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