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\Protocol\Relay;
  39 use Friendica\Util\DateTimeFormat;
  40 use Friendica\Util\Network;
  41 use Friendica\Util\Strings;
  42 use Friendica\Util\XML;
  43
  44 /**
  45  * This class handles GServer related functions
  46  */
  47 class GServer
  48 {
  49         // Directory types
  50         const DT_NONE = 0;
  51         const DT_POCO = 1;
  52         const DT_MASTODON = 2;
  53
  54         // Methods to detect server types
  55
  56         // Non endpoint specific methods
  57         const DETECT_MANUAL = 0;
  58         const DETECT_HEADER = 1;
  59         const DETECT_BODY = 2;
  60
  61         // Implementation specific endpoints
  62         const DETECT_FRIENDIKA = 10;
  63         const DETECT_FRIENDICA = 11;
  64         const DETECT_STATUSNET = 12;
  65         const DETECT_GNUSOCIAL = 13;
  66         const DETECT_CONFIG_JSON = 14; // Statusnet, GNU Social, Older Hubzilla/Redmatrix
  67         const DETECT_SITEINFO_JSON = 15; // Newer Hubzilla
  68         const DETECT_MASTODON_API = 16;
  69         const DETECT_STATUS_PHP = 17; // Nextcloud
  70         const DETECT_V1_CONFIG = 18;
  71         const DETECT_PUMPIO = 19;
  72
  73         // Standardized endpoints
  74         const DETECT_STATISTICS_JSON = 100;
  75         const DETECT_NODEINFO_1 = 101;
  76         const DETECT_NODEINFO_2 = 102;
  77
  78         /**
  79          * Check for the existance of a server and adds it in the background if not existant
  80          *
  81          * @param string $url
  82          * @param boolean $only_nodeinfo
  83          * @return void
  84          */
  85         public static function add(string $url, bool $only_nodeinfo = false)
  86         {
  87                 if (self::getID($url, false)) {
  88                         return;
  89                 }
  90
  91                 Worker::add(PRIORITY_LOW, 'UpdateGServer', $url, $only_nodeinfo);
  92         }
  93
  94         /**
  95          * Get the ID for the given server URL
  96          *
  97          * @param string $url
  98          * @param boolean $no_check Don't check if the server hadn't been found
  99          * @return int gserver id
 100          */
 101         public static function getID(string $url, bool $no_check = false)
 102         {
 103                 if (empty($url)) {
 104                         return null;
 105                 }
 106
 107                 $url = self::cleanURL($url);
 108
 109                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
 110                 if (DBA::isResult($gserver)) {
 111                         Logger::info('Got ID for URL', ['id' => $gserver['id'], 'url' => $url, 'callstack' => System::callstack(20)]);
 112                         return $gserver['id'];
 113                 }
 114
 115                 if ($no_check || !self::check($url)) {
 116                         return null;
 117                 }
 118
 119                 return self::getID($url, true);
 120         }
 121
 122         /**
 123          * Retrieves all the servers which base domain are matching the provided domain pattern
 124          *
 125          * The pattern is a simple fnmatch() pattern with ? for single wildcard and * for multiple wildcard
 126          *
 127          * @param string $pattern
 128          * @return array
 129          * @throws Exception
 130          */
 131         public static function listByDomainPattern(string $pattern): array
 132         {
 133                 $likePattern = 'http://' . strtr($pattern, ['_' => '\_', '%' => '\%', '?' => '_', '*' => '%']);
 134
 135                 // The SUBSTRING_INDEX returns everything before the eventual third /, which effectively trims an
 136                 // eventual server path and keep only the server domain which we're matching against the pattern.
 137                 $sql = "SELECT `gserver`.*, COUNT(*) AS `contacts`
 138                         FROM `gserver`
 139                         LEFT JOIN `contact` ON `gserver`.`id` = `contact`.`gsid`
 140                         WHERE SUBSTRING_INDEX(`gserver`.`nurl`, '/', 3) LIKE ?
 141                         AND NOT `gserver`.`failed`
 142                         GROUP BY `gserver`.`id`";
 143
 144                 $stmt = DI::dba()->p($sql, $likePattern);
 145
 146                 return DI::dba()->toArray($stmt);
 147         }
 148
 149         /**
 150          * Checks if the given server is reachable
 151          *
 152          * @param string  $profile URL of the given profile
 153          * @param string  $server  URL of the given server (If empty, taken from profile)
 154          * @param string  $network Network value that is used, when detection failed
 155          * @param boolean $force   Force an update.
 156          *
 157          * @return boolean 'true' if server seems vital
 158          */
 159         public static function reachable(string $profile, string $server = '', string $network = '', bool $force = false)
 160         {
 161                 if ($server == '') {
 162                         $contact = Contact::getByURL($profile, null, ['baseurl']);
 163                         if (!empty($contact['baseurl'])) {
 164                                 $server = $contact['baseurl'];
 165                         }
 166                 }
 167
 168                 if ($server == '') {
 169                         return true;
 170                 }
 171
 172                 return self::check($server, $network, $force);
 173         }
 174
 175         public static function getNextUpdateDate(bool $success, string $created = '', string $last_contact = '')
 176         {
 177                 // On successful contact process check again next week
 178                 if ($success) {
 179                         return DateTimeFormat::utc('now +7 day');
 180                 }
 181
 182                 $now = strtotime(DateTimeFormat::utcNow());
 183
 184                 if ($created > $last_contact) {
 185                         $contact_time = strtotime($created);
 186                 } else {
 187                         $contact_time = strtotime($last_contact);
 188                 }
 189
 190                 // If the last contact was less than 6 hours before then try again in 6 hours
 191                 if (($now - $contact_time) < (60 * 60 * 6)) {
 192                         return DateTimeFormat::utc('now +6 hour');
 193                 }
 194
 195                 // If the last contact was less than 12 hours before then try again in 12 hours
 196                 if (($now - $contact_time) < (60 * 60 * 12)) {
 197                         return DateTimeFormat::utc('now +12 hour');
 198                 }
 199
 200                 // If the last contact was less than 24 hours before then try tomorrow again
 201                 if (($now - $contact_time) < (60 * 60 * 24)) {
 202                         return DateTimeFormat::utc('now +1 day');
 203                 }
 204
 205                 // If the last contact was less than a week before then try again in a week
 206                 if (($now - $contact_time) < (60 * 60 * 24 * 7)) {
 207                         return DateTimeFormat::utc('now +1 week');
 208                 }
 209
 210                 // If the last contact was less than two weeks before then try again in two week
 211                 if (($now - $contact_time) < (60 * 60 * 24 * 14)) {
 212                         return DateTimeFormat::utc('now +2 week');
 213                 }
 214
 215                 // If the last contact was less than a month before then try again in a month
 216                 if (($now - $contact_time) < (60 * 60 * 24 * 30)) {
 217                         return DateTimeFormat::utc('now +1 month');
 218                 }
 219
 220                 // The system hadn't been successul contacted for more than a month, so try again in three months
 221                 return DateTimeFormat::utc('now +3 month');
 222         }
 223
 224         /**
 225          * Checks the state of the given server.
 226          *
 227          * @param string  $server_url    URL of the given server
 228          * @param string  $network       Network value that is used, when detection failed
 229          * @param boolean $force         Force an update.
 230          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 231          *
 232          * @return boolean 'true' if server seems vital
 233          */
 234         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false)
 235         {
 236                 $server_url = self::cleanURL($server_url);
 237                 if ($server_url == '') {
 238                         return false;
 239                 }
 240
 241                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
 242                 if (DBA::isResult($gserver)) {
 243                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
 244                                 $fields = ['created' => DateTimeFormat::utcNow()];
 245                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
 246                                 DBA::update('gserver', $fields, $condition);
 247                         }
 248
 249                         if (!$force && (strtotime($gserver['next_contact']) > time())) {
 250                                 Logger::info('No update needed', ['server' => $server_url]);
 251                                 return (!$gserver['failed']);
 252                         }
 253                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force]);
 254                 } else {
 255                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
 256                 }
 257
 258                 return self::detect($server_url, $network, $only_nodeinfo);
 259         }
 260
 261         /**
 262          * Set failed server status
 263          *
 264          * @param string $url
 265          */
 266         public static function setFailure(string $url)
 267         {
 268                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($url)]);
 269                 if (DBA::isResult($gserver)) {
 270                         $next_update = self::getNextUpdateDate(false, $gserver['created'], $gserver['last_contact']);
 271                         DBA::update('gserver', ['failed' => true, 'last_failure' => DateTimeFormat::utcNow(),
 272                         'next_contact' => $next_update, 'detection-method' => null],
 273                         ['nurl' => Strings::normaliseLink($url)]);
 274                         Logger::info('Set failed status for existing server', ['url' => $url]);
 275                         return;
 276                 }
 277                 DBA::insert('gserver', ['url' => $url, 'nurl' => Strings::normaliseLink($url),
 278                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
 279                         'failed' => true, 'last_failure' => DateTimeFormat::utcNow()]);
 280                 Logger::info('Set failed status for new server', ['url' => $url]);
 281         }
 282
 283         /**
 284          * Remove unwanted content from the given URL
 285          *
 286          * @param string $url
 287          * @return string cleaned URL
 288          */
 289         public static function cleanURL(string $url)
 290         {
 291                 $url = trim($url, '/');
 292                 $url = str_replace('/index.php', '', $url);
 293
 294                 $urlparts = parse_url($url);
 295                 unset($urlparts['user']);
 296                 unset($urlparts['pass']);
 297                 unset($urlparts['query']);
 298                 unset($urlparts['fragment']);
 299                 return Network::unparseURL($urlparts);
 300         }
 301
 302         /**
 303          * Return the base URL
 304          *
 305          * @param string $url
 306          * @return string base URL
 307          */
 308         private static function getBaseURL(string $url)
 309         {
 310                 $urlparts = parse_url(self::cleanURL($url));
 311                 unset($urlparts['path']);
 312                 return Network::unparseURL($urlparts);
 313         }
 314
 315         /**
 316          * Detect server data (type, protocol, version number, ...)
 317          * The detected data is then updated or inserted in the gserver table.
 318          *
 319          * @param string  $url           URL of the given server
 320          * @param string  $network       Network value that is used, when detection failed
 321          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 322          *
 323          * @return boolean 'true' if server could be detected
 324          */
 325         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false)
 326         {
 327                 Logger::info('Detect server type', ['server' => $url]);
 328                 $serverdata = ['detection-method' => self::DETECT_MANUAL];
 329
 330                 $original_url = $url;
 331
 332                 // Remove URL content that is not supposed to exist for a server url
 333                 $url = self::cleanURL($url);
 334
 335                 // Get base URL
 336                 $baseurl = self::getBaseURL($url);
 337
 338                 // If the URL missmatches, then we mark the old entry as failure
 339                 if ($url != $original_url) {
 340                         /// @todo What to do with "next_contact" here?
 341                         DBA::update('gserver', ['failed' => true, 'last_failure' => DateTimeFormat::utcNow()],
 342                                 ['nurl' => Strings::normaliseLink($original_url)]);
 343                 }
 344
 345                 // When a nodeinfo is present, we don't need to dig further
 346                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
 347                 $curlResult = DI::httpClient()->get($url . '/.well-known/nodeinfo', HttpClientAccept::JSON, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 348                 if ($curlResult->isTimeout()) {
 349                         self::setFailure($url);
 350                         return false;
 351                 }
 352
 353                 // On a redirect follow the new host but mark the old one as failure
 354                 if ($curlResult->isSuccess() && !empty($curlResult->getRedirectUrl()) && (parse_url($url, PHP_URL_HOST) != parse_url($curlResult->getRedirectUrl(), PHP_URL_HOST))) {
 355                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 356                         if (!empty($curlResult->getRedirectUrl()) && parse_url($url, PHP_URL_HOST) != parse_url($curlResult->getRedirectUrl(), PHP_URL_HOST)) {
 357                                 Logger::info('Found redirect. Mark old entry as failure', ['old' => $url, 'new' => $curlResult->getRedirectUrl()]);
 358                                 self::setFailure($url);
 359                                 self::detect($curlResult->getRedirectUrl(), $network, $only_nodeinfo);
 360                                 return false;
 361                         }
 362                 }
 363
 364                 $nodeinfo = self::fetchNodeinfo($url, $curlResult);
 365                 if ($only_nodeinfo && empty($nodeinfo)) {
 366                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 367                         self::setFailure($url);
 368                         return false;
 369                 }
 370
 371                 // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 372                 if (empty($nodeinfo)) {
 373                         $nodeinfo = self::fetchStatistics($url);
 374                 }
 375
 376                 // If that didn't work out well, we use some protocol specific endpoints
 377                 // For Friendica and Zot based networks we have to dive deeper to reveal more details
 378                 if (empty($nodeinfo['network']) || in_array($nodeinfo['network'], [Protocol::DFRN, Protocol::ZOT])) {
 379                         if (!empty($nodeinfo['detection-method'])) {
 380                                 $serverdata['detection-method'] = $nodeinfo['detection-method'];
 381
 382                                 foreach (['registered-users', 'active_users_monthly', 'active-halfyear-users', 'local-posts'] as $field) {
 383                                         if (!empty($nodeinfo[$field])) {
 384                                                 $serverdata[$field] = $nodeinfo[$field];
 385                                         }
 386                                 }
 387                         }
 388
 389                         // Fetch the landing page, possibly it reveals some data
 390                         if (empty($nodeinfo['network'])) {
 391                                 if ($baseurl == $url) {
 392                                         $basedata = $serverdata;
 393                                 } else {
 394                                         $basedata = ['detection-method' => self::DETECT_MANUAL];
 395                                 }
 396
 397                                 $curlResult = DI::httpClient()->get($baseurl, HttpClientAccept::HTML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 398                                 if ($curlResult->isSuccess()) {
 399                                         if (!empty($curlResult->getRedirectUrl()) && (parse_url($baseurl, PHP_URL_HOST) != parse_url($curlResult->getRedirectUrl(), PHP_URL_HOST))) {
 400                                                 Logger::info('Found redirect. Mark old entry as failure', ['old' => $url, 'new' => $curlResult->getRedirectUrl()]);
 401                                                 self::setFailure($url);
 402                                                 self::detect($curlResult->getRedirectUrl(), $network, $only_nodeinfo);
 403                                                 return false;
 404                                         }
 405
 406                                         $basedata = self::analyseRootHeader($curlResult, $basedata);
 407                                         $basedata = self::analyseRootBody($curlResult, $basedata, $baseurl);
 408                                 }
 409
 410                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody()) || self::invalidBody($curlResult->getBody())) {
 411                                         self::setFailure($url);
 412                                         return false;
 413                                 }
 414
 415                                 if ($baseurl == $url) {
 416                                         $serverdata = $basedata;
 417                                 } else {
 418                                         // When the base path doesn't seem to contain a social network we try the complete path.
 419                                         // Most detectable system have to be installed in the root directory.
 420                                         // We checked the base to avoid false positives.
 421                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 422                                         if ($curlResult->isSuccess()) {
 423                                                 $urldata = self::analyseRootHeader($curlResult, $serverdata);
 424                                                 $urldata = self::analyseRootBody($curlResult, $urldata, $url);
 425
 426                                                 $comparebase = $basedata;
 427                                                 unset($comparebase['info']);
 428                                                 unset($comparebase['site_name']);
 429                                                 $compareurl = $urldata;
 430                                                 unset($compareurl['info']);
 431                                                 unset($compareurl['site_name']);
 432
 433                                                 // We assume that no one will install the identical system in the root and a subfolder
 434                                                 if (!empty(array_diff($comparebase, $compareurl))) {
 435                                                         $serverdata = $urldata;
 436                                                 }
 437                                         }
 438                                 }
 439                         }
 440
 441                         if (empty($nodeinfo['network']) && (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ACTIVITYPUB))) {
 442                                 $serverdata = self::detectMastodonAlikes($url, $serverdata);
 443                         }
 444
 445                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 446                         // With this check we don't have to waste time and ressources for dead systems.
 447                         // Also this hopefully prevents us from receiving abuse messages.
 448                         $validHostMeta = self::validHostMeta($url);
 449
 450                         if (empty($serverdata['network']) && !$validHostMeta) {
 451                                 $serverdata = self::detectFromContacts($url, $serverdata);
 452                         }
 453
 454                         if (empty($serverdata['network']) && !$validHostMeta) {
 455                                 self::setFailure($url);
 456                                 return false;
 457                         }
 458
 459                         if (empty($serverdata['network']) || in_array($serverdata['network'], [Protocol::DFRN, Protocol::ACTIVITYPUB])) {
 460                                 $serverdata = self::detectFriendica($url, $serverdata);
 461                         }
 462
 463                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 464                         if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ZOT)) {
 465                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 466                         }
 467
 468                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations
 469                         if (empty($serverdata['network'])) {
 470                                 $serverdata = self::detectHubzilla($url, $serverdata);
 471                         }
 472
 473                         if (empty($serverdata['network']) || in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_BODY])) {
 474                                 $serverdata = self::detectPeertube($url, $serverdata);
 475                         }
 476
 477                         if (empty($serverdata['network'])) {
 478                                 $serverdata = self::detectNextcloud($url, $serverdata);
 479                         }
 480
 481                         if (empty($nodeinfo['network']) && empty($serverdata['network'])) {
 482                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 483                         }
 484
 485                         if (empty($serverdata['network'])) {
 486                                 $serverdata = self::detectPumpIO($url, $serverdata);
 487                         }
 488
 489                         $serverdata = array_merge($nodeinfo, $serverdata);
 490                 } else {
 491                         $serverdata = $nodeinfo;
 492                 }
 493
 494                 // Detect the directory type
 495                 $serverdata['directory-type'] = self::DT_NONE;
 496
 497                 $serverdata = self::checkMastodonDirectory($url, $serverdata);
 498
 499                 if ($serverdata['directory-type'] == self::DT_NONE) {
 500                         $serverdata = self::checkPoCo($url, $serverdata);
 501                 }
 502
 503                 // We can't detect the network type. Possibly it is some system that we don't know yet
 504                 if (empty($serverdata['network'])) {
 505                         $serverdata['network'] = Protocol::PHANTOM;
 506                 }
 507
 508                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 509                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 510                         $serverdata['network'] = $network;
 511                 }
 512
 513                 $serverdata['url'] = $url;
 514                 $serverdata['nurl'] = Strings::normaliseLink($url);
 515
 516                 if (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 517                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 518                 }
 519
 520                 if ($serverdata['network'] == Protocol::ACTIVITYPUB) {
 521                         $serverdata = self::fetchWeeklyUsage($url, $serverdata);
 522                 }
 523
 524                 $serverdata['registered-users'] = $serverdata['registered-users'] ?? 0;
 525
 526                 // On an active server there has to be at least a single user
 527                 if (!in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) && ($serverdata['registered-users'] == 0)) {
 528                         $serverdata['registered-users'] = 1;
 529                 } elseif (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 530                         $serverdata['registered-users'] = 0;
 531                 }
 532
 533                 $serverdata['next_contact'] = self::getNextUpdateDate(true);
 534
 535                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 536                 $serverdata['failed'] = false;
 537
 538                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 539                 if (!DBA::isResult($gserver)) {
 540                         $serverdata['created'] = DateTimeFormat::utcNow();
 541                         $ret = DBA::insert('gserver', $serverdata);
 542                         $id = DBA::lastInsertId();
 543                 } else {
 544                         $ret = DBA::update('gserver', $serverdata, ['nurl' => $serverdata['nurl']]);
 545                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 546                         if (DBA::isResult($gserver)) {
 547                                 $id = $gserver['id'];
 548                         }
 549                 }
 550
 551                 // Count the number of known contacts from this server
 552                 if (!empty($id) && !in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 553                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 554                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id, 'failed' => false]);
 555                         $max_users = max($apcontacts, $contacts);
 556                         if ($max_users > $serverdata['registered-users']) {
 557                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 558                                 DBA::update('gserver', ['registered-users' => $max_users], ['id' => $id]);
 559                         }
 560
 561                         if (empty($serverdata['active-month-users'])) {
 562                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 30 days')]);
 563                                 if ($contacts > 0) {
 564                                         Logger::info('Update monthly users', ['id' => $id, 'url' => $serverdata['nurl'], 'monthly-users' => $contacts]);
 565                                         DBA::update('gserver', ['active-month-users' => $contacts], ['id' => $id]);
 566                                 }
 567                         }
 568
 569                         if (empty($serverdata['active-halfyear-users'])) {
 570                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 180 days')]);
 571                                 if ($contacts > 0) {
 572                                         Logger::info('Update halfyear users', ['id' => $id, 'url' => $serverdata['nurl'], 'halfyear-users' => $contacts]);
 573                                         DBA::update('gserver', ['active-halfyear-users' => $contacts], ['id' => $id]);
 574                                 }
 575                         }
 576                 }
 577
 578                 if (!empty($serverdata['network']) && in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 579                         self::discoverRelay($url);
 580                 }
 581
 582                 return $ret;
 583         }
 584
 585         /**
 586          * Fetch relay data from a given server url
 587          *
 588          * @param string $server_url address of the server
 589          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 590          */
 591         private static function discoverRelay(string $server_url)
 592         {
 593                 Logger::info('Discover relay data', ['server' => $server_url]);
 594
 595                 $curlResult = DI::httpClient()->get($server_url . '/.well-known/x-social-relay', HttpClientAccept::JSON);
 596                 if (!$curlResult->isSuccess()) {
 597                         return;
 598                 }
 599
 600                 $data = json_decode($curlResult->getBody(), true);
 601                 if (!is_array($data)) {
 602                         return;
 603                 }
 604
 605                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 606                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
 607
 608                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 609                         $data['scope'] = '';
 610                         $data['subscribe'] = false;
 611                         $data['tags'] = [];
 612                 }
 613
 614                 $gserver = DBA::selectFirst('gserver', ['id', 'url', 'network', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 615                 if (!DBA::isResult($gserver)) {
 616                         return;
 617                 }
 618
 619                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 620                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 621                         DBA::update('gserver', $fields, ['id' => $gserver['id']]);
 622                 }
 623
 624                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 625
 626                 if ($data['scope'] == 'tags') {
 627                         // Avoid duplicates
 628                         $tags = [];
 629                         foreach ($data['tags'] as $tag) {
 630                                 $tag = mb_strtolower($tag);
 631                                 if (strlen($tag) < 100) {
 632                                         $tags[$tag] = $tag;
 633                                 }
 634                         }
 635
 636                         foreach ($tags as $tag) {
 637                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], Database::INSERT_IGNORE);
 638                         }
 639                 }
 640
 641                 // Create or update the relay contact
 642                 $fields = [];
 643                 if (isset($data['protocols'])) {
 644                         if (isset($data['protocols']['diaspora'])) {
 645                                 $fields['network'] = Protocol::DIASPORA;
 646
 647                                 if (isset($data['protocols']['diaspora']['receive'])) {
 648                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 649                                 } elseif (is_string($data['protocols']['diaspora'])) {
 650                                         $fields['batch'] = $data['protocols']['diaspora'];
 651                                 }
 652                         }
 653
 654                         if (isset($data['protocols']['dfrn'])) {
 655                                 $fields['network'] = Protocol::DFRN;
 656
 657                                 if (isset($data['protocols']['dfrn']['receive'])) {
 658                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 659                                 } elseif (is_string($data['protocols']['dfrn'])) {
 660                                         $fields['batch'] = $data['protocols']['dfrn'];
 661                                 }
 662                         }
 663
 664                         if (isset($data['protocols']['activitypub'])) {
 665                                 $fields['network'] = Protocol::ACTIVITYPUB;
 666
 667                                 if (!empty($data['protocols']['activitypub']['actor'])) {
 668                                         $fields['url'] = $data['protocols']['activitypub']['actor'];
 669                                 }
 670                                 if (!empty($data['protocols']['activitypub']['receive'])) {
 671                                         $fields['batch'] = $data['protocols']['activitypub']['receive'];
 672                                 }
 673                         }
 674                 }
 675
 676                 Logger::info('Discovery ended', ['server' => $server_url, 'data' => $fields]);
 677
 678                 Relay::updateContact($gserver, $fields);
 679         }
 680
 681         /**
 682          * Fetch server data from '/statistics.json' on the given server
 683          *
 684          * @param string $url URL of the given server
 685          *
 686          * @return array server data
 687          */
 688         private static function fetchStatistics(string $url)
 689         {
 690                 $curlResult = DI::httpClient()->get($url . '/statistics.json', HttpClientAccept::JSON);
 691                 if (!$curlResult->isSuccess()) {
 692                         return [];
 693                 }
 694
 695                 $data = json_decode($curlResult->getBody(), true);
 696                 if (empty($data)) {
 697                         return [];
 698                 }
 699
 700                 $serverdata = ['detection-method' => self::DETECT_STATISTICS_JSON];
 701
 702                 if (!empty($data['version'])) {
 703                         $serverdata['version'] = $data['version'];
 704                         // Version numbers on statistics.json are presented with additional info, e.g.:
 705                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 706                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 707                 }
 708
 709                 if (!empty($data['name'])) {
 710                         $serverdata['site_name'] = $data['name'];
 711                 }
 712
 713                 if (!empty($data['network'])) {
 714                         $serverdata['platform'] = strtolower($data['network']);
 715
 716                         if ($serverdata['platform'] == 'diaspora') {
 717                                 $serverdata['network'] = Protocol::DIASPORA;
 718                         } elseif ($serverdata['platform'] == 'friendica') {
 719                                 $serverdata['network'] = Protocol::DFRN;
 720                         } elseif ($serverdata['platform'] == 'hubzilla') {
 721                                 $serverdata['network'] = Protocol::ZOT;
 722                         } elseif ($serverdata['platform'] == 'redmatrix') {
 723                                 $serverdata['network'] = Protocol::ZOT;
 724                         }
 725                 }
 726
 727                 if (!empty($data['total_users'])) {
 728                         $serverdata['registered-users'] = max($data['total_users'], 1);
 729                 }
 730
 731                 if (!empty($data['active_users_monthly'])) {
 732                         $serverdata['active-month-users'] = max($data['active_users_monthly'], 0);
 733                 }
 734
 735                 if (!empty($data['active_users_halfyear'])) {
 736                         $serverdata['active-halfyear-users'] = max($data['active_users_halfyear'], 0);
 737                 }
 738
 739                 if (!empty($data['local_posts'])) {
 740                         $serverdata['local-posts'] = max($data['local_posts'], 0);
 741                 }
 742
 743                 if (!empty($data['registrations_open'])) {
 744                         $serverdata['register_policy'] = Register::OPEN;
 745                 } else {
 746                         $serverdata['register_policy'] = Register::CLOSED;
 747                 }
 748
 749                 return $serverdata;
 750         }
 751
 752         /**
 753          * Detect server type by using the nodeinfo data
 754          *
 755          * @param string                  $url        address of the server
 756          * @param ICanHandleHttpResponses $httpResult
 757          *
 758          * @return array Server data
 759          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 760          */
 761         private static function fetchNodeinfo(string $url, ICanHandleHttpResponses $httpResult)
 762         {
 763                 if (!$httpResult->isSuccess()) {
 764                         return [];
 765                 }
 766
 767                 $nodeinfo = json_decode($httpResult->getBody(), true);
 768
 769                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
 770                         return [];
 771                 }
 772
 773                 $nodeinfo1_url = '';
 774                 $nodeinfo2_url = '';
 775
 776                 foreach ($nodeinfo['links'] as $link) {
 777                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
 778                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
 779                                 continue;
 780                         }
 781                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 782                                 $nodeinfo1_url = $link['href'];
 783                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 784                                 $nodeinfo2_url = $link['href'];
 785                         }
 786                 }
 787
 788                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 789                         return [];
 790                 }
 791
 792                 $server = [];
 793
 794                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 795                 if (!empty($nodeinfo2_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo2_url, PHP_URL_HOST))) {
 796                         $server = self::parseNodeinfo2($nodeinfo2_url);
 797                 }
 798
 799                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 800                 if (empty($server) && !empty($nodeinfo1_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo1_url, PHP_URL_HOST))) {
 801                         $server = self::parseNodeinfo1($nodeinfo1_url);
 802                 }
 803
 804                 return $server;
 805         }
 806
 807         /**
 808          * Parses Nodeinfo 1
 809          *
 810          * @param string $nodeinfo_url address of the nodeinfo path
 811          * @return array Server data
 812          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 813          */
 814         private static function parseNodeinfo1(string $nodeinfo_url)
 815         {
 816                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 817                 if (!$curlResult->isSuccess()) {
 818                         return [];
 819                 }
 820
 821                 $nodeinfo = json_decode($curlResult->getBody(), true);
 822
 823                 if (!is_array($nodeinfo)) {
 824                         return [];
 825                 }
 826
 827                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
 828                         'register_policy' => Register::CLOSED];
 829
 830                 if (!empty($nodeinfo['openRegistrations'])) {
 831                         $server['register_policy'] = Register::OPEN;
 832                 }
 833
 834                 if (is_array($nodeinfo['software'])) {
 835                         if (!empty($nodeinfo['software']['name'])) {
 836                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 837                         }
 838
 839                         if (!empty($nodeinfo['software']['version'])) {
 840                                 $server['version'] = $nodeinfo['software']['version'];
 841                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 842                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 843                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 844                         }
 845                 }
 846
 847                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 848                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 849                 }
 850
 851                 if (!empty($nodeinfo['usage']['users']['total'])) {
 852                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
 853                 }
 854
 855                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
 856                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
 857                 }
 858
 859                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
 860                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
 861                 }
 862
 863                 if (!empty($nodeinfo['usage']['localPosts'])) {
 864                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
 865                 }
 866
 867                 if (!empty($nodeinfo['usage']['localComments'])) {
 868                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
 869                 }
 870
 871                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
 872                         $protocols = [];
 873                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
 874                                 $protocols[$protocol] = true;
 875                         }
 876
 877                         if (!empty($protocols['friendica'])) {
 878                                 $server['network'] = Protocol::DFRN;
 879                         } elseif (!empty($protocols['activitypub'])) {
 880                                 $server['network'] = Protocol::ACTIVITYPUB;
 881                         } elseif (!empty($protocols['diaspora'])) {
 882                                 $server['network'] = Protocol::DIASPORA;
 883                         } elseif (!empty($protocols['ostatus'])) {
 884                                 $server['network'] = Protocol::OSTATUS;
 885                         } elseif (!empty($protocols['gnusocial'])) {
 886                                 $server['network'] = Protocol::OSTATUS;
 887                         } elseif (!empty($protocols['zot'])) {
 888                                 $server['network'] = Protocol::ZOT;
 889                         }
 890                 }
 891
 892                 if (empty($server)) {
 893                         return [];
 894                 }
 895
 896                 return $server;
 897         }
 898
 899         /**
 900          * Parses Nodeinfo 2
 901          *
 902          * @see https://git.feneas.org/jaywink/nodeinfo2
 903          * @param string $nodeinfo_url address of the nodeinfo path
 904          * @return array Server data
 905          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 906          */
 907         private static function parseNodeinfo2(string $nodeinfo_url)
 908         {
 909                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 910                 if (!$curlResult->isSuccess()) {
 911                         return [];
 912                 }
 913
 914                 $nodeinfo = json_decode($curlResult->getBody(), true);
 915
 916                 if (!is_array($nodeinfo)) {
 917                         return [];
 918                 }
 919
 920                 $server = ['detection-method' => self::DETECT_NODEINFO_2,
 921                         'register_policy' => Register::CLOSED];
 922
 923                 if (!empty($nodeinfo['openRegistrations'])) {
 924                         $server['register_policy'] = Register::OPEN;
 925                 }
 926
 927                 if (is_array($nodeinfo['software'])) {
 928                         if (!empty($nodeinfo['software']['name'])) {
 929                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 930                         }
 931
 932                         if (!empty($nodeinfo['software']['version'])) {
 933                                 $server['version'] = $nodeinfo['software']['version'];
 934                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 935                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 936                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 937
 938                                 // qoto advertises itself as Mastodon
 939                                 if (($server['platform'] == 'mastodon') && substr($nodeinfo['software']['version'], -5) == '-qoto') {
 940                                         $server['platform'] = 'qoto';
 941                                 }
 942                         }
 943                 }
 944
 945                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 946                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 947                 }
 948
 949                 if (!empty($nodeinfo['usage']['users']['total'])) {
 950                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
 951                 }
 952
 953                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
 954                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
 955                 }
 956
 957                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
 958                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
 959                 }
 960
 961                 if (!empty($nodeinfo['usage']['localPosts'])) {
 962                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
 963                 }
 964
 965                 if (!empty($nodeinfo['usage']['localComments'])) {
 966                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
 967                 }
 968
 969                 if (!empty($nodeinfo['protocols'])) {
 970                         $protocols = [];
 971                         foreach ($nodeinfo['protocols'] as $protocol) {
 972                                 if (is_string($protocol)) {
 973                                         $protocols[$protocol] = true;
 974                                 }
 975                         }
 976
 977                         if (!empty($protocols['dfrn'])) {
 978                                 $server['network'] = Protocol::DFRN;
 979                         } elseif (!empty($protocols['activitypub'])) {
 980                                 $server['network'] = Protocol::ACTIVITYPUB;
 981                         } elseif (!empty($protocols['diaspora'])) {
 982                                 $server['network'] = Protocol::DIASPORA;
 983                         } elseif (!empty($protocols['ostatus'])) {
 984                                 $server['network'] = Protocol::OSTATUS;
 985                         } elseif (!empty($protocols['gnusocial'])) {
 986                                 $server['network'] = Protocol::OSTATUS;
 987                         } elseif (!empty($protocols['zot'])) {
 988                                 $server['network'] = Protocol::ZOT;
 989                         }
 990                 }
 991
 992                 if (empty($server)) {
 993                         return [];
 994                 }
 995
 996                 return $server;
 997         }
 998
 999         /**
1000          * Fetch server information from a 'siteinfo.json' file on the given server
1001          *
1002          * @param string $url        URL of the given server
1003          * @param array  $serverdata array with server data
1004          *
1005          * @return array server data
1006          */
1007         private static function fetchSiteinfo(string $url, array $serverdata)
1008         {
1009                 $curlResult = DI::httpClient()->get($url . '/siteinfo.json', HttpClientAccept::JSON);
1010                 if (!$curlResult->isSuccess()) {
1011                         return $serverdata;
1012                 }
1013
1014                 $data = json_decode($curlResult->getBody(), true);
1015                 if (empty($data)) {
1016                         return $serverdata;
1017                 }
1018
1019                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1020                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
1021                 }
1022
1023                 if (!empty($data['url'])) {
1024                         $serverdata['platform'] = strtolower($data['platform']);
1025                         $serverdata['version'] = $data['version'];
1026                 }
1027
1028                 if (!empty($data['plugins'])) {
1029                         if (in_array('pubcrawl', $data['plugins'])) {
1030                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1031                         } elseif (in_array('diaspora', $data['plugins'])) {
1032                                 $serverdata['network'] = Protocol::DIASPORA;
1033                         } elseif (in_array('gnusoc', $data['plugins'])) {
1034                                 $serverdata['network'] = Protocol::OSTATUS;
1035                         } else {
1036                                 $serverdata['network'] = Protocol::ZOT;
1037                         }
1038                 }
1039
1040                 if (!empty($data['site_name'])) {
1041                         $serverdata['site_name'] = $data['site_name'];
1042                 }
1043
1044                 if (!empty($data['channels_total'])) {
1045                         $serverdata['registered-users'] = max($data['channels_total'], 1);
1046                 }
1047
1048                 if (!empty($data['channels_active_monthly'])) {
1049                         $serverdata['active-month-users'] = max($data['channels_active_monthly'], 0);
1050                 }
1051
1052                 if (!empty($data['channels_active_halfyear'])) {
1053                         $serverdata['active-halfyear-users'] = max($data['channels_active_halfyear'], 0);
1054                 }
1055
1056                 if (!empty($data['local_posts'])) {
1057                         $serverdata['local-posts'] = max($data['local_posts'], 0);
1058                 }
1059
1060                 if (!empty($data['local_comments'])) {
1061                         $serverdata['local-comments'] = max($data['local_comments'], 0);
1062                 }
1063
1064                 if (!empty($data['register_policy'])) {
1065                         switch ($data['register_policy']) {
1066                                 case 'REGISTER_OPEN':
1067                                         $serverdata['register_policy'] = Register::OPEN;
1068                                         break;
1069
1070                                 case 'REGISTER_APPROVE':
1071                                         $serverdata['register_policy'] = Register::APPROVE;
1072                                         break;
1073
1074                                 case 'REGISTER_CLOSED':
1075                                 default:
1076                                         $serverdata['register_policy'] = Register::CLOSED;
1077                                         break;
1078                         }
1079                 }
1080
1081                 return $serverdata;
1082         }
1083
1084         /**
1085          * Checks if the server contains a valid host meta file
1086          *
1087          * @param string $url URL of the given server
1088          *
1089          * @return boolean 'true' if the server seems to be vital
1090          */
1091         private static function validHostMeta(string $url)
1092         {
1093                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
1094                 $curlResult = DI::httpClient()->get($url . '/.well-known/host-meta', HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
1095                 if (!$curlResult->isSuccess()) {
1096                         return false;
1097                 }
1098
1099                 $xrd = XML::parseString($curlResult->getBody());
1100                 if (!is_object($xrd)) {
1101                         return false;
1102                 }
1103
1104                 $elements = XML::elementToArray($xrd);
1105                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
1106                         return false;
1107                 }
1108
1109                 $valid = false;
1110                 foreach ($elements['xrd']['link'] as $link) {
1111                         // When there is more than a single "link" element, the array looks slightly different
1112                         if (!empty($link['@attributes'])) {
1113                                 $link = $link['@attributes'];
1114                         }
1115
1116                         if (empty($link['rel']) || empty($link['template'])) {
1117                                 continue;
1118                         }
1119
1120                         if ($link['rel'] == 'lrdd') {
1121                                 // When the webfinger host is the same like the system host, it should be ok.
1122                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
1123                         }
1124                 }
1125
1126                 return $valid;
1127         }
1128
1129         /**
1130          * Detect the network of the given server via their known contacts
1131          *
1132          * @param string $url        URL of the given server
1133          * @param array  $serverdata array with server data
1134          *
1135          * @return array server data
1136          */
1137         private static function detectNetworkViaContacts(string $url, array $serverdata)
1138         {
1139                 $contacts = [];
1140
1141                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $serverdata['nurl']]]);
1142                 while ($apcontact = DBA::fetch($apcontacts)) {
1143                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
1144                 }
1145                 DBA::close($apcontacts);
1146
1147                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $serverdata['nurl']]]);
1148                 while ($pcontact = DBA::fetch($pcontacts)) {
1149                         $contacts[$pcontact['nurl']] = $pcontact['url'];
1150                 }
1151                 DBA::close($pcontacts);
1152
1153                 if (empty($contacts)) {
1154                         return $serverdata;
1155                 }
1156
1157                 $time = time();
1158                 foreach ($contacts as $contact) {
1159                         $probed = Contact::getByURL($contact, true);
1160                         if (!empty($probed) && !$probed['failed'] && in_array($probed['network'], Protocol::FEDERATED)) {
1161                                 $serverdata['network'] = $probed['network'];
1162                                 break;
1163                         } elseif ((time() - $time) > 10) {
1164                                 // To reduce the stress on remote systems we probe a maximum of 10 seconds
1165                                 break;
1166                         }
1167                 }
1168
1169                 return $serverdata;
1170         }
1171
1172         /**
1173          * Checks if the given server does have a '/poco' endpoint.
1174          * This is used for the 'PortableContact' functionality,
1175          * which is used by both Friendica and Hubzilla.
1176          *
1177          * @param string $url        URL of the given server
1178          * @param array  $serverdata array with server data
1179          *
1180          * @return array server data
1181          */
1182         private static function checkPoCo(string $url, array $serverdata)
1183         {
1184                 $serverdata['poco'] = '';
1185
1186                 $curlResult = DI::httpClient()->get($url . '/poco', HttpClientAccept::JSON);
1187                 if (!$curlResult->isSuccess()) {
1188                         return $serverdata;
1189                 }
1190
1191                 $data = json_decode($curlResult->getBody(), true);
1192                 if (empty($data)) {
1193                         return $serverdata;
1194                 }
1195
1196                 if (!empty($data['totalResults'])) {
1197                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1198                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers, 1);
1199                         $serverdata['directory-type'] = self::DT_POCO;
1200                         $serverdata['poco'] = $url . '/poco';
1201                 }
1202
1203                 return $serverdata;
1204         }
1205
1206         /**
1207          * Checks if the given server does have a Mastodon style directory endpoint.
1208          *
1209          * @param string $url        URL of the given server
1210          * @param array  $serverdata array with server data
1211          *
1212          * @return array server data
1213          */
1214         public static function checkMastodonDirectory(string $url, array $serverdata)
1215         {
1216                 $curlResult = DI::httpClient()->get($url . '/api/v1/directory?limit=1', HttpClientAccept::JSON);
1217                 if (!$curlResult->isSuccess()) {
1218                         return $serverdata;
1219                 }
1220
1221                 $data = json_decode($curlResult->getBody(), true);
1222                 if (empty($data)) {
1223                         return $serverdata;
1224                 }
1225
1226                 if (count($data) == 1) {
1227                         $serverdata['directory-type'] = self::DT_MASTODON;
1228                 }
1229
1230                 return $serverdata;
1231         }
1232
1233         /**
1234          * Detects Peertube via their known endpoint
1235          *
1236          * @param string $url        URL of the given server
1237          * @param array  $serverdata array with server data
1238          *
1239          * @return array server data
1240          */
1241         private static function detectPeertube(string $url, array $serverdata)
1242         {
1243                 $curlResult = DI::httpClient()->get($url . '/api/v1/config', HttpClientAccept::JSON);
1244                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1245                         return $serverdata;
1246                 }
1247
1248                 $data = json_decode($curlResult->getBody(), true);
1249                 if (empty($data)) {
1250                         return $serverdata;
1251                 }
1252
1253                 if (!empty($data['instance']) && !empty($data['serverVersion'])) {
1254                         $serverdata['platform'] = 'peertube';
1255                         $serverdata['version'] = $data['serverVersion'];
1256                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1257
1258                         if (!empty($data['instance']['name'])) {
1259                                 $serverdata['site_name'] = $data['instance']['name'];
1260                         }
1261
1262                         if (!empty($data['instance']['shortDescription'])) {
1263                                 $serverdata['info'] = $data['instance']['shortDescription'];
1264                         }
1265
1266                         if (!empty($data['signup'])) {
1267                                 if (!empty($data['signup']['allowed'])) {
1268                                         $serverdata['register_policy'] = Register::OPEN;
1269                                 }
1270                         }
1271
1272                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1273                                 $serverdata['detection-method'] = self::DETECT_V1_CONFIG;
1274                         }
1275                 }
1276
1277                 return $serverdata;
1278         }
1279
1280         /**
1281          * Detects the version number of a given server when it was a NextCloud installation
1282          *
1283          * @param string $url        URL of the given server
1284          * @param array  $serverdata array with server data
1285          *
1286          * @return array server data
1287          */
1288         private static function detectNextcloud(string $url, array $serverdata)
1289         {
1290                 $curlResult = DI::httpClient()->get($url . '/status.php', HttpClientAccept::JSON);
1291                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1292                         return $serverdata;
1293                 }
1294
1295                 $data = json_decode($curlResult->getBody(), true);
1296                 if (empty($data)) {
1297                         return $serverdata;
1298                 }
1299
1300                 if (!empty($data['version'])) {
1301                         $serverdata['platform'] = 'nextcloud';
1302                         $serverdata['version'] = $data['version'];
1303                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1304
1305                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1306                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1307                         }
1308                 }
1309
1310                 return $serverdata;
1311         }
1312
1313         private static function fetchWeeklyUsage(string $url, array $serverdata) {
1314                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance/activity', HttpClientAccept::JSON);
1315                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1316                         return $serverdata;
1317                 }
1318
1319                 $data = json_decode($curlResult->getBody(), true);
1320                 if (empty($data)) {
1321                         return $serverdata;
1322                 }
1323
1324                 $current_week = [];
1325                 foreach ($data as $week) {
1326                         // Use only data from a full week
1327                         if (empty($week['week']) || (time() - $week['week']) < 7 * 24 * 60 * 60) {
1328                                 continue;
1329                         }
1330
1331                         // Most likely the data is sorted correctly. But we better are safe than sorry
1332                         if (empty($current_week['week']) || ($current_week['week'] < $week['week'])) {
1333                                 $current_week = $week;
1334                         }
1335                 }
1336
1337                 if (!empty($current_week['logins'])) {
1338                         $serverdata['active-week-users'] = max($current_week['logins'], 0);
1339                 }
1340
1341                 return $serverdata;
1342         }
1343
1344         /**
1345          * Detects the server network type from contacts of that server
1346          *
1347          * @param string $url        URL of the given server
1348          * @param array  $serverdata array with server data
1349          *
1350          * @return array server data
1351          */
1352         private static function detectFromContacts(string $url, array $serverdata)
1353         {
1354                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
1355                 if (empty($gserver)) {
1356                         return $serverdata;
1357                 }
1358
1359                 $contact = Contact::selectFirst(['id'], ['uid' => 0, 'failed' => false, 'gsid' => $gserver['id']]);
1360
1361                 // Via probing we can be sure that the server is responding
1362                 if (!empty($contact['id']) && Contact::updateFromProbe($contact['id'])) {
1363                         $contact = Contact::selectFirst(['network', 'failed'], ['id' => $contact['id']]);
1364                         if (!$contact['failed'] && in_array($contact['network'], Protocol::FEDERATED)) {
1365                                 $serverdata['network'] = $contact['network'];
1366                         }
1367                 }
1368
1369                 return $serverdata;
1370         }
1371
1372         /**
1373          * Detects data from a given server url if it was a mastodon alike system
1374          *
1375          * @param string $url        URL of the given server
1376          * @param array  $serverdata array with server data
1377          *
1378          * @return array server data
1379          */
1380         private static function detectMastodonAlikes(string $url, array $serverdata)
1381         {
1382                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance', HttpClientAccept::JSON);
1383                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1384                         return $serverdata;
1385                 }
1386
1387                 $data = json_decode($curlResult->getBody(), true);
1388                 if (empty($data)) {
1389                         return $serverdata;
1390                 }
1391
1392                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1393                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1394                 }
1395
1396                 if (!empty($data['version'])) {
1397                         $serverdata['platform'] = 'mastodon';
1398                         $serverdata['version'] = $data['version'] ?? '';
1399                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1400                 }
1401
1402                 if (!empty($data['title'])) {
1403                         $serverdata['site_name'] = $data['title'];
1404                 }
1405
1406                 if (!empty($data['title']) && empty($serverdata['platform']) && empty($serverdata['network'])) {
1407                         $serverdata['platform'] = 'mastodon';
1408                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1409                 }
1410
1411                 if (!empty($data['description'])) {
1412                         $serverdata['info'] = trim($data['description']);
1413                 }
1414
1415                 if (!empty($data['stats']['user_count'])) {
1416                         $serverdata['registered-users'] = max($data['stats']['user_count'], 1);
1417                 }
1418
1419                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1420                         $serverdata['platform'] = strtolower($matches[1]);
1421                         $serverdata['version'] = $matches[2];
1422                 }
1423
1424                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1425                         $serverdata['platform'] = 'pleroma';
1426                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1427                 }
1428
1429                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1430                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1431                         $serverdata['platform'] = 'pleroma';
1432                 }
1433
1434                 return $serverdata;
1435         }
1436
1437         /**
1438          * Detects data from typical Hubzilla endpoints
1439          *
1440          * @param string $url        URL of the given server
1441          * @param array  $serverdata array with server data
1442          *
1443          * @return array server data
1444          */
1445         private static function detectHubzilla(string $url, array $serverdata)
1446         {
1447                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/config.json', HttpClientAccept::JSON);
1448                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1449                         return $serverdata;
1450                 }
1451
1452                 $data = json_decode($curlResult->getBody(), true);
1453                 if (empty($data) || empty($data['site'])) {
1454                         return $serverdata;
1455                 }
1456
1457                 if (!empty($data['site']['name'])) {
1458                         $serverdata['site_name'] = $data['site']['name'];
1459                 }
1460
1461                 if (!empty($data['site']['platform'])) {
1462                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1463                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1464                         $serverdata['network'] = Protocol::ZOT;
1465                 }
1466
1467                 if (!empty($data['site']['hubzilla'])) {
1468                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1469                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1470                         $serverdata['network'] = Protocol::ZOT;
1471                 }
1472
1473                 if (!empty($data['site']['redmatrix'])) {
1474                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1475                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1476                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1477                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1478                         }
1479
1480                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1481                         $serverdata['network'] = Protocol::ZOT;
1482                 }
1483
1484                 $private = false;
1485                 $inviteonly = false;
1486                 $closed = false;
1487
1488                 if (!empty($data['site']['closed'])) {
1489                         $closed = self::toBoolean($data['site']['closed']);
1490                 }
1491
1492                 if (!empty($data['site']['private'])) {
1493                         $private = self::toBoolean($data['site']['private']);
1494                 }
1495
1496                 if (!empty($data['site']['inviteonly'])) {
1497                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1498                 }
1499
1500                 if (!$closed && !$private and $inviteonly) {
1501                         $serverdata['register_policy'] = Register::APPROVE;
1502                 } elseif (!$closed && !$private) {
1503                         $serverdata['register_policy'] = Register::OPEN;
1504                 } else {
1505                         $serverdata['register_policy'] = Register::CLOSED;
1506                 }
1507
1508                 if (!empty($serverdata['network']) && in_array($serverdata['detection-method'],
1509                         [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1510                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1511                 }
1512
1513                 return $serverdata;
1514         }
1515
1516         /**
1517          * Converts input value to a boolean value
1518          *
1519          * @param string|integer $val
1520          *
1521          * @return boolean
1522          */
1523         private static function toBoolean($val)
1524         {
1525                 if (($val == 'true') || ($val == 1)) {
1526                         return true;
1527                 } elseif (($val == 'false') || ($val == 0)) {
1528                         return false;
1529                 }
1530
1531                 return $val;
1532         }
1533
1534         /**
1535          * Detect if the URL belongs to a pump.io server
1536          *
1537          * @param string $url        URL of the given server
1538          * @param array  $serverdata array with server data
1539          *
1540          * @return array server data
1541          */
1542         private static function detectPumpIO(string $url, array $serverdata)
1543         {
1544                 $curlResult = DI::httpClient()->get($url . '/.well-known/host-meta.json', HttpClientAccept::JSON);
1545                 if (!$curlResult->isSuccess()) {
1546                         return $serverdata;
1547                 }
1548
1549                 $data = json_decode($curlResult->getBody(), true);
1550                 if (empty($data['links'])) {
1551                         return $serverdata;
1552
1553                 }
1554
1555                 // We are looking for some endpoints that are typical for pump.io
1556                 $trust = 0;
1557                 foreach ($data['links'] as $link) {
1558                         if (empty($link['rel'])) {
1559                                 continue;
1560                         }
1561                         if (in_array($link['rel'], ['registration_endpoint', 'dialback', 'http://apinamespace.org/activitypub/whoami'])) {
1562                                 ++$trust;
1563                         }
1564                 }
1565
1566                 if ($trust == 3) {
1567                         $serverdata['detection-method'] = self::DETECT_PUMPIO;
1568
1569                         $serverdata['platform'] = 'pumpio';
1570                         $serverdata['version']  = '';
1571                         $serverdata['network']  = Protocol::PUMPIO;
1572
1573                         $servers = $curlResult->getHeader('Server');
1574                         foreach ($servers as $server) {
1575                                 if (preg_match("#pump.io/(.*)\s#U", $server, $matches)) {
1576                                         $serverdata['version']  = $matches[1];
1577                                 }
1578                         }
1579                 }
1580
1581                 return $serverdata;
1582         }
1583
1584         /**
1585          * Detect if the URL belongs to a GNU Social server
1586          *
1587          * @param string $url        URL of the given server
1588          * @param array  $serverdata array with server data
1589          *
1590          * @return array server data
1591          */
1592         private static function detectGNUSocial(string $url, array $serverdata)
1593         {
1594                 // Test for GNU Social
1595                 $curlResult = DI::httpClient()->get($url . '/api/gnusocial/version.json', HttpClientAccept::JSON);
1596                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1597                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1598                         $serverdata['platform'] = 'gnusocial';
1599                         // Remove junk that some GNU Social servers return
1600                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1601                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1602                         $serverdata['version'] = trim($serverdata['version'], '"');
1603                         $serverdata['network'] = Protocol::OSTATUS;
1604
1605                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1606                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1607                         }
1608
1609                         return $serverdata;
1610                 }
1611
1612                 // Test for Statusnet
1613                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/version.json', HttpClientAccept::JSON);
1614                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1615                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1616
1617                         // Remove junk that some GNU Social servers return
1618                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1619                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1620                         $serverdata['version'] = trim($serverdata['version'], '"');
1621
1622                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1623                                 $serverdata['platform'] = 'pleroma';
1624                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1625                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1626                         } else {
1627                                 $serverdata['platform'] = 'statusnet';
1628                                 $serverdata['network'] = Protocol::OSTATUS;
1629                         }
1630
1631                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1632                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1633                         }
1634                 }
1635
1636                 return $serverdata;
1637         }
1638
1639         /**
1640          * Detect if the URL belongs to a Friendica server
1641          *
1642          * @param string $url        URL of the given server
1643          * @param array  $serverdata array with server data
1644          *
1645          * @return array server data
1646          */
1647         private static function detectFriendica(string $url, array $serverdata)
1648         {
1649                 // There is a bug in some versions of Friendica that will return an ActivityStream actor when the content type "application/json" is requested.
1650                 // Because of this me must not use ACCEPT_JSON here.
1651                 $curlResult = DI::httpClient()->get($url . '/friendica/json');
1652                 if (!$curlResult->isSuccess()) {
1653                         $curlResult = DI::httpClient()->get($url . '/friendika/json');
1654                         $friendika = true;
1655                         $platform = 'Friendika';
1656                 } else {
1657                         $friendika = false;
1658                         $platform = 'Friendica';
1659                 }
1660
1661                 if (!$curlResult->isSuccess()) {
1662                         return $serverdata;
1663                 }
1664
1665                 $data = json_decode($curlResult->getBody(), true);
1666                 if (empty($data) || empty($data['version'])) {
1667                         return $serverdata;
1668                 }
1669
1670                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1671                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1672                 }
1673
1674                 $serverdata['network'] = Protocol::DFRN;
1675                 $serverdata['version'] = $data['version'];
1676
1677                 if (!empty($data['no_scrape_url'])) {
1678                         $serverdata['noscrape'] = $data['no_scrape_url'];
1679                 }
1680
1681                 if (!empty($data['site_name'])) {
1682                         $serverdata['site_name'] = $data['site_name'];
1683                 }
1684
1685                 if (!empty($data['info'])) {
1686                         $serverdata['info'] = trim($data['info']);
1687                 }
1688
1689                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1690                 switch ($register_policy) {
1691                         case 'REGISTER_OPEN':
1692                                 $serverdata['register_policy'] = Register::OPEN;
1693                                 break;
1694
1695                         case 'REGISTER_APPROVE':
1696                                 $serverdata['register_policy'] = Register::APPROVE;
1697                                 break;
1698
1699                         case 'REGISTER_CLOSED':
1700                         case 'REGISTER_INVITATION':
1701                                 $serverdata['register_policy'] = Register::CLOSED;
1702                                 break;
1703                         default:
1704                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1705                                 $serverdata['register_policy'] = Register::CLOSED;
1706                                 break;
1707                 }
1708
1709                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1710
1711                 return $serverdata;
1712         }
1713
1714         /**
1715          * Analyses the landing page of a given server for hints about type and system of that server
1716          *
1717          * @param object $curlResult result of curl execution
1718          * @param array  $serverdata array with server data
1719          * @param string $url        Server URL
1720          *
1721          * @return array server data
1722          */
1723         private static function analyseRootBody($curlResult, array $serverdata, string $url)
1724         {
1725                 if (empty($curlResult->getBody())) {
1726                         return $serverdata;
1727                 }
1728
1729                 // Using only body information we cannot safely detect a lot of systems.
1730                 // So we define a list of platforms that we can detect safely.
1731                 $valid_platforms = ['friendica', 'friendika', 'diaspora', 'mastodon', 'hubzilla', 'misskey', 'peertube', 'wordpress', 'write.as'];
1732
1733                 $doc = new DOMDocument();
1734                 @$doc->loadHTML($curlResult->getBody());
1735                 $xpath = new DOMXPath($doc);
1736
1737                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1738                 if (!empty($title)) {
1739                         $serverdata['site_name'] = $title;
1740                 }
1741
1742                 $list = $xpath->query('//meta[@name]');
1743
1744                 foreach ($list as $node) {
1745                         $attr = [];
1746                         if ($node->attributes->length) {
1747                                 foreach ($node->attributes as $attribute) {
1748                                         $value = trim($attribute->value);
1749                                         if (empty($value)) {
1750                                                 continue;
1751                                         }
1752
1753                                         $attr[$attribute->name] = $value;
1754                                 }
1755
1756                                 if (empty($attr['name']) || empty($attr['content'])) {
1757                                         continue;
1758                                 }
1759                         }
1760
1761                         if ($attr['name'] == 'description') {
1762                                 $serverdata['info'] = $attr['content'];
1763                         }
1764
1765                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1766                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad'])) {
1767                                 $serverdata['platform'] = strtolower($attr['content']);
1768                                 if (in_array($attr['content'], ['Misskey', 'Write.as'])) {
1769                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1770                                 }
1771                         }
1772                         if (($attr['name'] == 'generator') && (empty($serverdata['platform']) || (substr(strtolower($attr['content']), 0, 9) == 'wordpress'))) {
1773                                 $serverdata['platform'] = strtolower($attr['content']);
1774                                 $version_part = explode(' ', $attr['content']);
1775
1776                                 if (count($version_part) == 2) {
1777                                         if (in_array($version_part[0], ['WordPress'])) {
1778                                                 $serverdata['platform'] = 'wordpress';
1779                                                 $serverdata['version'] = $version_part[1];
1780
1781                                                 // We still do need a reliable test if some AP plugin is activated
1782                                                 // By now we just check in a later process for some known contacts
1783                                                 $serverdata['network'] = Protocol::FEED;
1784
1785                                                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1786                                                         $serverdata['detection-method'] = self::DETECT_BODY;
1787                                                 }
1788                                         }
1789                                         if (in_array($version_part[0], ['Friendika', 'Friendica'])) {
1790                                                 $serverdata['platform'] = strtolower($version_part[0]);
1791                                                 $serverdata['version'] = $version_part[1];
1792                                                 $serverdata['network'] = Protocol::DFRN;
1793                                         }
1794                                 }
1795                         }
1796                 }
1797
1798                 $list = $xpath->query('//meta[@property]');
1799
1800                 foreach ($list as $node) {
1801                         $attr = [];
1802                         if ($node->attributes->length) {
1803                                 foreach ($node->attributes as $attribute) {
1804                                         $value = trim($attribute->value);
1805                                         if (empty($value)) {
1806                                                 continue;
1807                                         }
1808
1809                                         $attr[$attribute->name] = $value;
1810                                 }
1811
1812                                 if (empty($attr['property']) || empty($attr['content'])) {
1813                                         continue;
1814                                 }
1815                         }
1816
1817                         if ($attr['property'] == 'og:site_name') {
1818                                 $serverdata['site_name'] = $attr['content'];
1819                         }
1820
1821                         if ($attr['property'] == 'og:description') {
1822                                 $serverdata['info'] = $attr['content'];
1823                         }
1824
1825                         if ($attr['property'] == 'og:platform') {
1826                                 $serverdata['platform'] = strtolower($attr['content']);
1827
1828                                 if (in_array($attr['content'], ['PeerTube'])) {
1829                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1830                                 }
1831                         }
1832
1833                         if ($attr['property'] == 'generator') {
1834                                 $serverdata['platform'] = strtolower($attr['content']);
1835
1836                                 if (in_array($attr['content'], ['hubzilla'])) {
1837                                         // We later check which compatible protocol modules are loaded.
1838                                         $serverdata['network'] = Protocol::ZOT;
1839                                 }
1840                         }
1841                 }
1842
1843                 if (!empty($serverdata['platform']) && in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_BODY]) && !in_array($serverdata['platform'], $valid_platforms)) {
1844                         $serverdata['network'] = Protocol::PHANTOM;
1845                         $serverdata['version'] = '';
1846                         $serverdata['detection-method'] = self::DETECT_MANUAL;
1847                 } elseif (!empty($serverdata['network']) && ($serverdata['detection-method'] == self::DETECT_MANUAL)) {
1848                         $serverdata['detection-method'] = self::DETECT_BODY;
1849                 }
1850
1851                 return $serverdata;
1852         }
1853
1854         /**
1855          * Analyses the header data of a given server for hints about type and system of that server
1856          *
1857          * @param object $curlResult result of curl execution
1858          * @param array  $serverdata array with server data
1859          *
1860          * @return array server data
1861          */
1862         private static function analyseRootHeader($curlResult, array $serverdata)
1863         {
1864                 if ($curlResult->getHeader('server') == 'Mastodon') {
1865                         $serverdata['platform'] = 'mastodon';
1866                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1867                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
1868                         $serverdata['platform'] = 'diaspora';
1869                         $serverdata['network'] = Protocol::DIASPORA;
1870                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version')[0] ?? '';
1871                 } elseif ($curlResult->inHeader('x-friendica-version')) {
1872                         $serverdata['platform'] = 'friendica';
1873                         $serverdata['network'] = Protocol::DFRN;
1874                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version')[0] ?? '';
1875                 } else {
1876                         return $serverdata;
1877                 }
1878
1879                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1880                         $serverdata['detection-method'] = self::DETECT_HEADER;
1881                 }
1882
1883                 return $serverdata;
1884         }
1885
1886         /**
1887          * Test if the body contains valid content
1888          *
1889          * @param string $body
1890          * @return boolean
1891          */
1892         private static function invalidBody(string $body)
1893         {
1894                 // Currently we only test for a HTML element.
1895                 // Possibly we enhance this in the future.
1896                 return !strpos($body, '>');
1897         }
1898
1899         /**
1900          * Update GServer entries
1901          */
1902         public static function discover()
1903         {
1904                 // Update the server list
1905                 self::discoverFederation();
1906
1907                 $no_of_queries = 5;
1908
1909                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
1910
1911                 if ($requery_days == 0) {
1912                         $requery_days = 7;
1913                 }
1914
1915                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
1916
1917                 $gservers = DBA::select('gserver', ['id', 'url', 'nurl', 'network', 'poco', 'directory-type'],
1918                         ["NOT `failed` AND `directory-type` != ? AND `last_poco_query` < ?", GServer::DT_NONE, $last_update],
1919                         ['order' => ['RAND()']]);
1920
1921                 while ($gserver = DBA::fetch($gservers)) {
1922                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
1923                         Worker::add(PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
1924
1925                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
1926                         Worker::add(PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
1927
1928                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1929                         DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1930
1931                         if (--$no_of_queries == 0) {
1932                                 break;
1933                         }
1934                 }
1935
1936                 DBA::close($gservers);
1937         }
1938
1939         /**
1940          * Discover federated servers
1941          */
1942         private static function discoverFederation()
1943         {
1944                 $last = DI::config()->get('poco', 'last_federation_discovery');
1945
1946                 if ($last) {
1947                         $next = $last + (24 * 60 * 60);
1948
1949                         if ($next > time()) {
1950                                 return;
1951                         }
1952                 }
1953
1954                 // Discover federated servers
1955                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
1956                 foreach ($protocols as $protocol) {
1957                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
1958                         $curlResult = DI::httpClient()->fetch('https://the-federation.info/graphql?query=' . urlencode($query), HttpClientAccept::JSON);
1959                         if (!empty($curlResult)) {
1960                                 $data = json_decode($curlResult, true);
1961                                 if (!empty($data['data']['nodes'])) {
1962                                         foreach ($data['data']['nodes'] as $server) {
1963                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
1964                                                 self::add('https://' . $server['host'], true);
1965                                         }
1966                                 }
1967                         }
1968                 }
1969
1970                 // Disvover Mastodon servers
1971                 $accesstoken = DI::config()->get('system', 'instances_social_key');
1972
1973                 if (!empty($accesstoken)) {
1974                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
1975                         $curlResult = DI::httpClient()->get($api, HttpClientAccept::JSON, [HttpClientOptions::HEADERS => ['Authorization' => ['Bearer ' . $accesstoken]]]);
1976                         if ($curlResult->isSuccess()) {
1977                                 $servers = json_decode($curlResult->getBody(), true);
1978
1979                                 foreach ($servers['instances'] as $server) {
1980                                         $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
1981                                         self::add($url);
1982                                 }
1983                         }
1984                 }
1985
1986                 DI::config()->set('poco', 'last_federation_discovery', time());
1987         }
1988
1989         /**
1990          * Set the protocol for the given server
1991          *
1992          * @param int $gsid     Server id
1993          * @param int $protocol Protocol id
1994          * @return void
1995          * @throws Exception
1996          */
1997         public static function setProtocol(int $gsid, int $protocol)
1998         {
1999                 if (empty($gsid)) {
2000                         return;
2001                 }
2002
2003                 $gserver = DBA::selectFirst('gserver', ['protocol', 'url'], ['id' => $gsid]);
2004                 if (!DBA::isResult($gserver)) {
2005                         return;
2006                 }
2007
2008                 $old = $gserver['protocol'];
2009
2010                 if (!is_null($old)) {
2011                         /*
2012                         The priority for the protocols is:
2013                                 1. ActivityPub
2014                                 2. DFRN via Diaspora
2015                                 3. Legacy DFRN
2016                                 4. Diaspora
2017                                 5. OStatus
2018                         */
2019
2020                         // We don't need to change it when nothing is to be changed
2021                         if ($old == $protocol) {
2022                                 return;
2023                         }
2024
2025                         // We don't want to mark a server as OStatus when it had been marked with any other protocol before
2026                         if ($protocol == Post\DeliveryData::OSTATUS) {
2027                                 return;
2028                         }
2029
2030                         // If the server is marked as ActivityPub then we won't change it to anything different
2031                         if ($old == Post\DeliveryData::ACTIVITYPUB) {
2032                                 return;
2033                         }
2034
2035                         // Don't change it to anything lower than DFRN if the new one wasn't ActivityPub
2036                         if (($old == Post\DeliveryData::DFRN) && ($protocol != Post\DeliveryData::ACTIVITYPUB)) {
2037                                 return;
2038                         }
2039
2040                         // Don't change it to Diaspora when it is a legacy DFRN server
2041                         if (($old == Post\DeliveryData::LEGACY_DFRN) && ($protocol == Post\DeliveryData::DIASPORA)) {
2042                                 return;
2043                         }
2044                 }
2045
2046                 Logger::info('Protocol for server', ['protocol' => $protocol, 'old' => $old, 'id' => $gsid, 'url' => $gserver['url'], 'callstack' => System::callstack(20)]);
2047                 DBA::update('gserver', ['protocol' => $protocol], ['id' => $gsid]);
2048         }
2049
2050         /**
2051          * Fetch the protocol of the given server
2052          *
2053          * @param int $gsid Server id
2054          * @return int
2055          * @throws Exception
2056          */
2057         public static function getProtocol(int $gsid)
2058         {
2059                 if (empty($gsid)) {
2060                         return null;
2061                 }
2062
2063                 $gserver = DBA::selectFirst('gserver', ['protocol'], ['id' => $gsid]);
2064                 if (DBA::isResult($gserver)) {
2065                         return $gserver['protocol'];
2066                 }
2067
2068                 return null;
2069         }
2070 }