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