src/Model/GServer.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2022, 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\System;
  30 use Friendica\Core\Worker;
  31 use Friendica\Database\Database;
  32 use Friendica\Database\DBA;
  33 use Friendica\DI;
  34 use Friendica\Module\Register;
  35 use Friendica\Network\HTTPClient\Client\HttpClientAccept;
  36 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  37 use Friendica\Network\HTTPClient\Capability\ICanHandleHttpResponses;
  38 use Friendica\Network\Probe;
  39 use Friendica\Protocol\ActivityPub;
  40 use Friendica\Protocol\Relay;
  41 use Friendica\Util\DateTimeFormat;
  42 use Friendica\Util\JsonLD;
  43 use Friendica\Util\Network;
  44 use Friendica\Util\Strings;
  45 use Friendica\Util\XML;
  46 use Friendica\Network\HTTPException;
  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_PUMPIO = 19; // Deprecated
  85         const DETECT_SYSTEM_ACTOR = 20; // Mistpark, Osada, Roadhouse, Zap
  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 existance 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, false)) {
 104                         return;
 105                 }
 106
 107                 Worker::add(Worker::PRIORITY_LOW, 'UpdateGServer', $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, 'callstack' => System::callstack(20)]);
 129                         return $gserver['id'];
 130                 }
 131
 132                 if ($no_check || !self::check($url)) {
 133                         return null;
 134                 }
 135
 136                 return self::getID($url, true);
 137         }
 138
 139         /**
 140          * Retrieves all the servers which base domain are matching the provided domain pattern
 141          *
 142          * The pattern is a simple fnmatch() pattern with ? for single wildcard and * for multiple wildcard
 143          *
 144          * @param string $pattern
 145          *
 146          * @return array
 147          *
 148          * @throws Exception
 149          */
 150         public static function listByDomainPattern(string $pattern): array
 151         {
 152                 $likePattern = 'http://' . strtr($pattern, ['_' => '\_', '%' => '\%', '?' => '_', '*' => '%']);
 153
 154                 // The SUBSTRING_INDEX returns everything before the eventual third /, which effectively trims an
 155                 // eventual server path and keep only the server domain which we're matching against the pattern.
 156                 $sql = "SELECT `gserver`.*, COUNT(*) AS `contacts`
 157                         FROM `gserver`
 158                         LEFT JOIN `contact` ON `gserver`.`id` = `contact`.`gsid`
 159                         WHERE SUBSTRING_INDEX(`gserver`.`nurl`, '/', 3) LIKE ?
 160                         AND NOT `gserver`.`failed`
 161                         GROUP BY `gserver`.`id`";
 162
 163                 $stmt = DI::dba()->p($sql, $likePattern);
 164
 165                 return DI::dba()->toArray($stmt);
 166         }
 167
 168         /**
 169          * Checks if the given server is reachable
 170          *
 171          * @param array $contact Contact that should be checked
 172          *
 173          * @return boolean 'true' if server seems vital
 174          */
 175         public static function reachable(array $contact): bool
 176         {
 177                 if (!empty($contact['gsid'])) {
 178                         $gsid = $contact['gsid'];
 179                 } elseif (!empty($contact['baseurl'])) {
 180                         $server = $contact['baseurl'];
 181                 } elseif ($contact['network'] == Protocol::DIASPORA) {
 182                         $parts = parse_url($contact['url']);
 183                         unset($parts['path']);
 184                         $server = (string)Uri::fromParts($parts);
 185                 } else {
 186                         return true;
 187                 }
 188
 189                 if (!empty($gsid)) {
 190                         $condition = ['id' => $gsid];
 191                 } else {
 192                         $condition = ['nurl' => Strings::normaliseLink($server)];
 193                 }
 194
 195                 $gserver = DBA::selectFirst('gserver', ['url', 'next_contact', 'failed'], $condition);
 196                 if (empty($gserver)) {
 197                         $reachable = true;
 198                 } else {
 199                         $reachable = !$gserver['failed'];
 200                         $server    = $gserver['url'];
 201                 }
 202
 203                 if (!empty($server) && (empty($gserver) || strtotime($gserver['next_contact']) < time())) {
 204                         Worker::add(Worker::PRIORITY_LOW, 'UpdateGServer', $server, false);
 205                 }
 206
 207                 return $reachable;
 208         }
 209
 210         /**
 211          * Calculate the next update day
 212          *
 213          * @param bool $success
 214          * @param string $created
 215          * @param string $last_contact
 216          * @param bool $undetected
 217          *
 218          * @return string
 219          *
 220          * @throws Exception
 221          */
 222         public static function getNextUpdateDate(bool $success, string $created = '', string $last_contact = '', bool $undetected = false): string
 223         {
 224                 // On successful contact process check again next week when it is a detected system.
 225                 // When we haven't detected the system, it could be a static website or a really old system.
 226                 if ($success) {
 227                         return DateTimeFormat::utc($undetected ? 'now +1 month' : 'now +7 day');
 228                 }
 229
 230                 $now = strtotime(DateTimeFormat::utcNow());
 231
 232                 if ($created > $last_contact) {
 233                         $contact_time = strtotime($created);
 234                 } else {
 235                         $contact_time = strtotime($last_contact);
 236                 }
 237
 238                 // If the last contact was less than 6 hours before then try again in 6 hours
 239                 if (($now - $contact_time) < (60 * 60 * 6)) {
 240                         return DateTimeFormat::utc('now +6 hour');
 241                 }
 242
 243                 // If the last contact was less than 12 hours before then try again in 12 hours
 244                 if (($now - $contact_time) < (60 * 60 * 12)) {
 245                         return DateTimeFormat::utc('now +12 hour');
 246                 }
 247
 248                 // If the last contact was less than 24 hours before then try tomorrow again
 249                 if (($now - $contact_time) < (60 * 60 * 24)) {
 250                         return DateTimeFormat::utc('now +1 day');
 251                 }
 252
 253                 // If the last contact was less than a week before then try again in a week
 254                 if (($now - $contact_time) < (60 * 60 * 24 * 7)) {
 255                         return DateTimeFormat::utc('now +1 week');
 256                 }
 257
 258                 // If the last contact was less than two weeks before then try again in two week
 259                 if (($now - $contact_time) < (60 * 60 * 24 * 14)) {
 260                         return DateTimeFormat::utc('now +2 week');
 261                 }
 262
 263                 // If the last contact was less than a month before then try again in a month
 264                 if (($now - $contact_time) < (60 * 60 * 24 * 30)) {
 265                         return DateTimeFormat::utc('now +1 month');
 266                 }
 267
 268                 // The system hadn't been successul contacted for more than a month, so try again in three months
 269                 return DateTimeFormat::utc('now +3 month');
 270         }
 271
 272         /**
 273          * Checks the state of the given server.
 274          *
 275          * @param string  $server_url    URL of the given server
 276          * @param string  $network       Network value that is used, when detection failed
 277          * @param boolean $force         Force an update.
 278          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 279          *
 280          * @return boolean 'true' if server seems vital
 281          */
 282         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false): bool
 283         {
 284                 $server_url = self::cleanURL($server_url);
 285                 if ($server_url == '') {
 286                         return false;
 287                 }
 288
 289                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
 290                 if (DBA::isResult($gserver)) {
 291                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
 292                                 $fields = ['created' => DateTimeFormat::utcNow()];
 293                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
 294                                 self::update($fields, $condition);
 295                         }
 296
 297                         if (!$force && (strtotime($gserver['next_contact']) > time())) {
 298                                 Logger::info('No update needed', ['server' => $server_url]);
 299                                 return (!$gserver['failed']);
 300                         }
 301                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force]);
 302                 } else {
 303                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
 304                 }
 305
 306                 return self::detect($server_url, $network, $only_nodeinfo);
 307         }
 308
 309         /**
 310          * Set failed server status
 311          *
 312          * @param string $url
 313          */
 314         public static function setFailure(string $url)
 315         {
 316                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($url)]);
 317                 if (DBA::isResult($gserver)) {
 318                         $next_update = self::getNextUpdateDate(false, $gserver['created'], $gserver['last_contact']);
 319                         self::update(['url' => $url, 'failed' => true, 'last_failure' => DateTimeFormat::utcNow(),
 320                         'next_contact' => $next_update, 'network' => Protocol::PHANTOM, 'detection-method' => null],
 321                         ['nurl' => Strings::normaliseLink($url)]);
 322                         Logger::info('Set failed status for existing server', ['url' => $url]);
 323                         return;
 324                 }
 325                 self::insert(['url' => $url, 'nurl' => Strings::normaliseLink($url),
 326                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
 327                         'failed' => true, 'last_failure' => DateTimeFormat::utcNow()]);
 328                 Logger::info('Set failed status for new server', ['url' => $url]);
 329         }
 330
 331         /**
 332          * Remove unwanted content from the given URL
 333          *
 334          * @param string $dirtyUrl
 335          *
 336          * @return string cleaned URL
 337          * @throws Exception
 338          * @deprecated since 2023.03 Use cleanUri instead
 339          */
 340         public static function cleanURL(string $dirtyUrl): string
 341         {
 342                 try {
 343                         return (string)self::cleanUri(new Uri($dirtyUrl));
 344                 } catch (\Throwable $e) {
 345                         Logger::warning('Invalid URL', ['dirtyUrl' => $dirtyUrl]);
 346                         return '';
 347                 }
 348         }
 349
 350         /**
 351          * Remove unwanted content from the given URI
 352          *
 353          * @param UriInterface $dirtyUri
 354          *
 355          * @return UriInterface cleaned URI
 356          * @throws Exception
 357          */
 358         public static function cleanUri(UriInterface $dirtyUri): string
 359         {
 360                 return $dirtyUri
 361                         ->withUserInfo('')
 362                         ->withQuery('')
 363                         ->withFragment('')
 364                         ->withPath(
 365                                 preg_replace(
 366                                         '#(?:^|/)index\.php#',
 367                                         '',
 368                                         rtrim($dirtyUri->getPath(), '/')
 369                                 )
 370                         );
 371         }
 372
 373         /**
 374          * Detect server data (type, protocol, version number, ...)
 375          * The detected data is then updated or inserted in the gserver table.
 376          *
 377          * @param string  $url           URL of the given server
 378          * @param string  $network       Network value that is used, when detection failed
 379          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 380          *
 381          * @return boolean 'true' if server could be detected
 382          */
 383         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false): bool
 384         {
 385                 Logger::info('Detect server type', ['server' => $url]);
 386
 387                 $original_url = $url;
 388
 389                 // Remove URL content that is not supposed to exist for a server url
 390                 $url = rtrim(self::cleanURL($url), '/');
 391                 if (empty($url)) {
 392                         Logger::notice('Empty URL.');
 393                         return false;
 394                 }
 395
 396                 // If the URL missmatches, then we mark the old entry as failure
 397                 if (!Strings::compareLink($url, $original_url)) {
 398                         self::setFailure($original_url);
 399                         if (!self::getID($url, true)) {
 400                                 self::detect($url, $network, $only_nodeinfo);
 401                         }
 402                         return false;
 403                 }
 404
 405                 $valid_url = Network::isUrlValid($url);
 406                 if (!$valid_url) {
 407                         self::setFailure($url);
 408                         return false;
 409                 } else {
 410                         $valid_url = rtrim($valid_url, '/');
 411                 }
 412
 413                 if (!Strings::compareLink($url, $valid_url)) {
 414                         // We only follow redirects when the path stays the same or the target url has no path.
 415                         // Some systems have got redirects on their landing page to a single account page. This check handles it.
 416                         if (((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH))) ||
 417                                 (((parse_url($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)))) {
 418                                 Logger::debug('Found redirect. Mark old entry as failure', ['old' => $url, 'new' => $valid_url]);
 419                                 self::setFailure($url);
 420                                 if (!self::getID($valid_url, true)) {
 421                                         self::detect($valid_url, $network, $only_nodeinfo);
 422                                 }
 423                                 return false;
 424                         }
 425
 426                         if ((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) != parse_url($valid_url, PHP_URL_PATH)) &&
 427                                 (parse_url($url, PHP_URL_PATH) == '')) {
 428                                 Logger::debug('Found redirect. Mark old entry as failure and redirect to the basepath.', ['old' => $url, 'new' => $valid_url]);
 429                                 $parts = parse_url($valid_url);
 430                                 unset($parts['path']);
 431                                 $valid_url = (string)Uri::fromParts($parts);
 432
 433                                 self::setFailure($url);
 434                                 if (!self::getID($valid_url, true)) {
 435                                         self::detect($valid_url, $network, $only_nodeinfo);
 436                                 }
 437                                 return false;
 438                         }
 439                         Logger::debug('Found redirect, but ignore it.', ['old' => $url, 'new' => $valid_url]);
 440                 }
 441
 442                 if ((parse_url($url, PHP_URL_HOST) == parse_url($valid_url, PHP_URL_HOST)) &&
 443                         (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH)) &&
 444                         (parse_url($url, PHP_URL_SCHEME) != parse_url($valid_url, PHP_URL_SCHEME))) {
 445                         $url = $valid_url;
 446                 }
 447
 448                 $in_webroot = empty(parse_url($url, PHP_URL_PATH));
 449
 450                 // When a nodeinfo is present, we don't need to dig further
 451                 $curlResult = DI::httpClient()->get($url . '/.well-known/x-nodeinfo2', HttpClientAccept::JSON);
 452                 if ($curlResult->isTimeout()) {
 453                         self::setFailure($url);
 454                         return false;
 455                 }
 456
 457                 $serverdata = self::parseNodeinfo210($curlResult);
 458                 if (empty($serverdata)) {
 459                         $curlResult = DI::httpClient()->get($url . '/.well-known/nodeinfo', HttpClientAccept::JSON);
 460                         $serverdata = self::fetchNodeinfo($url, $curlResult);
 461                 }
 462
 463                 if ($only_nodeinfo && empty($serverdata)) {
 464                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 465                         self::setFailure($url);
 466                         return false;
 467                 } elseif (empty($serverdata)) {
 468                         $serverdata = ['detection-method' => self::DETECT_MANUAL, 'network' => Protocol::PHANTOM, 'platform' => '', 'version' => '', 'site_name' => '', 'info' => ''];
 469                 }
 470
 471                 // When there is no Nodeinfo, then use some protocol specific endpoints
 472                 if ($serverdata['network'] == Protocol::PHANTOM) {
 473                         if ($in_webroot) {
 474                                 // Fetch the landing page, possibly it reveals some data
 475                                 $accept = 'application/activity+json,application/ld+json,application/json,*/*;q=0.9';
 476                                 $curlResult = DI::httpClient()->get($url, $accept);
 477                                 if (!$curlResult->isSuccess() && $curlResult->getReturnCode() == '406') {
 478                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 479                                         $html_fetched = true;
 480                                 } else {
 481                                         $html_fetched = false;
 482                                 }
 483
 484                                 if ($curlResult->isSuccess()) {
 485                                         $json = json_decode($curlResult->getBody(), true);
 486                                         if (!empty($json) && is_array($json)) {
 487                                                 $data = self::fetchDataFromSystemActor($json, $serverdata);
 488                                                 $serverdata = $data['server'];
 489                                                 $systemactor = $data['actor'];
 490                                                 if (!$html_fetched && !in_array($serverdata['detection-method'], [self::DETECT_SYSTEM_ACTOR, self::DETECT_AP_COLLECTION])) {
 491                                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 492                                                 }
 493                                         } elseif (!$html_fetched && (strlen($curlResult->getBody()) < 1000)) {
 494                                                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 495                                         }
 496
 497                                         if ($serverdata['detection-method'] != self::DETECT_SYSTEM_ACTOR) {
 498                                                 $serverdata = self::analyseRootHeader($curlResult, $serverdata);
 499                                                 $serverdata = self::analyseRootBody($curlResult, $serverdata);
 500                                         }
 501                                 }
 502
 503                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
 504                                         self::setFailure($url);
 505                                         return false;
 506                                 }
 507
 508                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 509                                         $serverdata = self::detectMastodonAlikes($url, $serverdata);
 510                                 }
 511                         }
 512
 513                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 514                         // With this check we don't have to waste time and ressources for dead systems.
 515                         // Also this hopefully prevents us from receiving abuse messages.
 516                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 517                                 $validHostMeta = self::validHostMeta($url);
 518                         } else {
 519                                 $validHostMeta = false;
 520                         }
 521
 522                         if ($validHostMeta) {
 523                                 if (in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY])) {
 524                                         $serverdata['detection-method'] = self::DETECT_HOST_META;
 525                                 }
 526
 527                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 528                                         $serverdata = self::detectFriendica($url, $serverdata);
 529                                 }
 530
 531                                 // The following systems have to be installed in the root directory.
 532                                 if ($in_webroot) {
 533                                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 534                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 535                                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 536                                         }
 537
 538                                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations, so we check other endpoints as well
 539                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 540                                                 $serverdata = self::detectHubzilla($url, $serverdata);
 541                                         }
 542
 543                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 544                                                 $serverdata = self::detectPeertube($url, $serverdata);
 545                                         }
 546
 547                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 548                                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 549                                         }
 550                                 }
 551                         } elseif (in_array($serverdata['platform'], ['friendica', 'friendika']) && in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_SYSTEM_ACTOR]))) {
 552                                 $serverdata = self::detectFriendica($url, $serverdata);
 553                         }
 554
 555                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 556                                 $serverdata = self::detectNextcloud($url, $serverdata, $validHostMeta);
 557                         }
 558
 559                         // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 560                         // Since this endpoint is only rarely used, we query it at a later time
 561                         if (in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_FRIENDICA, self::DETECT_CONFIG_JSON]))) {
 562                                 $serverdata = self::fetchStatistics($url, $serverdata);
 563                         }
 564                 }
 565
 566                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 567                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 568                         $serverdata['network'] = $network;
 569                 }
 570
 571                 // Most servers aren't installed in a subdirectory, so we declare this entry as failed
 572                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty(parse_url($url, PHP_URL_PATH)) && in_array($serverdata['detection-method'], [self::DETECT_MANUAL])) {
 573                         self::setFailure($url);
 574                         return false;
 575                 }
 576
 577                 $serverdata['url'] = $url;
 578                 $serverdata['nurl'] = Strings::normaliseLink($url);
 579
 580                 // We have to prevent an endless loop here.
 581                 // When a server is new, then there is no gserver entry yet.
 582                 // But in "detectNetworkViaContacts" it could happen that a contact is updated,
 583                 // and this can call this function here as well.
 584                 if (self::getID($url, true) && (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) ||
 585                         in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META]))) {
 586                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 587                 }
 588
 589                 // Detect the directory type
 590                 $serverdata['directory-type'] = self::DT_NONE;
 591
 592                 if (in_array($serverdata['network'], Protocol::FEDERATED)) {
 593                         $serverdata = self::checkMastodonDirectory($url, $serverdata);
 594
 595                         if ($serverdata['directory-type'] == self::DT_NONE) {
 596                                 $serverdata = self::checkPoCo($url, $serverdata);
 597                         }
 598                 }
 599
 600                 if ($serverdata['network'] == Protocol::ACTIVITYPUB) {
 601                         $serverdata = self::fetchWeeklyUsage($url, $serverdata);
 602                 }
 603
 604                 $serverdata['registered-users'] = $serverdata['registered-users'] ?? 0;
 605
 606                 // Numbers above a reasonable value (10 millions) are ignored
 607                 if ($serverdata['registered-users'] > 10000000) {
 608                         $serverdata['registered-users'] = 0;
 609                 }
 610
 611                 // On an active server there has to be at least a single user
 612                 if (!in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) && ($serverdata['registered-users'] <= 0)) {
 613                         $serverdata['registered-users'] = 1;
 614                 } elseif (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 615                         $serverdata['registered-users'] = 0;
 616                 }
 617
 618                 $serverdata['next_contact'] = self::getNextUpdateDate(true, '', '', in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]));
 619
 620                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 621                 $serverdata['failed'] = false;
 622
 623                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 624                 if (!DBA::isResult($gserver)) {
 625                         $serverdata['created'] = DateTimeFormat::utcNow();
 626                         $ret = self::insert($serverdata);
 627                         $id = DBA::lastInsertId();
 628                 } else {
 629                         $ret = self::update($serverdata, ['nurl' => $serverdata['nurl']]);
 630                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 631                         if (DBA::isResult($gserver)) {
 632                                 $id = $gserver['id'];
 633                         }
 634                 }
 635
 636                 // Count the number of known contacts from this server
 637                 if (!empty($id) && !in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 638                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 639                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id, 'failed' => false]);
 640                         $max_users = max($apcontacts, $contacts);
 641                         if ($max_users > $serverdata['registered-users']) {
 642                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 643                                 self::update(['registered-users' => $max_users], ['id' => $id]);
 644                         }
 645
 646                         if (empty($serverdata['active-month-users'])) {
 647                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 30 days')]);
 648                                 if ($contacts > 0) {
 649                                         Logger::info('Update monthly users', ['id' => $id, 'url' => $serverdata['nurl'], 'monthly-users' => $contacts]);
 650                                         self::update(['active-month-users' => $contacts], ['id' => $id]);
 651                                 }
 652                         }
 653
 654                         if (empty($serverdata['active-halfyear-users'])) {
 655                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 180 days')]);
 656                                 if ($contacts > 0) {
 657                                         Logger::info('Update halfyear users', ['id' => $id, 'url' => $serverdata['nurl'], 'halfyear-users' => $contacts]);
 658                                         self::update(['active-halfyear-users' => $contacts], ['id' => $id]);
 659                                 }
 660                         }
 661                 }
 662
 663                 if (in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 664                         self::discoverRelay($url);
 665                 }
 666
 667                 if (!empty($systemactor)) {
 668                         $contact = Contact::getByURL($systemactor, true, ['gsid', 'baseurl', 'id', 'network', 'url', 'name']);
 669                         Logger::debug('Fetched system actor',  ['url' => $url, 'gsid' => $id, 'contact' => $contact]);
 670                 }
 671
 672                 return $ret;
 673         }
 674
 675         /**
 676          * Fetch relay data from a given server url
 677          *
 678          * @param string $server_url address of the server
 679          *
 680          * @return void
 681          *
 682          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 683          */
 684         private static function discoverRelay(string $server_url)
 685         {
 686                 Logger::info('Discover relay data', ['server' => $server_url]);
 687
 688                 $curlResult = DI::httpClient()->get($server_url . '/.well-known/x-social-relay', HttpClientAccept::JSON);
 689                 if (!$curlResult->isSuccess()) {
 690                         return;
 691                 }
 692
 693                 $data = json_decode($curlResult->getBody(), true);
 694                 if (!is_array($data)) {
 695                         return;
 696                 }
 697
 698                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 699                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
 700
 701                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 702                         $data['scope'] = '';
 703                         $data['subscribe'] = false;
 704                         $data['tags'] = [];
 705                 }
 706
 707                 $gserver = DBA::selectFirst('gserver', ['id', 'url', 'network', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 708                 if (!DBA::isResult($gserver)) {
 709                         return;
 710                 }
 711
 712                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 713                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 714                         self::update($fields, ['id' => $gserver['id']]);
 715                 }
 716
 717                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 718
 719                 if ($data['scope'] == 'tags') {
 720                         // Avoid duplicates
 721                         $tags = [];
 722                         foreach ($data['tags'] as $tag) {
 723                                 $tag = mb_strtolower($tag);
 724                                 if (strlen($tag) < 100) {
 725                                         $tags[$tag] = $tag;
 726                                 }
 727                         }
 728
 729                         foreach ($tags as $tag) {
 730                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], Database::INSERT_IGNORE);
 731                         }
 732                 }
 733
 734                 // Create or update the relay contact
 735                 $fields = [];
 736                 if (isset($data['protocols'])) {
 737                         if (isset($data['protocols']['diaspora'])) {
 738                                 $fields['network'] = Protocol::DIASPORA;
 739
 740                                 if (isset($data['protocols']['diaspora']['receive'])) {
 741                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 742                                 } elseif (is_string($data['protocols']['diaspora'])) {
 743                                         $fields['batch'] = $data['protocols']['diaspora'];
 744                                 }
 745                         }
 746
 747                         if (isset($data['protocols']['dfrn'])) {
 748                                 $fields['network'] = Protocol::DFRN;
 749
 750                                 if (isset($data['protocols']['dfrn']['receive'])) {
 751                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 752                                 } elseif (is_string($data['protocols']['dfrn'])) {
 753                                         $fields['batch'] = $data['protocols']['dfrn'];
 754                                 }
 755                         }
 756
 757                         if (isset($data['protocols']['activitypub'])) {
 758                                 $fields['network'] = Protocol::ACTIVITYPUB;
 759
 760                                 if (!empty($data['protocols']['activitypub']['actor'])) {
 761                                         $fields['url'] = $data['protocols']['activitypub']['actor'];
 762                                 }
 763                                 if (!empty($data['protocols']['activitypub']['receive'])) {
 764                                         $fields['batch'] = $data['protocols']['activitypub']['receive'];
 765                                 }
 766                         }
 767                 }
 768
 769                 Logger::info('Discovery ended', ['server' => $server_url, 'data' => $fields]);
 770
 771                 Relay::updateContact($gserver, $fields);
 772         }
 773
 774         /**
 775          * Fetch server data from '/statistics.json' on the given server
 776          *
 777          * @param string $url URL of the given server
 778          *
 779          * @return array server data
 780          */
 781         private static function fetchStatistics(string $url, array $serverdata): array
 782         {
 783                 $curlResult = DI::httpClient()->get($url . '/statistics.json', HttpClientAccept::JSON);
 784                 if (!$curlResult->isSuccess()) {
 785                         return $serverdata;
 786                 }
 787
 788                 $data = json_decode($curlResult->getBody(), true);
 789                 if (empty($data)) {
 790                         return $serverdata;
 791                 }
 792
 793                 // Some AP enabled systems return activity data that we don't expect here.
 794                 if (strpos($curlResult->getContentType(), 'application/activity+json') !== false) {
 795                         return $serverdata;
 796                 }
 797
 798                 $valid = false;
 799                 $old_serverdata = $serverdata;
 800
 801                 $serverdata['detection-method'] = self::DETECT_STATISTICS_JSON;
 802
 803                 if (!empty($data['version'])) {
 804                         $valid = true;
 805                         $serverdata['version'] = $data['version'];
 806                         // Version numbers on statistics.json are presented with additional info, e.g.:
 807                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 808                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 809                 }
 810
 811                 if (!empty($data['name'])) {
 812                         $valid = true;
 813                         $serverdata['site_name'] = $data['name'];
 814                 }
 815
 816                 if (!empty($data['network'])) {
 817                         $valid = true;
 818                         $serverdata['platform'] = strtolower($data['network']);
 819
 820                         if ($serverdata['platform'] == 'diaspora') {
 821                                 $serverdata['network'] = Protocol::DIASPORA;
 822                         } elseif ($serverdata['platform'] == 'friendica') {
 823                                 $serverdata['network'] = Protocol::DFRN;
 824                         } elseif ($serverdata['platform'] == 'hubzilla') {
 825                                 $serverdata['network'] = Protocol::ZOT;
 826                         } elseif ($serverdata['platform'] == 'redmatrix') {
 827                                 $serverdata['network'] = Protocol::ZOT;
 828                         }
 829                 }
 830
 831                 if (!empty($data['total_users'])) {
 832                         $valid = true;
 833                         $serverdata['registered-users'] = max($data['total_users'], 1);
 834                 }
 835
 836                 if (!empty($data['active_users_monthly'])) {
 837                         $valid = true;
 838                         $serverdata['active-month-users'] = max($data['active_users_monthly'], 0);
 839                 }
 840
 841                 if (!empty($data['active_users_halfyear'])) {
 842                         $valid = true;
 843                         $serverdata['active-halfyear-users'] = max($data['active_users_halfyear'], 0);
 844                 }
 845
 846                 if (!empty($data['local_posts'])) {
 847                         $valid = true;
 848                         $serverdata['local-posts'] = max($data['local_posts'], 0);
 849                 }
 850
 851                 if (!empty($data['registrations_open'])) {
 852                         $serverdata['register_policy'] = Register::OPEN;
 853                 } else {
 854                         $serverdata['register_policy'] = Register::CLOSED;
 855                 }
 856
 857                 if (!$valid) {
 858                         return $old_serverdata;
 859                 }
 860
 861                 return $serverdata;
 862         }
 863
 864         /**
 865          * Detect server type by using the nodeinfo data
 866          *
 867          * @param string                  $url        address of the server
 868          * @param ICanHandleHttpResponses $httpResult
 869          *
 870          * @return array Server data
 871          *
 872          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 873          */
 874         private static function fetchNodeinfo(string $url, ICanHandleHttpResponses $httpResult): array
 875         {
 876                 if (!$httpResult->isSuccess()) {
 877                         return [];
 878                 }
 879
 880                 $nodeinfo = json_decode($httpResult->getBody(), true);
 881
 882                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
 883                         return [];
 884                 }
 885
 886                 $nodeinfo1_url = '';
 887                 $nodeinfo2_url = '';
 888
 889                 foreach ($nodeinfo['links'] as $link) {
 890                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
 891                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
 892                                 continue;
 893                         }
 894                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 895                                 $nodeinfo1_url = $link['href'];
 896                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 897                                 $nodeinfo2_url = $link['href'];
 898                         }
 899                 }
 900
 901                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 902                         return [];
 903                 }
 904
 905                 $server = [];
 906
 907                 if (!empty($nodeinfo2_url)) {
 908                         $server = self::parseNodeinfo2($nodeinfo2_url);
 909                 }
 910
 911                 if (empty($server) && !empty($nodeinfo1_url)) {
 912                         $server = self::parseNodeinfo1($nodeinfo1_url);
 913                 }
 914
 915                 return $server;
 916         }
 917
 918         /**
 919          * Parses Nodeinfo 1
 920          *
 921          * @param string $nodeinfo_url address of the nodeinfo path
 922          *
 923          * @return array Server data
 924          *
 925          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 926          */
 927         private static function parseNodeinfo1(string $nodeinfo_url): array
 928         {
 929                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 930                 if (!$curlResult->isSuccess()) {
 931                         return [];
 932                 }
 933
 934                 $nodeinfo = json_decode($curlResult->getBody(), true);
 935
 936                 if (!is_array($nodeinfo)) {
 937                         return [];
 938                 }
 939
 940                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
 941                         'register_policy' => Register::CLOSED];
 942
 943                 if (!empty($nodeinfo['openRegistrations'])) {
 944                         $server['register_policy'] = Register::OPEN;
 945                 }
 946
 947                 if (is_array($nodeinfo['software'])) {
 948                         if (!empty($nodeinfo['software']['name'])) {
 949                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 950                         }
 951
 952                         if (!empty($nodeinfo['software']['version'])) {
 953                                 $server['version'] = $nodeinfo['software']['version'];
 954                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 955                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 956                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 957                         }
 958                 }
 959
 960                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 961                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 962                 }
 963
 964                 if (!empty($nodeinfo['usage']['users']['total'])) {
 965                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
 966                 }
 967
 968                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
 969                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
 970                 }
 971
 972                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
 973                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
 974                 }
 975
 976                 if (!empty($nodeinfo['usage']['localPosts'])) {
 977                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
 978                 }
 979
 980                 if (!empty($nodeinfo['usage']['localComments'])) {
 981                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
 982                 }
 983
 984                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
 985                         $protocols = [];
 986                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
 987                                 $protocols[$protocol] = true;
 988                         }
 989
 990                         if (!empty($protocols['friendica'])) {
 991                                 $server['network'] = Protocol::DFRN;
 992                         } elseif (!empty($protocols['activitypub'])) {
 993                                 $server['network'] = Protocol::ACTIVITYPUB;
 994                         } elseif (!empty($protocols['diaspora'])) {
 995                                 $server['network'] = Protocol::DIASPORA;
 996                         } elseif (!empty($protocols['ostatus'])) {
 997                                 $server['network'] = Protocol::OSTATUS;
 998                         } elseif (!empty($protocols['gnusocial'])) {
 999                                 $server['network'] = Protocol::OSTATUS;
1000                         } elseif (!empty($protocols['zot'])) {
1001                                 $server['network'] = Protocol::ZOT;
1002                         }
1003                 }
1004
1005                 if (empty($server)) {
1006                         return [];
1007                 }
1008
1009                 if (empty($server['network'])) {
1010                         $server['network'] = Protocol::PHANTOM;
1011                 }
1012
1013                 return $server;
1014         }
1015
1016         /**
1017          * Parses Nodeinfo 2
1018          *
1019          * @see https://git.feneas.org/jaywink/nodeinfo2
1020          *
1021          * @param string $nodeinfo_url address of the nodeinfo path
1022          *
1023          * @return array Server data
1024          *
1025          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1026          */
1027         private static function parseNodeinfo2(string $nodeinfo_url): array
1028         {
1029                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
1030                 if (!$curlResult->isSuccess()) {
1031                         return [];
1032                 }
1033
1034                 $nodeinfo = json_decode($curlResult->getBody(), true);
1035                 if (!is_array($nodeinfo)) {
1036                         return [];
1037                 }
1038
1039                 $server = [
1040                         'detection-method' => self::DETECT_NODEINFO_2,
1041                         'register_policy' => Register::CLOSED,
1042                         'platform' => 'unknown',
1043                 ];
1044
1045                 if (!empty($nodeinfo['openRegistrations'])) {
1046                         $server['register_policy'] = Register::OPEN;
1047                 }
1048
1049                 if (!empty($nodeinfo['software'])) {
1050                         if (isset($nodeinfo['software']['name'])) {
1051                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
1052                         }
1053
1054                         if (!empty($nodeinfo['software']['version']) && isset($server['platform'])) {
1055                                 $server['version'] = $nodeinfo['software']['version'];
1056                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1057                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1058                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1059
1060                                 // qoto advertises itself as Mastodon
1061                                 if (($server['platform'] == 'mastodon') && substr($nodeinfo['software']['version'], -5) == '-qoto') {
1062                                         $server['platform'] = 'qoto';
1063                                 }
1064                         }
1065                 }
1066
1067                 if (!empty($nodeinfo['metadata']['nodeName'])) {
1068                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
1069                 }
1070
1071                 if (!empty($nodeinfo['usage']['users']['total'])) {
1072                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1073                 }
1074
1075                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1076                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1077                 }
1078
1079                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1080                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1081                 }
1082
1083                 if (!empty($nodeinfo['usage']['localPosts'])) {
1084                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1085                 }
1086
1087                 if (!empty($nodeinfo['usage']['localComments'])) {
1088                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1089                 }
1090
1091                 if (!empty($nodeinfo['protocols'])) {
1092                         $protocols = [];
1093                         foreach ($nodeinfo['protocols'] as $protocol) {
1094                                 if (is_string($protocol)) {
1095                                         $protocols[$protocol] = true;
1096                                 }
1097                         }
1098
1099                         if (!empty($protocols['dfrn'])) {
1100                                 $server['network'] = Protocol::DFRN;
1101                         } elseif (!empty($protocols['activitypub'])) {
1102                                 $server['network'] = Protocol::ACTIVITYPUB;
1103                         } elseif (!empty($protocols['diaspora'])) {
1104                                 $server['network'] = Protocol::DIASPORA;
1105                         } elseif (!empty($protocols['ostatus'])) {
1106                                 $server['network'] = Protocol::OSTATUS;
1107                         } elseif (!empty($protocols['gnusocial'])) {
1108                                 $server['network'] = Protocol::OSTATUS;
1109                         } elseif (!empty($protocols['zot'])) {
1110                                 $server['network'] = Protocol::ZOT;
1111                         }
1112                 }
1113
1114                 if (empty($server)) {
1115                         return [];
1116                 }
1117
1118                 if (empty($server['network'])) {
1119                         $server['network'] = Protocol::PHANTOM;
1120                 }
1121
1122                 return $server;
1123         }
1124
1125         /**
1126          * Parses NodeInfo2 protocol 1.0
1127          *
1128          * @see https://github.com/jaywink/nodeinfo2/blob/master/PROTOCOL.md
1129          *
1130          * @param string $nodeinfo_url address of the nodeinfo path
1131          *
1132          * @return array Server data
1133          *
1134          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1135          */
1136         private static function parseNodeinfo210(ICanHandleHttpResponses $httpResult): array
1137         {
1138                 if (!$httpResult->isSuccess()) {
1139                         return [];
1140                 }
1141
1142                 $nodeinfo = json_decode($httpResult->getBody(), true);
1143
1144                 if (!is_array($nodeinfo)) {
1145                         return [];
1146                 }
1147
1148                 $server = ['detection-method' => self::DETECT_NODEINFO_210,
1149                         'register_policy' => Register::CLOSED];
1150
1151                 if (!empty($nodeinfo['openRegistrations'])) {
1152                         $server['register_policy'] = Register::OPEN;
1153                 }
1154
1155                 if (!empty($nodeinfo['server'])) {
1156                         if (!empty($nodeinfo['server']['software'])) {
1157                                 $server['platform'] = strtolower($nodeinfo['server']['software']);
1158                         }
1159
1160                         if (!empty($nodeinfo['server']['version'])) {
1161                                 $server['version'] = $nodeinfo['server']['version'];
1162                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1163                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1164                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1165                         }
1166
1167                         if (!empty($nodeinfo['server']['name'])) {
1168                                 $server['site_name'] = $nodeinfo['server']['name'];
1169                         }
1170                 }
1171
1172                 if (!empty($nodeinfo['usage']['users']['total'])) {
1173                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1174                 }
1175
1176                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1177                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1178                 }
1179
1180                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1181                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1182                 }
1183
1184                 if (!empty($nodeinfo['usage']['localPosts'])) {
1185                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1186                 }
1187
1188                 if (!empty($nodeinfo['usage']['localComments'])) {
1189                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1190                 }
1191
1192                 if (!empty($nodeinfo['protocols'])) {
1193                         $protocols = [];
1194                         foreach ($nodeinfo['protocols'] as $protocol) {
1195                                 if (is_string($protocol)) {
1196                                         $protocols[$protocol] = true;
1197                                 }
1198                         }
1199
1200                         if (!empty($protocols['dfrn'])) {
1201                                 $server['network'] = Protocol::DFRN;
1202                         } elseif (!empty($protocols['activitypub'])) {
1203                                 $server['network'] = Protocol::ACTIVITYPUB;
1204                         } elseif (!empty($protocols['diaspora'])) {
1205                                 $server['network'] = Protocol::DIASPORA;
1206                         } elseif (!empty($protocols['ostatus'])) {
1207                                 $server['network'] = Protocol::OSTATUS;
1208                         } elseif (!empty($protocols['gnusocial'])) {
1209                                 $server['network'] = Protocol::OSTATUS;
1210                         } elseif (!empty($protocols['zot'])) {
1211                                 $server['network'] = Protocol::ZOT;
1212                         }
1213                 }
1214
1215                 if (empty($server) || empty($server['platform'])) {
1216                         return [];
1217                 }
1218
1219                 if (empty($server['network'])) {
1220                         $server['network'] = Protocol::PHANTOM;
1221                 }
1222
1223                 return $server;
1224         }
1225
1226         /**
1227          * Fetch server information from a 'siteinfo.json' file on the given server
1228          *
1229          * @param string $url        URL of the given server
1230          * @param array  $serverdata array with server data
1231          *
1232          * @return array server data
1233          */
1234         private static function fetchSiteinfo(string $url, array $serverdata): array
1235         {
1236                 $curlResult = DI::httpClient()->get($url . '/siteinfo.json', HttpClientAccept::JSON);
1237                 if (!$curlResult->isSuccess()) {
1238                         return $serverdata;
1239                 }
1240
1241                 $data = json_decode($curlResult->getBody(), true);
1242                 if (empty($data)) {
1243                         return $serverdata;
1244                 }
1245
1246                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1247                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
1248                 }
1249
1250                 if (!empty($data['url'])) {
1251                         $serverdata['platform'] = strtolower($data['platform']);
1252                         $serverdata['version'] = $data['version'] ?? 'N/A';
1253                 }
1254
1255                 if (!empty($data['plugins'])) {
1256                         if (in_array('pubcrawl', $data['plugins'])) {
1257                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1258                         } elseif (in_array('diaspora', $data['plugins'])) {
1259                                 $serverdata['network'] = Protocol::DIASPORA;
1260                         } elseif (in_array('gnusoc', $data['plugins'])) {
1261                                 $serverdata['network'] = Protocol::OSTATUS;
1262                         } else {
1263                                 $serverdata['network'] = Protocol::ZOT;
1264                         }
1265                 }
1266
1267                 if (!empty($data['site_name'])) {
1268                         $serverdata['site_name'] = $data['site_name'];
1269                 }
1270
1271                 if (!empty($data['channels_total'])) {
1272                         $serverdata['registered-users'] = max($data['channels_total'], 1);
1273                 }
1274
1275                 if (!empty($data['channels_active_monthly'])) {
1276                         $serverdata['active-month-users'] = max($data['channels_active_monthly'], 0);
1277                 }
1278
1279                 if (!empty($data['channels_active_halfyear'])) {
1280                         $serverdata['active-halfyear-users'] = max($data['channels_active_halfyear'], 0);
1281                 }
1282
1283                 if (!empty($data['local_posts'])) {
1284                         $serverdata['local-posts'] = max($data['local_posts'], 0);
1285                 }
1286
1287                 if (!empty($data['local_comments'])) {
1288                         $serverdata['local-comments'] = max($data['local_comments'], 0);
1289                 }
1290
1291                 if (!empty($data['register_policy'])) {
1292                         switch ($data['register_policy']) {
1293                                 case 'REGISTER_OPEN':
1294                                         $serverdata['register_policy'] = Register::OPEN;
1295                                         break;
1296
1297                                 case 'REGISTER_APPROVE':
1298                                         $serverdata['register_policy'] = Register::APPROVE;
1299                                         break;
1300
1301                                 case 'REGISTER_CLOSED':
1302                                 default:
1303                                         $serverdata['register_policy'] = Register::CLOSED;
1304                                         break;
1305                         }
1306                 }
1307
1308                 return $serverdata;
1309         }
1310
1311         /**
1312          * Fetches server data via an ActivityPub account with url of that server
1313          *
1314          * @param string $url        URL of the given server
1315          * @param array  $serverdata array with server data
1316          *
1317          * @return array server data
1318          *
1319          * @throws Exception
1320          */
1321         private static function fetchDataFromSystemActor(array $data, array $serverdata): array
1322         {
1323                 if (empty($data)) {
1324                         return ['server' => $serverdata, 'actor' => ''];
1325                 }
1326
1327                 $actor = JsonLD::compact($data, false);
1328                 if (in_array(JsonLD::fetchElement($actor, '@type'), ActivityPub\Receiver::ACCOUNT_TYPES)) {
1329                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1330                         $serverdata['site_name'] = JsonLD::fetchElement($actor, 'as:name', '@value');
1331                         $serverdata['info'] = JsonLD::fetchElement($actor, 'as:summary', '@value');
1332                         if (!empty($actor['as:generator'])) {
1333                                 $generator = explode(' ', JsonLD::fetchElement($actor['as:generator'], 'as:name', '@value'));
1334                                 $serverdata['platform'] = strtolower(array_shift($generator));
1335                                 $serverdata['detection-method'] = self::DETECT_SYSTEM_ACTOR;
1336                         } else {
1337                                 $serverdata['detection-method'] = self::DETECT_AP_ACTOR;
1338                         }
1339                         return ['server' => $serverdata, 'actor' => $actor['@id']];
1340                 } elseif ((JsonLD::fetchElement($actor, '@type') == 'as:Collection')) {
1341                         // By now only Ktistec seems to provide collections this way
1342                         $serverdata['platform'] = 'ktistec';
1343                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1344                         $serverdata['detection-method'] = self::DETECT_AP_COLLECTION;
1345
1346                         $actors = JsonLD::fetchElementArray($actor, 'as:items');
1347                         if (!empty($actors) && !empty($actors[0]['@id'])) {
1348                                 $actor_url = $actor['@id'] . $actors[0]['@id'];
1349                         } else {
1350                                 $actor_url = '';
1351                         }
1352
1353                         return ['server' => $serverdata, 'actor' => $actor_url];
1354                 }
1355                 return ['server' => $serverdata, 'actor' => ''];
1356         }
1357
1358         /**
1359          * Checks if the server contains a valid host meta file
1360          *
1361          * @param string $url URL of the given server
1362          *
1363          * @return boolean 'true' if the server seems to be vital
1364          */
1365         private static function validHostMeta(string $url): bool
1366         {
1367                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
1368                 $curlResult = DI::httpClient()->get($url . Probe::HOST_META, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
1369                 if (!$curlResult->isSuccess()) {
1370                         return false;
1371                 }
1372
1373                 $xrd = XML::parseString($curlResult->getBody(), true);
1374                 if (!is_object($xrd)) {
1375                         return false;
1376                 }
1377
1378                 $elements = XML::elementToArray($xrd);
1379                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
1380                         return false;
1381                 }
1382
1383                 $valid = false;
1384                 foreach ($elements['xrd']['link'] as $link) {
1385                         // When there is more than a single "link" element, the array looks slightly different
1386                         if (!empty($link['@attributes'])) {
1387                                 $link = $link['@attributes'];
1388                         }
1389
1390                         if (empty($link['rel']) || empty($link['template'])) {
1391                                 continue;
1392                         }
1393
1394                         if ($link['rel'] == 'lrdd') {
1395                                 // When the webfinger host is the same like the system host, it should be ok.
1396                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
1397                         }
1398                 }
1399
1400                 return $valid;
1401         }
1402
1403         /**
1404          * Detect the network of the given server via their known contacts
1405          *
1406          * @param string $url        URL of the given server
1407          * @param array  $serverdata array with server data
1408          *
1409          * @return array server data
1410          */
1411         private static function detectNetworkViaContacts(string $url, array $serverdata): array
1412         {
1413                 $contacts = [];
1414
1415                 $nurl = Strings::normaliseLink($url);
1416
1417                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $nurl]]);
1418                 while ($apcontact = DBA::fetch($apcontacts)) {
1419                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
1420                 }
1421                 DBA::close($apcontacts);
1422
1423                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $nurl]]);
1424                 while ($pcontact = DBA::fetch($pcontacts)) {
1425                         $contacts[$pcontact['nurl']] = $pcontact['url'];
1426                 }
1427                 DBA::close($pcontacts);
1428
1429                 if (empty($contacts)) {
1430                         return $serverdata;
1431                 }
1432
1433                 $time = time();
1434                 foreach ($contacts as $contact) {
1435                         // Endlosschleife verhindern wegen gsid!
1436                         $data = Probe::uri($contact);
1437                         if (in_array($data['network'], Protocol::FEDERATED)) {
1438                                 $serverdata['network'] = $data['network'];
1439
1440                                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1441                                         $serverdata['detection-method'] = self::DETECT_CONTACTS;
1442                                 }
1443                                 break;
1444                         } elseif ((time() - $time) > 10) {
1445                                 // To reduce the stress on remote systems we probe a maximum of 10 seconds
1446                                 break;
1447                         }
1448                 }
1449
1450                 return $serverdata;
1451         }
1452
1453         /**
1454          * Checks if the given server does have a '/poco' endpoint.
1455          * This is used for the 'PortableContact' functionality,
1456          * which is used by both Friendica and Hubzilla.
1457          *
1458          * @param string $url        URL of the given server
1459          * @param array  $serverdata array with server data
1460          *
1461          * @return array server data
1462          */
1463         private static function checkPoCo(string $url, array $serverdata): array
1464         {
1465                 $serverdata['poco'] = '';
1466
1467                 $curlResult = DI::httpClient()->get($url . '/poco', HttpClientAccept::JSON);
1468                 if (!$curlResult->isSuccess()) {
1469                         return $serverdata;
1470                 }
1471
1472                 $data = json_decode($curlResult->getBody(), true);
1473                 if (empty($data)) {
1474                         return $serverdata;
1475                 }
1476
1477                 if (!empty($data['totalResults'])) {
1478                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1479                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers, 1);
1480                         $serverdata['directory-type'] = self::DT_POCO;
1481                         $serverdata['poco'] = $url . '/poco';
1482                 }
1483
1484                 return $serverdata;
1485         }
1486
1487         /**
1488          * Checks if the given server does have a Mastodon style directory endpoint.
1489          *
1490          * @param string $url        URL of the given server
1491          * @param array  $serverdata array with server data
1492          *
1493          * @return array server data
1494          */
1495         public static function checkMastodonDirectory(string $url, array $serverdata): array
1496         {
1497                 $curlResult = DI::httpClient()->get($url . '/api/v1/directory?limit=1', HttpClientAccept::JSON);
1498                 if (!$curlResult->isSuccess()) {
1499                         return $serverdata;
1500                 }
1501
1502                 $data = json_decode($curlResult->getBody(), true);
1503                 if (empty($data)) {
1504                         return $serverdata;
1505                 }
1506
1507                 if (count($data) == 1) {
1508                         $serverdata['directory-type'] = self::DT_MASTODON;
1509                 }
1510
1511                 return $serverdata;
1512         }
1513
1514         /**
1515          * Detects Peertube via their known endpoint
1516          *
1517          * @param string $url        URL of the given server
1518          * @param array  $serverdata array with server data
1519          *
1520          * @return array server data
1521          */
1522         private static function detectPeertube(string $url, array $serverdata): array
1523         {
1524                 $curlResult = DI::httpClient()->get($url . '/api/v1/config', HttpClientAccept::JSON);
1525                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1526                         return $serverdata;
1527                 }
1528
1529                 $data = json_decode($curlResult->getBody(), true);
1530                 if (empty($data)) {
1531                         return $serverdata;
1532                 }
1533
1534                 if (!empty($data['instance']) && !empty($data['serverVersion'])) {
1535                         $serverdata['platform'] = 'peertube';
1536                         $serverdata['version'] = $data['serverVersion'];
1537                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1538
1539                         if (!empty($data['instance']['name'])) {
1540                                 $serverdata['site_name'] = $data['instance']['name'];
1541                         }
1542
1543                         if (!empty($data['instance']['shortDescription'])) {
1544                                 $serverdata['info'] = $data['instance']['shortDescription'];
1545                         }
1546
1547                         if (!empty($data['signup'])) {
1548                                 if (!empty($data['signup']['allowed'])) {
1549                                         $serverdata['register_policy'] = Register::OPEN;
1550                                 }
1551                         }
1552
1553                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1554                                 $serverdata['detection-method'] = self::DETECT_V1_CONFIG;
1555                         }
1556                 }
1557
1558                 return $serverdata;
1559         }
1560
1561         /**
1562          * Detects the version number of a given server when it was a NextCloud installation
1563          *
1564          * @param string $url        URL of the given server
1565          * @param array  $serverdata array with server data
1566          * @param bool   $validHostMeta
1567          *
1568          * @return array server data
1569          */
1570         private static function detectNextcloud(string $url, array $serverdata, bool $validHostMeta): array
1571         {
1572                 $curlResult = DI::httpClient()->get($url . '/status.php', HttpClientAccept::JSON);
1573                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1574                         return $serverdata;
1575                 }
1576
1577                 $data = json_decode($curlResult->getBody(), true);
1578                 if (empty($data)) {
1579                         return $serverdata;
1580                 }
1581
1582                 if (!empty($data['version'])) {
1583                         $serverdata['platform'] = 'nextcloud';
1584                         $serverdata['version'] = $data['version'];
1585
1586                         if ($validHostMeta) {
1587                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1588                         }
1589
1590                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1591                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1592                         }
1593                 }
1594
1595                 return $serverdata;
1596         }
1597
1598         /**
1599          * Fetches weekly usage data
1600          *
1601          * @param string $url        URL of the given server
1602          * @param array  $serverdata array with server data
1603          *
1604          * @return array server data
1605          */
1606         private static function fetchWeeklyUsage(string $url, array $serverdata): array
1607         {
1608                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance/activity', HttpClientAccept::JSON);
1609                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1610                         return $serverdata;
1611                 }
1612
1613                 $data = json_decode($curlResult->getBody(), true);
1614                 if (empty($data)) {
1615                         return $serverdata;
1616                 }
1617
1618                 $current_week = [];
1619                 foreach ($data as $week) {
1620                         // Use only data from a full week
1621                         if (empty($week['week']) || (time() - $week['week']) < 7 * 24 * 60 * 60) {
1622                                 continue;
1623                         }
1624
1625                         // Most likely the data is sorted correctly. But we better are safe than sorry
1626                         if (empty($current_week['week']) || ($current_week['week'] < $week['week'])) {
1627                                 $current_week = $week;
1628                         }
1629                 }
1630
1631                 if (!empty($current_week['logins'])) {
1632                         $serverdata['active-week-users'] = max($current_week['logins'], 0);
1633                 }
1634
1635                 return $serverdata;
1636         }
1637
1638         /**
1639          * Detects data from a given server url if it was a mastodon alike system
1640          *
1641          * @param string $url        URL of the given server
1642          * @param array  $serverdata array with server data
1643          *
1644          * @return array server data
1645          */
1646         private static function detectMastodonAlikes(string $url, array $serverdata): array
1647         {
1648                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance', HttpClientAccept::JSON);
1649                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1650                         return $serverdata;
1651                 }
1652
1653                 $data = json_decode($curlResult->getBody(), true);
1654                 if (empty($data)) {
1655                         return $serverdata;
1656                 }
1657
1658                 $valid = false;
1659
1660                 if (!empty($data['version'])) {
1661                         $serverdata['platform'] = 'mastodon';
1662                         $serverdata['version'] = $data['version'] ?? '';
1663                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1664                         $valid = true;
1665                 }
1666
1667                 if (!empty($data['title'])) {
1668                         $serverdata['site_name'] = $data['title'];
1669                 }
1670
1671                 if (!empty($data['title']) && empty($serverdata['platform']) && ($serverdata['network'] == Protocol::PHANTOM)) {
1672                         $serverdata['platform'] = 'mastodon';
1673                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1674                         $valid = true;
1675                 }
1676
1677                 if (!empty($data['description'])) {
1678                         $serverdata['info'] = trim($data['description']);
1679                 }
1680
1681                 if (!empty($data['stats']['user_count'])) {
1682                         $serverdata['registered-users'] = max($data['stats']['user_count'], 1);
1683                 }
1684
1685                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1686                         $serverdata['platform'] = strtolower($matches[1]);
1687                         $serverdata['version'] = $matches[2];
1688                         $valid = true;
1689                 }
1690
1691                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1692                         $serverdata['platform'] = 'pleroma';
1693                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1694                         $valid = true;
1695                 }
1696
1697                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1698                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1699                         $serverdata['platform'] = 'pleroma';
1700                         $valid = true;
1701                 }
1702
1703                 if ($valid && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1704                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1705                 }
1706
1707                 return $serverdata;
1708         }
1709
1710         /**
1711          * Detects data from typical Hubzilla endpoints
1712          *
1713          * @param string $url        URL of the given server
1714          * @param array  $serverdata array with server data
1715          *
1716          * @return array server data
1717          */
1718         private static function detectHubzilla(string $url, array $serverdata): array
1719         {
1720                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/config.json', HttpClientAccept::JSON);
1721                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1722                         return $serverdata;
1723                 }
1724
1725                 $data = json_decode($curlResult->getBody(), true);
1726                 if (empty($data) || empty($data['site'])) {
1727                         return $serverdata;
1728                 }
1729
1730                 if (!empty($data['site']['name'])) {
1731                         $serverdata['site_name'] = $data['site']['name'];
1732                 }
1733
1734                 if (!empty($data['site']['platform'])) {
1735                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1736                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1737                         $serverdata['network'] = Protocol::ZOT;
1738                 }
1739
1740                 if (!empty($data['site']['hubzilla'])) {
1741                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1742                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1743                         $serverdata['network'] = Protocol::ZOT;
1744                 }
1745
1746                 if (!empty($data['site']['redmatrix'])) {
1747                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1748                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1749                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1750                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1751                         }
1752
1753                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1754                         $serverdata['network'] = Protocol::ZOT;
1755                 }
1756
1757                 $private = false;
1758                 $inviteonly = false;
1759                 $closed = false;
1760
1761                 if (!empty($data['site']['closed'])) {
1762                         $closed = self::toBoolean($data['site']['closed']);
1763                 }
1764
1765                 if (!empty($data['site']['private'])) {
1766                         $private = self::toBoolean($data['site']['private']);
1767                 }
1768
1769                 if (!empty($data['site']['inviteonly'])) {
1770                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1771                 }
1772
1773                 if (!$closed && !$private and $inviteonly) {
1774                         $serverdata['register_policy'] = Register::APPROVE;
1775                 } elseif (!$closed && !$private) {
1776                         $serverdata['register_policy'] = Register::OPEN;
1777                 } else {
1778                         $serverdata['register_policy'] = Register::CLOSED;
1779                 }
1780
1781                 if (($serverdata['network'] != Protocol::PHANTOM) && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1782                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1783                 }
1784
1785                 return $serverdata;
1786         }
1787
1788         /**
1789          * Converts input value to a boolean value
1790          *
1791          * @param string|integer $val
1792          *
1793          * @return boolean
1794          */
1795         private static function toBoolean($val): bool
1796         {
1797                 if (($val == 'true') || ($val == 1)) {
1798                         return true;
1799                 } elseif (($val == 'false') || ($val == 0)) {
1800                         return false;
1801                 }
1802
1803                 return $val;
1804         }
1805
1806         /**
1807          * Detect if the URL belongs to a GNU Social server
1808          *
1809          * @param string $url        URL of the given server
1810          * @param array  $serverdata array with server data
1811          *
1812          * @return array server data
1813          */
1814         private static function detectGNUSocial(string $url, array $serverdata): array
1815         {
1816                 // Test for GNU Social
1817                 $curlResult = DI::httpClient()->get($url . '/api/gnusocial/version.json', HttpClientAccept::JSON);
1818                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1819                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1820                         $serverdata['platform'] = 'gnusocial';
1821                         // Remove junk that some GNU Social servers return
1822                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1823                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1824                         $serverdata['version'] = trim($serverdata['version'], '"');
1825                         $serverdata['network'] = Protocol::OSTATUS;
1826
1827                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1828                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1829                         }
1830
1831                         return $serverdata;
1832                 }
1833
1834                 // Test for Statusnet
1835                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/version.json', HttpClientAccept::JSON);
1836                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1837                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1838
1839                         // Remove junk that some GNU Social servers return
1840                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1841                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1842                         $serverdata['version'] = trim($serverdata['version'], '"');
1843
1844                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1845                                 $serverdata['platform'] = 'pleroma';
1846                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1847                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1848                         } else {
1849                                 $serverdata['platform'] = 'statusnet';
1850                                 $serverdata['network'] = Protocol::OSTATUS;
1851                         }
1852
1853                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1854                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1855                         }
1856                 }
1857
1858                 return $serverdata;
1859         }
1860
1861         /**
1862          * Detect if the URL belongs to a Friendica server
1863          *
1864          * @param string $url        URL of the given server
1865          * @param array  $serverdata array with server data
1866          *
1867          * @return array server data
1868          */
1869         private static function detectFriendica(string $url, array $serverdata): array
1870         {
1871                 // There is a bug in some versions of Friendica that will return an ActivityStream actor when the content type "application/json" is requested.
1872                 // Because of this me must not use ACCEPT_JSON here.
1873                 $curlResult = DI::httpClient()->get($url . '/friendica/json');
1874                 if (!$curlResult->isSuccess()) {
1875                         $curlResult = DI::httpClient()->get($url . '/friendika/json');
1876                         $friendika = true;
1877                         $platform = 'Friendika';
1878                 } else {
1879                         $friendika = false;
1880                         $platform = 'Friendica';
1881                 }
1882
1883                 if (!$curlResult->isSuccess()) {
1884                         return $serverdata;
1885                 }
1886
1887                 $data = json_decode($curlResult->getBody(), true);
1888                 if (empty($data) || empty($data['version'])) {
1889                         return $serverdata;
1890                 }
1891
1892                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1893                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1894                 }
1895
1896                 $serverdata['network'] = Protocol::DFRN;
1897                 $serverdata['version'] = $data['version'];
1898
1899                 if (!empty($data['no_scrape_url'])) {
1900                         $serverdata['noscrape'] = $data['no_scrape_url'];
1901                 }
1902
1903                 if (!empty($data['site_name'])) {
1904                         $serverdata['site_name'] = $data['site_name'];
1905                 }
1906
1907                 if (!empty($data['info'])) {
1908                         $serverdata['info'] = trim($data['info']);
1909                 }
1910
1911                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1912                 switch ($register_policy) {
1913                         case 'REGISTER_OPEN':
1914                                 $serverdata['register_policy'] = Register::OPEN;
1915                                 break;
1916
1917                         case 'REGISTER_APPROVE':
1918                                 $serverdata['register_policy'] = Register::APPROVE;
1919                                 break;
1920
1921                         case 'REGISTER_CLOSED':
1922                         case 'REGISTER_INVITATION':
1923                                 $serverdata['register_policy'] = Register::CLOSED;
1924                                 break;
1925                         default:
1926                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1927                                 $serverdata['register_policy'] = Register::CLOSED;
1928                                 break;
1929                 }
1930
1931                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1932
1933                 return $serverdata;
1934         }
1935
1936         /**
1937          * Analyses the landing page of a given server for hints about type and system of that server
1938          *
1939          * @param object $curlResult result of curl execution
1940          * @param array  $serverdata array with server data
1941          *
1942          * @return array server data
1943          */
1944         private static function analyseRootBody($curlResult, array $serverdata): array
1945         {
1946                 if (empty($curlResult->getBody())) {
1947                         return $serverdata;
1948                 }
1949
1950                 if (file_exists(__DIR__ . '/../../static/platforms.config.php')) {
1951                         require __DIR__ . '/../../static/platforms.config.php';
1952                 } else {
1953                         throw new HTTPException\InternalServerErrorException('Invalid platform file');
1954                 }
1955
1956                 $platforms = array_merge($ap_platforms, $dfrn_platforms, $zap_platforms, $platforms);
1957
1958                 $doc = new DOMDocument();
1959                 @$doc->loadHTML($curlResult->getBody());
1960                 $xpath = new DOMXPath($doc);
1961                 $assigned = false;
1962
1963                 // We can only detect honk via some HTML element on their page
1964                 if ($xpath->query('//div[@id="honksonpage"]')->count() == 1) {
1965                         $serverdata['platform'] = 'honk';
1966                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1967                         $assigned = true;
1968                 }
1969
1970                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1971                 if (!empty($title)) {
1972                         $serverdata['site_name'] = $title;
1973                 }
1974
1975                 $list = $xpath->query('//meta[@name]');
1976
1977                 foreach ($list as $node) {
1978                         $attr = [];
1979                         if ($node->attributes->length) {
1980                                 foreach ($node->attributes as $attribute) {
1981                                         $value = trim($attribute->value);
1982                                         if (empty($value)) {
1983                                                 continue;
1984                                         }
1985
1986                                         $attr[$attribute->name] = $value;
1987                                 }
1988
1989                                 if (empty($attr['name']) || empty($attr['content'])) {
1990                                         continue;
1991                                 }
1992                         }
1993
1994                         if ($attr['name'] == 'description') {
1995                                 $serverdata['info'] = $attr['content'];
1996                         }
1997
1998                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1999                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad', 'generator'])) {
2000                                 $platform = str_ireplace(array_keys($platforms), array_values($platforms), $attr['content']);
2001                                 $platform = str_replace('/', ' ', $platform);
2002                                 $platform_parts = explode(' ', $platform);
2003                                 if ((count($platform_parts) >= 2) && in_array(strtolower($platform_parts[0]), array_values($platforms))) {
2004                                         $platform = $platform_parts[0];
2005                                         $serverdata['version'] = $platform_parts[1];
2006                                 }
2007                                 if (in_array($platform, array_values($dfrn_platforms))) {
2008                                         $serverdata['network'] = Protocol::DFRN;
2009                                 } elseif (in_array($platform, array_values($ap_platforms))) {
2010                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2011                                 } elseif (in_array($platform, array_values($zap_platforms))) {
2012                                         $serverdata['network'] = Protocol::ZOT;
2013                                 }
2014                                 if (in_array($platform, array_values($platforms))) {
2015                                         $serverdata['platform'] = $platform;
2016                                         $assigned = true;
2017                                 }
2018                         }
2019                 }
2020
2021                 $list = $xpath->query('//meta[@property]');
2022
2023                 foreach ($list as $node) {
2024                         $attr = [];
2025                         if ($node->attributes->length) {
2026                                 foreach ($node->attributes as $attribute) {
2027                                         $value = trim($attribute->value);
2028                                         if (empty($value)) {
2029                                                 continue;
2030                                         }
2031
2032                                         $attr[$attribute->name] = $value;
2033                                 }
2034
2035                                 if (empty($attr['property']) || empty($attr['content'])) {
2036                                         continue;
2037                                 }
2038                         }
2039
2040                         if ($attr['property'] == 'og:site_name') {
2041                                 $serverdata['site_name'] = $attr['content'];
2042                         }
2043
2044                         if ($attr['property'] == 'og:description') {
2045                                 $serverdata['info'] = $attr['content'];
2046                         }
2047
2048                         if (in_array($attr['property'], ['og:platform', 'generator'])) {
2049                                 if (in_array($attr['content'], array_keys($platforms))) {
2050                                         $serverdata['platform'] = $platforms[$attr['content']];
2051                                         $assigned = true;
2052                                 }
2053
2054                                 if (in_array($attr['content'], array_keys($ap_platforms))) {
2055                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2056                                 } elseif (in_array($attr['content'], array_values($zap_platforms))) {
2057                                         $serverdata['network'] = Protocol::ZOT;
2058                                 }
2059                         }
2060                 }
2061
2062                 $list = $xpath->query('//link[@rel="me"]');
2063                 foreach ($list as $node) {
2064                         foreach ($node->attributes as $attribute) {
2065                                 if (parse_url(trim($attribute->value), PHP_URL_HOST) == 'micro.blog') {
2066                                         $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2067                                         $serverdata['platform'] = 'microblog';
2068                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2069                                         $assigned = true;
2070                                 }
2071                         }
2072                 }
2073
2074                 if ($serverdata['platform'] != 'microblog') {
2075                         $list = $xpath->query('//link[@rel="micropub"]');
2076                         foreach ($list as $node) {
2077                                 foreach ($node->attributes as $attribute) {
2078                                         if (trim($attribute->value) == 'https://micro.blog/micropub') {
2079                                                 $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2080                                                 $serverdata['platform'] = 'microblog';
2081                                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
2082                                                 $assigned = true;
2083                                         }
2084                                 }
2085                         }
2086                 }
2087
2088                 if ($assigned && in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER])) {
2089                         $serverdata['detection-method'] = self::DETECT_BODY;
2090                 }
2091
2092                 return $serverdata;
2093         }
2094
2095         /**
2096          * Analyses the header data of a given server for hints about type and system of that server
2097          *
2098          * @param object $curlResult result of curl execution
2099          * @param array  $serverdata array with server data
2100          *
2101          * @return array server data
2102          */
2103         private static function analyseRootHeader($curlResult, array $serverdata): array
2104         {
2105                 if ($curlResult->getHeader('server') == 'Mastodon') {
2106                         $serverdata['platform'] = 'mastodon';
2107                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2108                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
2109                         $serverdata['platform'] = 'diaspora';
2110                         $serverdata['network'] = Protocol::DIASPORA;
2111                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version')[0] ?? '';
2112                 } elseif ($curlResult->inHeader('x-friendica-version')) {
2113                         $serverdata['platform'] = 'friendica';
2114                         $serverdata['network'] = Protocol::DFRN;
2115                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version')[0] ?? '';
2116                 } else {
2117                         return $serverdata;
2118                 }
2119
2120                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
2121                         $serverdata['detection-method'] = self::DETECT_HEADER;
2122                 }
2123
2124                 return $serverdata;
2125         }
2126
2127         /**
2128          * Update GServer entries
2129          */
2130         public static function discover()
2131         {
2132                 // Update the server list
2133                 self::discoverFederation();
2134
2135                 $no_of_queries = 5;
2136
2137                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
2138
2139                 if ($requery_days == 0) {
2140                         $requery_days = 7;
2141                 }
2142
2143                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
2144
2145                 $gservers = DBA::select('gserver', ['id', 'url', 'nurl', 'network', 'poco', 'directory-type'],
2146                         ["NOT `failed` AND `directory-type` != ? AND `last_poco_query` < ?", GServer::DT_NONE, $last_update],
2147                         ['order' => ['RAND()']]);
2148
2149                 while ($gserver = DBA::fetch($gservers)) {
2150                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2151                         Worker::add(Worker::PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
2152
2153                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2154                         Worker::add(Worker::PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
2155
2156                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
2157                         self::update($fields, ['nurl' => $gserver['nurl']]);
2158
2159                         if (--$no_of_queries == 0) {
2160                                 break;
2161                         }
2162                 }
2163
2164                 DBA::close($gservers);
2165         }
2166
2167         /**
2168          * Discover federated servers
2169          */
2170         private static function discoverFederation()
2171         {
2172                 $last = DI::config()->get('poco', 'last_federation_discovery');
2173
2174                 if ($last) {
2175                         $next = $last + (24 * 60 * 60);
2176
2177                         if ($next > time()) {
2178                                 return;
2179                         }
2180                 }
2181
2182                 // Discover federated servers
2183                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
2184                 foreach ($protocols as $protocol) {
2185                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
2186                         $curlResult = DI::httpClient()->fetch('https://the-federation.info/graphql?query=' . urlencode($query), HttpClientAccept::JSON);
2187                         if (!empty($curlResult)) {
2188                                 $data = json_decode($curlResult, true);
2189                                 if (!empty($data['data']['nodes'])) {
2190                                         foreach ($data['data']['nodes'] as $server) {
2191                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
2192                                                 self::add('https://' . $server['host'], true);
2193                                         }
2194                                 }
2195                         }
2196                 }
2197
2198                 // Disvover Mastodon servers
2199                 $accesstoken = DI::config()->get('system', 'instances_social_key');
2200
2201                 if (!empty($accesstoken)) {
2202                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
2203                         $curlResult = DI::httpClient()->get($api, HttpClientAccept::JSON, [HttpClientOptions::HEADERS => ['Authorization' => ['Bearer ' . $accesstoken]]]);
2204                         if ($curlResult->isSuccess()) {
2205                                 $servers = json_decode($curlResult->getBody(), true);
2206
2207                                 if (!empty($servers['instances'])) {
2208                                         foreach ($servers['instances'] as $server) {
2209                                                 $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
2210                                                 self::add($url);
2211                                         }
2212                                 }
2213                         }
2214                 }
2215
2216                 DI::config()->set('poco', 'last_federation_discovery', time());
2217         }
2218
2219         /**
2220          * Set the protocol for the given server
2221          *
2222          * @param int $gsid     Server id
2223          * @param int $protocol Protocol id
2224          *
2225          * @throws Exception
2226          */
2227         public static function setProtocol(int $gsid, int $protocol)
2228         {
2229                 if (empty($gsid)) {
2230                         return;
2231                 }
2232
2233                 $gserver = DBA::selectFirst('gserver', ['protocol', 'url'], ['id' => $gsid]);
2234                 if (!DBA::isResult($gserver)) {
2235                         return;
2236                 }
2237
2238                 $old = $gserver['protocol'];
2239
2240                 if (!is_null($old)) {
2241                         /*
2242                         The priority for the protocols is:
2243                                 1. ActivityPub
2244                                 2. DFRN via Diaspora
2245                                 3. Legacy DFRN
2246                                 4. Diaspora
2247                                 5. OStatus
2248                         */
2249
2250                         // We don't need to change it when nothing is to be changed
2251                         if ($old == $protocol) {
2252                                 return;
2253                         }
2254
2255                         // We don't want to mark a server as OStatus when it had been marked with any other protocol before
2256                         if ($protocol == Post\DeliveryData::OSTATUS) {
2257                                 return;
2258                         }
2259
2260                         // If the server is marked as ActivityPub then we won't change it to anything different
2261                         if ($old == Post\DeliveryData::ACTIVITYPUB) {
2262                                 return;
2263                         }
2264
2265                         // Don't change it to anything lower than DFRN if the new one wasn't ActivityPub
2266                         if (($old == Post\DeliveryData::DFRN) && ($protocol != Post\DeliveryData::ACTIVITYPUB)) {
2267                                 return;
2268                         }
2269
2270                         // Don't change it to Diaspora when it is a legacy DFRN server
2271                         if (($old == Post\DeliveryData::LEGACY_DFRN) && ($protocol == Post\DeliveryData::DIASPORA)) {
2272                                 return;
2273                         }
2274                 }
2275
2276                 Logger::info('Protocol for server', ['protocol' => $protocol, 'old' => $old, 'id' => $gsid, 'url' => $gserver['url'], 'callstack' => System::callstack(20)]);
2277                 self::update(['protocol' => $protocol], ['id' => $gsid]);
2278         }
2279
2280         /**
2281          * Fetch the protocol of the given server
2282          *
2283          * @param int $gsid Server id
2284          *
2285          * @return ?int One of Post\DeliveryData protocol constants or null if unknown or gserver is missing
2286          *
2287          * @throws Exception
2288          */
2289         public static function getProtocol(int $gsid): ?int
2290         {
2291                 if (empty($gsid)) {
2292                         return null;
2293                 }
2294
2295                 $gserver = DBA::selectFirst('gserver', ['protocol'], ['id' => $gsid]);
2296                 if (DBA::isResult($gserver)) {
2297                         return $gserver['protocol'];
2298                 }
2299
2300                 return null;
2301         }
2302
2303         /**
2304          * Update rows in the gserver table.
2305          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2306          *
2307          * @param array $fields
2308          * @param array $condition
2309          *
2310          * @return bool
2311          *
2312          * @throws Exception
2313          */
2314         public static function update(array $fields, array $condition): bool
2315         {
2316                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2317
2318                 return DBA::update('gserver', $fields, $condition);
2319         }
2320
2321         /**
2322          * Insert a row into the gserver table.
2323          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2324          *
2325          * @param array $fields
2326          * @param int   $duplicate_mode What to do on a duplicated entry
2327          *
2328          * @return bool
2329          *
2330          * @throws Exception
2331          */
2332         public static function insert(array $fields, int $duplicate_mode = Database::INSERT_DEFAULT): bool
2333         {
2334                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2335
2336                 return DBA::insert('gserver', $fields, $duplicate_mode);
2337         }
2338 }