if ($curlResult->getHeader('server') == 'Mastodon') {
1535                         $serverdata['platform'] = 'mastodon';
1536                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1537                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
1538                         $serverdata['platform'] = 'diaspora';
1539                         $serverdata['network'] = Protocol::DIASPORA;
1540                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version');
1541                 } elseif ($curlResult->inHeader('x-friendica-version')) {
1542                         $serverdata['platform'] = 'friendica';
1543                         $serverdata['network'] = Protocol::DFRN;
1544                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version');
1545                 } else {
1546                         return $serverdata;
1547                 }
1548
1549                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1550                         $serverdata['detection-method'] = self::DETECT_HEADER;
1551                 }
1552
1553                 return $serverdata;
1554         }
1555
1556         /**
1557          * Test if the body contains valid content
1558          *
1559          * @param string $body
1560          * @return boolean
1561          */
1562         private static function invalidBody(string $body)
1563         {
1564                 // Currently we only test for a HTML element.
1565                 // Possibly we enhance this in the future.
1566                 return !strpos($body, '>');
1567         }
1568
1569         /**
1570          * Update GServer entries
1571          */
1572         public static function discover()
1573         {
1574                 // Update the server list
1575                 self::discoverFederation();
1576
1577                 $no_of_queries = 5;
1578
1579                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
1580
1581                 if ($requery_days == 0) {
1582                         $requery_days = 7;
1583                 }
1584
1585                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
1586
1587                 $gservers = DBA::p("SELECT `id`, `url`, `nurl`, `network`, `poco`, `directory-type`
1588                         FROM `gserver`
1589                         WHERE NOT `failed`
1590                         AND `directory-type` != ?
1591                         AND `last_poco_query` < ?
1592                         ORDER BY RAND()", self::DT_NONE, $last_update
1593                 );
1594
1595                 while ($gserver = DBA::fetch($gservers)) {
1596                         if (!GServer::check($gserver['url'], $gserver['network'])) {
1597                                 // The server is not reachable? Okay, then we will try it later
1598                                 $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1599                                 DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1600                                 continue;
1601                         }
1602
1603                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
1604                         Worker::add(PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
1605
1606                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
1607                         Worker::add(PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
1608
1609                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1610                         DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1611
1612                         if (--$no_of_queries == 0) {
1613                                 break;
1614                         }
1615                 }
1616
1617                 DBA::close($gservers);
1618         }
1619
1620         /**
1621          * Discover federated servers
1622          */
1623         private static function discoverFederation()
1624         {
1625                 $last = DI::config()->get('poco', 'last_federation_discovery');
1626
1627                 if ($last) {
1628                         $next = $last + (24 * 60 * 60);
1629
1630                         if ($next > time()) {
1631                                 return;
1632                         }
1633                 }
1634
1635                 // Discover federated servers
1636                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
1637                 foreach ($protocols as $protocol) {
1638                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
1639                         $curlResult = DI::httpRequest()->fetch('https://the-federation.info/graphql?query=' . urlencode($query));
1640                         if (!empty($curlResult)) {
1641                                 $data = json_decode($curlResult, true);
1642                                 if (!empty($data['data']['nodes'])) {
1643                                         foreach ($data['data']['nodes'] as $server) {
1644                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
1645                                                 Worker::add(PRIORITY_LOW, 'UpdateGServer', 'https://' . $server['host'], true);
1646                                         }
1647                                 }
1648                         }
1649                 }
1650
1651                 // Disvover Mastodon servers
1652                 $accesstoken = DI::config()->get('system', 'instances_social_key');
1653
1654                 if (!empty($accesstoken)) {
1655                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
1656                         $header = ['Authorization: Bearer '.$accesstoken];
1657                         $curlResult = DI::httpRequest()->get($api, ['header' => $header]);
1658
1659                         if ($curlResult->isSuccess()) {
1660                                 $servers = json_decode($curlResult->getBody(), true);
1661
1662                                 foreach ($servers['instances'] as $server) {
1663                                         $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
1664                                         Worker::add(PRIORITY_LOW, 'UpdateGServer', $url);
1665                                 }
1666                         }
1667                 }
1668
1669                 DI::config()->set('poco', 'last_federation_discovery', time());
1670         }
1671 }