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
  49 /**
  50  * This class handles GServer related functions
  51  */
  52 class GServer
  53 {
  54         // Directory types
  55         const DT_NONE = 0;
  56         const DT_POCO = 1;
  57         const DT_MASTODON = 2;
  58
  59         // Methods to detect server types
  60
  61         // Non endpoint specific methods
  62         const DETECT_MANUAL = 0;
  63         const DETECT_HEADER = 1;
  64         const DETECT_BODY = 2;
  65         const DETECT_HOST_META = 3;
  66         const DETECT_CONTACTS = 4;
  67         const DETECT_AP_ACTOR = 5;
  68         const DETECT_AP_COLLECTION = 6;
  69
  70         const DETECT_UNSPECIFIC = [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META, self::DETECT_CONTACTS, self::DETECT_AP_ACTOR];
  71
  72         // Implementation specific endpoints
  73         const DETECT_FRIENDIKA = 10;
  74         const DETECT_FRIENDICA = 11;
  75         const DETECT_STATUSNET = 12;
  76         const DETECT_GNUSOCIAL = 13;
  77         const DETECT_CONFIG_JSON = 14; // Statusnet, GNU Social, Older Hubzilla/Redmatrix
  78         const DETECT_SITEINFO_JSON = 15; // Newer Hubzilla
  79         const DETECT_MASTODON_API = 16;
  80         const DETECT_STATUS_PHP = 17; // Nextcloud
  81         const DETECT_V1_CONFIG = 18;
  82         const DETECT_PUMPIO = 19; // Deprecated
  83         const DETECT_SYSTEM_ACTOR = 20; // Mistpark, Osada, Roadhouse, Zap
  84
  85         // Standardized endpoints
  86         const DETECT_STATISTICS_JSON = 100;
  87         const DETECT_NODEINFO_1 = 101;
  88         const DETECT_NODEINFO_2 = 102;
  89         const DETECT_NODEINFO_210 = 103;
  90
  91         /**
  92          * Check for the existance of a server and adds it in the background if not existant
  93          *
  94          * @param string $url
  95          * @param boolean $only_nodeinfo
  96          *
  97          * @return void
  98          */
  99         public static function add(string $url, bool $only_nodeinfo = false)
 100         {
 101                 if (self::getID($url, false)) {
 102                         return;
 103                 }
 104
 105                 Worker::add(PRIORITY_LOW, 'UpdateGServer', $url, $only_nodeinfo);
 106         }
 107
 108         /**
 109          * Get the ID for the given server URL
 110          *
 111          * @param string $url
 112          * @param boolean $no_check Don't check if the server hadn't been found
 113          *
 114          * @return int|null gserver id or NULL on empty URL or failed check
 115          */
 116         public static function getID(string $url, bool $no_check = false): ?int
 117         {
 118                 if (empty($url)) {
 119                         return null;
 120                 }
 121
 122                 $url = self::cleanURL($url);
 123
 124                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
 125                 if (DBA::isResult($gserver)) {
 126                         Logger::debug('Got ID for URL', ['id' => $gserver['id'], 'url' => $url, 'callstack' => System::callstack(20)]);
 127                         return $gserver['id'];
 128                 }
 129
 130                 if ($no_check || !self::check($url)) {
 131                         return null;
 132                 }
 133
 134                 return self::getID($url, true);
 135         }
 136
 137         /**
 138          * Retrieves all the servers which base domain are matching the provided domain pattern
 139          *
 140          * The pattern is a simple fnmatch() pattern with ? for single wildcard and * for multiple wildcard
 141          *
 142          * @param string $pattern
 143          *
 144          * @return array
 145          *
 146          * @throws Exception
 147          */
 148         public static function listByDomainPattern(string $pattern): array
 149         {
 150                 $likePattern = 'http://' . strtr($pattern, ['_' => '\_', '%' => '\%', '?' => '_', '*' => '%']);
 151
 152                 // The SUBSTRING_INDEX returns everything before the eventual third /, which effectively trims an
 153                 // eventual server path and keep only the server domain which we're matching against the pattern.
 154                 $sql = "SELECT `gserver`.*, COUNT(*) AS `contacts`
 155                         FROM `gserver`
 156                         LEFT JOIN `contact` ON `gserver`.`id` = `contact`.`gsid`
 157                         WHERE SUBSTRING_INDEX(`gserver`.`nurl`, '/', 3) LIKE ?
 158                         AND NOT `gserver`.`failed`
 159                         GROUP BY `gserver`.`id`";
 160
 161                 $stmt = DI::dba()->p($sql, $likePattern);
 162
 163                 return DI::dba()->toArray($stmt);
 164         }
 165
 166         /**
 167          * Checks if the given server is reachable
 168          *
 169          * @param string  $profile URL of the given profile
 170          * @param string  $server  URL of the given server (If empty, taken from profile)
 171          * @param string  $network Network value that is used, when detection failed
 172          * @param boolean $force   Force an update.
 173          *
 174          * @return boolean 'true' if server seems vital
 175          */
 176         public static function reachable(string $profile, string $server = '', string $network = '', bool $force = false): bool
 177         {
 178                 if ($server == '') {
 179                         $contact = Contact::getByURL($profile, null, ['baseurl']);
 180                         if (!empty($contact['baseurl'])) {
 181                                 $server = $contact['baseurl'];
 182                         }
 183                 }
 184
 185                 if ($server == '') {
 186                         return true;
 187                 }
 188
 189                 return self::check($server, $network, $force);
 190         }
 191
 192         /**
 193          * Calculate the next update day
 194          *
 195          * @param bool $success
 196          * @param string $created
 197          * @param string $last_contact
 198          * @param bool $undetected
 199          *
 200          * @return string
 201          *
 202          * @throws Exception
 203          */
 204         public static function getNextUpdateDate(bool $success, string $created = '', string $last_contact = '', bool $undetected = false): string
 205         {
 206                 // On successful contact process check again next week when it is a detected system.
 207                 // When we haven't detected the system, it could be a static website or a really old system.
 208                 if ($success) {
 209                         return DateTimeFormat::utc($undetected ? 'now +1 month' : 'now +7 day');
 210                 }
 211
 212                 $now = strtotime(DateTimeFormat::utcNow());
 213
 214                 if ($created > $last_contact) {
 215                         $contact_time = strtotime($created);
 216                 } else {
 217                         $contact_time = strtotime($last_contact);
 218                 }
 219
 220                 // If the last contact was less than 6 hours before then try again in 6 hours
 221                 if (($now - $contact_time) < (60 * 60 * 6)) {
 222                         return DateTimeFormat::utc('now +6 hour');
 223                 }
 224
 225                 // If the last contact was less than 12 hours before then try again in 12 hours
 226                 if (($now - $contact_time) < (60 * 60 * 12)) {
 227                         return DateTimeFormat::utc('now +12 hour');
 228                 }
 229
 230                 // If the last contact was less than 24 hours before then try tomorrow again
 231                 if (($now - $contact_time) < (60 * 60 * 24)) {
 232                         return DateTimeFormat::utc('now +1 day');
 233                 }
 234
 235                 // If the last contact was less than a week before then try again in a week
 236                 if (($now - $contact_time) < (60 * 60 * 24 * 7)) {
 237                         return DateTimeFormat::utc('now +1 week');
 238                 }
 239
 240                 // If the last contact was less than two weeks before then try again in two week
 241                 if (($now - $contact_time) < (60 * 60 * 24 * 14)) {
 242                         return DateTimeFormat::utc('now +2 week');
 243                 }
 244
 245                 // If the last contact was less than a month before then try again in a month
 246                 if (($now - $contact_time) < (60 * 60 * 24 * 30)) {
 247                         return DateTimeFormat::utc('now +1 month');
 248                 }
 249
 250                 // The system hadn't been successul contacted for more than a month, so try again in three months
 251                 return DateTimeFormat::utc('now +3 month');
 252         }
 253
 254         /**
 255          * Checks the state of the given server.
 256          *
 257          * @param string  $server_url    URL of the given server
 258          * @param string  $network       Network value that is used, when detection failed
 259          * @param boolean $force         Force an update.
 260          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 261          *
 262          * @return boolean 'true' if server seems vital
 263          */
 264         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false): bool
 265         {
 266                 $server_url = self::cleanURL($server_url);
 267                 if ($server_url == '') {
 268                         return false;
 269                 }
 270
 271                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
 272                 if (DBA::isResult($gserver)) {
 273                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
 274                                 $fields = ['created' => DateTimeFormat::utcNow()];
 275                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
 276                                 self::update($fields, $condition);
 277                         }
 278
 279                         if (!$force && (strtotime($gserver['next_contact']) > time())) {
 280                                 Logger::info('No update needed', ['server' => $server_url]);
 281                                 return (!$gserver['failed']);
 282                         }
 283                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force]);
 284                 } else {
 285                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
 286                 }
 287
 288                 return self::detect($server_url, $network, $only_nodeinfo);
 289         }
 290
 291         /**
 292          * Set failed server status
 293          *
 294          * @param string $url
 295          */
 296         public static function setFailure(string $url)
 297         {
 298                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($url)]);
 299                 if (DBA::isResult($gserver)) {
 300                         $next_update = self::getNextUpdateDate(false, $gserver['created'], $gserver['last_contact']);
 301                         self::update(['url' => $url, 'failed' => true, 'last_failure' => DateTimeFormat::utcNow(),
 302                         'next_contact' => $next_update, 'network' => Protocol::PHANTOM, 'detection-method' => null],
 303                         ['nurl' => Strings::normaliseLink($url)]);
 304                         Logger::info('Set failed status for existing server', ['url' => $url]);
 305                         return;
 306                 }
 307                 DBA::insert('gserver', ['url' => $url, 'nurl' => Strings::normaliseLink($url),
 308                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
 309                         'failed' => true, 'last_failure' => DateTimeFormat::utcNow()]);
 310                 Logger::info('Set failed status for new server', ['url' => $url]);
 311         }
 312
 313         /**
 314          * Remove unwanted content from the given URL
 315          *
 316          * @param string $url
 317          *
 318          * @return string cleaned URL
 319          */
 320         public static function cleanURL(string $url): string
 321         {
 322                 $url = trim($url, '/');
 323                 $url = str_replace('/index.php', '', $url);
 324
 325                 $urlparts = parse_url($url);
 326                 unset($urlparts['user']);
 327                 unset($urlparts['pass']);
 328                 unset($urlparts['query']);
 329                 unset($urlparts['fragment']);
 330                 return (string)Uri::fromParts($urlparts);
 331         }
 332
 333         /**
 334          * Detect server data (type, protocol, version number, ...)
 335          * The detected data is then updated or inserted in the gserver table.
 336          *
 337          * @param string  $url           URL of the given server
 338          * @param string  $network       Network value that is used, when detection failed
 339          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 340          *
 341          * @return boolean 'true' if server could be detected
 342          */
 343         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false): bool
 344         {
 345                 Logger::info('Detect server type', ['server' => $url]);
 346
 347                 $original_url = $url;
 348
 349                 // Remove URL content that is not supposed to exist for a server url
 350                 $url = rtrim(self::cleanURL($url), '/');
 351                 if (empty($url)) {
 352                         Logger::notice('Empty URL.');
 353                         return false;
 354                 }
 355
 356                 // If the URL missmatches, then we mark the old entry as failure
 357                 if (!Strings::compareLink($url, $original_url)) {
 358                         self::setFailure($original_url);
 359                         if (!self::getID($url, true)) {
 360                                 self::detect($url, $network, $only_nodeinfo);
 361                         }
 362                         return false;
 363                 }
 364
 365                 $valid_url = Network::isUrlValid($url);
 366                 if (!$valid_url) {
 367                         self::setFailure($url);
 368                         return false;
 369                 } else {
 370                         $valid_url = rtrim($valid_url, '/');
 371                 }
 372
 373                 if (!Strings::compareLink($url, $valid_url)) {
 374                         // We only follow redirects when the path stays the same or the target url has no path.
 375                         // Some systems have got redirects on their landing page to a single account page. This check handles it.
 376                         if (((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH))) ||
 377                                 (((parse_url($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)))) {
 378                                 Logger::debug('Found redirect. Mark old entry as failure', ['old' => $url, 'new' => $valid_url]);
 379                                 self::setFailure($url);
 380                                 if (!self::getID($valid_url, true)) {
 381                                         self::detect($valid_url, $network, $only_nodeinfo);
 382                                 }
 383                                 return false;
 384                         }
 385
 386                         if ((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) != parse_url($valid_url, PHP_URL_PATH)) &&
 387                                 (parse_url($url, PHP_URL_PATH) == '')) {
 388                                 Logger::debug('Found redirect. Mark old entry as failure and redirect to the basepath.', ['old' => $url, 'new' => $valid_url]);
 389                                 $parts = parse_url($valid_url);
 390                                 unset($parts['path']);
 391                                 $valid_url = (string)Uri::fromParts($parts);
 392
 393                                 self::setFailure($url);
 394                                 if (!self::getID($valid_url, true)) {
 395                                         self::detect($valid_url, $network, $only_nodeinfo);
 396                                 }
 397                                 return false;
 398                         }
 399                         Logger::debug('Found redirect, but ignore it.', ['old' => $url, 'new' => $valid_url]);
 400                 }
 401
 402                 if ((parse_url($url, PHP_URL_HOST) == parse_url($valid_url, PHP_URL_HOST)) &&
 403                         (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH)) &&
 404                         (parse_url($url, PHP_URL_SCHEME) != parse_url($valid_url, PHP_URL_SCHEME))) {
 405                         $url = $valid_url;
 406                 }
 407
 408                 $in_webroot = empty(parse_url($url, PHP_URL_PATH));
 409
 410                 // When a nodeinfo is present, we don't need to dig further
 411                 $curlResult = DI::httpClient()->get($url . '/.well-known/x-nodeinfo2', HttpClientAccept::JSON);
 412                 if ($curlResult->isTimeout()) {
 413                         self::setFailure($url);
 414                         return false;
 415                 }
 416
 417                 $serverdata = self::parseNodeinfo210($curlResult);
 418                 if (empty($serverdata)) {
 419                         $curlResult = DI::httpClient()->get($url . '/.well-known/nodeinfo', HttpClientAccept::JSON);
 420                         $serverdata = self::fetchNodeinfo($url, $curlResult);
 421                 }
 422
 423                 if ($only_nodeinfo && empty($serverdata)) {
 424                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 425                         self::setFailure($url);
 426                         return false;
 427                 } elseif (empty($serverdata)) {
 428                         $serverdata = ['detection-method' => self::DETECT_MANUAL, 'network' => Protocol::PHANTOM, 'platform' => '', 'version' => '', 'site_name' => '', 'info' => ''];
 429                 }
 430
 431                 // When there is no Nodeinfo, then use some protocol specific endpoints
 432                 if ($serverdata['network'] == Protocol::PHANTOM) {
 433                         if ($in_webroot) {
 434                                 // Fetch the landing page, possibly it reveals some data
 435                                 $accept = 'application/activity+json,application/ld+json,application/json,*/*;q=0.9';
 436                                 $curlResult = DI::httpClient()->get($url, $accept);
 437                                 if (!$curlResult->isSuccess() && $curlResult->getReturnCode() == '406') {
 438                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 439                                         $html_fetched = true;
 440                                 } else {
 441                                         $html_fetched = false;
 442                                 }
 443
 444                                 if ($curlResult->isSuccess()) {
 445                                         $json = json_decode($curlResult->getBody(), true);
 446                                         if (!empty($json) && is_array($json)) {
 447                                                 $data = self::fetchDataFromSystemActor($json, $serverdata);
 448                                                 $serverdata = $data['server'];
 449                                                 $systemactor = $data['actor'];
 450                                                 if (!$html_fetched && !in_array($serverdata['detection-method'], [self::DETECT_SYSTEM_ACTOR, self::DETECT_AP_COLLECTION])) {
 451                                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 452                                                 }
 453                                         } elseif (!$html_fetched && (strlen($curlResult->getBody()) < 1000)) {
 454                                                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 455                                         }
 456
 457                                         if ($serverdata['detection-method'] != self::DETECT_SYSTEM_ACTOR) {
 458                                                 $serverdata = self::analyseRootHeader($curlResult, $serverdata);
 459                                                 $serverdata = self::analyseRootBody($curlResult, $serverdata);
 460                                         }
 461                                 }
 462
 463                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
 464                                         self::setFailure($url);
 465                                         return false;
 466                                 }
 467
 468                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 469                                         $serverdata = self::detectMastodonAlikes($url, $serverdata);
 470                                 }
 471                         }
 472
 473                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 474                         // With this check we don't have to waste time and ressources for dead systems.
 475                         // Also this hopefully prevents us from receiving abuse messages.
 476                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 477                                 $validHostMeta = self::validHostMeta($url);
 478                         } else {
 479                                 $validHostMeta = false;
 480                         }
 481
 482                         if ($validHostMeta) {
 483                                 if (in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY])) {
 484                                         $serverdata['detection-method'] = self::DETECT_HOST_META;
 485                                 }
 486
 487                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 488                                         $serverdata = self::detectFriendica($url, $serverdata);
 489                                 }
 490
 491                                 // The following systems have to be installed in the root directory.
 492                                 if ($in_webroot) {
 493                                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 494                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 495                                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 496                                         }
 497
 498                                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations, so we check other endpoints as well
 499                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 500                                                 $serverdata = self::detectHubzilla($url, $serverdata);
 501                                         }
 502
 503                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 504                                                 $serverdata = self::detectPeertube($url, $serverdata);
 505                                         }
 506
 507                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 508                                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 509                                         }
 510                                 }
 511                         } elseif (in_array($serverdata['platform'], ['friendica', 'friendika']) && in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_SYSTEM_ACTOR]))) {
 512                                 $serverdata = self::detectFriendica($url, $serverdata);
 513                         }
 514
 515                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 516                                 $serverdata = self::detectNextcloud($url, $serverdata, $validHostMeta);
 517                         }
 518
 519                         // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 520                         // Since this endpoint is only rarely used, we query it at a later time
 521                         if (in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_FRIENDICA, self::DETECT_CONFIG_JSON]))) {
 522                                 $serverdata = self::fetchStatistics($url, $serverdata);
 523                         }
 524                 }
 525
 526                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 527                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 528                         $serverdata['network'] = $network;
 529                 }
 530
 531                 // Most servers aren't installed in a subdirectory, so we declare this entry as failed
 532                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty(parse_url($url, PHP_URL_PATH)) && in_array($serverdata['detection-method'], [self::DETECT_MANUAL])) {
 533                         self::setFailure($url);
 534                         return false;
 535                 }
 536
 537                 $serverdata['url'] = $url;
 538                 $serverdata['nurl'] = Strings::normaliseLink($url);
 539
 540                 // We have to prevent an endless loop here.
 541                 // When a server is new, then there is no gserver entry yet.
 542                 // But in "detectNetworkViaContacts" it could happen that a contact is updated,
 543                 // and this can call this function here as well.
 544                 if (self::getID($url, true) && (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) ||
 545                         in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META]))) {
 546                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 547                 }
 548
 549                 // Detect the directory type
 550                 $serverdata['directory-type'] = self::DT_NONE;
 551
 552                 if (in_array($serverdata['network'], Protocol::FEDERATED)) {
 553                         $serverdata = self::checkMastodonDirectory($url, $serverdata);
 554
 555                         if ($serverdata['directory-type'] == self::DT_NONE) {
 556                                 $serverdata = self::checkPoCo($url, $serverdata);
 557                         }
 558                 }
 559
 560                 if ($serverdata['network'] == Protocol::ACTIVITYPUB) {
 561                         $serverdata = self::fetchWeeklyUsage($url, $serverdata);
 562                 }
 563
 564                 $serverdata['registered-users'] = $serverdata['registered-users'] ?? 0;
 565
 566                 // Numbers above a reasonable value (10 millions) are ignored
 567                 if ($serverdata['registered-users'] > 10000000) {
 568                         $serverdata['registered-users'] = 0;
 569                 }
 570
 571                 // On an active server there has to be at least a single user
 572                 if (!in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) && ($serverdata['registered-users'] <= 0)) {
 573                         $serverdata['registered-users'] = 1;
 574                 } elseif (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 575                         $serverdata['registered-users'] = 0;
 576                 }
 577
 578                 $serverdata['next_contact'] = self::getNextUpdateDate(true, '', '', in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]));
 579
 580                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 581                 $serverdata['failed'] = false;
 582
 583                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 584                 if (!DBA::isResult($gserver)) {
 585                         $serverdata['created'] = DateTimeFormat::utcNow();
 586                         $ret = DBA::insert('gserver', $serverdata);
 587                         $id = DBA::lastInsertId();
 588                 } else {
 589                         $ret = self::update($serverdata, ['nurl' => $serverdata['nurl']]);
 590                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 591                         if (DBA::isResult($gserver)) {
 592                                 $id = $gserver['id'];
 593                         }
 594                 }
 595
 596                 // Count the number of known contacts from this server
 597                 if (!empty($id) && !in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 598                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 599                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id, 'failed' => false]);
 600                         $max_users = max($apcontacts, $contacts);
 601                         if ($max_users > $serverdata['registered-users']) {
 602                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 603                                 self::update(['registered-users' => $max_users], ['id' => $id]);
 604                         }
 605
 606                         if (empty($serverdata['active-month-users'])) {
 607                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 30 days')]);
 608                                 if ($contacts > 0) {
 609                                         Logger::info('Update monthly users', ['id' => $id, 'url' => $serverdata['nurl'], 'monthly-users' => $contacts]);
 610                                         self::update(['active-month-users' => $contacts], ['id' => $id]);
 611                                 }
 612                         }
 613
 614                         if (empty($serverdata['active-halfyear-users'])) {
 615                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 180 days')]);
 616                                 if ($contacts > 0) {
 617                                         Logger::info('Update halfyear users', ['id' => $id, 'url' => $serverdata['nurl'], 'halfyear-users' => $contacts]);
 618                                         self::update(['active-halfyear-users' => $contacts], ['id' => $id]);
 619                                 }
 620                         }
 621                 }
 622
 623                 if (in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 624                         self::discoverRelay($url);
 625                 }
 626
 627                 if (!empty($systemactor)) {
 628                         $contact = Contact::getByURL($systemactor, true, ['gsid', 'baseurl', 'id', 'network', 'url', 'name']);
 629                         Logger::debug('Fetched system actor',  ['url' => $url, 'gsid' => $id, 'contact' => $contact]);
 630                 }
 631
 632                 return $ret;
 633         }
 634
 635         /**
 636          * Fetch relay data from a given server url
 637          *
 638          * @param string $server_url address of the server
 639          *
 640          * @return void
 641          *
 642          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 643          */
 644         private static function discoverRelay(string $server_url)
 645         {
 646                 Logger::info('Discover relay data', ['server' => $server_url]);
 647
 648                 $curlResult = DI::httpClient()->get($server_url . '/.well-known/x-social-relay', HttpClientAccept::JSON);
 649                 if (!$curlResult->isSuccess()) {
 650                         return;
 651                 }
 652
 653                 $data = json_decode($curlResult->getBody(), true);
 654                 if (!is_array($data)) {
 655                         return;
 656                 }
 657
 658                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 659                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
 660
 661                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 662                         $data['scope'] = '';
 663                         $data['subscribe'] = false;
 664                         $data['tags'] = [];
 665                 }
 666
 667                 $gserver = DBA::selectFirst('gserver', ['id', 'url', 'network', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 668                 if (!DBA::isResult($gserver)) {
 669                         return;
 670                 }
 671
 672                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 673                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 674                         self::update($fields, ['id' => $gserver['id']]);
 675                 }
 676
 677                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 678
 679                 if ($data['scope'] == 'tags') {
 680                         // Avoid duplicates
 681                         $tags = [];
 682                         foreach ($data['tags'] as $tag) {
 683                                 $tag = mb_strtolower($tag);
 684                                 if (strlen($tag) < 100) {
 685                                         $tags[$tag] = $tag;
 686                                 }
 687                         }
 688
 689                         foreach ($tags as $tag) {
 690                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], Database::INSERT_IGNORE);
 691                         }
 692                 }
 693
 694                 // Create or update the relay contact
 695                 $fields = [];
 696                 if (isset($data['protocols'])) {
 697                         if (isset($data['protocols']['diaspora'])) {
 698                                 $fields['network'] = Protocol::DIASPORA;
 699
 700                                 if (isset($data['protocols']['diaspora']['receive'])) {
 701                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 702                                 } elseif (is_string($data['protocols']['diaspora'])) {
 703                                         $fields['batch'] = $data['protocols']['diaspora'];
 704                                 }
 705                         }
 706
 707                         if (isset($data['protocols']['dfrn'])) {
 708                                 $fields['network'] = Protocol::DFRN;
 709
 710                                 if (isset($data['protocols']['dfrn']['receive'])) {
 711                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 712                                 } elseif (is_string($data['protocols']['dfrn'])) {
 713                                         $fields['batch'] = $data['protocols']['dfrn'];
 714                                 }
 715                         }
 716
 717                         if (isset($data['protocols']['activitypub'])) {
 718                                 $fields['network'] = Protocol::ACTIVITYPUB;
 719
 720                                 if (!empty($data['protocols']['activitypub']['actor'])) {
 721                                         $fields['url'] = $data['protocols']['activitypub']['actor'];
 722                                 }
 723                                 if (!empty($data['protocols']['activitypub']['receive'])) {
 724                                         $fields['batch'] = $data['protocols']['activitypub']['receive'];
 725                                 }
 726                         }
 727                 }
 728
 729                 Logger::info('Discovery ended', ['server' => $server_url, 'data' => $fields]);
 730
 731                 Relay::updateContact($gserver, $fields);
 732         }
 733
 734         /**
 735          * Fetch server data from '/statistics.json' on the given server
 736          *
 737          * @param string $url URL of the given server
 738          *
 739          * @return array server data
 740          */
 741         private static function fetchStatistics(string $url, array $serverdata): array
 742         {
 743                 $curlResult = DI::httpClient()->get($url . '/statistics.json', HttpClientAccept::JSON);
 744                 if (!$curlResult->isSuccess()) {
 745                         return $serverdata;
 746                 }
 747
 748                 $data = json_decode($curlResult->getBody(), true);
 749                 if (empty($data)) {
 750                         return $serverdata;
 751                 }
 752
 753                 // Some AP enabled systems return activity data that we don't expect here.
 754                 if (strpos($curlResult->getContentType(), 'application/activity+json') !== false) {
 755                         return $serverdata;
 756                 }
 757
 758                 $valid = false;
 759                 $old_serverdata = $serverdata;
 760
 761                 $serverdata['detection-method'] = self::DETECT_STATISTICS_JSON;
 762
 763                 if (!empty($data['version'])) {
 764                         $valid = true;
 765                         $serverdata['version'] = $data['version'];
 766                         // Version numbers on statistics.json are presented with additional info, e.g.:
 767                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 768                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 769                 }
 770
 771                 if (!empty($data['name'])) {
 772                         $valid = true;
 773                         $serverdata['site_name'] = $data['name'];
 774                 }
 775
 776                 if (!empty($data['network'])) {
 777                         $valid = true;
 778                         $serverdata['platform'] = strtolower($data['network']);
 779
 780                         if ($serverdata['platform'] == 'diaspora') {
 781                                 $serverdata['network'] = Protocol::DIASPORA;
 782                         } elseif ($serverdata['platform'] == 'friendica') {
 783                                 $serverdata['network'] = Protocol::DFRN;
 784                         } elseif ($serverdata['platform'] == 'hubzilla') {
 785                                 $serverdata['network'] = Protocol::ZOT;
 786                         } elseif ($serverdata['platform'] == 'redmatrix') {
 787                                 $serverdata['network'] = Protocol::ZOT;
 788                         }
 789                 }
 790
 791                 if (!empty($data['total_users'])) {
 792                         $valid = true;
 793                         $serverdata['registered-users'] = max($data['total_users'], 1);
 794                 }
 795
 796                 if (!empty($data['active_users_monthly'])) {
 797                         $valid = true;
 798                         $serverdata['active-month-users'] = max($data['active_users_monthly'], 0);
 799                 }
 800
 801                 if (!empty($data['active_users_halfyear'])) {
 802                         $valid = true;
 803                         $serverdata['active-halfyear-users'] = max($data['active_users_halfyear'], 0);
 804                 }
 805
 806                 if (!empty($data['local_posts'])) {
 807                         $valid = true;
 808                         $serverdata['local-posts'] = max($data['local_posts'], 0);
 809                 }
 810
 811                 if (!empty($data['registrations_open'])) {
 812                         $serverdata['register_policy'] = Register::OPEN;
 813                 } else {
 814                         $serverdata['register_policy'] = Register::CLOSED;
 815                 }
 816
 817                 if (!$valid) {
 818                         return $old_serverdata;
 819                 }
 820
 821                 return $serverdata;
 822         }
 823
 824         /**
 825          * Detect server type by using the nodeinfo data
 826          *
 827          * @param string                  $url        address of the server
 828          * @param ICanHandleHttpResponses $httpResult
 829          *
 830          * @return array Server data
 831          *
 832          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 833          */
 834         private static function fetchNodeinfo(string $url, ICanHandleHttpResponses $httpResult): array
 835         {
 836                 if (!$httpResult->isSuccess()) {
 837                         return [];
 838                 }
 839
 840                 $nodeinfo = json_decode($httpResult->getBody(), true);
 841
 842                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
 843                         return [];
 844                 }
 845
 846                 $nodeinfo1_url = '';
 847                 $nodeinfo2_url = '';
 848
 849                 foreach ($nodeinfo['links'] as $link) {
 850                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
 851                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
 852                                 continue;
 853                         }
 854                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 855                                 $nodeinfo1_url = $link['href'];
 856                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 857                                 $nodeinfo2_url = $link['href'];
 858                         }
 859                 }
 860
 861                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 862                         return [];
 863                 }
 864
 865                 $server = [];
 866
 867                 if (!empty($nodeinfo2_url)) {
 868                         $server = self::parseNodeinfo2($nodeinfo2_url);
 869                 }
 870
 871                 if (empty($server) && !empty($nodeinfo1_url)) {
 872                         $server = self::parseNodeinfo1($nodeinfo1_url);
 873                 }
 874
 875                 return $server;
 876         }
 877
 878         /**
 879          * Parses Nodeinfo 1
 880          *
 881          * @param string $nodeinfo_url address of the nodeinfo path
 882          *
 883          * @return array Server data
 884          *
 885          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 886          */
 887         private static function parseNodeinfo1(string $nodeinfo_url): array
 888         {
 889                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 890                 if (!$curlResult->isSuccess()) {
 891                         return [];
 892                 }
 893
 894                 $nodeinfo = json_decode($curlResult->getBody(), true);
 895
 896                 if (!is_array($nodeinfo)) {
 897                         return [];
 898                 }
 899
 900                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
 901                         'register_policy' => Register::CLOSED];
 902
 903                 if (!empty($nodeinfo['openRegistrations'])) {
 904                         $server['register_policy'] = Register::OPEN;
 905                 }
 906
 907                 if (is_array($nodeinfo['software'])) {
 908                         if (!empty($nodeinfo['software']['name'])) {
 909                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 910                         }
 911
 912                         if (!empty($nodeinfo['software']['version'])) {
 913                                 $server['version'] = $nodeinfo['software']['version'];
 914                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 915                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 916                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 917                         }
 918                 }
 919
 920                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 921                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 922                 }
 923
 924                 if (!empty($nodeinfo['usage']['users']['total'])) {
 925                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
 926                 }
 927
 928                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
 929                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
 930                 }
 931
 932                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
 933                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
 934                 }
 935
 936                 if (!empty($nodeinfo['usage']['localPosts'])) {
 937                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
 938                 }
 939
 940                 if (!empty($nodeinfo['usage']['localComments'])) {
 941                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
 942                 }
 943
 944                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
 945                         $protocols = [];
 946                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
 947                                 $protocols[$protocol] = true;
 948                         }
 949
 950                         if (!empty($protocols['friendica'])) {
 951                                 $server['network'] = Protocol::DFRN;
 952                         } elseif (!empty($protocols['activitypub'])) {
 953                                 $server['network'] = Protocol::ACTIVITYPUB;
 954                         } elseif (!empty($protocols['diaspora'])) {
 955                                 $server['network'] = Protocol::DIASPORA;
 956                         } elseif (!empty($protocols['ostatus'])) {
 957                                 $server['network'] = Protocol::OSTATUS;
 958                         } elseif (!empty($protocols['gnusocial'])) {
 959                                 $server['network'] = Protocol::OSTATUS;
 960                         } elseif (!empty($protocols['zot'])) {
 961                                 $server['network'] = Protocol::ZOT;
 962                         }
 963                 }
 964
 965                 if (empty($server)) {
 966                         return [];
 967                 }
 968
 969                 if (empty($server['network'])) {
 970                         $server['network'] = Protocol::PHANTOM;
 971                 }
 972
 973                 return $server;
 974         }
 975
 976         /**
 977          * Parses Nodeinfo 2
 978          *
 979          * @see https://git.feneas.org/jaywink/nodeinfo2
 980          *
 981          * @param string $nodeinfo_url address of the nodeinfo path
 982          *
 983          * @return array Server data
 984          *
 985          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 986          */
 987         private static function parseNodeinfo2(string $nodeinfo_url): array
 988         {
 989                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 990                 if (!$curlResult->isSuccess()) {
 991                         return [];
 992                 }
 993
 994                 $nodeinfo = json_decode($curlResult->getBody(), true);
 995                 if (!is_array($nodeinfo)) {
 996                         return [];
 997                 }
 998
 999                 $server = [
1000                         'detection-method' => self::DETECT_NODEINFO_2,
1001                         'register_policy' => Register::CLOSED,
1002                         'platform' => 'unknown',
1003                 ];
1004
1005                 if (!empty($nodeinfo['openRegistrations'])) {
1006                         $server['register_policy'] = Register::OPEN;
1007                 }
1008
1009                 if (!empty($nodeinfo['software'])) {
1010                         if (isset($nodeinfo['software']['name'])) {
1011                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
1012                         }
1013
1014                         if (!empty($nodeinfo['software']['version']) && isset($server['platform'])) {
1015                                 $server['version'] = $nodeinfo['software']['version'];
1016                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1017                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1018                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1019
1020                                 // qoto advertises itself as Mastodon
1021                                 if (($server['platform'] == 'mastodon') && substr($nodeinfo['software']['version'], -5) == '-qoto') {
1022                                         $server['platform'] = 'qoto';
1023                                 }
1024                         }
1025                 }
1026
1027                 if (!empty($nodeinfo['metadata']['nodeName'])) {
1028                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
1029                 }
1030
1031                 if (!empty($nodeinfo['usage']['users']['total'])) {
1032                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1033                 }
1034
1035                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1036                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1037                 }
1038
1039                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1040                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1041                 }
1042
1043                 if (!empty($nodeinfo['usage']['localPosts'])) {
1044                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1045                 }
1046
1047                 if (!empty($nodeinfo['usage']['localComments'])) {
1048                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1049                 }
1050
1051                 if (!empty($nodeinfo['protocols'])) {
1052                         $protocols = [];
1053                         foreach ($nodeinfo['protocols'] as $protocol) {
1054                                 if (is_string($protocol)) {
1055                                         $protocols[$protocol] = true;
1056                                 }
1057                         }
1058
1059                         if (!empty($protocols['dfrn'])) {
1060                                 $server['network'] = Protocol::DFRN;
1061                         } elseif (!empty($protocols['activitypub'])) {
1062                                 $server['network'] = Protocol::ACTIVITYPUB;
1063                         } elseif (!empty($protocols['diaspora'])) {
1064                                 $server['network'] = Protocol::DIASPORA;
1065                         } elseif (!empty($protocols['ostatus'])) {
1066                                 $server['network'] = Protocol::OSTATUS;
1067                         } elseif (!empty($protocols['gnusocial'])) {
1068                                 $server['network'] = Protocol::OSTATUS;
1069                         } elseif (!empty($protocols['zot'])) {
1070                                 $server['network'] = Protocol::ZOT;
1071                         }
1072                 }
1073
1074                 if (empty($server)) {
1075                         return [];
1076                 }
1077
1078                 if (empty($server['network'])) {
1079                         $server['network'] = Protocol::PHANTOM;
1080                 }
1081
1082                 return $server;
1083         }
1084
1085         /**
1086          * Parses NodeInfo2 protocol 1.0
1087          *
1088          * @see https://github.com/jaywink/nodeinfo2/blob/master/PROTOCOL.md
1089          *
1090          * @param string $nodeinfo_url address of the nodeinfo path
1091          *
1092          * @return array Server data
1093          *
1094          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1095          */
1096         private static function parseNodeinfo210(ICanHandleHttpResponses $httpResult): array
1097         {
1098                 if (!$httpResult->isSuccess()) {
1099                         return [];
1100                 }
1101
1102                 $nodeinfo = json_decode($httpResult->getBody(), true);
1103
1104                 if (!is_array($nodeinfo)) {
1105                         return [];
1106                 }
1107
1108                 $server = ['detection-method' => self::DETECT_NODEINFO_210,
1109                         'register_policy' => Register::CLOSED];
1110
1111                 if (!empty($nodeinfo['openRegistrations'])) {
1112                         $server['register_policy'] = Register::OPEN;
1113                 }
1114
1115                 if (!empty($nodeinfo['server'])) {
1116                         if (!empty($nodeinfo['server']['software'])) {
1117                                 $server['platform'] = strtolower($nodeinfo['server']['software']);
1118                         }
1119
1120                         if (!empty($nodeinfo['server']['version'])) {
1121                                 $server['version'] = $nodeinfo['server']['version'];
1122                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1123                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1124                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1125                         }
1126
1127                         if (!empty($nodeinfo['server']['name'])) {
1128                                 $server['site_name'] = $nodeinfo['server']['name'];
1129                         }
1130                 }
1131
1132                 if (!empty($nodeinfo['usage']['users']['total'])) {
1133                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1134                 }
1135
1136                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1137                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1138                 }
1139
1140                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1141                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1142                 }
1143
1144                 if (!empty($nodeinfo['usage']['localPosts'])) {
1145                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1146                 }
1147
1148                 if (!empty($nodeinfo['usage']['localComments'])) {
1149                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1150                 }
1151
1152                 if (!empty($nodeinfo['protocols'])) {
1153                         $protocols = [];
1154                         foreach ($nodeinfo['protocols'] as $protocol) {
1155                                 if (is_string($protocol)) {
1156                                         $protocols[$protocol] = true;
1157                                 }
1158                         }
1159
1160                         if (!empty($protocols['dfrn'])) {
1161                                 $server['network'] = Protocol::DFRN;
1162                         } elseif (!empty($protocols['activitypub'])) {
1163                                 $server['network'] = Protocol::ACTIVITYPUB;
1164                         } elseif (!empty($protocols['diaspora'])) {
1165                                 $server['network'] = Protocol::DIASPORA;
1166                         } elseif (!empty($protocols['ostatus'])) {
1167                                 $server['network'] = Protocol::OSTATUS;
1168                         } elseif (!empty($protocols['gnusocial'])) {
1169                                 $server['network'] = Protocol::OSTATUS;
1170                         } elseif (!empty($protocols['zot'])) {
1171                                 $server['network'] = Protocol::ZOT;
1172                         }
1173                 }
1174
1175                 if (empty($server) || empty($server['platform'])) {
1176                         return [];
1177                 }
1178
1179                 if (empty($server['network'])) {
1180                         $server['network'] = Protocol::PHANTOM;
1181                 }
1182
1183                 return $server;
1184         }
1185
1186         /**
1187          * Fetch server information from a 'siteinfo.json' file on the given server
1188          *
1189          * @param string $url        URL of the given server
1190          * @param array  $serverdata array with server data
1191          *
1192          * @return array server data
1193          */
1194         private static function fetchSiteinfo(string $url, array $serverdata): array
1195         {
1196                 $curlResult = DI::httpClient()->get($url . '/siteinfo.json', HttpClientAccept::JSON);
1197                 if (!$curlResult->isSuccess()) {
1198                         return $serverdata;
1199                 }
1200
1201                 $data = json_decode($curlResult->getBody(), true);
1202                 if (empty($data)) {
1203                         return $serverdata;
1204                 }
1205
1206                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1207                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
1208                 }
1209
1210                 if (!empty($data['url'])) {
1211                         $serverdata['platform'] = strtolower($data['platform']);
1212                         $serverdata['version'] = $data['version'];
1213                 }
1214
1215                 if (!empty($data['plugins'])) {
1216                         if (in_array('pubcrawl', $data['plugins'])) {
1217                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1218                         } elseif (in_array('diaspora', $data['plugins'])) {
1219                                 $serverdata['network'] = Protocol::DIASPORA;
1220                         } elseif (in_array('gnusoc', $data['plugins'])) {
1221                                 $serverdata['network'] = Protocol::OSTATUS;
1222                         } else {
1223                                 $serverdata['network'] = Protocol::ZOT;
1224                         }
1225                 }
1226
1227                 if (!empty($data['site_name'])) {
1228                         $serverdata['site_name'] = $data['site_name'];
1229                 }
1230
1231                 if (!empty($data['channels_total'])) {
1232                         $serverdata['registered-users'] = max($data['channels_total'], 1);
1233                 }
1234
1235                 if (!empty($data['channels_active_monthly'])) {
1236                         $serverdata['active-month-users'] = max($data['channels_active_monthly'], 0);
1237                 }
1238
1239                 if (!empty($data['channels_active_halfyear'])) {
1240                         $serverdata['active-halfyear-users'] = max($data['channels_active_halfyear'], 0);
1241                 }
1242
1243                 if (!empty($data['local_posts'])) {
1244                         $serverdata['local-posts'] = max($data['local_posts'], 0);
1245                 }
1246
1247                 if (!empty($data['local_comments'])) {
1248                         $serverdata['local-comments'] = max($data['local_comments'], 0);
1249                 }
1250
1251                 if (!empty($data['register_policy'])) {
1252                         switch ($data['register_policy']) {
1253                                 case 'REGISTER_OPEN':
1254                                         $serverdata['register_policy'] = Register::OPEN;
1255                                         break;
1256
1257                                 case 'REGISTER_APPROVE':
1258                                         $serverdata['register_policy'] = Register::APPROVE;
1259                                         break;
1260
1261                                 case 'REGISTER_CLOSED':
1262                                 default:
1263                                         $serverdata['register_policy'] = Register::CLOSED;
1264                                         break;
1265                         }
1266                 }
1267
1268                 return $serverdata;
1269         }
1270
1271         /**
1272          * Fetches server data via an ActivityPub account with url of that server
1273          *
1274          * @param string $url        URL of the given server
1275          * @param array  $serverdata array with server data
1276          *
1277          * @return array server data
1278          *
1279          * @throws Exception
1280          */
1281         private static function fetchDataFromSystemActor(array $data, array $serverdata): array
1282         {
1283                 if (empty($data)) {
1284                         return ['server' => $serverdata, 'actor' => ''];
1285                 }
1286
1287                 $actor = JsonLD::compact($data, false);
1288                 if (in_array(JsonLD::fetchElement($actor, '@type'), ActivityPub\Receiver::ACCOUNT_TYPES)) {
1289                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1290                         $serverdata['site_name'] = JsonLD::fetchElement($actor, 'as:name', '@value');
1291                         $serverdata['info'] = JsonLD::fetchElement($actor, 'as:summary', '@value');
1292                         if (!empty($actor['as:generator'])) {
1293                                 $generator = explode(' ', JsonLD::fetchElement($actor['as:generator'], 'as:name', '@value'));
1294                                 $serverdata['platform'] = strtolower(array_shift($generator));
1295                                 $serverdata['detection-method'] = self::DETECT_SYSTEM_ACTOR;
1296                         } else {
1297                                 $serverdata['detection-method'] = self::DETECT_AP_ACTOR;
1298                         }
1299                         return ['server' => $serverdata, 'actor' => $actor['@id']];
1300                 } elseif ((JsonLD::fetchElement($actor, '@type') == 'as:Collection')) {
1301                         // By now only Ktistec seems to provide collections this way
1302                         $serverdata['platform'] = 'ktistec';
1303                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1304                         $serverdata['detection-method'] = self::DETECT_AP_COLLECTION;
1305
1306                         $actors = JsonLD::fetchElementArray($actor, 'as:items');
1307                         if (!empty($actors) && !empty($actors[0]['@id'])) {
1308                                 $actor_url = $actor['@id'] . $actors[0]['@id'];
1309                         } else {
1310                                 $actor_url = '';
1311                         }
1312
1313                         return ['server' => $serverdata, 'actor' => $actor_url];
1314                 }
1315                 return ['server' => $serverdata, 'actor' => ''];
1316         }
1317
1318         /**
1319          * Checks if the server contains a valid host meta file
1320          *
1321          * @param string $url URL of the given server
1322          *
1323          * @return boolean 'true' if the server seems to be vital
1324          */
1325         private static function validHostMeta(string $url): bool
1326         {
1327                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
1328                 $curlResult = DI::httpClient()->get($url . '/.well-known/host-meta', HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
1329                 if (!$curlResult->isSuccess()) {
1330                         return false;
1331                 }
1332
1333                 $xrd = XML::parseString($curlResult->getBody());
1334                 if (!is_object($xrd)) {
1335                         return false;
1336                 }
1337
1338                 $elements = XML::elementToArray($xrd);
1339                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
1340                         return false;
1341                 }
1342
1343                 $valid = false;
1344                 foreach ($elements['xrd']['link'] as $link) {
1345                         // When there is more than a single "link" element, the array looks slightly different
1346                         if (!empty($link['@attributes'])) {
1347                                 $link = $link['@attributes'];
1348                         }
1349
1350                         if (empty($link['rel']) || empty($link['template'])) {
1351                                 continue;
1352                         }
1353
1354                         if ($link['rel'] == 'lrdd') {
1355                                 // When the webfinger host is the same like the system host, it should be ok.
1356                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
1357                         }
1358                 }
1359
1360                 return $valid;
1361         }
1362
1363         /**
1364          * Detect the network of the given server via their known contacts
1365          *
1366          * @param string $url        URL of the given server
1367          * @param array  $serverdata array with server data
1368          *
1369          * @return array server data
1370          */
1371         private static function detectNetworkViaContacts(string $url, array $serverdata): array
1372         {
1373                 $contacts = [];
1374
1375                 $nurl = Strings::normaliseLink($url);
1376
1377                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $nurl]]);
1378                 while ($apcontact = DBA::fetch($apcontacts)) {
1379                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
1380                 }
1381                 DBA::close($apcontacts);
1382
1383                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $nurl]]);
1384                 while ($pcontact = DBA::fetch($pcontacts)) {
1385                         $contacts[$pcontact['nurl']] = $pcontact['url'];
1386                 }
1387                 DBA::close($pcontacts);
1388
1389                 if (empty($contacts)) {
1390                         return $serverdata;
1391                 }
1392
1393                 $time = time();
1394                 foreach ($contacts as $contact) {
1395                         // Endlosschleife verhindern wegen gsid!
1396                         $data = Probe::uri($contact);
1397                         if (in_array($data['network'], Protocol::FEDERATED)) {
1398                                 $serverdata['network'] = $data['network'];
1399
1400                                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1401                                         $serverdata['detection-method'] = self::DETECT_CONTACTS;
1402                                 }
1403                                 break;
1404                         } elseif ((time() - $time) > 10) {
1405                                 // To reduce the stress on remote systems we probe a maximum of 10 seconds
1406                                 break;
1407                         }
1408                 }
1409
1410                 return $serverdata;
1411         }
1412
1413         /**
1414          * Checks if the given server does have a '/poco' endpoint.
1415          * This is used for the 'PortableContact' functionality,
1416          * which is used by both Friendica and Hubzilla.
1417          *
1418          * @param string $url        URL of the given server
1419          * @param array  $serverdata array with server data
1420          *
1421          * @return array server data
1422          */
1423         private static function checkPoCo(string $url, array $serverdata): array
1424         {
1425                 $serverdata['poco'] = '';
1426
1427                 $curlResult = DI::httpClient()->get($url . '/poco', HttpClientAccept::JSON);
1428                 if (!$curlResult->isSuccess()) {
1429                         return $serverdata;
1430                 }
1431
1432                 $data = json_decode($curlResult->getBody(), true);
1433                 if (empty($data)) {
1434                         return $serverdata;
1435                 }
1436
1437                 if (!empty($data['totalResults'])) {
1438                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1439                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers, 1);
1440                         $serverdata['directory-type'] = self::DT_POCO;
1441                         $serverdata['poco'] = $url . '/poco';
1442                 }
1443
1444                 return $serverdata;
1445         }
1446
1447         /**
1448          * Checks if the given server does have a Mastodon style directory endpoint.
1449          *
1450          * @param string $url        URL of the given server
1451          * @param array  $serverdata array with server data
1452          *
1453          * @return array server data
1454          */
1455         public static function checkMastodonDirectory(string $url, array $serverdata): array
1456         {
1457                 $curlResult = DI::httpClient()->get($url . '/api/v1/directory?limit=1', HttpClientAccept::JSON);
1458                 if (!$curlResult->isSuccess()) {
1459                         return $serverdata;
1460                 }
1461
1462                 $data = json_decode($curlResult->getBody(), true);
1463                 if (empty($data)) {
1464                         return $serverdata;
1465                 }
1466
1467                 if (count($data) == 1) {
1468                         $serverdata['directory-type'] = self::DT_MASTODON;
1469                 }
1470
1471                 return $serverdata;
1472         }
1473
1474         /**
1475          * Detects Peertube via their known endpoint
1476          *
1477          * @param string $url        URL of the given server
1478          * @param array  $serverdata array with server data
1479          *
1480          * @return array server data
1481          */
1482         private static function detectPeertube(string $url, array $serverdata): array
1483         {
1484                 $curlResult = DI::httpClient()->get($url . '/api/v1/config', HttpClientAccept::JSON);
1485                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1486                         return $serverdata;
1487                 }
1488
1489                 $data = json_decode($curlResult->getBody(), true);
1490                 if (empty($data)) {
1491                         return $serverdata;
1492                 }
1493
1494                 if (!empty($data['instance']) && !empty($data['serverVersion'])) {
1495                         $serverdata['platform'] = 'peertube';
1496                         $serverdata['version'] = $data['serverVersion'];
1497                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1498
1499                         if (!empty($data['instance']['name'])) {
1500                                 $serverdata['site_name'] = $data['instance']['name'];
1501                         }
1502
1503                         if (!empty($data['instance']['shortDescription'])) {
1504                                 $serverdata['info'] = $data['instance']['shortDescription'];
1505                         }
1506
1507                         if (!empty($data['signup'])) {
1508                                 if (!empty($data['signup']['allowed'])) {
1509                                         $serverdata['register_policy'] = Register::OPEN;
1510                                 }
1511                         }
1512
1513                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1514                                 $serverdata['detection-method'] = self::DETECT_V1_CONFIG;
1515                         }
1516                 }
1517
1518                 return $serverdata;
1519         }
1520
1521         /**
1522          * Detects the version number of a given server when it was a NextCloud installation
1523          *
1524          * @param string $url        URL of the given server
1525          * @param array  $serverdata array with server data
1526          * @param bool   $validHostMeta
1527          *
1528          * @return array server data
1529          */
1530         private static function detectNextcloud(string $url, array $serverdata, bool $validHostMeta): array
1531         {
1532                 $curlResult = DI::httpClient()->get($url . '/status.php', HttpClientAccept::JSON);
1533                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1534                         return $serverdata;
1535                 }
1536
1537                 $data = json_decode($curlResult->getBody(), true);
1538                 if (empty($data)) {
1539                         return $serverdata;
1540                 }
1541
1542                 if (!empty($data['version'])) {
1543                         $serverdata['platform'] = 'nextcloud';
1544                         $serverdata['version'] = $data['version'];
1545
1546                         if ($validHostMeta) {
1547                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1548                         }
1549
1550                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1551                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1552                         }
1553                 }
1554
1555                 return $serverdata;
1556         }
1557
1558         /**
1559          * Fetches weekly usage data
1560          *
1561          * @param string $url        URL of the given server
1562          * @param array  $serverdata array with server data
1563          *
1564          * @return array server data
1565          */
1566         private static function fetchWeeklyUsage(string $url, array $serverdata): array
1567         {
1568                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance/activity', HttpClientAccept::JSON);
1569                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1570                         return $serverdata;
1571                 }
1572
1573                 $data = json_decode($curlResult->getBody(), true);
1574                 if (empty($data)) {
1575                         return $serverdata;
1576                 }
1577
1578                 $current_week = [];
1579                 foreach ($data as $week) {
1580                         // Use only data from a full week
1581                         if (empty($week['week']) || (time() - $week['week']) < 7 * 24 * 60 * 60) {
1582                                 continue;
1583                         }
1584
1585                         // Most likely the data is sorted correctly. But we better are safe than sorry
1586                         if (empty($current_week['week']) || ($current_week['week'] < $week['week'])) {
1587                                 $current_week = $week;
1588                         }
1589                 }
1590
1591                 if (!empty($current_week['logins'])) {
1592                         $serverdata['active-week-users'] = max($current_week['logins'], 0);
1593                 }
1594
1595                 return $serverdata;
1596         }
1597
1598         /**
1599          * Detects data from a given server url if it was a mastodon alike system
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 detectMastodonAlikes(string $url, array $serverdata): array
1607         {
1608                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance', 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                 $valid = false;
1619
1620                 if (!empty($data['version'])) {
1621                         $serverdata['platform'] = 'mastodon';
1622                         $serverdata['version'] = $data['version'] ?? '';
1623                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1624                         $valid = true;
1625                 }
1626
1627                 if (!empty($data['title'])) {
1628                         $serverdata['site_name'] = $data['title'];
1629                 }
1630
1631                 if (!empty($data['title']) && empty($serverdata['platform']) && ($serverdata['network'] == Protocol::PHANTOM)) {
1632                         $serverdata['platform'] = 'mastodon';
1633                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1634                         $valid = true;
1635                 }
1636
1637                 if (!empty($data['description'])) {
1638                         $serverdata['info'] = trim($data['description']);
1639                 }
1640
1641                 if (!empty($data['stats']['user_count'])) {
1642                         $serverdata['registered-users'] = max($data['stats']['user_count'], 1);
1643                 }
1644
1645                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1646                         $serverdata['platform'] = strtolower($matches[1]);
1647                         $serverdata['version'] = $matches[2];
1648                         $valid = true;
1649                 }
1650
1651                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1652                         $serverdata['platform'] = 'pleroma';
1653                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1654                         $valid = true;
1655                 }
1656
1657                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1658                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1659                         $serverdata['platform'] = 'pleroma';
1660                         $valid = true;
1661                 }
1662
1663                 if ($valid && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1664                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1665                 }
1666
1667                 return $serverdata;
1668         }
1669
1670         /**
1671          * Detects data from typical Hubzilla endpoints
1672          *
1673          * @param string $url        URL of the given server
1674          * @param array  $serverdata array with server data
1675          *
1676          * @return array server data
1677          */
1678         private static function detectHubzilla(string $url, array $serverdata): array
1679         {
1680                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/config.json', HttpClientAccept::JSON);
1681                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1682                         return $serverdata;
1683                 }
1684
1685                 $data = json_decode($curlResult->getBody(), true);
1686                 if (empty($data) || empty($data['site'])) {
1687                         return $serverdata;
1688                 }
1689
1690                 if (!empty($data['site']['name'])) {
1691                         $serverdata['site_name'] = $data['site']['name'];
1692                 }
1693
1694                 if (!empty($data['site']['platform'])) {
1695                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1696                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1697                         $serverdata['network'] = Protocol::ZOT;
1698                 }
1699
1700                 if (!empty($data['site']['hubzilla'])) {
1701                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1702                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1703                         $serverdata['network'] = Protocol::ZOT;
1704                 }
1705
1706                 if (!empty($data['site']['redmatrix'])) {
1707                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1708                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1709                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1710                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1711                         }
1712
1713                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1714                         $serverdata['network'] = Protocol::ZOT;
1715                 }
1716
1717                 $private = false;
1718                 $inviteonly = false;
1719                 $closed = false;
1720
1721                 if (!empty($data['site']['closed'])) {
1722                         $closed = self::toBoolean($data['site']['closed']);
1723                 }
1724
1725                 if (!empty($data['site']['private'])) {
1726                         $private = self::toBoolean($data['site']['private']);
1727                 }
1728
1729                 if (!empty($data['site']['inviteonly'])) {
1730                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1731                 }
1732
1733                 if (!$closed && !$private and $inviteonly) {
1734                         $serverdata['register_policy'] = Register::APPROVE;
1735                 } elseif (!$closed && !$private) {
1736                         $serverdata['register_policy'] = Register::OPEN;
1737                 } else {
1738                         $serverdata['register_policy'] = Register::CLOSED;
1739                 }
1740
1741                 if (($serverdata['network'] != Protocol::PHANTOM) && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1742                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1743                 }
1744
1745                 return $serverdata;
1746         }
1747
1748         /**
1749          * Converts input value to a boolean value
1750          *
1751          * @param string|integer $val
1752          *
1753          * @return boolean
1754          */
1755         private static function toBoolean($val): bool
1756         {
1757                 if (($val == 'true') || ($val == 1)) {
1758                         return true;
1759                 } elseif (($val == 'false') || ($val == 0)) {
1760                         return false;
1761                 }
1762
1763                 return $val;
1764         }
1765
1766         /**
1767          * Detect if the URL belongs to a GNU Social server
1768          *
1769          * @param string $url        URL of the given server
1770          * @param array  $serverdata array with server data
1771          *
1772          * @return array server data
1773          */
1774         private static function detectGNUSocial(string $url, array $serverdata): array
1775         {
1776                 // Test for GNU Social
1777                 $curlResult = DI::httpClient()->get($url . '/api/gnusocial/version.json', HttpClientAccept::JSON);
1778                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1779                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1780                         $serverdata['platform'] = 'gnusocial';
1781                         // Remove junk that some GNU Social servers return
1782                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1783                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1784                         $serverdata['version'] = trim($serverdata['version'], '"');
1785                         $serverdata['network'] = Protocol::OSTATUS;
1786
1787                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1788                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1789                         }
1790
1791                         return $serverdata;
1792                 }
1793
1794                 // Test for Statusnet
1795                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/version.json', HttpClientAccept::JSON);
1796                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1797                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1798
1799                         // Remove junk that some GNU Social servers return
1800                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1801                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1802                         $serverdata['version'] = trim($serverdata['version'], '"');
1803
1804                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1805                                 $serverdata['platform'] = 'pleroma';
1806                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1807                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1808                         } else {
1809                                 $serverdata['platform'] = 'statusnet';
1810                                 $serverdata['network'] = Protocol::OSTATUS;
1811                         }
1812
1813                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1814                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1815                         }
1816                 }
1817
1818                 return $serverdata;
1819         }
1820
1821         /**
1822          * Detect if the URL belongs to a Friendica server
1823          *
1824          * @param string $url        URL of the given server
1825          * @param array  $serverdata array with server data
1826          *
1827          * @return array server data
1828          */
1829         private static function detectFriendica(string $url, array $serverdata): array
1830         {
1831                 // There is a bug in some versions of Friendica that will return an ActivityStream actor when the content type "application/json" is requested.
1832                 // Because of this me must not use ACCEPT_JSON here.
1833                 $curlResult = DI::httpClient()->get($url . '/friendica/json');
1834                 if (!$curlResult->isSuccess()) {
1835                         $curlResult = DI::httpClient()->get($url . '/friendika/json');
1836                         $friendika = true;
1837                         $platform = 'Friendika';
1838                 } else {
1839                         $friendika = false;
1840                         $platform = 'Friendica';
1841                 }
1842
1843                 if (!$curlResult->isSuccess()) {
1844                         return $serverdata;
1845                 }
1846
1847                 $data = json_decode($curlResult->getBody(), true);
1848                 if (empty($data) || empty($data['version'])) {
1849                         return $serverdata;
1850                 }
1851
1852                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1853                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1854                 }
1855
1856                 $serverdata['network'] = Protocol::DFRN;
1857                 $serverdata['version'] = $data['version'];
1858
1859                 if (!empty($data['no_scrape_url'])) {
1860                         $serverdata['noscrape'] = $data['no_scrape_url'];
1861                 }
1862
1863                 if (!empty($data['site_name'])) {
1864                         $serverdata['site_name'] = $data['site_name'];
1865                 }
1866
1867                 if (!empty($data['info'])) {
1868                         $serverdata['info'] = trim($data['info']);
1869                 }
1870
1871                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1872                 switch ($register_policy) {
1873                         case 'REGISTER_OPEN':
1874                                 $serverdata['register_policy'] = Register::OPEN;
1875                                 break;
1876
1877                         case 'REGISTER_APPROVE':
1878                                 $serverdata['register_policy'] = Register::APPROVE;
1879                                 break;
1880
1881                         case 'REGISTER_CLOSED':
1882                         case 'REGISTER_INVITATION':
1883                                 $serverdata['register_policy'] = Register::CLOSED;
1884                                 break;
1885                         default:
1886                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1887                                 $serverdata['register_policy'] = Register::CLOSED;
1888                                 break;
1889                 }
1890
1891                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1892
1893                 return $serverdata;
1894         }
1895
1896         /**
1897          * Analyses the landing page of a given server for hints about type and system of that server
1898          *
1899          * @param object $curlResult result of curl execution
1900          * @param array  $serverdata array with server data
1901          *
1902          * @return array server data
1903          */
1904         private static function analyseRootBody($curlResult, array $serverdata): array
1905         {
1906                 if (empty($curlResult->getBody())) {
1907                         return $serverdata;
1908                 }
1909
1910                 if (file_exists(__DIR__ . '/../../static/platforms.config.php')) {
1911                         require __DIR__ . '/../../static/platforms.config.php';
1912                 } else {
1913                         throw new HTTPException\InternalServerErrorException('Invalid platform file');
1914                 }
1915
1916                 $platforms = array_merge($ap_platforms, $dfrn_platforms, $zap_platforms, $platforms);
1917
1918                 $doc = new DOMDocument();
1919                 @$doc->loadHTML($curlResult->getBody());
1920                 $xpath = new DOMXPath($doc);
1921                 $assigned = false;
1922
1923                 // We can only detect honk via some HTML element on their page
1924                 if ($xpath->query('//div[@id="honksonpage"]')->count() == 1) {
1925                         $serverdata['platform'] = 'honk';
1926                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1927                         $assigned = true;
1928                 }
1929
1930                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1931                 if (!empty($title)) {
1932                         $serverdata['site_name'] = $title;
1933                 }
1934
1935                 $list = $xpath->query('//meta[@name]');
1936
1937                 foreach ($list as $node) {
1938                         $attr = [];
1939                         if ($node->attributes->length) {
1940                                 foreach ($node->attributes as $attribute) {
1941                                         $value = trim($attribute->value);
1942                                         if (empty($value)) {
1943                                                 continue;
1944                                         }
1945
1946                                         $attr[$attribute->name] = $value;
1947                                 }
1948
1949                                 if (empty($attr['name']) || empty($attr['content'])) {
1950                                         continue;
1951                                 }
1952                         }
1953
1954                         if ($attr['name'] == 'description') {
1955                                 $serverdata['info'] = $attr['content'];
1956                         }
1957
1958                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1959                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad', 'generator'])) {
1960                                 $platform = str_ireplace(array_keys($platforms), array_values($platforms), $attr['content']);
1961                                 $platform = str_replace('/', ' ', $platform);
1962                                 $platform_parts = explode(' ', $platform);
1963                                 if ((count($platform_parts) >= 2) && in_array(strtolower($platform_parts[0]), array_values($platforms))) {
1964                                         $platform = $platform_parts[0];
1965                                         $serverdata['version'] = $platform_parts[1];
1966                                 }
1967                                 if (in_array($platform, array_values($dfrn_platforms))) {
1968                                         $serverdata['network'] = Protocol::DFRN;
1969                                 } elseif (in_array($platform, array_values($ap_platforms))) {
1970                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1971                                 } elseif (in_array($platform, array_values($zap_platforms))) {
1972                                         $serverdata['network'] = Protocol::ZOT;
1973                                 }
1974                                 if (in_array($platform, array_values($platforms))) {
1975                                         $serverdata['platform'] = $platform;
1976                                         $assigned = true;
1977                                 }
1978                         }
1979                 }
1980
1981                 $list = $xpath->query('//meta[@property]');
1982
1983                 foreach ($list as $node) {
1984                         $attr = [];
1985                         if ($node->attributes->length) {
1986                                 foreach ($node->attributes as $attribute) {
1987                                         $value = trim($attribute->value);
1988                                         if (empty($value)) {
1989                                                 continue;
1990                                         }
1991
1992                                         $attr[$attribute->name] = $value;
1993                                 }
1994
1995                                 if (empty($attr['property']) || empty($attr['content'])) {
1996                                         continue;
1997                                 }
1998                         }
1999
2000                         if ($attr['property'] == 'og:site_name') {
2001                                 $serverdata['site_name'] = $attr['content'];
2002                         }
2003
2004                         if ($attr['property'] == 'og:description') {
2005                                 $serverdata['info'] = $attr['content'];
2006                         }
2007
2008                         if (in_array($attr['property'], ['og:platform', 'generator'])) {
2009                                 if (in_array($attr['content'], array_keys($platforms))) {
2010                                         $serverdata['platform'] = $platforms[$attr['content']];
2011                                         $assigned = true;
2012                                 }
2013
2014                                 if (in_array($attr['content'], array_keys($ap_platforms))) {
2015                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2016                                 } elseif (in_array($attr['content'], array_values($zap_platforms))) {
2017                                         $serverdata['network'] = Protocol::ZOT;
2018                                 }
2019                         }
2020                 }
2021
2022                 $list = $xpath->query('//link[@rel="me"]');
2023                 foreach ($list as $node) {
2024                         foreach ($node->attributes as $attribute) {
2025                                 if (parse_url(trim($attribute->value), PHP_URL_HOST) == 'micro.blog') {
2026                                         $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2027                                         $serverdata['platform'] = 'microblog';
2028                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2029                                         $assigned = true;
2030                                 }
2031                         }
2032                 }
2033
2034                 if ($serverdata['platform'] != 'microblog') {
2035                         $list = $xpath->query('//link[@rel="micropub"]');
2036                         foreach ($list as $node) {
2037                                 foreach ($node->attributes as $attribute) {
2038                                         if (trim($attribute->value) == 'https://micro.blog/micropub') {
2039                                                 $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2040                                                 $serverdata['platform'] = 'microblog';
2041                                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
2042                                                 $assigned = true;
2043                                         }
2044                                 }
2045                         }
2046                 }
2047
2048                 if ($assigned && in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER])) {
2049                         $serverdata['detection-method'] = self::DETECT_BODY;
2050                 }
2051
2052                 return $serverdata;
2053         }
2054
2055         /**
2056          * Analyses the header data of a given server for hints about type and system of that server
2057          *
2058          * @param object $curlResult result of curl execution
2059          * @param array  $serverdata array with server data
2060          *
2061          * @return array server data
2062          */
2063         private static function analyseRootHeader($curlResult, array $serverdata): array
2064         {
2065                 if ($curlResult->getHeader('server') == 'Mastodon') {
2066                         $serverdata['platform'] = 'mastodon';
2067                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2068                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
2069                         $serverdata['platform'] = 'diaspora';
2070                         $serverdata['network'] = Protocol::DIASPORA;
2071                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version')[0] ?? '';
2072                 } elseif ($curlResult->inHeader('x-friendica-version')) {
2073                         $serverdata['platform'] = 'friendica';
2074                         $serverdata['network'] = Protocol::DFRN;
2075                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version')[0] ?? '';
2076                 } else {
2077                         return $serverdata;
2078                 }
2079
2080                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
2081                         $serverdata['detection-method'] = self::DETECT_HEADER;
2082                 }
2083
2084                 return $serverdata;
2085         }
2086
2087         /**
2088          * Update GServer entries
2089          */
2090         public static function discover()
2091         {
2092                 // Update the server list
2093                 self::discoverFederation();
2094
2095                 $no_of_queries = 5;
2096
2097                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
2098
2099                 if ($requery_days == 0) {
2100                         $requery_days = 7;
2101                 }
2102
2103                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
2104
2105                 $gservers = DBA::select('gserver', ['id', 'url', 'nurl', 'network', 'poco', 'directory-type'],
2106                         ["NOT `failed` AND `directory-type` != ? AND `last_poco_query` < ?", GServer::DT_NONE, $last_update],
2107                         ['order' => ['RAND()']]);
2108
2109                 while ($gserver = DBA::fetch($gservers)) {
2110                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2111                         Worker::add(PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
2112
2113                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2114                         Worker::add(PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
2115
2116                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
2117                         self::update($fields, ['nurl' => $gserver['nurl']]);
2118
2119                         if (--$no_of_queries == 0) {
2120                                 break;
2121                         }
2122                 }
2123
2124                 DBA::close($gservers);
2125         }
2126
2127         /**
2128          * Discover federated servers
2129          */
2130         private static function discoverFederation()
2131         {
2132                 $last = DI::config()->get('poco', 'last_federation_discovery');
2133
2134                 if ($last) {
2135                         $next = $last + (24 * 60 * 60);
2136
2137                         if ($next > time()) {
2138                                 return;
2139                         }
2140                 }
2141
2142                 // Discover federated servers
2143                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
2144                 foreach ($protocols as $protocol) {
2145                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
2146                         $curlResult = DI::httpClient()->fetch('https://the-federation.info/graphql?query=' . urlencode($query), HttpClientAccept::JSON);
2147                         if (!empty($curlResult)) {
2148                                 $data = json_decode($curlResult, true);
2149                                 if (!empty($data['data']['nodes'])) {
2150                                         foreach ($data['data']['nodes'] as $server) {
2151                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
2152                                                 self::add('https://' . $server['host'], true);
2153                                         }
2154                                 }
2155                         }
2156                 }
2157
2158                 // Disvover Mastodon servers
2159                 $accesstoken = DI::config()->get('system', 'instances_social_key');
2160
2161                 if (!empty($accesstoken)) {
2162                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
2163                         $curlResult = DI::httpClient()->get($api, HttpClientAccept::JSON, [HttpClientOptions::HEADERS => ['Authorization' => ['Bearer ' . $accesstoken]]]);
2164                         if ($curlResult->isSuccess()) {
2165                                 $servers = json_decode($curlResult->getBody(), true);
2166
2167                                 foreach ($servers['instances'] as $server) {
2168                                         $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
2169                                         self::add($url);
2170                                 }
2171                         }
2172                 }
2173
2174                 DI::config()->set('poco', 'last_federation_discovery', time());
2175         }
2176
2177         /**
2178          * Set the protocol for the given server
2179          *
2180          * @param int $gsid     Server id
2181          * @param int $protocol Protocol id
2182          *
2183          * @throws Exception
2184          */
2185         public static function setProtocol(int $gsid, int $protocol)
2186         {
2187                 if (empty($gsid)) {
2188                         return;
2189                 }
2190
2191                 $gserver = DBA::selectFirst('gserver', ['protocol', 'url'], ['id' => $gsid]);
2192                 if (!DBA::isResult($gserver)) {
2193                         return;
2194                 }
2195
2196                 $old = $gserver['protocol'];
2197
2198                 if (!is_null($old)) {
2199                         /*
2200                         The priority for the protocols is:
2201                                 1. ActivityPub
2202                                 2. DFRN via Diaspora
2203                                 3. Legacy DFRN
2204                                 4. Diaspora
2205                                 5. OStatus
2206                         */
2207
2208                         // We don't need to change it when nothing is to be changed
2209                         if ($old == $protocol) {
2210                                 return;
2211                         }
2212
2213                         // We don't want to mark a server as OStatus when it had been marked with any other protocol before
2214                         if ($protocol == Post\DeliveryData::OSTATUS) {
2215                                 return;
2216                         }
2217
2218                         // If the server is marked as ActivityPub then we won't change it to anything different
2219                         if ($old == Post\DeliveryData::ACTIVITYPUB) {
2220                                 return;
2221                         }
2222
2223                         // Don't change it to anything lower than DFRN if the new one wasn't ActivityPub
2224                         if (($old == Post\DeliveryData::DFRN) && ($protocol != Post\DeliveryData::ACTIVITYPUB)) {
2225                                 return;
2226                         }
2227
2228                         // Don't change it to Diaspora when it is a legacy DFRN server
2229                         if (($old == Post\DeliveryData::LEGACY_DFRN) && ($protocol == Post\DeliveryData::DIASPORA)) {
2230                                 return;
2231                         }
2232                 }
2233
2234                 Logger::info('Protocol for server', ['protocol' => $protocol, 'old' => $old, 'id' => $gsid, 'url' => $gserver['url'], 'callstack' => System::callstack(20)]);
2235                 self::update(['protocol' => $protocol], ['id' => $gsid]);
2236         }
2237
2238         /**
2239          * Fetch the protocol of the given server
2240          *
2241          * @param int $gsid Server id
2242          *
2243          * @return ?int One of Post\DeliveryData protocol constants or null if unknown or gserver is missing
2244          *
2245          * @throws Exception
2246          */
2247         public static function getProtocol(int $gsid): ?int
2248         {
2249                 if (empty($gsid)) {
2250                         return null;
2251                 }
2252
2253                 $gserver = DBA::selectFirst('gserver', ['protocol'], ['id' => $gsid]);
2254                 if (DBA::isResult($gserver)) {
2255                         return $gserver['protocol'];
2256                 }
2257
2258                 return null;
2259         }
2260
2261         /**
2262          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2263          *
2264          * @param array $fields
2265          * @param array $condition
2266          *
2267          * @return bool
2268          *
2269          * @throws Exception
2270          */
2271         public static function update(array $fields, array $condition): bool
2272         {
2273                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2274
2275                 return DBA::update('gserver', $fields, $condition);
2276         }
2277 }