src/Model/GServer.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2023, the Friendica project
   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 Exception;
  27 use Friendica\Core\Logger;
  28 use Friendica\Core\Protocol;
  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\HTTPClient\Client\HttpClientAccept;
  35 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  36 use Friendica\Network\HTTPClient\Capability\ICanHandleHttpResponses;
  37 use Friendica\Network\Probe;
  38 use Friendica\Protocol\ActivityPub;
  39 use Friendica\Protocol\Relay;
  40 use Friendica\Util\DateTimeFormat;
  41 use Friendica\Util\JsonLD;
  42 use Friendica\Util\Network;
  43 use Friendica\Util\Strings;
  44 use Friendica\Util\XML;
  45 use Friendica\Network\HTTPException;
  46 use Friendica\Worker\UpdateGServer;
  47 use GuzzleHttp\Psr7\Uri;
  48 use Psr\Http\Message\UriInterface;
  49
  50 /**
  51  * This class handles GServer related functions
  52  */
  53 class GServer
  54 {
  55         // Directory types
  56         const DT_NONE = 0;
  57         const DT_POCO = 1;
  58         const DT_MASTODON = 2;
  59
  60         // Methods to detect server types
  61
  62         // Non endpoint specific methods
  63         const DETECT_MANUAL = 0;
  64         const DETECT_HEADER = 1;
  65         const DETECT_BODY = 2;
  66         const DETECT_HOST_META = 3;
  67         const DETECT_CONTACTS = 4;
  68         const DETECT_AP_ACTOR = 5;
  69         const DETECT_AP_COLLECTION = 6;
  70
  71         const DETECT_UNSPECIFIC = [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META, self::DETECT_CONTACTS, self::DETECT_AP_ACTOR];
  72
  73         // Implementation specific endpoints
  74         // @todo Possibly add Lemmy detection via the endpoint /api/v3/site
  75         const DETECT_FRIENDIKA = 10;
  76         const DETECT_FRIENDICA = 11;
  77         const DETECT_STATUSNET = 12;
  78         const DETECT_GNUSOCIAL = 13;
  79         const DETECT_CONFIG_JSON = 14; // Statusnet, GNU Social, Older Hubzilla/Redmatrix
  80         const DETECT_SITEINFO_JSON = 15; // Newer Hubzilla
  81         const DETECT_MASTODON_API = 16;
  82         const DETECT_STATUS_PHP = 17; // Nextcloud
  83         const DETECT_V1_CONFIG = 18;
  84         const DETECT_SYSTEM_ACTOR = 20; // Mistpark, Osada, Roadhouse, Zap
  85         const DETECT_THREADS = 21;
  86
  87         // Standardized endpoints
  88         const DETECT_STATISTICS_JSON = 100;
  89         const DETECT_NODEINFO_1 = 101;
  90         const DETECT_NODEINFO_2 = 102;
  91         const DETECT_NODEINFO_210 = 103;
  92
  93         /**
  94          * Check for the existence of a server and adds it in the background if not existant
  95          *
  96          * @param string $url
  97          * @param boolean $only_nodeinfo
  98          *
  99          * @return void
 100          */
 101         public static function add(string $url, bool $only_nodeinfo = false)
 102         {
 103                 if (self::getID($url)) {
 104                         return;
 105                 }
 106
 107                 UpdateGServer::add(Worker::PRIORITY_LOW, $url, $only_nodeinfo);
 108         }
 109
 110         /**
 111          * Get the ID for the given server URL
 112          *
 113          * @param string $url
 114          * @param boolean $no_check Don't check if the server hadn't been found
 115          *
 116          * @return int|null gserver id or NULL on empty URL or failed check
 117          */
 118         public static function getID(string $url, bool $no_check = false): ?int
 119         {
 120                 $url = self::cleanURL($url);
 121
 122                 if (empty($url)) {
 123                         return null;
 124                 }
 125
 126                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
 127                 if (DBA::isResult($gserver)) {
 128                         Logger::debug('Got ID for URL', ['id' => $gserver['id'], 'url' => $url]);
 129
 130                         if (Network::isUrlBlocked($url)) {
 131                                 self::setBlockedById($gserver['id']);
 132                         } else {
 133                                 self::setUnblockedById($gserver['id']);
 134                         }
 135
 136                         return $gserver['id'];
 137                 }
 138
 139                 if ($no_check || !self::check($url)) {
 140                         return null;
 141                 }
 142
 143                 return self::getID($url, true);
 144         }
 145
 146         /**
 147          * Retrieves all the servers which base domain are matching the provided domain pattern
 148          *
 149          * The pattern is a simple fnmatch() pattern with ? for single wildcard and * for multiple wildcard
 150          *
 151          * @param string $pattern
 152          *
 153          * @return array
 154          *
 155          * @throws Exception
 156          */
 157         public static function listByDomainPattern(string $pattern): array
 158         {
 159                 $likePattern = 'http://' . strtr($pattern, ['_' => '\_', '%' => '\%', '?' => '_', '*' => '%']);
 160
 161                 // The SUBSTRING_INDEX returns everything before the eventual third /, which effectively trims an
 162                 // eventual server path and keep only the server domain which we're matching against the pattern.
 163                 $sql = "SELECT `gserver`.*, COUNT(*) AS `contacts`
 164                         FROM `gserver`
 165                         LEFT JOIN `contact` ON `gserver`.`id` = `contact`.`gsid`
 166                         WHERE SUBSTRING_INDEX(`gserver`.`nurl`, '/', 3) LIKE ?
 167                         AND NOT `gserver`.`failed`
 168                         GROUP BY `gserver`.`id`";
 169
 170                 $stmt = DI::dba()->p($sql, $likePattern);
 171
 172                 return DI::dba()->toArray($stmt);
 173         }
 174
 175         /**
 176          * Checks if the given server array is unreachable for a long time now
 177          *
 178          * @param integer $gsid
 179          * @return boolean
 180          */
 181         private static function isDefunct(array $gserver): bool
 182         {
 183                 return ($gserver['failed'] || in_array($gserver['network'], Protocol::FEDERATED)) &&
 184                         ($gserver['last_contact'] >= $gserver['created']) &&
 185                         ($gserver['last_contact'] < $gserver['last_failure']) &&
 186                         ($gserver['last_contact'] < DateTimeFormat::utc('now - 90 days'));
 187         }
 188
 189         /**
 190          * Checks if the given server id is unreachable for a long time now
 191          *
 192          * @param integer $gsid
 193          * @return boolean
 194          */
 195         public static function isDefunctById(int $gsid): bool
 196         {
 197                 $gserver = DBA::selectFirst('gserver', ['url', 'next_contact', 'last_contact', 'last_failure', 'created', 'failed', 'network'], ['id' => $gsid]);
 198                 if (empty($gserver)) {
 199                         return false;
 200                 } else {
 201                         if (strtotime($gserver['next_contact']) < time()) {
 202                                 UpdateGServer::add(Worker::PRIORITY_LOW, $gserver['url']);
 203                         }
 204
 205                         return self::isDefunct($gserver);
 206                 }
 207         }
 208
 209         /**
 210          * Checks if the given server id is reachable
 211          *
 212          * @param integer $gsid
 213          * @return boolean
 214          */
 215         public static function isReachableById(int $gsid): bool
 216         {
 217                 $gserver = DBA::selectFirst('gserver', ['url', 'next_contact', 'failed', 'network'], ['id' => $gsid]);
 218                 if (empty($gserver)) {
 219                         return true;
 220                 } else {
 221                         if (strtotime($gserver['next_contact']) < time()) {
 222                                 UpdateGServer::add(Worker::PRIORITY_LOW, $gserver['url']);
 223                         }
 224
 225                         return !$gserver['failed'] && in_array($gserver['network'], Protocol::FEDERATED);
 226                 }
 227         }
 228
 229         /**
 230          * Checks if the given server is reachable
 231          *
 232          * @param array $contact Contact that should be checked
 233          *
 234          * @return boolean 'true' if server seems vital
 235          */
 236         public static function reachable(array $contact): bool
 237         {
 238                 if (!empty($contact['gsid'])) {
 239                         $gsid = $contact['gsid'];
 240                 } elseif (!empty($contact['baseurl'])) {
 241                         $server = $contact['baseurl'];
 242                 } elseif ($contact['network'] == Protocol::DIASPORA) {
 243                         $parts = (array)parse_url($contact['url']);
 244                         unset($parts['path']);
 245                         $server = (string)Uri::fromParts($parts);
 246                 } else {
 247                         return true;
 248                 }
 249
 250                 if (!empty($gsid)) {
 251                         $condition = ['id' => $gsid];
 252                 } else {
 253                         $condition = ['nurl' => Strings::normaliseLink($server)];
 254                 }
 255
 256                 $gserver = DBA::selectFirst('gserver', ['url', 'next_contact', 'failed', 'network'], $condition);
 257                 if (empty($gserver)) {
 258                         $reachable = true;
 259                 } else {
 260                         $reachable = !$gserver['failed'] && in_array($gserver['network'], Protocol::FEDERATED);
 261                         $server    = $gserver['url'];
 262                 }
 263
 264                 if (!empty($server) && (empty($gserver) || strtotime($gserver['next_contact']) < time())) {
 265                         UpdateGServer::add(Worker::PRIORITY_LOW, $server);
 266                 }
 267
 268                 return $reachable;
 269         }
 270
 271         /**
 272          * Calculate the next update day
 273          *
 274          * @param bool $success
 275          * @param string $created
 276          * @param string $last_contact
 277          * @param bool $undetected
 278          *
 279          * @return string
 280          *
 281          * @throws Exception
 282          */
 283         public static function getNextUpdateDate(bool $success, string $created = '', string $last_contact = '', bool $undetected = false): string
 284         {
 285                 // On successful contact process check again next week when it is a detected system.
 286                 // When we haven't detected the system, it could be a static website or a really old system.
 287                 if ($success) {
 288                         return DateTimeFormat::utc($undetected ? 'now +1 month' : 'now +7 day');
 289                 }
 290
 291                 $now = strtotime(DateTimeFormat::utcNow());
 292
 293                 if ($created > $last_contact) {
 294                         $contact_time = strtotime($created);
 295                 } else {
 296                         $contact_time = strtotime($last_contact);
 297                 }
 298
 299                 // If the last contact was less than 6 hours before then try again in 6 hours
 300                 if (($now - $contact_time) < (60 * 60 * 6)) {
 301                         return DateTimeFormat::utc('now +6 hour');
 302                 }
 303
 304                 // If the last contact was less than 12 hours before then try again in 12 hours
 305                 if (($now - $contact_time) < (60 * 60 * 12)) {
 306                         return DateTimeFormat::utc('now +12 hour');
 307                 }
 308
 309                 // If the last contact was less than 24 hours before then try tomorrow again
 310                 if (($now - $contact_time) < (60 * 60 * 24)) {
 311                         return DateTimeFormat::utc('now +1 day');
 312                 }
 313
 314                 // If the last contact was less than a week before then try again in a week
 315                 if (($now - $contact_time) < (60 * 60 * 24 * 7)) {
 316                         return DateTimeFormat::utc('now +1 week');
 317                 }
 318
 319                 // If the last contact was less than two weeks before then try again in two week
 320                 if (($now - $contact_time) < (60 * 60 * 24 * 14)) {
 321                         return DateTimeFormat::utc('now +2 week');
 322                 }
 323
 324                 // If the last contact was less than a month before then try again in a month
 325                 if (($now - $contact_time) < (60 * 60 * 24 * 30)) {
 326                         return DateTimeFormat::utc('now +1 month');
 327                 }
 328
 329                 // The system hadn't been successful contacted for more than a month, so try again in three months
 330                 return DateTimeFormat::utc('now +3 month');
 331         }
 332
 333         /**
 334          * Checks the state of the given server.
 335          *
 336          * @param string  $server_url    URL of the given server
 337          * @param string  $network       Network value that is used, when detection failed
 338          * @param boolean $force         Force an update.
 339          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 340          *
 341          * @return boolean 'true' if server seems vital
 342          */
 343         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false): bool
 344         {
 345                 $server_url = self::cleanURL($server_url);
 346                 if ($server_url == '') {
 347                         return false;
 348                 }
 349
 350                 if (Network::isUrlBlocked($server_url)) {
 351                         Logger::info('Server is blocked', ['url' => $server_url]);
 352                         self::setBlockedByUrl($server_url);
 353                         return false;
 354                 }
 355
 356                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
 357                 if (DBA::isResult($gserver)) {
 358                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
 359                                 $fields = ['created' => DateTimeFormat::utcNow()];
 360                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
 361                                 self::update($fields, $condition);
 362                         }
 363
 364                         if (!$force && (strtotime($gserver['next_contact']) > time())) {
 365                                 Logger::info('No update needed', ['server' => $server_url]);
 366                                 return (!$gserver['failed']);
 367                         }
 368                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force]);
 369                 } else {
 370                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
 371                 }
 372
 373                 return self::detect($server_url, $network, $only_nodeinfo);
 374         }
 375
 376         /**
 377          * Reset failed server status by gserver id
 378          *
 379          * @param int    $gsid
 380          * @param string $network
 381          */
 382         public static function setReachableById(int $gsid, string $network)
 383         {
 384                 $gserver = DBA::selectFirst('gserver', ['url', 'failed', 'next_contact', 'network'], ['id' => $gsid]);
 385                 if (!DBA::isResult($gserver)) {
 386                         return;
 387                 }
 388
 389                 $blocked = Network::isUrlBlocked($gserver['url']);
 390                 if ($gserver['failed']) {
 391                         $fields = ['failed' => false, 'blocked' => $blocked, 'last_contact' => DateTimeFormat::utcNow()];
 392                         if (!empty($network) && !in_array($gserver['network'], Protocol::FEDERATED)) {
 393                                 $fields['network'] = $network;
 394                         }
 395                         self::update($fields, ['id' => $gsid]);
 396                         Logger::info('Reset failed status for server', ['url' => $gserver['url']]);
 397
 398                         if (strtotime($gserver['next_contact']) < time()) {
 399                                 UpdateGServer::add(Worker::PRIORITY_LOW, $gserver['url']);
 400                         }
 401                 } elseif ($blocked) {
 402                         self::setBlockedById($gsid);
 403                 } else {
 404                         self::setUnblockedById($gsid);
 405                 }
 406         }
 407
 408         /**
 409          * Set failed server status by gserver id
 410          *
 411          * @param int $gsid
 412          */
 413         public static function setFailureById(int $gsid)
 414         {
 415                 $gserver = DBA::selectFirst('gserver', ['url', 'failed', 'next_contact'], ['id' => $gsid]);
 416                 if (DBA::isResult($gserver) && !$gserver['failed']) {
 417                         self::update(['failed' => true, 'blocked' => Network::isUrlBlocked($gserver['url']), 'last_failure' => DateTimeFormat::utcNow()], ['id' => $gsid]);
 418                         Logger::info('Set failed status for server', ['url' => $gserver['url']]);
 419
 420                         if (strtotime($gserver['next_contact']) < time()) {
 421                                 UpdateGServer::add(Worker::PRIORITY_LOW, $gserver['url']);
 422                         }
 423                 }
 424         }
 425
 426         public static function setUnblockedById(int $gsid)
 427         {
 428                 $gserver = DBA::selectFirst('gserver', ['url'], ["(`blocked` OR `blocked` IS NULL) AND `id` = ?", $gsid]);
 429                 if (DBA::isResult($gserver)) {
 430                         self::update(['blocked' => false], ['id' => $gsid]);
 431                         Logger::info('Set unblocked status for server', ['url' => $gserver['url']]);
 432                 }
 433         }
 434
 435         public static function setBlockedById(int $gsid)
 436         {
 437                 $gserver = DBA::selectFirst('gserver', ['url'], ["(NOT `blocked` OR `blocked` IS NULL) AND `id` = ?", $gsid]);
 438                 if (DBA::isResult($gserver)) {
 439                         self::update(['blocked' => true, 'failed' => true], ['id' => $gsid]);
 440                         Logger::info('Set blocked status for server', ['url' => $gserver['url']]);
 441                 }
 442         }
 443
 444         public static function setBlockedByUrl(string $url)
 445         {
 446                 $gserver = DBA::selectFirst('gserver', ['url', 'id'], ["(NOT `blocked` OR `blocked` IS NULL) AND `nurl` = ?", Strings::normaliseLink($url)]);
 447                 if (DBA::isResult($gserver)) {
 448                         self::update(['blocked' => true, 'failed' => true], ['id' => $gserver['id']]);
 449                         Logger::info('Set blocked status for server', ['url' => $gserver['url']]);
 450                 }
 451         }
 452
 453         /**
 454          * Set failed server status
 455          *
 456          * @param string $url
 457          * @return void
 458          */
 459         public static function setFailureByUrl(string $url)
 460         {
 461                 $nurl = Strings::normaliseLink($url);
 462
 463                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => $nurl]);
 464                 if (DBA::isResult($gserver)) {
 465                         $next_update = self::getNextUpdateDate(false, $gserver['created'], $gserver['last_contact']);
 466                         self::update(['url' => $url, 'failed' => true, 'blocked' => Network::isUrlBlocked($url), 'last_failure' => DateTimeFormat::utcNow(),
 467                         'next_contact' => $next_update, 'network' => Protocol::PHANTOM, 'detection-method' => null],
 468                         ['nurl' => $nurl]);
 469                         Logger::info('Set failed status for existing server', ['url' => $url]);
 470                         if (self::isDefunct($gserver)) {
 471                                 self::archiveContacts($gserver['id']);
 472                         }
 473                         return;
 474                 }
 475
 476                 self::insert(['url' => $url, 'nurl' => $nurl,
 477                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
 478                         'failed' => true, 'last_failure' => DateTimeFormat::utcNow()]);
 479                 Logger::info('Set failed status for new server', ['url' => $url]);
 480         }
 481
 482         /**
 483          * Archive server related contacts and inboxes
 484          *
 485          * @param integer $gsid
 486          * @return void
 487          */
 488         private static function archiveContacts(int $gsid)
 489         {
 490                 Contact::update(['archive' => true], ['gsid' => $gsid]);
 491                 DBA::update('inbox-status', ['archive' => true], ['gsid' => $gsid]);
 492         }
 493
 494         /**
 495          * Remove unwanted content from the given URL
 496          *
 497          * @param string $dirtyUrl
 498          *
 499          * @return string cleaned URL
 500          * @throws Exception
 501          * @deprecated since 2023.03 Use cleanUri instead
 502          */
 503         public static function cleanURL(string $dirtyUrl): string
 504         {
 505                 try {
 506                         return (string)self::cleanUri(new Uri($dirtyUrl));
 507                 } catch (\Throwable $e) {
 508                         Logger::warning('Invalid URL', ['dirtyUrl' => $dirtyUrl]);
 509                         return '';
 510                 }
 511         }
 512
 513         /**
 514          * Remove unwanted content from the given URI
 515          *
 516          * @param UriInterface $dirtyUri
 517          *
 518          * @return UriInterface cleaned URI
 519          * @throws Exception
 520          */
 521         public static function cleanUri(UriInterface $dirtyUri): string
 522         {
 523                 return $dirtyUri
 524                         ->withUserInfo('')
 525                         ->withQuery('')
 526                         ->withFragment('')
 527                         ->withPath(
 528                                 preg_replace(
 529                                         '#(?:^|/)index\.php#',
 530                                         '',
 531                                         rtrim($dirtyUri->getPath(), '/')
 532                                 )
 533                         );
 534         }
 535
 536         /**
 537          * Detect server data (type, protocol, version number, ...)
 538          * The detected data is then updated or inserted in the gserver table.
 539          *
 540          * @param string  $url           URL of the given server
 541          * @param string  $network       Network value that is used, when detection failed
 542          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 543          *
 544          * @return boolean 'true' if server could be detected
 545          */
 546         private static function detect(string $url, string $network = '', bool $only_nodeinfo = false): bool
 547         {
 548                 Logger::info('Detect server type', ['server' => $url]);
 549
 550                 $original_url = $url;
 551
 552                 // Remove URL content that is not supposed to exist for a server url
 553                 $url = rtrim(self::cleanURL($url), '/');
 554                 if (empty($url)) {
 555                         Logger::notice('Empty URL.');
 556                         return false;
 557                 }
 558
 559                 // If the URL mismatches, then we mark the old entry as failure
 560                 if (!Strings::compareLink($url, $original_url)) {
 561                         self::setFailureByUrl($original_url);
 562                         if (!self::getID($url, true) && !Network::isUrlBlocked($url)) {
 563                                 self::detect($url, $network, $only_nodeinfo);
 564                         }
 565                         return false;
 566                 }
 567
 568                 $valid_url = Network::isUrlValid($url);
 569                 if (!$valid_url) {
 570                         self::setFailureByUrl($url);
 571                         return false;
 572                 } else {
 573                         $valid_url = rtrim($valid_url, '/');
 574                 }
 575
 576                 if (!Strings::compareLink($url, $valid_url)) {
 577                         // We only follow redirects when the path stays the same or the target url has no path.
 578                         // Some systems have got redirects on their landing page to a single account page. This check handles it.
 579                         if (((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH))) ||
 580                                 (((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) || (parse_url($url, PHP_URL_PATH) != parse_url($valid_url, PHP_URL_PATH))) && empty(parse_url($valid_url, PHP_URL_PATH)))) {
 581                                 Logger::debug('Found redirect. Mark old entry as failure', ['old' => $url, 'new' => $valid_url]);
 582                                 self::setFailureByUrl($url);
 583                                 if (!self::getID($valid_url, true) && !Network::isUrlBlocked($valid_url)) {
 584                                         self::detect($valid_url, $network, $only_nodeinfo);
 585                                 }
 586                                 return false;
 587                         }
 588
 589                         if ((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) != parse_url($valid_url, PHP_URL_PATH)) &&
 590                                 (parse_url($url, PHP_URL_PATH) == '')) {
 591                                 Logger::debug('Found redirect. Mark old entry as failure and redirect to the basepath.', ['old' => $url, 'new' => $valid_url]);
 592                                 $parts = (array)parse_url($valid_url);
 593                                 unset($parts['path']);
 594                                 $valid_url = (string)Uri::fromParts($parts);
 595
 596                                 self::setFailureByUrl($url);
 597                                 if (!self::getID($valid_url, true) && !Network::isUrlBlocked($valid_url)) {
 598                                         self::detect($valid_url, $network, $only_nodeinfo);
 599                                 }
 600                                 return false;
 601                         }
 602                         Logger::debug('Found redirect, but ignore it.', ['old' => $url, 'new' => $valid_url]);
 603                 }
 604
 605                 if ((parse_url($url, PHP_URL_HOST) == parse_url($valid_url, PHP_URL_HOST)) &&
 606                         (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH)) &&
 607                         (parse_url($url, PHP_URL_SCHEME) != parse_url($valid_url, PHP_URL_SCHEME))) {
 608                         $url = $valid_url;
 609                 }
 610
 611                 $in_webroot = empty(parse_url($url, PHP_URL_PATH));
 612
 613                 // When a nodeinfo is present, we don't need to dig further
 614                 $curlResult = DI::httpClient()->get($url . '/.well-known/x-nodeinfo2', HttpClientAccept::JSON);
 615                 if ($curlResult->isTimeout()) {
 616                         self::setFailureByUrl($url);
 617                         return false;
 618                 }
 619
 620                 if (!empty($network) && !in_array($network, Protocol::NATIVE_SUPPORT)) {
 621                         $serverdata = ['detection-method' => self::DETECT_MANUAL, 'network' => $network, 'platform' => '', 'version' => '', 'site_name' => '', 'info' => ''];
 622                 } else {
 623                         $serverdata = self::parseNodeinfo210($curlResult);
 624                         if (empty($serverdata)) {
 625                                 $curlResult = DI::httpClient()->get($url . '/.well-known/nodeinfo', HttpClientAccept::JSON);
 626                                 $serverdata = self::fetchNodeinfo($url, $curlResult);
 627                         }
 628                 }
 629
 630                 if ($only_nodeinfo && empty($serverdata)) {
 631                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 632                         self::setFailureByUrl($url);
 633                         return false;
 634                 } elseif (empty($serverdata)) {
 635                         $serverdata = ['detection-method' => self::DETECT_MANUAL, 'network' => Protocol::PHANTOM, 'platform' => '', 'version' => '', 'site_name' => '', 'info' => ''];
 636                 }
 637
 638                 // When there is no Nodeinfo, then use some protocol specific endpoints
 639                 if ($serverdata['network'] == Protocol::PHANTOM) {
 640                         if ($in_webroot) {
 641                                 // Fetch the landing page, possibly it reveals some data
 642                                 $accept = 'application/activity+json,application/ld+json,application/json,*/*;q=0.9';
 643                                 $curlResult = DI::httpClient()->get($url, $accept);
 644                                 if (!$curlResult->isSuccess() && $curlResult->getReturnCode() == '406') {
 645                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 646                                         $html_fetched = true;
 647                                 } else {
 648                                         $html_fetched = false;
 649                                 }
 650
 651                                 if ($curlResult->isSuccess()) {
 652                                         $json = json_decode($curlResult->getBody(), true);
 653                                         if (!empty($json) && is_array($json)) {
 654                                                 $data = self::fetchDataFromSystemActor($json, $serverdata);
 655                                                 $serverdata = $data['server'];
 656                                                 $systemactor = $data['actor'];
 657                                                 if (!$html_fetched && !in_array($serverdata['detection-method'], [self::DETECT_SYSTEM_ACTOR, self::DETECT_AP_COLLECTION])) {
 658                                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 659                                                 }
 660                                         } elseif (!$html_fetched && (strlen($curlResult->getBody()) < 1000)) {
 661                                                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 662                                         }
 663
 664                                         if ($serverdata['detection-method'] != self::DETECT_SYSTEM_ACTOR) {
 665                                                 $serverdata = self::analyseRootHeader($curlResult, $serverdata);
 666                                                 $serverdata = self::analyseRootBody($curlResult, $serverdata);
 667                                         }
 668                                 }
 669
 670                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
 671                                         self::setFailureByUrl($url);
 672                                         return false;
 673                                 }
 674
 675                                 if (in_array($url, ['https://www.threads.net', 'https://threads.net'])) {
 676                                         $serverdata['detection-method'] = self::DETECT_THREADS;
 677                                         $serverdata['network']          = Protocol::ACTIVITYPUB;
 678                                         $serverdata['platform']         = 'threads';
 679                                 }
 680
 681                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 682                                         $serverdata = self::detectMastodonAlikes($url, $serverdata);
 683                                 }
 684                         }
 685
 686                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 687                         // With this check we don't have to waste time and resources for dead systems.
 688                         // Also this hopefully prevents us from receiving abuse messages.
 689                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 690                                 $validHostMeta = self::validHostMeta($url);
 691                         } else {
 692                                 $validHostMeta = false;
 693                         }
 694
 695                         if ($validHostMeta) {
 696                                 if (in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY])) {
 697                                         $serverdata['detection-method'] = self::DETECT_HOST_META;
 698                                 }
 699
 700                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 701                                         $serverdata = self::detectFriendica($url, $serverdata);
 702                                 }
 703
 704                                 // The following systems have to be installed in the root directory.
 705                                 if ($in_webroot) {
 706                                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 707                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 708                                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 709                                         }
 710
 711                                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations, so we check other endpoints as well
 712                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 713                                                 $serverdata = self::detectHubzilla($url, $serverdata);
 714                                         }
 715
 716                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 717                                                 $serverdata = self::detectPeertube($url, $serverdata);
 718                                         }
 719
 720                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 721                                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 722                                         }
 723                                 }
 724                         } elseif (in_array($serverdata['platform'], ['friendica', 'friendika']) && in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_SYSTEM_ACTOR]))) {
 725                                 $serverdata = self::detectFriendica($url, $serverdata);
 726                         }
 727
 728                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 729                                 $serverdata = self::detectNextcloud($url, $serverdata, $validHostMeta);
 730                         }
 731
 732                         // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 733                         // Since this endpoint is only rarely used, we query it at a later time
 734                         if (in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_FRIENDICA, self::DETECT_CONFIG_JSON]))) {
 735                                 $serverdata = self::fetchStatistics($url, $serverdata);
 736                         }
 737                 }
 738
 739                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 740                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 741                         $serverdata['network'] = $network;
 742                 }
 743
 744                 // Most servers aren't installed in a subdirectory, so we declare this entry as failed
 745                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty(parse_url($url, PHP_URL_PATH)) && in_array($serverdata['detection-method'], [self::DETECT_MANUAL])) {
 746                         self::setFailureByUrl($url);
 747                         return false;
 748                 }
 749
 750                 $serverdata['url'] = $url;
 751                 $serverdata['nurl'] = Strings::normaliseLink($url);
 752
 753                 // We have to prevent an endless loop here.
 754                 // When a server is new, then there is no gserver entry yet.
 755                 // But in "detectNetworkViaContacts" it could happen that a contact is updated,
 756                 // and this can call this function here as well.
 757                 if (self::getID($url, true) && (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) ||
 758                         in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META]))) {
 759                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 760                 }
 761
 762                 if (($serverdata['network'] == Protocol::PHANTOM) && in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_BODY])) {
 763                         self::setFailureByUrl($url);
 764                         return false;
 765                 }
 766
 767                 if (empty($serverdata['version']) && in_array($serverdata['platform'], ['osada']) && in_array($serverdata['detection-method'], [self::DETECT_CONTACTS, self::DETECT_BODY])) {
 768                         $serverdata['version'] = self::getNomadVersion($url);
 769                 }
 770
 771                 // Detect the directory type
 772                 $serverdata['directory-type'] = self::DT_NONE;
 773
 774                 if (in_array($serverdata['network'], Protocol::FEDERATED)) {
 775                         $serverdata = self::checkMastodonDirectory($url, $serverdata);
 776
 777                         if ($serverdata['directory-type'] == self::DT_NONE) {
 778                                 $serverdata = self::checkPoCo($url, $serverdata);
 779                         }
 780                 }
 781
 782                 if ($serverdata['network'] == Protocol::ACTIVITYPUB) {
 783                         $serverdata = self::fetchWeeklyUsage($url, $serverdata);
 784                 }
 785
 786                 $serverdata['registered-users'] = $serverdata['registered-users'] ?? 0;
 787
 788                 // Numbers above a reasonable value (10 millions) are ignored
 789                 if ($serverdata['registered-users'] > 10000000) {
 790                         $serverdata['registered-users'] = 0;
 791                 }
 792
 793                 // On an active server there has to be at least a single user
 794                 if (!in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) && ($serverdata['registered-users'] <= 0)) {
 795                         $serverdata['registered-users'] = 1;
 796                 } elseif (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 797                         $serverdata['registered-users'] = 0;
 798                 }
 799
 800                 $serverdata['next_contact'] = self::getNextUpdateDate(true, '', '', in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]));
 801                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 802                 $serverdata['failed']       = false;
 803                 $serverdata['blocked']      = false;
 804
 805                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 806                 if (!DBA::isResult($gserver)) {
 807                         $serverdata['created'] = DateTimeFormat::utcNow();
 808                         $ret = self::insert($serverdata);
 809                         $id = DBA::lastInsertId();
 810                 } else {
 811                         $ret = self::update($serverdata, ['nurl' => $serverdata['nurl']]);
 812                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 813                         if (DBA::isResult($gserver)) {
 814                                 $id = $gserver['id'];
 815                         }
 816                 }
 817
 818                 // Count the number of known contacts from this server
 819                 if (!empty($id) && !in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 820                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 821                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id, 'failed' => false]);
 822                         $max_users = max($apcontacts, $contacts);
 823                         if ($max_users > $serverdata['registered-users']) {
 824                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 825                                 self::update(['registered-users' => $max_users], ['id' => $id]);
 826                         }
 827
 828                         if (empty($serverdata['active-month-users'])) {
 829                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 30 days')]);
 830                                 if ($contacts > 0) {
 831                                         Logger::info('Update monthly users', ['id' => $id, 'url' => $serverdata['nurl'], 'monthly-users' => $contacts]);
 832                                         self::update(['active-month-users' => $contacts], ['id' => $id]);
 833                                 }
 834                         }
 835
 836                         if (empty($serverdata['active-halfyear-users'])) {
 837                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 180 days')]);
 838                                 if ($contacts > 0) {
 839                                         Logger::info('Update halfyear users', ['id' => $id, 'url' => $serverdata['nurl'], 'halfyear-users' => $contacts]);
 840                                         self::update(['active-halfyear-users' => $contacts], ['id' => $id]);
 841                                 }
 842                         }
 843                 }
 844
 845                 if (in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 846                         self::discoverRelay($url);
 847                 }
 848
 849                 if (!empty($systemactor)) {
 850                         $contact = Contact::getByURL($systemactor, true, ['gsid', 'baseurl', 'id', 'network', 'url', 'name']);
 851                         Logger::debug('Fetched system actor',  ['url' => $url, 'gsid' => $id, 'contact' => $contact]);
 852                 }
 853
 854                 return $ret;
 855         }
 856
 857         /**
 858          * Fetch relay data from a given server url
 859          *
 860          * @param string $server_url address of the server
 861          *
 862          * @return void
 863          *
 864          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 865          */
 866         private static function discoverRelay(string $server_url)
 867         {
 868                 Logger::info('Discover relay data', ['server' => $server_url]);
 869
 870                 $curlResult = DI::httpClient()->get($server_url . '/.well-known/x-social-relay', HttpClientAccept::JSON);
 871                 if (!$curlResult->isSuccess()) {
 872                         return;
 873                 }
 874
 875                 $data = json_decode($curlResult->getBody(), true);
 876                 if (!is_array($data)) {
 877                         return;
 878                 }
 879
 880                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 881                 $data['subscribe'] = (bool)($data['subscribe'] ?? false);
 882
 883                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 884                         $data['scope'] = '';
 885                         $data['subscribe'] = false;
 886                         $data['tags'] = [];
 887                 }
 888
 889                 $gserver = DBA::selectFirst('gserver', ['id', 'url', 'network', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 890                 if (!DBA::isResult($gserver)) {
 891                         return;
 892                 }
 893
 894                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 895                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 896                         self::update($fields, ['id' => $gserver['id']]);
 897                 }
 898
 899                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 900
 901                 if ($data['scope'] == 'tags') {
 902                         // Avoid duplicates
 903                         $tags = [];
 904                         foreach ($data['tags'] as $tag) {
 905                                 $tag = mb_strtolower($tag);
 906                                 if (strlen($tag) < 100) {
 907                                         $tags[$tag] = $tag;
 908                                 }
 909                         }
 910
 911                         foreach ($tags as $tag) {
 912                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], Database::INSERT_IGNORE);
 913                         }
 914                 }
 915
 916                 // Create or update the relay contact
 917                 $fields = [];
 918                 if (isset($data['protocols'])) {
 919                         if (isset($data['protocols']['diaspora'])) {
 920                                 $fields['network'] = Protocol::DIASPORA;
 921
 922                                 if (isset($data['protocols']['diaspora']['receive'])) {
 923                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 924                                 } elseif (is_string($data['protocols']['diaspora'])) {
 925                                         $fields['batch'] = $data['protocols']['diaspora'];
 926                                 }
 927                         }
 928
 929                         if (isset($data['protocols']['dfrn'])) {
 930                                 $fields['network'] = Protocol::DFRN;
 931
 932                                 if (isset($data['protocols']['dfrn']['receive'])) {
 933                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 934                                 } elseif (is_string($data['protocols']['dfrn'])) {
 935                                         $fields['batch'] = $data['protocols']['dfrn'];
 936                                 }
 937                         }
 938
 939                         if (isset($data['protocols']['activitypub'])) {
 940                                 $fields['network'] = Protocol::ACTIVITYPUB;
 941
 942                                 if (!empty($data['protocols']['activitypub']['actor'])) {
 943                                         $fields['url'] = $data['protocols']['activitypub']['actor'];
 944                                 }
 945                                 if (!empty($data['protocols']['activitypub']['receive'])) {
 946                                         $fields['batch'] = $data['protocols']['activitypub']['receive'];
 947                                 }
 948                         }
 949                 }
 950
 951                 Logger::info('Discovery ended', ['server' => $server_url, 'data' => $fields]);
 952
 953                 Relay::updateContact($gserver, $fields);
 954         }
 955
 956         /**
 957          * Fetch server data from '/statistics.json' on the given server
 958          *
 959          * @param string $url URL of the given server
 960          *
 961          * @return array server data
 962          */
 963         private static function fetchStatistics(string $url, array $serverdata): array
 964         {
 965                 $curlResult = DI::httpClient()->get($url . '/statistics.json', HttpClientAccept::JSON);
 966                 if (!$curlResult->isSuccess()) {
 967                         return $serverdata;
 968                 }
 969
 970                 $data = json_decode($curlResult->getBody(), true);
 971                 if (empty($data)) {
 972                         return $serverdata;
 973                 }
 974
 975                 // Some AP enabled systems return activity data that we don't expect here.
 976                 if (strpos($curlResult->getContentType(), 'application/activity+json') !== false) {
 977                         return $serverdata;
 978                 }
 979
 980                 $valid = false;
 981                 $old_serverdata = $serverdata;
 982
 983                 $serverdata['detection-method'] = self::DETECT_STATISTICS_JSON;
 984
 985                 if (!empty($data['version'])) {
 986                         $valid = true;
 987                         $serverdata['version'] = $data['version'];
 988                         // Version numbers on statistics.json are presented with additional info, e.g.:
 989                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 990                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 991                 }
 992
 993                 if (!empty($data['name'])) {
 994                         $valid = true;
 995                         $serverdata['site_name'] = $data['name'];
 996                 }
 997
 998                 if (!empty($data['network'])) {
 999                         $valid = true;
1000                         $serverdata['platform'] = strtolower($data['network']);
1001
1002                         if ($serverdata['platform'] == 'diaspora') {
1003                                 $serverdata['network'] = Protocol::DIASPORA;
1004                         } elseif ($serverdata['platform'] == 'friendica') {
1005                                 $serverdata['network'] = Protocol::DFRN;
1006                         } elseif ($serverdata['platform'] == 'hubzilla') {
1007                                 $serverdata['network'] = Protocol::ZOT;
1008                         } elseif ($serverdata['platform'] == 'redmatrix') {
1009                                 $serverdata['network'] = Protocol::ZOT;
1010                         }
1011                 }
1012
1013                 if (!empty($data['total_users'])) {
1014                         $valid = true;
1015                         $serverdata['registered-users'] = max($data['total_users'], 1);
1016                 }
1017
1018                 if (!empty($data['active_users_monthly'])) {
1019                         $valid = true;
1020                         $serverdata['active-month-users'] = max($data['active_users_monthly'], 0);
1021                 }
1022
1023                 if (!empty($data['active_users_halfyear'])) {
1024                         $valid = true;
1025                         $serverdata['active-halfyear-users'] = max($data['active_users_halfyear'], 0);
1026                 }
1027
1028                 if (!empty($data['local_posts'])) {
1029                         $valid = true;
1030                         $serverdata['local-posts'] = max($data['local_posts'], 0);
1031                 }
1032
1033                 if (!empty($data['registrations_open'])) {
1034                         $serverdata['register_policy'] = Register::OPEN;
1035                 } else {
1036                         $serverdata['register_policy'] = Register::CLOSED;
1037                 }
1038
1039                 if (!$valid) {
1040                         return $old_serverdata;
1041                 }
1042
1043                 return $serverdata;
1044         }
1045
1046         /**
1047          * Detect server type by using the nodeinfo data
1048          *
1049          * @param string                  $url        address of the server
1050          * @param ICanHandleHttpResponses $httpResult
1051          *
1052          * @return array Server data
1053          *
1054          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1055          */
1056         private static function fetchNodeinfo(string $url, ICanHandleHttpResponses $httpResult): array
1057         {
1058                 if (!$httpResult->isSuccess()) {
1059                         return [];
1060                 }
1061
1062                 $nodeinfo = json_decode($httpResult->getBody(), true);
1063
1064                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
1065                         return [];
1066                 }
1067
1068                 $nodeinfo1_url = '';
1069                 $nodeinfo2_url = '';
1070
1071                 foreach ($nodeinfo['links'] as $link) {
1072                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
1073                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
1074                                 continue;
1075                         }
1076
1077                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
1078                                 $nodeinfo1_url = Network::addBasePath($link['href'], $httpResult->getUrl());
1079                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
1080                                 $nodeinfo2_url = Network::addBasePath($link['href'], $httpResult->getUrl());
1081                         }
1082                 }
1083
1084                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
1085                         return [];
1086                 }
1087
1088                 $server = [];
1089
1090                 if (!empty($nodeinfo2_url)) {
1091                         $server = self::parseNodeinfo2($nodeinfo2_url);
1092                 }
1093
1094                 if (empty($server) && !empty($nodeinfo1_url)) {
1095                         $server = self::parseNodeinfo1($nodeinfo1_url);
1096                 }
1097
1098                 return $server;
1099         }
1100
1101         /**
1102          * Parses Nodeinfo 1
1103          *
1104          * @param string $nodeinfo_url address of the nodeinfo path
1105          *
1106          * @return array Server data
1107          *
1108          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1109          */
1110         private static function parseNodeinfo1(string $nodeinfo_url): array
1111         {
1112                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
1113                 if (!$curlResult->isSuccess()) {
1114                         return [];
1115                 }
1116
1117                 $nodeinfo = json_decode($curlResult->getBody(), true);
1118
1119                 if (!is_array($nodeinfo)) {
1120                         return [];
1121                 }
1122
1123                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
1124                         'register_policy' => Register::CLOSED];
1125
1126                 if (!empty($nodeinfo['openRegistrations'])) {
1127                         $server['register_policy'] = Register::OPEN;
1128                 }
1129
1130                 if (is_array($nodeinfo['software'])) {
1131                         if (!empty($nodeinfo['software']['name'])) {
1132                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
1133                         }
1134
1135                         if (!empty($nodeinfo['software']['version'])) {
1136                                 $server['version'] = $nodeinfo['software']['version'];
1137                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1138                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1139                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1140                         }
1141                 }
1142
1143                 if (!empty($nodeinfo['metadata']['nodeName'])) {
1144                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
1145                 }
1146
1147                 if (!empty($nodeinfo['usage']['users']['total'])) {
1148                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1149                 }
1150
1151                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1152                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1153                 }
1154
1155                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1156                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1157                 }
1158
1159                 if (!empty($nodeinfo['usage']['localPosts'])) {
1160                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1161                 }
1162
1163                 if (!empty($nodeinfo['usage']['localComments'])) {
1164                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1165                 }
1166
1167                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
1168                         $protocols = [];
1169                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
1170                                 $protocols[$protocol] = true;
1171                         }
1172
1173                         if (!empty($protocols['friendica'])) {
1174                                 $server['network'] = Protocol::DFRN;
1175                         } elseif (!empty($protocols['activitypub'])) {
1176                                 $server['network'] = Protocol::ACTIVITYPUB;
1177                         } elseif (!empty($protocols['diaspora'])) {
1178                                 $server['network'] = Protocol::DIASPORA;
1179                         } elseif (!empty($protocols['ostatus'])) {
1180                                 $server['network'] = Protocol::OSTATUS;
1181                         } elseif (!empty($protocols['gnusocial'])) {
1182                                 $server['network'] = Protocol::OSTATUS;
1183                         } elseif (!empty($protocols['zot'])) {
1184                                 $server['network'] = Protocol::ZOT;
1185                         }
1186                 }
1187
1188                 if (empty($server)) {
1189                         return [];
1190                 }
1191
1192                 if (empty($server['network'])) {
1193                         $server['network'] = Protocol::PHANTOM;
1194                 }
1195
1196                 return $server;
1197         }
1198
1199         /**
1200          * Parses Nodeinfo 2
1201          *
1202          * @see https://git.feneas.org/jaywink/nodeinfo2
1203          *
1204          * @param string $nodeinfo_url address of the nodeinfo path
1205          *
1206          * @return array Server data
1207          *
1208          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1209          */
1210         private static function parseNodeinfo2(string $nodeinfo_url): array
1211         {
1212                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
1213                 if (!$curlResult->isSuccess()) {
1214                         return [];
1215                 }
1216
1217                 $nodeinfo = json_decode($curlResult->getBody(), true);
1218                 if (!is_array($nodeinfo)) {
1219                         return [];
1220                 }
1221
1222                 $server = [
1223                         'detection-method' => self::DETECT_NODEINFO_2,
1224                         'register_policy' => Register::CLOSED,
1225                         'platform' => 'unknown',
1226                 ];
1227
1228                 if (!empty($nodeinfo['openRegistrations'])) {
1229                         $server['register_policy'] = Register::OPEN;
1230                 }
1231
1232                 if (!empty($nodeinfo['software'])) {
1233                         if (isset($nodeinfo['software']['name'])) {
1234                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
1235                         }
1236
1237                         if (!empty($nodeinfo['software']['version']) && isset($server['platform'])) {
1238                                 $server['version'] = $nodeinfo['software']['version'];
1239                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1240                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1241                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1242
1243                                 // qoto advertises itself as Mastodon
1244                                 if (($server['platform'] == 'mastodon') && substr($nodeinfo['software']['version'], -5) == '-qoto') {
1245                                         $server['platform'] = 'qoto';
1246                                 }
1247                         }
1248                 }
1249
1250                 if (!empty($nodeinfo['metadata']['nodeName'])) {
1251                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
1252                 }
1253
1254                 if (!empty($nodeinfo['usage']['users']['total'])) {
1255                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1256                 }
1257
1258                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1259                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1260                 }
1261
1262                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1263                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1264                 }
1265
1266                 if (!empty($nodeinfo['usage']['localPosts'])) {
1267                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1268                 }
1269
1270                 if (!empty($nodeinfo['usage']['localComments'])) {
1271                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1272                 }
1273
1274                 if (!empty($nodeinfo['protocols'])) {
1275                         $protocols = [];
1276                         if (is_string($nodeinfo['protocols'])) {
1277                                 $protocols[$nodeinfo['protocols']] = true;
1278                         } else {
1279                                 foreach ($nodeinfo['protocols'] as $protocol) {
1280                                         if (is_string($protocol)) {
1281                                                 $protocols[$protocol] = true;
1282                                         }
1283                                 }
1284                         }
1285
1286                         if (!empty($protocols['dfrn'])) {
1287                                 $server['network'] = Protocol::DFRN;
1288                         } elseif (!empty($protocols['activitypub'])) {
1289                                 $server['network'] = Protocol::ACTIVITYPUB;
1290                         } elseif (!empty($protocols['diaspora'])) {
1291                                 $server['network'] = Protocol::DIASPORA;
1292                         } elseif (!empty($protocols['ostatus'])) {
1293                                 $server['network'] = Protocol::OSTATUS;
1294                         } elseif (!empty($protocols['gnusocial'])) {
1295                                 $server['network'] = Protocol::OSTATUS;
1296                         } elseif (!empty($protocols['zot'])) {
1297                                 $server['network'] = Protocol::ZOT;
1298                         }
1299                 }
1300
1301                 if (empty($server)) {
1302                         return [];
1303                 }
1304
1305                 if (empty($server['network'])) {
1306                         $server['network'] = Protocol::PHANTOM;
1307                 }
1308
1309                 return $server;
1310         }
1311
1312         /**
1313          * Parses NodeInfo2 protocol 1.0
1314          *
1315          * @see https://github.com/jaywink/nodeinfo2/blob/master/PROTOCOL.md
1316          *
1317          * @param string $nodeinfo_url address of the nodeinfo path
1318          *
1319          * @return array Server data
1320          *
1321          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1322          */
1323         private static function parseNodeinfo210(ICanHandleHttpResponses $httpResult): array
1324         {
1325                 if (!$httpResult->isSuccess()) {
1326                         return [];
1327                 }
1328
1329                 $nodeinfo = json_decode($httpResult->getBody(), true);
1330
1331                 if (!is_array($nodeinfo)) {
1332                         return [];
1333                 }
1334
1335                 $server = ['detection-method' => self::DETECT_NODEINFO_210,
1336                         'register_policy' => Register::CLOSED];
1337
1338                 if (!empty($nodeinfo['openRegistrations'])) {
1339                         $server['register_policy'] = Register::OPEN;
1340                 }
1341
1342                 if (!empty($nodeinfo['server'])) {
1343                         if (!empty($nodeinfo['server']['software'])) {
1344                                 $server['platform'] = strtolower($nodeinfo['server']['software']);
1345                         }
1346
1347                         if (!empty($nodeinfo['server']['version'])) {
1348                                 $server['version'] = $nodeinfo['server']['version'];
1349                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1350                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1351                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1352                         }
1353
1354                         if (!empty($nodeinfo['server']['name'])) {
1355                                 $server['site_name'] = $nodeinfo['server']['name'];
1356                         }
1357                 }
1358
1359                 if (!empty($nodeinfo['usage']['users']['total'])) {
1360                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1361                 }
1362
1363                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1364                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1365                 }
1366
1367                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1368                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1369                 }
1370
1371                 if (!empty($nodeinfo['usage']['localPosts'])) {
1372                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1373                 }
1374
1375                 if (!empty($nodeinfo['usage']['localComments'])) {
1376                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1377                 }
1378
1379                 if (!empty($nodeinfo['protocols'])) {
1380                         $protocols = [];
1381                         if (is_string($nodeinfo['protocols'])) {
1382                                 $protocols[$nodeinfo['protocols']] = true;
1383                         } else {
1384                                 foreach ($nodeinfo['protocols'] as $protocol) {
1385                                         if (is_string($protocol)) {
1386                                                 $protocols[$protocol] = true;
1387                                         }
1388                                 }
1389                         }
1390
1391                         if (!empty($protocols['dfrn'])) {
1392                                 $server['network'] = Protocol::DFRN;
1393                         } elseif (!empty($protocols['activitypub'])) {
1394                                 $server['network'] = Protocol::ACTIVITYPUB;
1395                         } elseif (!empty($protocols['diaspora'])) {
1396                                 $server['network'] = Protocol::DIASPORA;
1397                         } elseif (!empty($protocols['ostatus'])) {
1398                                 $server['network'] = Protocol::OSTATUS;
1399                         } elseif (!empty($protocols['gnusocial'])) {
1400                                 $server['network'] = Protocol::OSTATUS;
1401                         } elseif (!empty($protocols['zot'])) {
1402                                 $server['network'] = Protocol::ZOT;
1403                         }
1404                 }
1405
1406                 if (empty($server) || empty($server['platform'])) {
1407                         return [];
1408                 }
1409
1410                 if (empty($server['network'])) {
1411                         $server['network'] = Protocol::PHANTOM;
1412                 }
1413
1414                 return $server;
1415         }
1416
1417         /**
1418          * Fetch server information from a 'siteinfo.json' file on the given server
1419          *
1420          * @param string $url        URL of the given server
1421          * @param array  $serverdata array with server data
1422          *
1423          * @return array server data
1424          */
1425         private static function fetchSiteinfo(string $url, array $serverdata): array
1426         {
1427                 $curlResult = DI::httpClient()->get($url . '/siteinfo.json', HttpClientAccept::JSON);
1428                 if (!$curlResult->isSuccess()) {
1429                         return $serverdata;
1430                 }
1431
1432                 $data = json_decode($curlResult->getBody(), true);
1433                 if (empty($data)) {
1434                         return $serverdata;
1435                 }
1436
1437                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1438                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
1439                 }
1440
1441                 if (!empty($data['platform'])) {
1442                         $serverdata['platform'] = strtolower($data['platform']);
1443                         $serverdata['version'] = $data['version'] ?? 'N/A';
1444                 }
1445
1446                 if (!empty($data['plugins'])) {
1447                         if (in_array('pubcrawl', $data['plugins'])) {
1448                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1449                         } elseif (in_array('diaspora', $data['plugins'])) {
1450                                 $serverdata['network'] = Protocol::DIASPORA;
1451                         } elseif (in_array('gnusoc', $data['plugins'])) {
1452                                 $serverdata['network'] = Protocol::OSTATUS;
1453                         } else {
1454                                 $serverdata['network'] = Protocol::ZOT;
1455                         }
1456                 }
1457
1458                 if (!empty($data['site_name'])) {
1459                         $serverdata['site_name'] = $data['site_name'];
1460                 }
1461
1462                 if (!empty($data['channels_total'])) {
1463                         $serverdata['registered-users'] = max($data['channels_total'], 1);
1464                 }
1465
1466                 if (!empty($data['channels_active_monthly'])) {
1467                         $serverdata['active-month-users'] = max($data['channels_active_monthly'], 0);
1468                 }
1469
1470                 if (!empty($data['channels_active_halfyear'])) {
1471                         $serverdata['active-halfyear-users'] = max($data['channels_active_halfyear'], 0);
1472                 }
1473
1474                 if (!empty($data['local_posts'])) {
1475                         $serverdata['local-posts'] = max($data['local_posts'], 0);
1476                 }
1477
1478                 if (!empty($data['local_comments'])) {
1479                         $serverdata['local-comments'] = max($data['local_comments'], 0);
1480                 }
1481
1482                 if (!empty($data['register_policy'])) {
1483                         switch ($data['register_policy']) {
1484                                 case 'REGISTER_OPEN':
1485                                         $serverdata['register_policy'] = Register::OPEN;
1486                                         break;
1487
1488                                 case 'REGISTER_APPROVE':
1489                                         $serverdata['register_policy'] = Register::APPROVE;
1490                                         break;
1491
1492                                 case 'REGISTER_CLOSED':
1493                                 default:
1494                                         $serverdata['register_policy'] = Register::CLOSED;
1495                                         break;
1496                         }
1497                 }
1498
1499                 return $serverdata;
1500         }
1501
1502         /**
1503          * Fetches server data via an ActivityPub account with url of that server
1504          *
1505          * @param string $url        URL of the given server
1506          * @param array  $serverdata array with server data
1507          *
1508          * @return array server data
1509          *
1510          * @throws Exception
1511          */
1512         private static function fetchDataFromSystemActor(array $data, array $serverdata): array
1513         {
1514                 if (empty($data)) {
1515                         return ['server' => $serverdata, 'actor' => ''];
1516                 }
1517
1518                 $actor = JsonLD::compact($data, false);
1519                 if (in_array(JsonLD::fetchElement($actor, '@type'), ActivityPub\Receiver::ACCOUNT_TYPES)) {
1520                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1521                         $serverdata['site_name'] = JsonLD::fetchElement($actor, 'as:name', '@value');
1522                         $serverdata['info'] = JsonLD::fetchElement($actor, 'as:summary', '@value');
1523                         if (self::isNomad($actor)) {
1524                                 $serverdata['platform'] = self::getNomadName($actor['@id']);
1525                                 $serverdata['version'] = self::getNomadVersion($actor['@id']);
1526                                 $serverdata['detection-method'] = self::DETECT_SYSTEM_ACTOR;
1527                         } elseif (!empty($actor['as:generator'])) {
1528                                 $generator = explode(' ', JsonLD::fetchElement($actor['as:generator'], 'as:name', '@value'));
1529                                 $serverdata['platform'] = strtolower(array_shift($generator));
1530                                 $serverdata['version'] = self::getNomadVersion($actor['@id']);
1531                                 $serverdata['detection-method'] = self::DETECT_SYSTEM_ACTOR;
1532                         } else {
1533                                 $serverdata['detection-method'] = self::DETECT_AP_ACTOR;
1534                         }
1535                         return ['server' => $serverdata, 'actor' => $actor['@id']];
1536                 } elseif ((JsonLD::fetchElement($actor, '@type') == 'as:Collection')) {
1537                         // By now only Ktistec seems to provide collections this way
1538                         $serverdata['platform'] = 'ktistec';
1539                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1540                         $serverdata['detection-method'] = self::DETECT_AP_COLLECTION;
1541
1542                         $actors = JsonLD::fetchElementArray($actor, 'as:items');
1543                         if (!empty($actors) && !empty($actors[0]['@id'])) {
1544                                 $actor_url = $actor['@id'] . $actors[0]['@id'];
1545                         } else {
1546                                 $actor_url = '';
1547                         }
1548
1549                         return ['server' => $serverdata, 'actor' => $actor_url];
1550                 }
1551                 return ['server' => $serverdata, 'actor' => ''];
1552         }
1553
1554         /**
1555          * Detect if the given actor is a nomad account
1556          *
1557          * @param array $actor
1558          * @return boolean
1559          */
1560         private static function isNomad(array $actor): bool
1561         {
1562                 $tags = JsonLD::fetchElementArray($actor, 'as:tag');
1563                 if (empty($tags)) {
1564                         return false;
1565                 }
1566
1567                 foreach ($tags as $tag) {
1568                         if ((($tag['as:name'] ?? '') == 'Protocol') && (($tag['sc:value'] ?? '') == 'nomad')) {
1569                                 return true;
1570                         }
1571                 }
1572                 return false;
1573         }
1574
1575         /**
1576          * Fetch the name of Nomad implementation
1577          *
1578          * @param string $url
1579          * @return string
1580          */
1581         private static function getNomadName(string $url): string
1582         {
1583                 $name = 'nomad';
1584                 $curlResult = DI::httpClient()->get($url . '/manifest', 'application/manifest+json');
1585                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1586                         return $name;
1587                 }
1588
1589                 $data = json_decode($curlResult->getBody(), true);
1590                 if (empty($data)) {
1591                         return $name;
1592                 }
1593
1594                 return $data['name'] ?? $name;
1595         }
1596
1597         /**
1598          * Fetch the version of the Nomad installation
1599          *
1600          * @param string $url
1601          * @return string
1602          */
1603         private static function getNomadVersion(string $url): string
1604         {
1605                 $curlResult = DI::httpClient()->get($url . '/api/z/1.0/version', HttpClientAccept::JSON);
1606                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1607                         return '';
1608                 }
1609
1610                 $data = json_decode($curlResult->getBody(), true);
1611                 if (empty($data)) {
1612                         return '';
1613                 }
1614                 return $data ?? '';
1615         }
1616
1617         /**
1618          * Checks if the server contains a valid host meta file
1619          *
1620          * @param string $url URL of the given server
1621          *
1622          * @return boolean 'true' if the server seems to be vital
1623          */
1624         private static function validHostMeta(string $url): bool
1625         {
1626                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
1627                 $curlResult = DI::httpClient()->get($url . Probe::HOST_META, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
1628                 if (!$curlResult->isSuccess()) {
1629                         return false;
1630                 }
1631
1632                 $xrd = XML::parseString($curlResult->getBody(), true);
1633                 if (!is_object($xrd)) {
1634                         return false;
1635                 }
1636
1637                 $elements = XML::elementToArray($xrd);
1638                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
1639                         return false;
1640                 }
1641
1642                 $valid = false;
1643                 foreach ($elements['xrd']['link'] as $link) {
1644                         // When there is more than a single "link" element, the array looks slightly different
1645                         if (!empty($link['@attributes'])) {
1646                                 $link = $link['@attributes'];
1647                         }
1648
1649                         if (empty($link['rel']) || empty($link['template'])) {
1650                                 continue;
1651                         }
1652
1653                         if ($link['rel'] == 'lrdd') {
1654                                 // When the webfinger host is the same like the system host, it should be ok.
1655                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
1656                         }
1657                 }
1658
1659                 return $valid;
1660         }
1661
1662         /**
1663          * Detect the network of the given server via their known contacts
1664          *
1665          * @param string $url        URL of the given server
1666          * @param array  $serverdata array with server data
1667          *
1668          * @return array server data
1669          */
1670         private static function detectNetworkViaContacts(string $url, array $serverdata): array
1671         {
1672                 $contacts = [];
1673
1674                 $nurl = Strings::normaliseLink($url);
1675
1676                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $nurl]]);
1677                 while ($apcontact = DBA::fetch($apcontacts)) {
1678                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
1679                 }
1680                 DBA::close($apcontacts);
1681
1682                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $nurl]]);
1683                 while ($pcontact = DBA::fetch($pcontacts)) {
1684                         $contacts[$pcontact['nurl']] = $pcontact['url'];
1685                 }
1686                 DBA::close($pcontacts);
1687
1688                 if (empty($contacts)) {
1689                         return $serverdata;
1690                 }
1691
1692                 $time = time();
1693                 foreach ($contacts as $contact) {
1694                         // Endlosschleife verhindern wegen gsid!
1695                         $data = Probe::uri($contact);
1696                         if (in_array($data['network'], Protocol::FEDERATED)) {
1697                                 $serverdata['network'] = $data['network'];
1698
1699                                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1700                                         $serverdata['detection-method'] = self::DETECT_CONTACTS;
1701                                 }
1702                                 break;
1703                         } elseif ((time() - $time) > 10) {
1704                                 // To reduce the stress on remote systems we probe a maximum of 10 seconds
1705                                 break;
1706                         }
1707                 }
1708
1709                 return $serverdata;
1710         }
1711
1712         /**
1713          * Checks if the given server does have a '/poco' endpoint.
1714          * This is used for the 'PortableContact' functionality,
1715          * which is used by both Friendica and Hubzilla.
1716          *
1717          * @param string $url        URL of the given server
1718          * @param array  $serverdata array with server data
1719          *
1720          * @return array server data
1721          */
1722         private static function checkPoCo(string $url, array $serverdata): array
1723         {
1724                 $serverdata['poco'] = '';
1725
1726                 $curlResult = DI::httpClient()->get($url . '/poco', HttpClientAccept::JSON);
1727                 if (!$curlResult->isSuccess()) {
1728                         return $serverdata;
1729                 }
1730
1731                 $data = json_decode($curlResult->getBody(), true);
1732                 if (empty($data)) {
1733                         return $serverdata;
1734                 }
1735
1736                 if (!empty($data['totalResults'])) {
1737                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1738                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers, 1);
1739                         $serverdata['directory-type'] = self::DT_POCO;
1740                         $serverdata['poco'] = $url . '/poco';
1741                 }
1742
1743                 return $serverdata;
1744         }
1745
1746         /**
1747          * Checks if the given server does have a Mastodon style directory endpoint.
1748          *
1749          * @param string $url        URL of the given server
1750          * @param array  $serverdata array with server data
1751          *
1752          * @return array server data
1753          */
1754         public static function checkMastodonDirectory(string $url, array $serverdata): array
1755         {
1756                 $curlResult = DI::httpClient()->get($url . '/api/v1/directory?limit=1', HttpClientAccept::JSON);
1757                 if (!$curlResult->isSuccess()) {
1758                         return $serverdata;
1759                 }
1760
1761                 $data = json_decode($curlResult->getBody(), true);
1762                 if (empty($data)) {
1763                         return $serverdata;
1764                 }
1765
1766                 if (count($data) == 1) {
1767                         $serverdata['directory-type'] = self::DT_MASTODON;
1768                 }
1769
1770                 return $serverdata;
1771         }
1772
1773         /**
1774          * Detects Peertube via their known endpoint
1775          *
1776          * @param string $url        URL of the given server
1777          * @param array  $serverdata array with server data
1778          *
1779          * @return array server data
1780          */
1781         private static function detectPeertube(string $url, array $serverdata): array
1782         {
1783                 $curlResult = DI::httpClient()->get($url . '/api/v1/config', HttpClientAccept::JSON);
1784                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1785                         return $serverdata;
1786                 }
1787
1788                 $data = json_decode($curlResult->getBody(), true);
1789                 if (empty($data)) {
1790                         return $serverdata;
1791                 }
1792
1793                 if (!empty($data['instance']) && !empty($data['serverVersion'])) {
1794                         $serverdata['platform'] = 'peertube';
1795                         $serverdata['version'] = $data['serverVersion'];
1796                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1797
1798                         if (!empty($data['instance']['name'])) {
1799                                 $serverdata['site_name'] = $data['instance']['name'];
1800                         }
1801
1802                         if (!empty($data['instance']['shortDescription'])) {
1803                                 $serverdata['info'] = $data['instance']['shortDescription'];
1804                         }
1805
1806                         if (!empty($data['signup'])) {
1807                                 if (!empty($data['signup']['allowed'])) {
1808                                         $serverdata['register_policy'] = Register::OPEN;
1809                                 }
1810                         }
1811
1812                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1813                                 $serverdata['detection-method'] = self::DETECT_V1_CONFIG;
1814                         }
1815                 }
1816
1817                 return $serverdata;
1818         }
1819
1820         /**
1821          * Detects the version number of a given server when it was a NextCloud installation
1822          *
1823          * @param string $url        URL of the given server
1824          * @param array  $serverdata array with server data
1825          * @param bool   $validHostMeta
1826          *
1827          * @return array server data
1828          */
1829         private static function detectNextcloud(string $url, array $serverdata, bool $validHostMeta): array
1830         {
1831                 $curlResult = DI::httpClient()->get($url . '/status.php', HttpClientAccept::JSON);
1832                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1833                         return $serverdata;
1834                 }
1835
1836                 $data = json_decode($curlResult->getBody(), true);
1837                 if (empty($data)) {
1838                         return $serverdata;
1839                 }
1840
1841                 if (!empty($data['version'])) {
1842                         $serverdata['platform'] = 'nextcloud';
1843                         $serverdata['version'] = $data['version'];
1844
1845                         if ($validHostMeta) {
1846                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1847                         }
1848
1849                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1850                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1851                         }
1852                 }
1853
1854                 return $serverdata;
1855         }
1856
1857         /**
1858          * Fetches weekly usage data
1859          *
1860          * @param string $url        URL of the given server
1861          * @param array  $serverdata array with server data
1862          *
1863          * @return array server data
1864          */
1865         private static function fetchWeeklyUsage(string $url, array $serverdata): array
1866         {
1867                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance/activity', HttpClientAccept::JSON);
1868                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1869                         return $serverdata;
1870                 }
1871
1872                 $data = json_decode($curlResult->getBody(), true);
1873                 if (empty($data)) {
1874                         return $serverdata;
1875                 }
1876
1877                 $current_week = [];
1878                 foreach ($data as $week) {
1879                         // Use only data from a full week
1880                         if (empty($week['week']) || (time() - $week['week']) < 7 * 24 * 60 * 60) {
1881                                 continue;
1882                         }
1883
1884                         // Most likely the data is sorted correctly. But we better are safe than sorry
1885                         if (empty($current_week['week']) || ($current_week['week'] < $week['week'])) {
1886                                 $current_week = $week;
1887                         }
1888                 }
1889
1890                 if (!empty($current_week['logins'])) {
1891                         $serverdata['active-week-users'] = max($current_week['logins'], 0);
1892                 }
1893
1894                 return $serverdata;
1895         }
1896
1897         /**
1898          * Detects data from a given server url if it was a mastodon alike system
1899          *
1900          * @param string $url        URL of the given server
1901          * @param array  $serverdata array with server data
1902          *
1903          * @return array server data
1904          */
1905         private static function detectMastodonAlikes(string $url, array $serverdata): array
1906         {
1907                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance', HttpClientAccept::JSON);
1908                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1909                         return $serverdata;
1910                 }
1911
1912                 $data = json_decode($curlResult->getBody(), true);
1913                 if (empty($data)) {
1914                         return $serverdata;
1915                 }
1916
1917                 $valid = false;
1918
1919                 if (!empty($data['version'])) {
1920                         $serverdata['platform'] = 'mastodon';
1921                         $serverdata['version'] = $data['version'] ?? '';
1922                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1923                         $valid = true;
1924                 }
1925
1926                 if (!empty($data['title'])) {
1927                         $serverdata['site_name'] = $data['title'];
1928                 }
1929
1930                 if (!empty($data['title']) && empty($serverdata['platform']) && ($serverdata['network'] == Protocol::PHANTOM)) {
1931                         $serverdata['platform'] = 'mastodon';
1932                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1933                         $valid = true;
1934                 }
1935
1936                 if (!empty($data['description'])) {
1937                         $serverdata['info'] = trim($data['description']);
1938                 }
1939
1940                 if (!empty($data['stats']['user_count'])) {
1941                         $serverdata['registered-users'] = max($data['stats']['user_count'], 1);
1942                 }
1943
1944                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1945                         $serverdata['platform'] = strtolower($matches[1]);
1946                         $serverdata['version'] = $matches[2];
1947                         $valid = true;
1948                 }
1949
1950                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1951                         $serverdata['platform'] = 'pleroma';
1952                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1953                         $valid = true;
1954                 }
1955
1956                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1957                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1958                         $serverdata['platform'] = 'pleroma';
1959                         $valid = true;
1960                 }
1961
1962                 if ($valid && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1963                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1964                 }
1965
1966                 return $serverdata;
1967         }
1968
1969         /**
1970          * Detects data from typical Hubzilla endpoints
1971          *
1972          * @param string $url        URL of the given server
1973          * @param array  $serverdata array with server data
1974          *
1975          * @return array server data
1976          */
1977         private static function detectHubzilla(string $url, array $serverdata): array
1978         {
1979                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/config.json', HttpClientAccept::JSON);
1980                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1981                         return $serverdata;
1982                 }
1983
1984                 $data = json_decode($curlResult->getBody(), true);
1985                 if (empty($data) || empty($data['site'])) {
1986                         return $serverdata;
1987                 }
1988
1989                 if (!empty($data['site']['name'])) {
1990                         $serverdata['site_name'] = $data['site']['name'];
1991                 }
1992
1993                 if (!empty($data['site']['platform'])) {
1994                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1995                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1996                         $serverdata['network'] = Protocol::ZOT;
1997                 }
1998
1999                 if (!empty($data['site']['hubzilla'])) {
2000                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
2001                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
2002                         $serverdata['network'] = Protocol::ZOT;
2003                 }
2004
2005                 if (!empty($data['site']['redmatrix'])) {
2006                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
2007                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
2008                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
2009                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
2010                         }
2011
2012                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
2013                         $serverdata['network'] = Protocol::ZOT;
2014                 }
2015
2016                 $private = false;
2017                 $inviteonly = false;
2018                 $closed = false;
2019
2020                 if (!empty($data['site']['closed'])) {
2021                         $closed = self::toBoolean($data['site']['closed']);
2022                 }
2023
2024                 if (!empty($data['site']['private'])) {
2025                         $private = self::toBoolean($data['site']['private']);
2026                 }
2027
2028                 if (!empty($data['site']['inviteonly'])) {
2029                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
2030                 }
2031
2032                 if (!$closed && !$private and $inviteonly) {
2033                         $serverdata['register_policy'] = Register::APPROVE;
2034                 } elseif (!$closed && !$private) {
2035                         $serverdata['register_policy'] = Register::OPEN;
2036                 } else {
2037                         $serverdata['register_policy'] = Register::CLOSED;
2038                 }
2039
2040                 if (($serverdata['network'] != Protocol::PHANTOM) && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
2041                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
2042                 }
2043
2044                 return $serverdata;
2045         }
2046
2047         /**
2048          * Converts input value to a boolean value
2049          *
2050          * @param string|integer $val
2051          *
2052          * @return boolean
2053          */
2054         private static function toBoolean($val): bool
2055         {
2056                 if (($val == 'true') || ($val == 1)) {
2057                         return true;
2058                 } elseif (($val == 'false') || ($val == 0)) {
2059                         return false;
2060                 }
2061
2062                 return $val;
2063         }
2064
2065         /**
2066          * Detect if the URL belongs to a GNU Social server
2067          *
2068          * @param string $url        URL of the given server
2069          * @param array  $serverdata array with server data
2070          *
2071          * @return array server data
2072          */
2073         private static function detectGNUSocial(string $url, array $serverdata): array
2074         {
2075                 // Test for GNU Social
2076                 $curlResult = DI::httpClient()->get($url . '/api/gnusocial/version.json', HttpClientAccept::JSON);
2077                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
2078                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
2079                         $serverdata['platform'] = 'gnusocial';
2080                         // Remove junk that some GNU Social servers return
2081                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
2082                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
2083                         $serverdata['version'] = trim($serverdata['version'], '"');
2084                         $serverdata['network'] = Protocol::OSTATUS;
2085
2086                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
2087                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
2088                         }
2089
2090                         return $serverdata;
2091                 }
2092
2093                 // Test for Statusnet
2094                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/version.json', HttpClientAccept::JSON);
2095                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
2096                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
2097
2098                         // Remove junk that some GNU Social servers return
2099                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
2100                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
2101                         $serverdata['version'] = trim($serverdata['version'], '"');
2102
2103                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
2104                                 $serverdata['platform'] = 'pleroma';
2105                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
2106                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
2107                         } else {
2108                                 $serverdata['platform'] = 'statusnet';
2109                                 $serverdata['network'] = Protocol::OSTATUS;
2110                         }
2111
2112                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
2113                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
2114                         }
2115                 }
2116
2117                 return $serverdata;
2118         }
2119
2120         /**
2121          * Detect if the URL belongs to a Friendica server
2122          *
2123          * @param string $url        URL of the given server
2124          * @param array  $serverdata array with server data
2125          *
2126          * @return array server data
2127          */
2128         private static function detectFriendica(string $url, array $serverdata): array
2129         {
2130                 // There is a bug in some versions of Friendica that will return an ActivityStream actor when the content type "application/json" is requested.
2131                 // Because of this me must not use ACCEPT_JSON here.
2132                 $curlResult = DI::httpClient()->get($url . '/friendica/json');
2133                 if (!$curlResult->isSuccess()) {
2134                         $curlResult = DI::httpClient()->get($url . '/friendika/json');
2135                         $friendika = true;
2136                         $platform = 'Friendika';
2137                 } else {
2138                         $friendika = false;
2139                         $platform = 'Friendica';
2140                 }
2141
2142                 if (!$curlResult->isSuccess()) {
2143                         return $serverdata;
2144                 }
2145
2146                 $data = json_decode($curlResult->getBody(), true);
2147                 if (empty($data) || empty($data['version'])) {
2148                         return $serverdata;
2149                 }
2150
2151                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
2152                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
2153                 }
2154
2155                 $serverdata['network'] = Protocol::DFRN;
2156                 $serverdata['version'] = $data['version'];
2157
2158                 if (!empty($data['no_scrape_url'])) {
2159                         $serverdata['noscrape'] = $data['no_scrape_url'];
2160                 }
2161
2162                 if (!empty($data['site_name'])) {
2163                         $serverdata['site_name'] = $data['site_name'];
2164                 }
2165
2166                 if (!empty($data['info'])) {
2167                         $serverdata['info'] = trim($data['info']);
2168                 }
2169
2170                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
2171                 switch ($register_policy) {
2172                         case 'REGISTER_OPEN':
2173                                 $serverdata['register_policy'] = Register::OPEN;
2174                                 break;
2175
2176                         case 'REGISTER_APPROVE':
2177                                 $serverdata['register_policy'] = Register::APPROVE;
2178                                 break;
2179
2180                         case 'REGISTER_CLOSED':
2181                         case 'REGISTER_INVITATION':
2182                                 $serverdata['register_policy'] = Register::CLOSED;
2183                                 break;
2184                         default:
2185                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
2186                                 $serverdata['register_policy'] = Register::CLOSED;
2187                                 break;
2188                 }
2189
2190                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
2191
2192                 return $serverdata;
2193         }
2194
2195         /**
2196          * Analyses the landing page of a given server for hints about type and system of that server
2197          *
2198          * @param object $curlResult result of curl execution
2199          * @param array  $serverdata array with server data
2200          *
2201          * @return array server data
2202          */
2203         private static function analyseRootBody($curlResult, array $serverdata): array
2204         {
2205                 if (empty($curlResult->getBody())) {
2206                         return $serverdata;
2207                 }
2208
2209                 if (file_exists(__DIR__ . '/../../static/platforms.config.php')) {
2210                         require __DIR__ . '/../../static/platforms.config.php';
2211                 } else {
2212                         throw new HTTPException\InternalServerErrorException('Invalid platform file');
2213                 }
2214
2215                 $platforms = array_merge($ap_platforms, $dfrn_platforms, $zap_platforms, $platforms);
2216
2217                 $doc = new DOMDocument();
2218                 @$doc->loadHTML($curlResult->getBody());
2219                 $xpath = new DOMXPath($doc);
2220                 $assigned = false;
2221
2222                 // We can only detect honk via some HTML element on their page
2223                 if ($xpath->query('//div[@id="honksonpage"]')->count() == 1) {
2224                         $serverdata['platform'] = 'honk';
2225                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2226                         $assigned = true;
2227                 }
2228
2229                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
2230                 if (!empty($title)) {
2231                         $serverdata['site_name'] = $title;
2232                 }
2233
2234                 $list = $xpath->query('//meta[@name]');
2235
2236                 foreach ($list as $node) {
2237                         $attr = [];
2238                         if ($node->attributes->length) {
2239                                 foreach ($node->attributes as $attribute) {
2240                                         $value = trim($attribute->value);
2241                                         if (empty($value)) {
2242                                                 continue;
2243                                         }
2244
2245                                         $attr[$attribute->name] = $value;
2246                                 }
2247
2248                                 if (empty($attr['name']) || empty($attr['content'])) {
2249                                         continue;
2250                                 }
2251                         }
2252
2253                         if ($attr['name'] == 'description') {
2254                                 $serverdata['info'] = $attr['content'];
2255                         }
2256
2257                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
2258                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad', 'generator'])) {
2259                                 $platform = str_ireplace(array_keys($platforms), array_values($platforms), $attr['content']);
2260                                 $platform = str_replace('/', ' ', $platform);
2261                                 $platform_parts = explode(' ', $platform);
2262                                 if ((count($platform_parts) >= 2) && in_array(strtolower($platform_parts[0]), array_values($platforms))) {
2263                                         $platform = $platform_parts[0];
2264                                         $serverdata['version'] = $platform_parts[1];
2265                                 }
2266                                 if (in_array($platform, array_values($dfrn_platforms))) {
2267                                         $serverdata['network'] = Protocol::DFRN;
2268                                 } elseif (in_array($platform, array_values($ap_platforms))) {
2269                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2270                                 } elseif (in_array($platform, array_values($zap_platforms))) {
2271                                         $serverdata['network'] = Protocol::ZOT;
2272                                 }
2273                                 if (in_array($platform, array_values($platforms))) {
2274                                         $serverdata['platform'] = $platform;
2275                                         $assigned = true;
2276                                 }
2277                         }
2278                 }
2279
2280                 $list = $xpath->query('//meta[@property]');
2281
2282                 foreach ($list as $node) {
2283                         $attr = [];
2284                         if ($node->attributes->length) {
2285                                 foreach ($node->attributes as $attribute) {
2286                                         $value = trim($attribute->value);
2287                                         if (empty($value)) {
2288                                                 continue;
2289                                         }
2290
2291                                         $attr[$attribute->name] = $value;
2292                                 }
2293
2294                                 if (empty($attr['property']) || empty($attr['content'])) {
2295                                         continue;
2296                                 }
2297                         }
2298
2299                         if ($attr['property'] == 'og:site_name') {
2300                                 $serverdata['site_name'] = $attr['content'];
2301                         }
2302
2303                         if ($attr['property'] == 'og:description') {
2304                                 $serverdata['info'] = $attr['content'];
2305                         }
2306
2307                         if (in_array($attr['property'], ['og:platform', 'generator'])) {
2308                                 if (in_array($attr['content'], array_keys($platforms))) {
2309                                         $serverdata['platform'] = $platforms[$attr['content']];
2310                                         $assigned = true;
2311                                 }
2312
2313                                 if (in_array($attr['content'], array_keys($ap_platforms))) {
2314                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2315                                 } elseif (in_array($attr['content'], array_values($zap_platforms))) {
2316                                         $serverdata['network'] = Protocol::ZOT;
2317                                 }
2318                         }
2319                 }
2320
2321                 $list = $xpath->query('//link[@rel="me"]');
2322                 foreach ($list as $node) {
2323                         foreach ($node->attributes as $attribute) {
2324                                 if (parse_url(trim($attribute->value), PHP_URL_HOST) == 'micro.blog') {
2325                                         $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2326                                         $serverdata['platform'] = 'microblog';
2327                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2328                                         $assigned = true;
2329                                 }
2330                         }
2331                 }
2332
2333                 if ($serverdata['platform'] != 'microblog') {
2334                         $list = $xpath->query('//link[@rel="micropub"]');
2335                         foreach ($list as $node) {
2336                                 foreach ($node->attributes as $attribute) {
2337                                         if (trim($attribute->value) == 'https://micro.blog/micropub') {
2338                                                 $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2339                                                 $serverdata['platform'] = 'microblog';
2340                                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
2341                                                 $assigned = true;
2342                                         }
2343                                 }
2344                         }
2345                 }
2346
2347                 if ($assigned && in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER])) {
2348                         $serverdata['detection-method'] = self::DETECT_BODY;
2349                 }
2350
2351                 return $serverdata;
2352         }
2353
2354         /**
2355          * Analyses the header data of a given server for hints about type and system of that server
2356          *
2357          * @param object $curlResult result of curl execution
2358          * @param array  $serverdata array with server data
2359          *
2360          * @return array server data
2361          */
2362         private static function analyseRootHeader($curlResult, array $serverdata): array
2363         {
2364                 if ($curlResult->getHeader('server') == 'Mastodon') {
2365                         $serverdata['platform'] = 'mastodon';
2366                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2367                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
2368                         $serverdata['platform'] = 'diaspora';
2369                         $serverdata['network'] = Protocol::DIASPORA;
2370                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version')[0] ?? '';
2371                 } elseif ($curlResult->inHeader('x-friendica-version')) {
2372                         $serverdata['platform'] = 'friendica';
2373                         $serverdata['network'] = Protocol::DFRN;
2374                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version')[0] ?? '';
2375                 } else {
2376                         return $serverdata;
2377                 }
2378
2379                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
2380                         $serverdata['detection-method'] = self::DETECT_HEADER;
2381                 }
2382
2383                 return $serverdata;
2384         }
2385
2386         /**
2387          * Update GServer entries
2388          */
2389         public static function discover()
2390         {
2391                 if (!DI::config('system', 'discover_servers')) {
2392                         return;
2393                 }
2394
2395                 // Update the server list
2396                 self::discoverFederation();
2397
2398                 $no_of_queries = 5;
2399
2400                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
2401
2402                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
2403
2404                 $gservers = DBA::select('gserver', ['id', 'url', 'nurl', 'network', 'poco', 'directory-type'],
2405                         ["NOT `blocked` AND NOT `failed` AND `directory-type` != ? AND `last_poco_query` < ?", GServer::DT_NONE, $last_update],
2406                         ['order' => ['RAND()']]);
2407
2408                 while ($gserver = DBA::fetch($gservers)) {
2409                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2410                         Worker::add(Worker::PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
2411
2412                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2413                         Worker::add(Worker::PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
2414
2415                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
2416                         self::update($fields, ['nurl' => $gserver['nurl']]);
2417
2418                         if (--$no_of_queries == 0) {
2419                                 break;
2420                         }
2421                 }
2422
2423                 DBA::close($gservers);
2424         }
2425
2426         /**
2427          * Discover federated servers
2428          */
2429         private static function discoverFederation()
2430         {
2431                 $last = DI::keyValue()->get('poco_last_federation_discovery');
2432
2433                 if ($last) {
2434                         $next = $last + (24 * 60 * 60);
2435
2436                         if ($next > time()) {
2437                                 return;
2438                         }
2439                 }
2440
2441                 // Discover federated servers
2442                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
2443                 foreach ($protocols as $protocol) {
2444                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
2445                         $curlResult = DI::httpClient()->fetch('https://the-federation.info/graphql?query=' . urlencode($query), HttpClientAccept::JSON);
2446                         if (!empty($curlResult)) {
2447                                 $data = json_decode($curlResult, true);
2448                                 if (!empty($data['data']['nodes'])) {
2449                                         foreach ($data['data']['nodes'] as $server) {
2450                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
2451                                                 self::add('https://' . $server['host'], true);
2452                                         }
2453                                 }
2454                         }
2455                 }
2456
2457                 // Discover Mastodon servers
2458                 $accesstoken = DI::config()->get('system', 'instances_social_key');
2459
2460                 if (!empty($accesstoken)) {
2461                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
2462                         $curlResult = DI::httpClient()->get($api, HttpClientAccept::JSON, [HttpClientOptions::HEADERS => ['Authorization' => ['Bearer ' . $accesstoken]]]);
2463                         if ($curlResult->isSuccess()) {
2464                                 $servers = json_decode($curlResult->getBody(), true);
2465
2466                                 if (!empty($servers['instances'])) {
2467                                         foreach ($servers['instances'] as $server) {
2468                                                 $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
2469                                                 self::add($url);
2470                                         }
2471                                 }
2472                         }
2473                 }
2474
2475                 DI::keyValue()->set('poco_last_federation_discovery', time());
2476         }
2477
2478         /**
2479          * Set the protocol for the given server
2480          *
2481          * @param int $gsid     Server id
2482          * @param int $protocol Protocol id
2483          *
2484          * @throws Exception
2485          */
2486         public static function setProtocol(int $gsid, int $protocol)
2487         {
2488                 if (empty($gsid)) {
2489                         return;
2490                 }
2491
2492                 $gserver = DBA::selectFirst('gserver', ['protocol', 'url'], ['id' => $gsid]);
2493                 if (!DBA::isResult($gserver)) {
2494                         return;
2495                 }
2496
2497                 $old = $gserver['protocol'];
2498
2499                 if (!is_null($old)) {
2500                         /*
2501                         The priority for the protocols is:
2502                                 1. ActivityPub
2503                                 2. DFRN via Diaspora
2504                                 3. Legacy DFRN
2505                                 4. Diaspora
2506                                 5. OStatus
2507                         */
2508
2509                         // We don't need to change it when nothing is to be changed
2510                         if ($old == $protocol) {
2511                                 return;
2512                         }
2513
2514                         // We don't want to mark a server as OStatus when it had been marked with any other protocol before
2515                         if ($protocol == Post\DeliveryData::OSTATUS) {
2516                                 return;
2517                         }
2518
2519                         // If the server is marked as ActivityPub then we won't change it to anything different
2520                         if ($old == Post\DeliveryData::ACTIVITYPUB) {
2521                                 return;
2522                         }
2523
2524                         // Don't change it to anything lower than DFRN if the new one wasn't ActivityPub
2525                         if (($old == Post\DeliveryData::DFRN) && ($protocol != Post\DeliveryData::ACTIVITYPUB)) {
2526                                 return;
2527                         }
2528
2529                         // Don't change it to Diaspora when it is a legacy DFRN server
2530                         if (($old == Post\DeliveryData::LEGACY_DFRN) && ($protocol == Post\DeliveryData::DIASPORA)) {
2531                                 return;
2532                         }
2533                 }
2534
2535                 Logger::info('Protocol for server', ['protocol' => $protocol, 'old' => $old, 'id' => $gsid, 'url' => $gserver['url']]);
2536                 self::update(['protocol' => $protocol], ['id' => $gsid]);
2537         }
2538
2539         /**
2540          * Fetch the protocol of the given server
2541          *
2542          * @param int $gsid Server id
2543          *
2544          * @return ?int One of Post\DeliveryData protocol constants or null if unknown or gserver is missing
2545          *
2546          * @throws Exception
2547          */
2548         public static function getProtocol(int $gsid): ?int
2549         {
2550                 if (empty($gsid)) {
2551                         return null;
2552                 }
2553
2554                 $gserver = DBA::selectFirst('gserver', ['protocol'], ['id' => $gsid]);
2555                 if (DBA::isResult($gserver)) {
2556                         return $gserver['protocol'];
2557                 }
2558
2559                 return null;
2560         }
2561
2562         /**
2563          * Update rows in the gserver table.
2564          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2565          *
2566          * @param array $fields
2567          * @param array $condition
2568          *
2569          * @return bool
2570          *
2571          * @throws Exception
2572          */
2573         public static function update(array $fields, array $condition): bool
2574         {
2575                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2576
2577                 return DBA::update('gserver', $fields, $condition);
2578         }
2579
2580         /**
2581          * Insert a row into the gserver table.
2582          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2583          *
2584          * @param array $fields
2585          * @param int   $duplicate_mode What to do on a duplicated entry
2586          *
2587          * @return bool
2588          *
2589          * @throws Exception
2590          */
2591         public static function insert(array $fields, int $duplicate_mode = Database::INSERT_DEFAULT): bool
2592         {
2593                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2594
2595                 return DBA::insert('gserver', $fields, $duplicate_mode);
2596         }
2597 }