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(Worker::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                 $url = self::cleanURL($url);
 119
 120                 if (empty($url)) {
 121                         return null;
 122                 }
 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                 self::insert(['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                 if (empty($urlparts)) {
 327                         return '';
 328                 }
 329
 330                 unset($urlparts['user']);
 331                 unset($urlparts['pass']);
 332                 unset($urlparts['query']);
 333                 unset($urlparts['fragment']);
 334                 return (string)Uri::fromParts($urlparts);
 335         }
 336
 337         /**
 338          * Detect server data (type, protocol, version number, ...)
 339          * The detected data is then updated or inserted in the gserver table.
 340          *
 341          * @param string  $url           URL of the given server
 342          * @param string  $network       Network value that is used, when detection failed
 343          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 344          *
 345          * @return boolean 'true' if server could be detected
 346          */
 347         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false): bool
 348         {
 349                 Logger::info('Detect server type', ['server' => $url]);
 350
 351                 $original_url = $url;
 352
 353                 // Remove URL content that is not supposed to exist for a server url
 354                 $url = rtrim(self::cleanURL($url), '/');
 355                 if (empty($url)) {
 356                         Logger::notice('Empty URL.');
 357                         return false;
 358                 }
 359
 360                 // If the URL missmatches, then we mark the old entry as failure
 361                 if (!Strings::compareLink($url, $original_url)) {
 362                         self::setFailure($original_url);
 363                         if (!self::getID($url, true)) {
 364                                 self::detect($url, $network, $only_nodeinfo);
 365                         }
 366                         return false;
 367                 }
 368
 369                 $valid_url = Network::isUrlValid($url);
 370                 if (!$valid_url) {
 371                         self::setFailure($url);
 372                         return false;
 373                 } else {
 374                         $valid_url = rtrim($valid_url, '/');
 375                 }
 376
 377                 if (!Strings::compareLink($url, $valid_url)) {
 378                         // We only follow redirects when the path stays the same or the target url has no path.
 379                         // Some systems have got redirects on their landing page to a single account page. This check handles it.
 380                         if (((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH))) ||
 381                                 (((parse_url($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)))) {
 382                                 Logger::debug('Found redirect. Mark old entry as failure', ['old' => $url, 'new' => $valid_url]);
 383                                 self::setFailure($url);
 384                                 if (!self::getID($valid_url, true)) {
 385                                         self::detect($valid_url, $network, $only_nodeinfo);
 386                                 }
 387                                 return false;
 388                         }
 389
 390                         if ((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) != parse_url($valid_url, PHP_URL_PATH)) &&
 391                                 (parse_url($url, PHP_URL_PATH) == '')) {
 392                                 Logger::debug('Found redirect. Mark old entry as failure and redirect to the basepath.', ['old' => $url, 'new' => $valid_url]);
 393                                 $parts = parse_url($valid_url);
 394                                 unset($parts['path']);
 395                                 $valid_url = (string)Uri::fromParts($parts);
 396
 397                                 self::setFailure($url);
 398                                 if (!self::getID($valid_url, true)) {
 399                                         self::detect($valid_url, $network, $only_nodeinfo);
 400                                 }
 401                                 return false;
 402                         }
 403                         Logger::debug('Found redirect, but ignore it.', ['old' => $url, 'new' => $valid_url]);
 404                 }
 405
 406                 if ((parse_url($url, PHP_URL_HOST) == parse_url($valid_url, PHP_URL_HOST)) &&
 407                         (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH)) &&
 408                         (parse_url($url, PHP_URL_SCHEME) != parse_url($valid_url, PHP_URL_SCHEME))) {
 409                         $url = $valid_url;
 410                 }
 411
 412                 $in_webroot = empty(parse_url($url, PHP_URL_PATH));
 413
 414                 // When a nodeinfo is present, we don't need to dig further
 415                 $curlResult = DI::httpClient()->get($url . '/.well-known/x-nodeinfo2', HttpClientAccept::JSON);
 416                 if ($curlResult->isTimeout()) {
 417                         self::setFailure($url);
 418                         return false;
 419                 }
 420
 421                 $serverdata = self::parseNodeinfo210($curlResult);
 422                 if (empty($serverdata)) {
 423                         $curlResult = DI::httpClient()->get($url . '/.well-known/nodeinfo', HttpClientAccept::JSON);
 424                         $serverdata = self::fetchNodeinfo($url, $curlResult);
 425                 }
 426
 427                 if ($only_nodeinfo && empty($serverdata)) {
 428                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 429                         self::setFailure($url);
 430                         return false;
 431                 } elseif (empty($serverdata)) {
 432                         $serverdata = ['detection-method' => self::DETECT_MANUAL, 'network' => Protocol::PHANTOM, 'platform' => '', 'version' => '', 'site_name' => '', 'info' => ''];
 433                 }
 434
 435                 // When there is no Nodeinfo, then use some protocol specific endpoints
 436                 if ($serverdata['network'] == Protocol::PHANTOM) {
 437                         if ($in_webroot) {
 438                                 // Fetch the landing page, possibly it reveals some data
 439                                 $accept = 'application/activity+json,application/ld+json,application/json,*/*;q=0.9';
 440                                 $curlResult = DI::httpClient()->get($url, $accept);
 441                                 if (!$curlResult->isSuccess() && $curlResult->getReturnCode() == '406') {
 442                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 443                                         $html_fetched = true;
 444                                 } else {
 445                                         $html_fetched = false;
 446                                 }
 447
 448                                 if ($curlResult->isSuccess()) {
 449                                         $json = json_decode($curlResult->getBody(), true);
 450                                         if (!empty($json) && is_array($json)) {
 451                                                 $data = self::fetchDataFromSystemActor($json, $serverdata);
 452                                                 $serverdata = $data['server'];
 453                                                 $systemactor = $data['actor'];
 454                                                 if (!$html_fetched && !in_array($serverdata['detection-method'], [self::DETECT_SYSTEM_ACTOR, self::DETECT_AP_COLLECTION])) {
 455                                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 456                                                 }
 457                                         } elseif (!$html_fetched && (strlen($curlResult->getBody()) < 1000)) {
 458                                                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 459                                         }
 460
 461                                         if ($serverdata['detection-method'] != self::DETECT_SYSTEM_ACTOR) {
 462                                                 $serverdata = self::analyseRootHeader($curlResult, $serverdata);
 463                                                 $serverdata = self::analyseRootBody($curlResult, $serverdata);
 464                                         }
 465                                 }
 466
 467                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
 468                                         self::setFailure($url);
 469                                         return false;
 470                                 }
 471
 472                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 473                                         $serverdata = self::detectMastodonAlikes($url, $serverdata);
 474                                 }
 475                         }
 476
 477                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 478                         // With this check we don't have to waste time and ressources for dead systems.
 479                         // Also this hopefully prevents us from receiving abuse messages.
 480                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 481                                 $validHostMeta = self::validHostMeta($url);
 482                         } else {
 483                                 $validHostMeta = false;
 484                         }
 485
 486                         if ($validHostMeta) {
 487                                 if (in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY])) {
 488                                         $serverdata['detection-method'] = self::DETECT_HOST_META;
 489                                 }
 490
 491                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 492                                         $serverdata = self::detectFriendica($url, $serverdata);
 493                                 }
 494
 495                                 // The following systems have to be installed in the root directory.
 496                                 if ($in_webroot) {
 497                                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 498                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 499                                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 500                                         }
 501
 502                                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations, so we check other endpoints as well
 503                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 504                                                 $serverdata = self::detectHubzilla($url, $serverdata);
 505                                         }
 506
 507                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 508                                                 $serverdata = self::detectPeertube($url, $serverdata);
 509                                         }
 510
 511                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 512                                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 513                                         }
 514                                 }
 515                         } elseif (in_array($serverdata['platform'], ['friendica', 'friendika']) && in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_SYSTEM_ACTOR]))) {
 516                                 $serverdata = self::detectFriendica($url, $serverdata);
 517                         }
 518
 519                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 520                                 $serverdata = self::detectNextcloud($url, $serverdata, $validHostMeta);
 521                         }
 522
 523                         // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 524                         // Since this endpoint is only rarely used, we query it at a later time
 525                         if (in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_FRIENDICA, self::DETECT_CONFIG_JSON]))) {
 526                                 $serverdata = self::fetchStatistics($url, $serverdata);
 527                         }
 528                 }
 529
 530                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 531                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 532                         $serverdata['network'] = $network;
 533                 }
 534
 535                 // Most servers aren't installed in a subdirectory, so we declare this entry as failed
 536                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty(parse_url($url, PHP_URL_PATH)) && in_array($serverdata['detection-method'], [self::DETECT_MANUAL])) {
 537                         self::setFailure($url);
 538                         return false;
 539                 }
 540
 541                 $serverdata['url'] = $url;
 542                 $serverdata['nurl'] = Strings::normaliseLink($url);
 543
 544                 // We have to prevent an endless loop here.
 545                 // When a server is new, then there is no gserver entry yet.
 546                 // But in "detectNetworkViaContacts" it could happen that a contact is updated,
 547                 // and this can call this function here as well.
 548                 if (self::getID($url, true) && (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) ||
 549                         in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META]))) {
 550                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 551                 }
 552
 553                 // Detect the directory type
 554                 $serverdata['directory-type'] = self::DT_NONE;
 555
 556                 if (in_array($serverdata['network'], Protocol::FEDERATED)) {
 557                         $serverdata = self::checkMastodonDirectory($url, $serverdata);
 558
 559                         if ($serverdata['directory-type'] == self::DT_NONE) {
 560                                 $serverdata = self::checkPoCo($url, $serverdata);
 561                         }
 562                 }
 563
 564                 if ($serverdata['network'] == Protocol::ACTIVITYPUB) {
 565                         $serverdata = self::fetchWeeklyUsage($url, $serverdata);
 566                 }
 567
 568                 $serverdata['registered-users'] = $serverdata['registered-users'] ?? 0;
 569
 570                 // Numbers above a reasonable value (10 millions) are ignored
 571                 if ($serverdata['registered-users'] > 10000000) {
 572                         $serverdata['registered-users'] = 0;
 573                 }
 574
 575                 // On an active server there has to be at least a single user
 576                 if (!in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) && ($serverdata['registered-users'] <= 0)) {
 577                         $serverdata['registered-users'] = 1;
 578                 } elseif (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 579                         $serverdata['registered-users'] = 0;
 580                 }
 581
 582                 $serverdata['next_contact'] = self::getNextUpdateDate(true, '', '', in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]));
 583
 584                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 585                 $serverdata['failed'] = false;
 586
 587                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 588                 if (!DBA::isResult($gserver)) {
 589                         $serverdata['created'] = DateTimeFormat::utcNow();
 590                         $ret = self::insert($serverdata);
 591                         $id = DBA::lastInsertId();
 592                 } else {
 593                         $ret = self::update($serverdata, ['nurl' => $serverdata['nurl']]);
 594                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 595                         if (DBA::isResult($gserver)) {
 596                                 $id = $gserver['id'];
 597                         }
 598                 }
 599
 600                 // Count the number of known contacts from this server
 601                 if (!empty($id) && !in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 602                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 603                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id, 'failed' => false]);
 604                         $max_users = max($apcontacts, $contacts);
 605                         if ($max_users > $serverdata['registered-users']) {
 606                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 607                                 self::update(['registered-users' => $max_users], ['id' => $id]);
 608                         }
 609
 610                         if (empty($serverdata['active-month-users'])) {
 611                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 30 days')]);
 612                                 if ($contacts > 0) {
 613                                         Logger::info('Update monthly users', ['id' => $id, 'url' => $serverdata['nurl'], 'monthly-users' => $contacts]);
 614                                         self::update(['active-month-users' => $contacts], ['id' => $id]);
 615                                 }
 616                         }
 617
 618                         if (empty($serverdata['active-halfyear-users'])) {
 619                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 180 days')]);
 620                                 if ($contacts > 0) {
 621                                         Logger::info('Update halfyear users', ['id' => $id, 'url' => $serverdata['nurl'], 'halfyear-users' => $contacts]);
 622                                         self::update(['active-halfyear-users' => $contacts], ['id' => $id]);
 623                                 }
 624                         }
 625                 }
 626
 627                 if (in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 628                         self::discoverRelay($url);
 629                 }
 630
 631                 if (!empty($systemactor)) {
 632                         $contact = Contact::getByURL($systemactor, true, ['gsid', 'baseurl', 'id', 'network', 'url', 'name']);
 633                         Logger::debug('Fetched system actor',  ['url' => $url, 'gsid' => $id, 'contact' => $contact]);
 634                 }
 635
 636                 return $ret;
 637         }
 638
 639         /**
 640          * Fetch relay data from a given server url
 641          *
 642          * @param string $server_url address of the server
 643          *
 644          * @return void
 645          *
 646          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 647          */
 648         private static function discoverRelay(string $server_url)
 649         {
 650                 Logger::info('Discover relay data', ['server' => $server_url]);
 651
 652                 $curlResult = DI::httpClient()->get($server_url . '/.well-known/x-social-relay', HttpClientAccept::JSON);
 653                 if (!$curlResult->isSuccess()) {
 654                         return;
 655                 }
 656
 657                 $data = json_decode($curlResult->getBody(), true);
 658                 if (!is_array($data)) {
 659                         return;
 660                 }
 661
 662                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 663                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
 664
 665                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 666                         $data['scope'] = '';
 667                         $data['subscribe'] = false;
 668                         $data['tags'] = [];
 669                 }
 670
 671                 $gserver = DBA::selectFirst('gserver', ['id', 'url', 'network', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 672                 if (!DBA::isResult($gserver)) {
 673                         return;
 674                 }
 675
 676                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 677                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 678                         self::update($fields, ['id' => $gserver['id']]);
 679                 }
 680
 681                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 682
 683                 if ($data['scope'] == 'tags') {
 684                         // Avoid duplicates
 685                         $tags = [];
 686                         foreach ($data['tags'] as $tag) {
 687                                 $tag = mb_strtolower($tag);
 688                                 if (strlen($tag) < 100) {
 689                                         $tags[$tag] = $tag;
 690                                 }
 691                         }
 692
 693                         foreach ($tags as $tag) {
 694                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], Database::INSERT_IGNORE);
 695                         }
 696                 }
 697
 698                 // Create or update the relay contact
 699                 $fields = [];
 700                 if (isset($data['protocols'])) {
 701                         if (isset($data['protocols']['diaspora'])) {
 702                                 $fields['network'] = Protocol::DIASPORA;
 703
 704                                 if (isset($data['protocols']['diaspora']['receive'])) {
 705                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 706                                 } elseif (is_string($data['protocols']['diaspora'])) {
 707                                         $fields['batch'] = $data['protocols']['diaspora'];
 708                                 }
 709                         }
 710
 711                         if (isset($data['protocols']['dfrn'])) {
 712                                 $fields['network'] = Protocol::DFRN;
 713
 714                                 if (isset($data['protocols']['dfrn']['receive'])) {
 715                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 716                                 } elseif (is_string($data['protocols']['dfrn'])) {
 717                                         $fields['batch'] = $data['protocols']['dfrn'];
 718                                 }
 719                         }
 720
 721                         if (isset($data['protocols']['activitypub'])) {
 722                                 $fields['network'] = Protocol::ACTIVITYPUB;
 723
 724                                 if (!empty($data['protocols']['activitypub']['actor'])) {
 725                                         $fields['url'] = $data['protocols']['activitypub']['actor'];
 726                                 }
 727                                 if (!empty($data['protocols']['activitypub']['receive'])) {
 728                                         $fields['batch'] = $data['protocols']['activitypub']['receive'];
 729                                 }
 730                         }
 731                 }
 732
 733                 Logger::info('Discovery ended', ['server' => $server_url, 'data' => $fields]);
 734
 735                 Relay::updateContact($gserver, $fields);
 736         }
 737
 738         /**
 739          * Fetch server data from '/statistics.json' on the given server
 740          *
 741          * @param string $url URL of the given server
 742          *
 743          * @return array server data
 744          */
 745         private static function fetchStatistics(string $url, array $serverdata): array
 746         {
 747                 $curlResult = DI::httpClient()->get($url . '/statistics.json', HttpClientAccept::JSON);
 748                 if (!$curlResult->isSuccess()) {
 749                         return $serverdata;
 750                 }
 751
 752                 $data = json_decode($curlResult->getBody(), true);
 753                 if (empty($data)) {
 754                         return $serverdata;
 755                 }
 756
 757                 // Some AP enabled systems return activity data that we don't expect here.
 758                 if (strpos($curlResult->getContentType(), 'application/activity+json') !== false) {
 759                         return $serverdata;
 760                 }
 761
 762                 $valid = false;
 763                 $old_serverdata = $serverdata;
 764
 765                 $serverdata['detection-method'] = self::DETECT_STATISTICS_JSON;
 766
 767                 if (!empty($data['version'])) {
 768                         $valid = true;
 769                         $serverdata['version'] = $data['version'];
 770                         // Version numbers on statistics.json are presented with additional info, e.g.:
 771                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 772                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 773                 }
 774
 775                 if (!empty($data['name'])) {
 776                         $valid = true;
 777                         $serverdata['site_name'] = $data['name'];
 778                 }
 779
 780                 if (!empty($data['network'])) {
 781                         $valid = true;
 782                         $serverdata['platform'] = strtolower($data['network']);
 783
 784                         if ($serverdata['platform'] == 'diaspora') {
 785                                 $serverdata['network'] = Protocol::DIASPORA;
 786                         } elseif ($serverdata['platform'] == 'friendica') {
 787                                 $serverdata['network'] = Protocol::DFRN;
 788                         } elseif ($serverdata['platform'] == 'hubzilla') {
 789                                 $serverdata['network'] = Protocol::ZOT;
 790                         } elseif ($serverdata['platform'] == 'redmatrix') {
 791                                 $serverdata['network'] = Protocol::ZOT;
 792                         }
 793                 }
 794
 795                 if (!empty($data['total_users'])) {
 796                         $valid = true;
 797                         $serverdata['registered-users'] = max($data['total_users'], 1);
 798                 }
 799
 800                 if (!empty($data['active_users_monthly'])) {
 801                         $valid = true;
 802                         $serverdata['active-month-users'] = max($data['active_users_monthly'], 0);
 803                 }
 804
 805                 if (!empty($data['active_users_halfyear'])) {
 806                         $valid = true;
 807                         $serverdata['active-halfyear-users'] = max($data['active_users_halfyear'], 0);
 808                 }
 809
 810                 if (!empty($data['local_posts'])) {
 811                         $valid = true;
 812                         $serverdata['local-posts'] = max($data['local_posts'], 0);
 813                 }
 814
 815                 if (!empty($data['registrations_open'])) {
 816                         $serverdata['register_policy'] = Register::OPEN;
 817                 } else {
 818                         $serverdata['register_policy'] = Register::CLOSED;
 819                 }
 820
 821                 if (!$valid) {
 822                         return $old_serverdata;
 823                 }
 824
 825                 return $serverdata;
 826         }
 827
 828         /**
 829          * Detect server type by using the nodeinfo data
 830          *
 831          * @param string                  $url        address of the server
 832          * @param ICanHandleHttpResponses $httpResult
 833          *
 834          * @return array Server data
 835          *
 836          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 837          */
 838         private static function fetchNodeinfo(string $url, ICanHandleHttpResponses $httpResult): array
 839         {
 840                 if (!$httpResult->isSuccess()) {
 841                         return [];
 842                 }
 843
 844                 $nodeinfo = json_decode($httpResult->getBody(), true);
 845
 846                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
 847                         return [];
 848                 }
 849
 850                 $nodeinfo1_url = '';
 851                 $nodeinfo2_url = '';
 852
 853                 foreach ($nodeinfo['links'] as $link) {
 854                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
 855                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
 856                                 continue;
 857                         }
 858                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 859                                 $nodeinfo1_url = $link['href'];
 860                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 861                                 $nodeinfo2_url = $link['href'];
 862                         }
 863                 }
 864
 865                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 866                         return [];
 867                 }
 868
 869                 $server = [];
 870
 871                 if (!empty($nodeinfo2_url)) {
 872                         $server = self::parseNodeinfo2($nodeinfo2_url);
 873                 }
 874
 875                 if (empty($server) && !empty($nodeinfo1_url)) {
 876                         $server = self::parseNodeinfo1($nodeinfo1_url);
 877                 }
 878
 879                 return $server;
 880         }
 881
 882         /**
 883          * Parses Nodeinfo 1
 884          *
 885          * @param string $nodeinfo_url address of the nodeinfo path
 886          *
 887          * @return array Server data
 888          *
 889          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 890          */
 891         private static function parseNodeinfo1(string $nodeinfo_url): array
 892         {
 893                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 894                 if (!$curlResult->isSuccess()) {
 895                         return [];
 896                 }
 897
 898                 $nodeinfo = json_decode($curlResult->getBody(), true);
 899
 900                 if (!is_array($nodeinfo)) {
 901                         return [];
 902                 }
 903
 904                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
 905                         'register_policy' => Register::CLOSED];
 906
 907                 if (!empty($nodeinfo['openRegistrations'])) {
 908                         $server['register_policy'] = Register::OPEN;
 909                 }
 910
 911                 if (is_array($nodeinfo['software'])) {
 912                         if (!empty($nodeinfo['software']['name'])) {
 913                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 914                         }
 915
 916                         if (!empty($nodeinfo['software']['version'])) {
 917                                 $server['version'] = $nodeinfo['software']['version'];
 918                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 919                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 920                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 921                         }
 922                 }
 923
 924                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 925                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 926                 }
 927
 928                 if (!empty($nodeinfo['usage']['users']['total'])) {
 929                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
 930                 }
 931
 932                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
 933                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
 934                 }
 935
 936                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
 937                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
 938                 }
 939
 940                 if (!empty($nodeinfo['usage']['localPosts'])) {
 941                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
 942                 }
 943
 944                 if (!empty($nodeinfo['usage']['localComments'])) {
 945                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
 946                 }
 947
 948                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
 949                         $protocols = [];
 950                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
 951                                 $protocols[$protocol] = true;
 952                         }
 953
 954                         if (!empty($protocols['friendica'])) {
 955                                 $server['network'] = Protocol::DFRN;
 956                         } elseif (!empty($protocols['activitypub'])) {
 957                                 $server['network'] = Protocol::ACTIVITYPUB;
 958                         } elseif (!empty($protocols['diaspora'])) {
 959                                 $server['network'] = Protocol::DIASPORA;
 960                         } elseif (!empty($protocols['ostatus'])) {
 961                                 $server['network'] = Protocol::OSTATUS;
 962                         } elseif (!empty($protocols['gnusocial'])) {
 963                                 $server['network'] = Protocol::OSTATUS;
 964                         } elseif (!empty($protocols['zot'])) {
 965                                 $server['network'] = Protocol::ZOT;
 966                         }
 967                 }
 968
 969                 if (empty($server)) {
 970                         return [];
 971                 }
 972
 973                 if (empty($server['network'])) {
 974                         $server['network'] = Protocol::PHANTOM;
 975                 }
 976
 977                 return $server;
 978         }
 979
 980         /**
 981          * Parses Nodeinfo 2
 982          *
 983          * @see https://git.feneas.org/jaywink/nodeinfo2
 984          *
 985          * @param string $nodeinfo_url address of the nodeinfo path
 986          *
 987          * @return array Server data
 988          *
 989          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 990          */
 991         private static function parseNodeinfo2(string $nodeinfo_url): array
 992         {
 993                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 994                 if (!$curlResult->isSuccess()) {
 995                         return [];
 996                 }
 997
 998                 $nodeinfo = json_decode($curlResult->getBody(), true);
 999                 if (!is_array($nodeinfo)) {
1000                         return [];
1001                 }
1002
1003                 $server = [
1004                         'detection-method' => self::DETECT_NODEINFO_2,
1005                         'register_policy' => Register::CLOSED,
1006                         'platform' => 'unknown',
1007                 ];
1008
1009                 if (!empty($nodeinfo['openRegistrations'])) {
1010                         $server['register_policy'] = Register::OPEN;
1011                 }
1012
1013                 if (!empty($nodeinfo['software'])) {
1014                         if (isset($nodeinfo['software']['name'])) {
1015                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
1016                         }
1017
1018                         if (!empty($nodeinfo['software']['version']) && isset($server['platform'])) {
1019                                 $server['version'] = $nodeinfo['software']['version'];
1020                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1021                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1022                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1023
1024                                 // qoto advertises itself as Mastodon
1025                                 if (($server['platform'] == 'mastodon') && substr($nodeinfo['software']['version'], -5) == '-qoto') {
1026                                         $server['platform'] = 'qoto';
1027                                 }
1028                         }
1029                 }
1030
1031                 if (!empty($nodeinfo['metadata']['nodeName'])) {
1032                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
1033                 }
1034
1035                 if (!empty($nodeinfo['usage']['users']['total'])) {
1036                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1037                 }
1038
1039                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1040                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1041                 }
1042
1043                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1044                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1045                 }
1046
1047                 if (!empty($nodeinfo['usage']['localPosts'])) {
1048                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1049                 }
1050
1051                 if (!empty($nodeinfo['usage']['localComments'])) {
1052                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1053                 }
1054
1055                 if (!empty($nodeinfo['protocols'])) {
1056                         $protocols = [];
1057                         foreach ($nodeinfo['protocols'] as $protocol) {
1058                                 if (is_string($protocol)) {
1059                                         $protocols[$protocol] = true;
1060                                 }
1061                         }
1062
1063                         if (!empty($protocols['dfrn'])) {
1064                                 $server['network'] = Protocol::DFRN;
1065                         } elseif (!empty($protocols['activitypub'])) {
1066                                 $server['network'] = Protocol::ACTIVITYPUB;
1067                         } elseif (!empty($protocols['diaspora'])) {
1068                                 $server['network'] = Protocol::DIASPORA;
1069                         } elseif (!empty($protocols['ostatus'])) {
1070                                 $server['network'] = Protocol::OSTATUS;
1071                         } elseif (!empty($protocols['gnusocial'])) {
1072                                 $server['network'] = Protocol::OSTATUS;
1073                         } elseif (!empty($protocols['zot'])) {
1074                                 $server['network'] = Protocol::ZOT;
1075                         }
1076                 }
1077
1078                 if (empty($server)) {
1079                         return [];
1080                 }
1081
1082                 if (empty($server['network'])) {
1083                         $server['network'] = Protocol::PHANTOM;
1084                 }
1085
1086                 return $server;
1087         }
1088
1089         /**
1090          * Parses NodeInfo2 protocol 1.0
1091          *
1092          * @see https://github.com/jaywink/nodeinfo2/blob/master/PROTOCOL.md
1093          *
1094          * @param string $nodeinfo_url address of the nodeinfo path
1095          *
1096          * @return array Server data
1097          *
1098          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1099          */
1100         private static function parseNodeinfo210(ICanHandleHttpResponses $httpResult): array
1101         {
1102                 if (!$httpResult->isSuccess()) {
1103                         return [];
1104                 }
1105
1106                 $nodeinfo = json_decode($httpResult->getBody(), true);
1107
1108                 if (!is_array($nodeinfo)) {
1109                         return [];
1110                 }
1111
1112                 $server = ['detection-method' => self::DETECT_NODEINFO_210,
1113                         'register_policy' => Register::CLOSED];
1114
1115                 if (!empty($nodeinfo['openRegistrations'])) {
1116                         $server['register_policy'] = Register::OPEN;
1117                 }
1118
1119                 if (!empty($nodeinfo['server'])) {
1120                         if (!empty($nodeinfo['server']['software'])) {
1121                                 $server['platform'] = strtolower($nodeinfo['server']['software']);
1122                         }
1123
1124                         if (!empty($nodeinfo['server']['version'])) {
1125                                 $server['version'] = $nodeinfo['server']['version'];
1126                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1127                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1128                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1129                         }
1130
1131                         if (!empty($nodeinfo['server']['name'])) {
1132                                 $server['site_name'] = $nodeinfo['server']['name'];
1133                         }
1134                 }
1135
1136                 if (!empty($nodeinfo['usage']['users']['total'])) {
1137                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1138                 }
1139
1140                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1141                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1142                 }
1143
1144                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1145                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1146                 }
1147
1148                 if (!empty($nodeinfo['usage']['localPosts'])) {
1149                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1150                 }
1151
1152                 if (!empty($nodeinfo['usage']['localComments'])) {
1153                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1154                 }
1155
1156                 if (!empty($nodeinfo['protocols'])) {
1157                         $protocols = [];
1158                         foreach ($nodeinfo['protocols'] as $protocol) {
1159                                 if (is_string($protocol)) {
1160                                         $protocols[$protocol] = true;
1161                                 }
1162                         }
1163
1164                         if (!empty($protocols['dfrn'])) {
1165                                 $server['network'] = Protocol::DFRN;
1166                         } elseif (!empty($protocols['activitypub'])) {
1167                                 $server['network'] = Protocol::ACTIVITYPUB;
1168                         } elseif (!empty($protocols['diaspora'])) {
1169                                 $server['network'] = Protocol::DIASPORA;
1170                         } elseif (!empty($protocols['ostatus'])) {
1171                                 $server['network'] = Protocol::OSTATUS;
1172                         } elseif (!empty($protocols['gnusocial'])) {
1173                                 $server['network'] = Protocol::OSTATUS;
1174                         } elseif (!empty($protocols['zot'])) {
1175                                 $server['network'] = Protocol::ZOT;
1176                         }
1177                 }
1178
1179                 if (empty($server) || empty($server['platform'])) {
1180                         return [];
1181                 }
1182
1183                 if (empty($server['network'])) {
1184                         $server['network'] = Protocol::PHANTOM;
1185                 }
1186
1187                 return $server;
1188         }
1189
1190         /**
1191          * Fetch server information from a 'siteinfo.json' file on the given server
1192          *
1193          * @param string $url        URL of the given server
1194          * @param array  $serverdata array with server data
1195          *
1196          * @return array server data
1197          */
1198         private static function fetchSiteinfo(string $url, array $serverdata): array
1199         {
1200                 $curlResult = DI::httpClient()->get($url . '/siteinfo.json', HttpClientAccept::JSON);
1201                 if (!$curlResult->isSuccess()) {
1202                         return $serverdata;
1203                 }
1204
1205                 $data = json_decode($curlResult->getBody(), true);
1206                 if (empty($data)) {
1207                         return $serverdata;
1208                 }
1209
1210                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1211                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
1212                 }
1213
1214                 if (!empty($data['url'])) {
1215                         $serverdata['platform'] = strtolower($data['platform']);
1216                         $serverdata['version'] = $data['version'] ?? 'N/A';
1217                 }
1218
1219                 if (!empty($data['plugins'])) {
1220                         if (in_array('pubcrawl', $data['plugins'])) {
1221                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1222                         } elseif (in_array('diaspora', $data['plugins'])) {
1223                                 $serverdata['network'] = Protocol::DIASPORA;
1224                         } elseif (in_array('gnusoc', $data['plugins'])) {
1225                                 $serverdata['network'] = Protocol::OSTATUS;
1226                         } else {
1227                                 $serverdata['network'] = Protocol::ZOT;
1228                         }
1229                 }
1230
1231                 if (!empty($data['site_name'])) {
1232                         $serverdata['site_name'] = $data['site_name'];
1233                 }
1234
1235                 if (!empty($data['channels_total'])) {
1236                         $serverdata['registered-users'] = max($data['channels_total'], 1);
1237                 }
1238
1239                 if (!empty($data['channels_active_monthly'])) {
1240                         $serverdata['active-month-users'] = max($data['channels_active_monthly'], 0);
1241                 }
1242
1243                 if (!empty($data['channels_active_halfyear'])) {
1244                         $serverdata['active-halfyear-users'] = max($data['channels_active_halfyear'], 0);
1245                 }
1246
1247                 if (!empty($data['local_posts'])) {
1248                         $serverdata['local-posts'] = max($data['local_posts'], 0);
1249                 }
1250
1251                 if (!empty($data['local_comments'])) {
1252                         $serverdata['local-comments'] = max($data['local_comments'], 0);
1253                 }
1254
1255                 if (!empty($data['register_policy'])) {
1256                         switch ($data['register_policy']) {
1257                                 case 'REGISTER_OPEN':
1258                                         $serverdata['register_policy'] = Register::OPEN;
1259                                         break;
1260
1261                                 case 'REGISTER_APPROVE':
1262                                         $serverdata['register_policy'] = Register::APPROVE;
1263                                         break;
1264
1265                                 case 'REGISTER_CLOSED':
1266                                 default:
1267                                         $serverdata['register_policy'] = Register::CLOSED;
1268                                         break;
1269                         }
1270                 }
1271
1272                 return $serverdata;
1273         }
1274
1275         /**
1276          * Fetches server data via an ActivityPub account with url of that server
1277          *
1278          * @param string $url        URL of the given server
1279          * @param array  $serverdata array with server data
1280          *
1281          * @return array server data
1282          *
1283          * @throws Exception
1284          */
1285         private static function fetchDataFromSystemActor(array $data, array $serverdata): array
1286         {
1287                 if (empty($data)) {
1288                         return ['server' => $serverdata, 'actor' => ''];
1289                 }
1290
1291                 $actor = JsonLD::compact($data, false);
1292                 if (in_array(JsonLD::fetchElement($actor, '@type'), ActivityPub\Receiver::ACCOUNT_TYPES)) {
1293                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1294                         $serverdata['site_name'] = JsonLD::fetchElement($actor, 'as:name', '@value');
1295                         $serverdata['info'] = JsonLD::fetchElement($actor, 'as:summary', '@value');
1296                         if (!empty($actor['as:generator'])) {
1297                                 $generator = explode(' ', JsonLD::fetchElement($actor['as:generator'], 'as:name', '@value'));
1298                                 $serverdata['platform'] = strtolower(array_shift($generator));
1299                                 $serverdata['detection-method'] = self::DETECT_SYSTEM_ACTOR;
1300                         } else {
1301                                 $serverdata['detection-method'] = self::DETECT_AP_ACTOR;
1302                         }
1303                         return ['server' => $serverdata, 'actor' => $actor['@id']];
1304                 } elseif ((JsonLD::fetchElement($actor, '@type') == 'as:Collection')) {
1305                         // By now only Ktistec seems to provide collections this way
1306                         $serverdata['platform'] = 'ktistec';
1307                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1308                         $serverdata['detection-method'] = self::DETECT_AP_COLLECTION;
1309
1310                         $actors = JsonLD::fetchElementArray($actor, 'as:items');
1311                         if (!empty($actors) && !empty($actors[0]['@id'])) {
1312                                 $actor_url = $actor['@id'] . $actors[0]['@id'];
1313                         } else {
1314                                 $actor_url = '';
1315                         }
1316
1317                         return ['server' => $serverdata, 'actor' => $actor_url];
1318                 }
1319                 return ['server' => $serverdata, 'actor' => ''];
1320         }
1321
1322         /**
1323          * Checks if the server contains a valid host meta file
1324          *
1325          * @param string $url URL of the given server
1326          *
1327          * @return boolean 'true' if the server seems to be vital
1328          */
1329         private static function validHostMeta(string $url): bool
1330         {
1331                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
1332                 $curlResult = DI::httpClient()->get($url . Probe::HOST_META, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
1333                 if (!$curlResult->isSuccess()) {
1334                         return false;
1335                 }
1336
1337                 $xrd = XML::parseString($curlResult->getBody(), true);
1338                 if (!is_object($xrd)) {
1339                         return false;
1340                 }
1341
1342                 $elements = XML::elementToArray($xrd);
1343                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
1344                         return false;
1345                 }
1346
1347                 $valid = false;
1348                 foreach ($elements['xrd']['link'] as $link) {
1349                         // When there is more than a single "link" element, the array looks slightly different
1350                         if (!empty($link['@attributes'])) {
1351                                 $link = $link['@attributes'];
1352                         }
1353
1354                         if (empty($link['rel']) || empty($link['template'])) {
1355                                 continue;
1356                         }
1357
1358                         if ($link['rel'] == 'lrdd') {
1359                                 // When the webfinger host is the same like the system host, it should be ok.
1360                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
1361                         }
1362                 }
1363
1364                 return $valid;
1365         }
1366
1367         /**
1368          * Detect the network of the given server via their known contacts
1369          *
1370          * @param string $url        URL of the given server
1371          * @param array  $serverdata array with server data
1372          *
1373          * @return array server data
1374          */
1375         private static function detectNetworkViaContacts(string $url, array $serverdata): array
1376         {
1377                 $contacts = [];
1378
1379                 $nurl = Strings::normaliseLink($url);
1380
1381                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $nurl]]);
1382                 while ($apcontact = DBA::fetch($apcontacts)) {
1383                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
1384                 }
1385                 DBA::close($apcontacts);
1386
1387                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $nurl]]);
1388                 while ($pcontact = DBA::fetch($pcontacts)) {
1389                         $contacts[$pcontact['nurl']] = $pcontact['url'];
1390                 }
1391                 DBA::close($pcontacts);
1392
1393                 if (empty($contacts)) {
1394                         return $serverdata;
1395                 }
1396
1397                 $time = time();
1398                 foreach ($contacts as $contact) {
1399                         // Endlosschleife verhindern wegen gsid!
1400                         $data = Probe::uri($contact);
1401                         if (in_array($data['network'], Protocol::FEDERATED)) {
1402                                 $serverdata['network'] = $data['network'];
1403
1404                                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1405                                         $serverdata['detection-method'] = self::DETECT_CONTACTS;
1406                                 }
1407                                 break;
1408                         } elseif ((time() - $time) > 10) {
1409                                 // To reduce the stress on remote systems we probe a maximum of 10 seconds
1410                                 break;
1411                         }
1412                 }
1413
1414                 return $serverdata;
1415         }
1416
1417         /**
1418          * Checks if the given server does have a '/poco' endpoint.
1419          * This is used for the 'PortableContact' functionality,
1420          * which is used by both Friendica and Hubzilla.
1421          *
1422          * @param string $url        URL of the given server
1423          * @param array  $serverdata array with server data
1424          *
1425          * @return array server data
1426          */
1427         private static function checkPoCo(string $url, array $serverdata): array
1428         {
1429                 $serverdata['poco'] = '';
1430
1431                 $curlResult = DI::httpClient()->get($url . '/poco', HttpClientAccept::JSON);
1432                 if (!$curlResult->isSuccess()) {
1433                         return $serverdata;
1434                 }
1435
1436                 $data = json_decode($curlResult->getBody(), true);
1437                 if (empty($data)) {
1438                         return $serverdata;
1439                 }
1440
1441                 if (!empty($data['totalResults'])) {
1442                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1443                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers, 1);
1444                         $serverdata['directory-type'] = self::DT_POCO;
1445                         $serverdata['poco'] = $url . '/poco';
1446                 }
1447
1448                 return $serverdata;
1449         }
1450
1451         /**
1452          * Checks if the given server does have a Mastodon style directory endpoint.
1453          *
1454          * @param string $url        URL of the given server
1455          * @param array  $serverdata array with server data
1456          *
1457          * @return array server data
1458          */
1459         public static function checkMastodonDirectory(string $url, array $serverdata): array
1460         {
1461                 $curlResult = DI::httpClient()->get($url . '/api/v1/directory?limit=1', HttpClientAccept::JSON);
1462                 if (!$curlResult->isSuccess()) {
1463                         return $serverdata;
1464                 }
1465
1466                 $data = json_decode($curlResult->getBody(), true);
1467                 if (empty($data)) {
1468                         return $serverdata;
1469                 }
1470
1471                 if (count($data) == 1) {
1472                         $serverdata['directory-type'] = self::DT_MASTODON;
1473                 }
1474
1475                 return $serverdata;
1476         }
1477
1478         /**
1479          * Detects Peertube via their known endpoint
1480          *
1481          * @param string $url        URL of the given server
1482          * @param array  $serverdata array with server data
1483          *
1484          * @return array server data
1485          */
1486         private static function detectPeertube(string $url, array $serverdata): array
1487         {
1488                 $curlResult = DI::httpClient()->get($url . '/api/v1/config', HttpClientAccept::JSON);
1489                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1490                         return $serverdata;
1491                 }
1492
1493                 $data = json_decode($curlResult->getBody(), true);
1494                 if (empty($data)) {
1495                         return $serverdata;
1496                 }
1497
1498                 if (!empty($data['instance']) && !empty($data['serverVersion'])) {
1499                         $serverdata['platform'] = 'peertube';
1500                         $serverdata['version'] = $data['serverVersion'];
1501                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1502
1503                         if (!empty($data['instance']['name'])) {
1504                                 $serverdata['site_name'] = $data['instance']['name'];
1505                         }
1506
1507                         if (!empty($data['instance']['shortDescription'])) {
1508                                 $serverdata['info'] = $data['instance']['shortDescription'];
1509                         }
1510
1511                         if (!empty($data['signup'])) {
1512                                 if (!empty($data['signup']['allowed'])) {
1513                                         $serverdata['register_policy'] = Register::OPEN;
1514                                 }
1515                         }
1516
1517                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1518                                 $serverdata['detection-method'] = self::DETECT_V1_CONFIG;
1519                         }
1520                 }
1521
1522                 return $serverdata;
1523         }
1524
1525         /**
1526          * Detects the version number of a given server when it was a NextCloud installation
1527          *
1528          * @param string $url        URL of the given server
1529          * @param array  $serverdata array with server data
1530          * @param bool   $validHostMeta
1531          *
1532          * @return array server data
1533          */
1534         private static function detectNextcloud(string $url, array $serverdata, bool $validHostMeta): array
1535         {
1536                 $curlResult = DI::httpClient()->get($url . '/status.php', HttpClientAccept::JSON);
1537                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1538                         return $serverdata;
1539                 }
1540
1541                 $data = json_decode($curlResult->getBody(), true);
1542                 if (empty($data)) {
1543                         return $serverdata;
1544                 }
1545
1546                 if (!empty($data['version'])) {
1547                         $serverdata['platform'] = 'nextcloud';
1548                         $serverdata['version'] = $data['version'];
1549
1550                         if ($validHostMeta) {
1551                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1552                         }
1553
1554                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1555                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1556                         }
1557                 }
1558
1559                 return $serverdata;
1560         }
1561
1562         /**
1563          * Fetches weekly usage data
1564          *
1565          * @param string $url        URL of the given server
1566          * @param array  $serverdata array with server data
1567          *
1568          * @return array server data
1569          */
1570         private static function fetchWeeklyUsage(string $url, array $serverdata): array
1571         {
1572                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance/activity', HttpClientAccept::JSON);
1573                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1574                         return $serverdata;
1575                 }
1576
1577                 $data = json_decode($curlResult->getBody(), true);
1578                 if (empty($data)) {
1579                         return $serverdata;
1580                 }
1581
1582                 $current_week = [];
1583                 foreach ($data as $week) {
1584                         // Use only data from a full week
1585                         if (empty($week['week']) || (time() - $week['week']) < 7 * 24 * 60 * 60) {
1586                                 continue;
1587                         }
1588
1589                         // Most likely the data is sorted correctly. But we better are safe than sorry
1590                         if (empty($current_week['week']) || ($current_week['week'] < $week['week'])) {
1591                                 $current_week = $week;
1592                         }
1593                 }
1594
1595                 if (!empty($current_week['logins'])) {
1596                         $serverdata['active-week-users'] = max($current_week['logins'], 0);
1597                 }
1598
1599                 return $serverdata;
1600         }
1601
1602         /**
1603          * Detects data from a given server url if it was a mastodon alike system
1604          *
1605          * @param string $url        URL of the given server
1606          * @param array  $serverdata array with server data
1607          *
1608          * @return array server data
1609          */
1610         private static function detectMastodonAlikes(string $url, array $serverdata): array
1611         {
1612                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance', HttpClientAccept::JSON);
1613                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1614                         return $serverdata;
1615                 }
1616
1617                 $data = json_decode($curlResult->getBody(), true);
1618                 if (empty($data)) {
1619                         return $serverdata;
1620                 }
1621
1622                 $valid = false;
1623
1624                 if (!empty($data['version'])) {
1625                         $serverdata['platform'] = 'mastodon';
1626                         $serverdata['version'] = $data['version'] ?? '';
1627                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1628                         $valid = true;
1629                 }
1630
1631                 if (!empty($data['title'])) {
1632                         $serverdata['site_name'] = $data['title'];
1633                 }
1634
1635                 if (!empty($data['title']) && empty($serverdata['platform']) && ($serverdata['network'] == Protocol::PHANTOM)) {
1636                         $serverdata['platform'] = 'mastodon';
1637                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1638                         $valid = true;
1639                 }
1640
1641                 if (!empty($data['description'])) {
1642                         $serverdata['info'] = trim($data['description']);
1643                 }
1644
1645                 if (!empty($data['stats']['user_count'])) {
1646                         $serverdata['registered-users'] = max($data['stats']['user_count'], 1);
1647                 }
1648
1649                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1650                         $serverdata['platform'] = strtolower($matches[1]);
1651                         $serverdata['version'] = $matches[2];
1652                         $valid = true;
1653                 }
1654
1655                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1656                         $serverdata['platform'] = 'pleroma';
1657                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1658                         $valid = true;
1659                 }
1660
1661                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1662                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1663                         $serverdata['platform'] = 'pleroma';
1664                         $valid = true;
1665                 }
1666
1667                 if ($valid && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1668                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1669                 }
1670
1671                 return $serverdata;
1672         }
1673
1674         /**
1675          * Detects data from typical Hubzilla endpoints
1676          *
1677          * @param string $url        URL of the given server
1678          * @param array  $serverdata array with server data
1679          *
1680          * @return array server data
1681          */
1682         private static function detectHubzilla(string $url, array $serverdata): array
1683         {
1684                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/config.json', HttpClientAccept::JSON);
1685                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1686                         return $serverdata;
1687                 }
1688
1689                 $data = json_decode($curlResult->getBody(), true);
1690                 if (empty($data) || empty($data['site'])) {
1691                         return $serverdata;
1692                 }
1693
1694                 if (!empty($data['site']['name'])) {
1695                         $serverdata['site_name'] = $data['site']['name'];
1696                 }
1697
1698                 if (!empty($data['site']['platform'])) {
1699                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1700                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1701                         $serverdata['network'] = Protocol::ZOT;
1702                 }
1703
1704                 if (!empty($data['site']['hubzilla'])) {
1705                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1706                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1707                         $serverdata['network'] = Protocol::ZOT;
1708                 }
1709
1710                 if (!empty($data['site']['redmatrix'])) {
1711                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1712                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1713                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1714                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1715                         }
1716
1717                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1718                         $serverdata['network'] = Protocol::ZOT;
1719                 }
1720
1721                 $private = false;
1722                 $inviteonly = false;
1723                 $closed = false;
1724
1725                 if (!empty($data['site']['closed'])) {
1726                         $closed = self::toBoolean($data['site']['closed']);
1727                 }
1728
1729                 if (!empty($data['site']['private'])) {
1730                         $private = self::toBoolean($data['site']['private']);
1731                 }
1732
1733                 if (!empty($data['site']['inviteonly'])) {
1734                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1735                 }
1736
1737                 if (!$closed && !$private and $inviteonly) {
1738                         $serverdata['register_policy'] = Register::APPROVE;
1739                 } elseif (!$closed && !$private) {
1740                         $serverdata['register_policy'] = Register::OPEN;
1741                 } else {
1742                         $serverdata['register_policy'] = Register::CLOSED;
1743                 }
1744
1745                 if (($serverdata['network'] != Protocol::PHANTOM) && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1746                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1747                 }
1748
1749                 return $serverdata;
1750         }
1751
1752         /**
1753          * Converts input value to a boolean value
1754          *
1755          * @param string|integer $val
1756          *
1757          * @return boolean
1758          */
1759         private static function toBoolean($val): bool
1760         {
1761                 if (($val == 'true') || ($val == 1)) {
1762                         return true;
1763                 } elseif (($val == 'false') || ($val == 0)) {
1764                         return false;
1765                 }
1766
1767                 return $val;
1768         }
1769
1770         /**
1771          * Detect if the URL belongs to a GNU Social server
1772          *
1773          * @param string $url        URL of the given server
1774          * @param array  $serverdata array with server data
1775          *
1776          * @return array server data
1777          */
1778         private static function detectGNUSocial(string $url, array $serverdata): array
1779         {
1780                 // Test for GNU Social
1781                 $curlResult = DI::httpClient()->get($url . '/api/gnusocial/version.json', HttpClientAccept::JSON);
1782                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1783                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1784                         $serverdata['platform'] = 'gnusocial';
1785                         // Remove junk that some GNU Social servers return
1786                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1787                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1788                         $serverdata['version'] = trim($serverdata['version'], '"');
1789                         $serverdata['network'] = Protocol::OSTATUS;
1790
1791                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1792                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1793                         }
1794
1795                         return $serverdata;
1796                 }
1797
1798                 // Test for Statusnet
1799                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/version.json', HttpClientAccept::JSON);
1800                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1801                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1802
1803                         // Remove junk that some GNU Social servers return
1804                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1805                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1806                         $serverdata['version'] = trim($serverdata['version'], '"');
1807
1808                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1809                                 $serverdata['platform'] = 'pleroma';
1810                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1811                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1812                         } else {
1813                                 $serverdata['platform'] = 'statusnet';
1814                                 $serverdata['network'] = Protocol::OSTATUS;
1815                         }
1816
1817                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1818                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1819                         }
1820                 }
1821
1822                 return $serverdata;
1823         }
1824
1825         /**
1826          * Detect if the URL belongs to a Friendica server
1827          *
1828          * @param string $url        URL of the given server
1829          * @param array  $serverdata array with server data
1830          *
1831          * @return array server data
1832          */
1833         private static function detectFriendica(string $url, array $serverdata): array
1834         {
1835                 // There is a bug in some versions of Friendica that will return an ActivityStream actor when the content type "application/json" is requested.
1836                 // Because of this me must not use ACCEPT_JSON here.
1837                 $curlResult = DI::httpClient()->get($url . '/friendica/json');
1838                 if (!$curlResult->isSuccess()) {
1839                         $curlResult = DI::httpClient()->get($url . '/friendika/json');
1840                         $friendika = true;
1841                         $platform = 'Friendika';
1842                 } else {
1843                         $friendika = false;
1844                         $platform = 'Friendica';
1845                 }
1846
1847                 if (!$curlResult->isSuccess()) {
1848                         return $serverdata;
1849                 }
1850
1851                 $data = json_decode($curlResult->getBody(), true);
1852                 if (empty($data) || empty($data['version'])) {
1853                         return $serverdata;
1854                 }
1855
1856                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1857                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1858                 }
1859
1860                 $serverdata['network'] = Protocol::DFRN;
1861                 $serverdata['version'] = $data['version'];
1862
1863                 if (!empty($data['no_scrape_url'])) {
1864                         $serverdata['noscrape'] = $data['no_scrape_url'];
1865                 }
1866
1867                 if (!empty($data['site_name'])) {
1868                         $serverdata['site_name'] = $data['site_name'];
1869                 }
1870
1871                 if (!empty($data['info'])) {
1872                         $serverdata['info'] = trim($data['info']);
1873                 }
1874
1875                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1876                 switch ($register_policy) {
1877                         case 'REGISTER_OPEN':
1878                                 $serverdata['register_policy'] = Register::OPEN;
1879                                 break;
1880
1881                         case 'REGISTER_APPROVE':
1882                                 $serverdata['register_policy'] = Register::APPROVE;
1883                                 break;
1884
1885                         case 'REGISTER_CLOSED':
1886                         case 'REGISTER_INVITATION':
1887                                 $serverdata['register_policy'] = Register::CLOSED;
1888                                 break;
1889                         default:
1890                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1891                                 $serverdata['register_policy'] = Register::CLOSED;
1892                                 break;
1893                 }
1894
1895                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1896
1897                 return $serverdata;
1898         }
1899
1900         /**
1901          * Analyses the landing page of a given server for hints about type and system of that server
1902          *
1903          * @param object $curlResult result of curl execution
1904          * @param array  $serverdata array with server data
1905          *
1906          * @return array server data
1907          */
1908         private static function analyseRootBody($curlResult, array $serverdata): array
1909         {
1910                 if (empty($curlResult->getBody())) {
1911                         return $serverdata;
1912                 }
1913
1914                 if (file_exists(__DIR__ . '/../../static/platforms.config.php')) {
1915                         require __DIR__ . '/../../static/platforms.config.php';
1916                 } else {
1917                         throw new HTTPException\InternalServerErrorException('Invalid platform file');
1918                 }
1919
1920                 $platforms = array_merge($ap_platforms, $dfrn_platforms, $zap_platforms, $platforms);
1921
1922                 $doc = new DOMDocument();
1923                 @$doc->loadHTML($curlResult->getBody());
1924                 $xpath = new DOMXPath($doc);
1925                 $assigned = false;
1926
1927                 // We can only detect honk via some HTML element on their page
1928                 if ($xpath->query('//div[@id="honksonpage"]')->count() == 1) {
1929                         $serverdata['platform'] = 'honk';
1930                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1931                         $assigned = true;
1932                 }
1933
1934                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1935                 if (!empty($title)) {
1936                         $serverdata['site_name'] = $title;
1937                 }
1938
1939                 $list = $xpath->query('//meta[@name]');
1940
1941                 foreach ($list as $node) {
1942                         $attr = [];
1943                         if ($node->attributes->length) {
1944                                 foreach ($node->attributes as $attribute) {
1945                                         $value = trim($attribute->value);
1946                                         if (empty($value)) {
1947                                                 continue;
1948                                         }
1949
1950                                         $attr[$attribute->name] = $value;
1951                                 }
1952
1953                                 if (empty($attr['name']) || empty($attr['content'])) {
1954                                         continue;
1955                                 }
1956                         }
1957
1958                         if ($attr['name'] == 'description') {
1959                                 $serverdata['info'] = $attr['content'];
1960                         }
1961
1962                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1963                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad', 'generator'])) {
1964                                 $platform = str_ireplace(array_keys($platforms), array_values($platforms), $attr['content']);
1965                                 $platform = str_replace('/', ' ', $platform);
1966                                 $platform_parts = explode(' ', $platform);
1967                                 if ((count($platform_parts) >= 2) && in_array(strtolower($platform_parts[0]), array_values($platforms))) {
1968                                         $platform = $platform_parts[0];
1969                                         $serverdata['version'] = $platform_parts[1];
1970                                 }
1971                                 if (in_array($platform, array_values($dfrn_platforms))) {
1972                                         $serverdata['network'] = Protocol::DFRN;
1973                                 } elseif (in_array($platform, array_values($ap_platforms))) {
1974                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1975                                 } elseif (in_array($platform, array_values($zap_platforms))) {
1976                                         $serverdata['network'] = Protocol::ZOT;
1977                                 }
1978                                 if (in_array($platform, array_values($platforms))) {
1979                                         $serverdata['platform'] = $platform;
1980                                         $assigned = true;
1981                                 }
1982                         }
1983                 }
1984
1985                 $list = $xpath->query('//meta[@property]');
1986
1987                 foreach ($list as $node) {
1988                         $attr = [];
1989                         if ($node->attributes->length) {
1990                                 foreach ($node->attributes as $attribute) {
1991                                         $value = trim($attribute->value);
1992                                         if (empty($value)) {
1993                                                 continue;
1994                                         }
1995
1996                                         $attr[$attribute->name] = $value;
1997                                 }
1998
1999                                 if (empty($attr['property']) || empty($attr['content'])) {
2000                                         continue;
2001                                 }
2002                         }
2003
2004                         if ($attr['property'] == 'og:site_name') {
2005                                 $serverdata['site_name'] = $attr['content'];
2006                         }
2007
2008                         if ($attr['property'] == 'og:description') {
2009                                 $serverdata['info'] = $attr['content'];
2010                         }
2011
2012                         if (in_array($attr['property'], ['og:platform', 'generator'])) {
2013                                 if (in_array($attr['content'], array_keys($platforms))) {
2014                                         $serverdata['platform'] = $platforms[$attr['content']];
2015                                         $assigned = true;
2016                                 }
2017
2018                                 if (in_array($attr['content'], array_keys($ap_platforms))) {
2019                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2020                                 } elseif (in_array($attr['content'], array_values($zap_platforms))) {
2021                                         $serverdata['network'] = Protocol::ZOT;
2022                                 }
2023                         }
2024                 }
2025
2026                 $list = $xpath->query('//link[@rel="me"]');
2027                 foreach ($list as $node) {
2028                         foreach ($node->attributes as $attribute) {
2029                                 if (parse_url(trim($attribute->value), PHP_URL_HOST) == 'micro.blog') {
2030                                         $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2031                                         $serverdata['platform'] = 'microblog';
2032                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2033                                         $assigned = true;
2034                                 }
2035                         }
2036                 }
2037
2038                 if ($serverdata['platform'] != 'microblog') {
2039                         $list = $xpath->query('//link[@rel="micropub"]');
2040                         foreach ($list as $node) {
2041                                 foreach ($node->attributes as $attribute) {
2042                                         if (trim($attribute->value) == 'https://micro.blog/micropub') {
2043                                                 $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2044                                                 $serverdata['platform'] = 'microblog';
2045                                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
2046                                                 $assigned = true;
2047                                         }
2048                                 }
2049                         }
2050                 }
2051
2052                 if ($assigned && in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER])) {
2053                         $serverdata['detection-method'] = self::DETECT_BODY;
2054                 }
2055
2056                 return $serverdata;
2057         }
2058
2059         /**
2060          * Analyses the header data of a given server for hints about type and system of that server
2061          *
2062          * @param object $curlResult result of curl execution
2063          * @param array  $serverdata array with server data
2064          *
2065          * @return array server data
2066          */
2067         private static function analyseRootHeader($curlResult, array $serverdata): array
2068         {
2069                 if ($curlResult->getHeader('server') == 'Mastodon') {
2070                         $serverdata['platform'] = 'mastodon';
2071                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2072                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
2073                         $serverdata['platform'] = 'diaspora';
2074                         $serverdata['network'] = Protocol::DIASPORA;
2075                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version')[0] ?? '';
2076                 } elseif ($curlResult->inHeader('x-friendica-version')) {
2077                         $serverdata['platform'] = 'friendica';
2078                         $serverdata['network'] = Protocol::DFRN;
2079                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version')[0] ?? '';
2080                 } else {
2081                         return $serverdata;
2082                 }
2083
2084                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
2085                         $serverdata['detection-method'] = self::DETECT_HEADER;
2086                 }
2087
2088                 return $serverdata;
2089         }
2090
2091         /**
2092          * Update GServer entries
2093          */
2094         public static function discover()
2095         {
2096                 // Update the server list
2097                 self::discoverFederation();
2098
2099                 $no_of_queries = 5;
2100
2101                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
2102
2103                 if ($requery_days == 0) {
2104                         $requery_days = 7;
2105                 }
2106
2107                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
2108
2109                 $gservers = DBA::select('gserver', ['id', 'url', 'nurl', 'network', 'poco', 'directory-type'],
2110                         ["NOT `failed` AND `directory-type` != ? AND `last_poco_query` < ?", GServer::DT_NONE, $last_update],
2111                         ['order' => ['RAND()']]);
2112
2113                 while ($gserver = DBA::fetch($gservers)) {
2114                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2115                         Worker::add(Worker::PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
2116
2117                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2118                         Worker::add(Worker::PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
2119
2120                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
2121                         self::update($fields, ['nurl' => $gserver['nurl']]);
2122
2123                         if (--$no_of_queries == 0) {
2124                                 break;
2125                         }
2126                 }
2127
2128                 DBA::close($gservers);
2129         }
2130
2131         /**
2132          * Discover federated servers
2133          */
2134         private static function discoverFederation()
2135         {
2136                 $last = DI::config()->get('poco', 'last_federation_discovery');
2137
2138                 if ($last) {
2139                         $next = $last + (24 * 60 * 60);
2140
2141                         if ($next > time()) {
2142                                 return;
2143                         }
2144                 }
2145
2146                 // Discover federated servers
2147                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
2148                 foreach ($protocols as $protocol) {
2149                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
2150                         $curlResult = DI::httpClient()->fetch('https://the-federation.info/graphql?query=' . urlencode($query), HttpClientAccept::JSON);
2151                         if (!empty($curlResult)) {
2152                                 $data = json_decode($curlResult, true);
2153                                 if (!empty($data['data']['nodes'])) {
2154                                         foreach ($data['data']['nodes'] as $server) {
2155                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
2156                                                 self::add('https://' . $server['host'], true);
2157                                         }
2158                                 }
2159                         }
2160                 }
2161
2162                 // Disvover Mastodon servers
2163                 $accesstoken = DI::config()->get('system', 'instances_social_key');
2164
2165                 if (!empty($accesstoken)) {
2166                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
2167                         $curlResult = DI::httpClient()->get($api, HttpClientAccept::JSON, [HttpClientOptions::HEADERS => ['Authorization' => ['Bearer ' . $accesstoken]]]);
2168                         if ($curlResult->isSuccess()) {
2169                                 $servers = json_decode($curlResult->getBody(), true);
2170
2171                                 if (!empty($servers['instances'])) {
2172                                         foreach ($servers['instances'] as $server) {
2173                                                 $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
2174                                                 self::add($url);
2175                                         }
2176                                 }
2177                         }
2178                 }
2179
2180                 DI::config()->set('poco', 'last_federation_discovery', time());
2181         }
2182
2183         /**
2184          * Set the protocol for the given server
2185          *
2186          * @param int $gsid     Server id
2187          * @param int $protocol Protocol id
2188          *
2189          * @throws Exception
2190          */
2191         public static function setProtocol(int $gsid, int $protocol)
2192         {
2193                 if (empty($gsid)) {
2194                         return;
2195                 }
2196
2197                 $gserver = DBA::selectFirst('gserver', ['protocol', 'url'], ['id' => $gsid]);
2198                 if (!DBA::isResult($gserver)) {
2199                         return;
2200                 }
2201
2202                 $old = $gserver['protocol'];
2203
2204                 if (!is_null($old)) {
2205                         /*
2206                         The priority for the protocols is:
2207                                 1. ActivityPub
2208                                 2. DFRN via Diaspora
2209                                 3. Legacy DFRN
2210                                 4. Diaspora
2211                                 5. OStatus
2212                         */
2213
2214                         // We don't need to change it when nothing is to be changed
2215                         if ($old == $protocol) {
2216                                 return;
2217                         }
2218
2219                         // We don't want to mark a server as OStatus when it had been marked with any other protocol before
2220                         if ($protocol == Post\DeliveryData::OSTATUS) {
2221                                 return;
2222                         }
2223
2224                         // If the server is marked as ActivityPub then we won't change it to anything different
2225                         if ($old == Post\DeliveryData::ACTIVITYPUB) {
2226                                 return;
2227                         }
2228
2229                         // Don't change it to anything lower than DFRN if the new one wasn't ActivityPub
2230                         if (($old == Post\DeliveryData::DFRN) && ($protocol != Post\DeliveryData::ACTIVITYPUB)) {
2231                                 return;
2232                         }
2233
2234                         // Don't change it to Diaspora when it is a legacy DFRN server
2235                         if (($old == Post\DeliveryData::LEGACY_DFRN) && ($protocol == Post\DeliveryData::DIASPORA)) {
2236                                 return;
2237                         }
2238                 }
2239
2240                 Logger::info('Protocol for server', ['protocol' => $protocol, 'old' => $old, 'id' => $gsid, 'url' => $gserver['url'], 'callstack' => System::callstack(20)]);
2241                 self::update(['protocol' => $protocol], ['id' => $gsid]);
2242         }
2243
2244         /**
2245          * Fetch the protocol of the given server
2246          *
2247          * @param int $gsid Server id
2248          *
2249          * @return ?int One of Post\DeliveryData protocol constants or null if unknown or gserver is missing
2250          *
2251          * @throws Exception
2252          */
2253         public static function getProtocol(int $gsid): ?int
2254         {
2255                 if (empty($gsid)) {
2256                         return null;
2257                 }
2258
2259                 $gserver = DBA::selectFirst('gserver', ['protocol'], ['id' => $gsid]);
2260                 if (DBA::isResult($gserver)) {
2261                         return $gserver['protocol'];
2262                 }
2263
2264                 return null;
2265         }
2266
2267         /**
2268          * Update rows in the gserver table.
2269          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2270          *
2271          * @param array $fields
2272          * @param array $condition
2273          *
2274          * @return bool
2275          *
2276          * @throws Exception
2277          */
2278         public static function update(array $fields, array $condition): bool
2279         {
2280                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2281
2282                 return DBA::update('gserver', $fields, $condition);
2283         }
2284
2285         /**
2286          * Insert a row into the gserver table.
2287          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2288          *
2289          * @param array $fields
2290          * @param int   $duplicate_mode What to do on a duplicated entry
2291          *
2292          * @return bool
2293          *
2294          * @throws Exception
2295          */
2296         public static function insert(array $fields, int $duplicate_mode = Database::INSERT_DEFAULT): bool
2297         {
2298                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2299
2300                 return DBA::insert('gserver', $fields, $duplicate_mode);
2301         }
2302 }