src/Model/GServer.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2022, the Friendica project
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Model;
  23
  24 use DOMDocument;
  25 use DOMXPath;
  26 use Exception;
  27 use Friendica\Core\Logger;
  28 use Friendica\Core\Protocol;
  29 use Friendica\Core\System;
  30 use Friendica\Core\Worker;
  31 use Friendica\Database\Database;
  32 use Friendica\Database\DBA;
  33 use Friendica\DI;
  34 use Friendica\Module\Register;
  35 use Friendica\Network\HTTPClient\Client\HttpClientAccept;
  36 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  37 use Friendica\Network\HTTPClient\Capability\ICanHandleHttpResponses;
  38 use Friendica\Network\Probe;
  39 use Friendica\Protocol\ActivityPub;
  40 use Friendica\Protocol\Relay;
  41 use Friendica\Util\DateTimeFormat;
  42 use Friendica\Util\JsonLD;
  43 use Friendica\Util\Network;
  44 use Friendica\Util\Strings;
  45 use Friendica\Util\XML;
  46 use Friendica\Network\HTTPException;
  47 use GuzzleHttp\Psr7\Uri;
  48
  49 /**
  50  * This class handles GServer related functions
  51  */
  52 class GServer
  53 {
  54         // Directory types
  55         const DT_NONE = 0;
  56         const DT_POCO = 1;
  57         const DT_MASTODON = 2;
  58
  59         // Methods to detect server types
  60
  61         // Non endpoint specific methods
  62         const DETECT_MANUAL = 0;
  63         const DETECT_HEADER = 1;
  64         const DETECT_BODY = 2;
  65         const DETECT_HOST_META = 3;
  66         const DETECT_CONTACTS = 4;
  67         const DETECT_AP_ACTOR = 5;
  68         const DETECT_AP_COLLECTION = 6;
  69
  70         const DETECT_UNSPECIFIC = [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META, self::DETECT_CONTACTS, self::DETECT_AP_ACTOR];
  71
  72         // Implementation specific endpoints
  73         const DETECT_FRIENDIKA = 10;
  74         const DETECT_FRIENDICA = 11;
  75         const DETECT_STATUSNET = 12;
  76         const DETECT_GNUSOCIAL = 13;
  77         const DETECT_CONFIG_JSON = 14; // Statusnet, GNU Social, Older Hubzilla/Redmatrix
  78         const DETECT_SITEINFO_JSON = 15; // Newer Hubzilla
  79         const DETECT_MASTODON_API = 16;
  80         const DETECT_STATUS_PHP = 17; // Nextcloud
  81         const DETECT_V1_CONFIG = 18;
  82         const DETECT_PUMPIO = 19; // Deprecated
  83         const DETECT_SYSTEM_ACTOR = 20; // Mistpark, Osada, Roadhouse, Zap
  84
  85         // Standardized endpoints
  86         const DETECT_STATISTICS_JSON = 100;
  87         const DETECT_NODEINFO_1 = 101;
  88         const DETECT_NODEINFO_2 = 102;
  89         const DETECT_NODEINFO_210 = 103;
  90
  91         /**
  92          * Check for the existance of a server and adds it in the background if not existant
  93          *
  94          * @param string $url
  95          * @param boolean $only_nodeinfo
  96          *
  97          * @return void
  98          */
  99         public static function add(string $url, bool $only_nodeinfo = false)
 100         {
 101                 if (self::getID($url, false)) {
 102                         return;
 103                 }
 104
 105                 Worker::add(PRIORITY_LOW, 'UpdateGServer', $url, $only_nodeinfo);
 106         }
 107
 108         /**
 109          * Get the ID for the given server URL
 110          *
 111          * @param string $url
 112          * @param boolean $no_check Don't check if the server hadn't been found
 113          *
 114          * @return int|null gserver id or NULL on empty URL or failed check
 115          */
 116         public static function getID(string $url, bool $no_check = false): ?int
 117         {
 118                 if (empty($url)) {
 119                         return null;
 120                 }
 121
 122                 $url = self::cleanURL($url);
 123
 124                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
 125                 if (DBA::isResult($gserver)) {
 126                         Logger::debug('Got ID for URL', ['id' => $gserver['id'], 'url' => $url, 'callstack' => System::callstack(20)]);
 127                         return $gserver['id'];
 128                 }
 129
 130                 if ($no_check || !self::check($url)) {
 131                         return null;
 132                 }
 133
 134                 return self::getID($url, true);
 135         }
 136
 137         /**
 138          * Retrieves all the servers which base domain are matching the provided domain pattern
 139          *
 140          * The pattern is a simple fnmatch() pattern with ? for single wildcard and * for multiple wildcard
 141          *
 142          * @param string $pattern
 143          *
 144          * @return array
 145          *
 146          * @throws Exception
 147          */
 148         public static function listByDomainPattern(string $pattern): array
 149         {
 150                 $likePattern = 'http://' . strtr($pattern, ['_' => '\_', '%' => '\%', '?' => '_', '*' => '%']);
 151
 152                 // The SUBSTRING_INDEX returns everything before the eventual third /, which effectively trims an
 153                 // eventual server path and keep only the server domain which we're matching against the pattern.
 154                 $sql = "SELECT `gserver`.*, COUNT(*) AS `contacts`
 155                         FROM `gserver`
 156                         LEFT JOIN `contact` ON `gserver`.`id` = `contact`.`gsid`
 157                         WHERE SUBSTRING_INDEX(`gserver`.`nurl`, '/', 3) LIKE ?
 158                         AND NOT `gserver`.`failed`
 159                         GROUP BY `gserver`.`id`";
 160
 161                 $stmt = DI::dba()->p($sql, $likePattern);
 162
 163                 return DI::dba()->toArray($stmt);
 164         }
 165
 166         /**
 167          * Checks if the given server is reachable
 168          *
 169          * @param string  $profile URL of the given profile
 170          * @param string  $server  URL of the given server (If empty, taken from profile)
 171          * @param string  $network Network value that is used, when detection failed
 172          * @param boolean $force   Force an update.
 173          *
 174          * @return boolean 'true' if server seems vital
 175          */
 176         public static function reachable(string $profile, string $server = '', string $network = '', bool $force = false): bool
 177         {
 178                 if ($server == '') {
 179                         $contact = Contact::getByURL($profile, null, ['baseurl']);
 180                         if (!empty($contact['baseurl'])) {
 181                                 $server = $contact['baseurl'];
 182                         }
 183                 }
 184
 185                 if ($server == '') {
 186                         return true;
 187                 }
 188
 189                 return self::check($server, $network, $force);
 190         }
 191
 192         /**
 193          * Calculate the next update day
 194          *
 195          * @param bool $success
 196          * @param string $created
 197          * @param string $last_contact
 198          * @param bool $undetected
 199          *
 200          * @return string
 201          *
 202          * @throws Exception
 203          */
 204         public static function getNextUpdateDate(bool $success, string $created = '', string $last_contact = '', bool $undetected = false): string
 205         {
 206                 // On successful contact process check again next week when it is a detected system.
 207                 // When we haven't detected the system, it could be a static website or a really old system.
 208                 if ($success) {
 209                         return DateTimeFormat::utc($undetected ? 'now +1 month' : 'now +7 day');
 210                 }
 211
 212                 $now = strtotime(DateTimeFormat::utcNow());
 213
 214                 if ($created > $last_contact) {
 215                         $contact_time = strtotime($created);
 216                 } else {
 217                         $contact_time = strtotime($last_contact);
 218                 }
 219
 220                 // If the last contact was less than 6 hours before then try again in 6 hours
 221                 if (($now - $contact_time) < (60 * 60 * 6)) {
 222                         return DateTimeFormat::utc('now +6 hour');
 223                 }
 224
 225                 // If the last contact was less than 12 hours before then try again in 12 hours
 226                 if (($now - $contact_time) < (60 * 60 * 12)) {
 227                         return DateTimeFormat::utc('now +12 hour');
 228                 }
 229
 230                 // If the last contact was less than 24 hours before then try tomorrow again
 231                 if (($now - $contact_time) < (60 * 60 * 24)) {
 232                         return DateTimeFormat::utc('now +1 day');
 233                 }
 234
 235                 // If the last contact was less than a week before then try again in a week
 236                 if (($now - $contact_time) < (60 * 60 * 24 * 7)) {
 237                         return DateTimeFormat::utc('now +1 week');
 238                 }
 239
 240                 // If the last contact was less than two weeks before then try again in two week
 241                 if (($now - $contact_time) < (60 * 60 * 24 * 14)) {
 242                         return DateTimeFormat::utc('now +2 week');
 243                 }
 244
 245                 // If the last contact was less than a month before then try again in a month
 246                 if (($now - $contact_time) < (60 * 60 * 24 * 30)) {
 247                         return DateTimeFormat::utc('now +1 month');
 248                 }
 249
 250                 // The system hadn't been successul contacted for more than a month, so try again in three months
 251                 return DateTimeFormat::utc('now +3 month');
 252         }
 253
 254         /**
 255          * Checks the state of the given server.
 256          *
 257          * @param string  $server_url    URL of the given server
 258          * @param string  $network       Network value that is used, when detection failed
 259          * @param boolean $force         Force an update.
 260          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 261          *
 262          * @return boolean 'true' if server seems vital
 263          */
 264         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false): bool
 265         {
 266                 $server_url = self::cleanURL($server_url);
 267                 if ($server_url == '') {
 268                         return false;
 269                 }
 270
 271                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
 272                 if (DBA::isResult($gserver)) {
 273                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
 274                                 $fields = ['created' => DateTimeFormat::utcNow()];
 275                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
 276                                 self::update($fields, $condition);
 277                         }
 278
 279                         if (!$force && (strtotime($gserver['next_contact']) > time())) {
 280                                 Logger::info('No update needed', ['server' => $server_url]);
 281                                 return (!$gserver['failed']);
 282                         }
 283                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force]);
 284                 } else {
 285                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
 286                 }
 287
 288                 return self::detect($server_url, $network, $only_nodeinfo);
 289         }
 290
 291         /**
 292          * Set failed server status
 293          *
 294          * @param string $url
 295          */
 296         public static function setFailure(string $url)
 297         {
 298                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($url)]);
 299                 if (DBA::isResult($gserver)) {
 300                         $next_update = self::getNextUpdateDate(false, $gserver['created'], $gserver['last_contact']);
 301                         self::update(['url' => $url, 'failed' => true, 'last_failure' => DateTimeFormat::utcNow(),
 302                         'next_contact' => $next_update, 'network' => Protocol::PHANTOM, 'detection-method' => null],
 303                         ['nurl' => Strings::normaliseLink($url)]);
 304                         Logger::info('Set failed status for existing server', ['url' => $url]);
 305                         return;
 306                 }
 307                 DBA::insert('gserver', ['url' => $url, 'nurl' => Strings::normaliseLink($url),
 308                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
 309                         'failed' => true, 'last_failure' => DateTimeFormat::utcNow()]);
 310                 Logger::info('Set failed status for new server', ['url' => $url]);
 311         }
 312
 313         /**
 314          * Remove unwanted content from the given URL
 315          *
 316          * @param string $url
 317          *
 318          * @return string cleaned URL
 319          */
 320         public static function cleanURL(string $url): string
 321         {
 322                 $url = trim($url, '/');
 323                 $url = str_replace('/index.php', '', $url);
 324
 325                 $urlparts = parse_url($url);
 326                 unset($urlparts['user']);
 327                 unset($urlparts['pass']);
 328                 unset($urlparts['query']);
 329                 unset($urlparts['fragment']);
 330                 return (string)Uri::fromParts($urlparts);
 331         }
 332
 333         /**
 334          * Detect server data (type, protocol, version number, ...)
 335          * The detected data is then updated or inserted in the gserver table.
 336          *
 337          * @param string  $url           URL of the given server
 338          * @param string  $network       Network value that is used, when detection failed
 339          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 340          *
 341          * @return boolean 'true' if server could be detected
 342          */
 343         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false): bool
 344         {
 345                 Logger::info('Detect server type', ['server' => $url]);
 346
 347                 $original_url = $url;
 348
 349                 // Remove URL content that is not supposed to exist for a server url
 350                 $url = rtrim(self::cleanURL($url), '/');
 351                 if (empty($url)) {
 352                         Logger::notice('Empty URL.');
 353                         return false;
 354                 }
 355
 356                 // If the URL missmatches, then we mark the old entry as failure
 357                 if (!Strings::compareLink($url, $original_url)) {
 358                         self::setFailure($original_url);
 359                         if (!self::getID($url, true)) {
 360                                 self::detect($url, $network, $only_nodeinfo);
 361                         }
 362                         return false;
 363                 }
 364
 365                 $valid_url = Network::isUrlValid($url);
 366                 if (!$valid_url) {
 367                         self::setFailure($url);
 368                         return false;
 369                 } else {
 370                         $valid_url = rtrim($valid_url, '/');
 371                 }
 372
 373                 if (!Strings::compareLink($url, $valid_url)) {
 374                         // We only follow redirects when the path stays the same or the target url has no path.
 375                         // Some systems have got redirects on their landing page to a single account page. This check handles it.
 376                         if (((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH))) ||
 377                                 (((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) || (parse_url($url, PHP_URL_PATH) != parse_url($valid_url, PHP_URL_PATH))) && empty(parse_url($valid_url, PHP_URL_PATH)))) {
 378                                 Logger::debug('Found redirect. Mark old entry as failure', ['old' => $url, 'new' => $valid_url]);
 379                                 self::setFailure($url);
 380                                 if (!self::getID($valid_url, true)) {
 381                                         self::detect($valid_url, $network, $only_nodeinfo);
 382                                 }
 383                                 return false;
 384                         }
 385
 386                         if ((parse_url($url, PHP_URL_HOST) != parse_url($valid_url, PHP_URL_HOST)) && (parse_url($url, PHP_URL_PATH) != parse_url($valid_url, PHP_URL_PATH)) &&
 387                                 (parse_url($url, PHP_URL_PATH) == '')) {
 388                                 Logger::debug('Found redirect. Mark old entry as failure and redirect to the basepath.', ['old' => $url, 'new' => $valid_url]);
 389                                 $parts = parse_url($valid_url);
 390                                 unset($parts['path']);
 391                                 $valid_url = (string)Uri::fromParts($parts);
 392
 393                                 self::setFailure($url);
 394                                 if (!self::getID($valid_url, true)) {
 395                                         self::detect($valid_url, $network, $only_nodeinfo);
 396                                 }
 397                                 return false;
 398                         }
 399                         Logger::debug('Found redirect, but ignore it.', ['old' => $url, 'new' => $valid_url]);
 400                 }
 401
 402                 if ((parse_url($url, PHP_URL_HOST) == parse_url($valid_url, PHP_URL_HOST)) &&
 403                         (parse_url($url, PHP_URL_PATH) == parse_url($valid_url, PHP_URL_PATH)) &&
 404                         (parse_url($url, PHP_URL_SCHEME) != parse_url($valid_url, PHP_URL_SCHEME))) {
 405                         $url = $valid_url;
 406                 }
 407
 408                 $in_webroot = empty(parse_url($url, PHP_URL_PATH));
 409
 410                 // When a nodeinfo is present, we don't need to dig further
 411                 $curlResult = DI::httpClient()->get($url . '/.well-known/x-nodeinfo2', HttpClientAccept::JSON);
 412                 if ($curlResult->isTimeout()) {
 413                         self::setFailure($url);
 414                         return false;
 415                 }
 416
 417                 $serverdata = self::parseNodeinfo210($curlResult);
 418                 if (empty($serverdata)) {
 419                         $curlResult = DI::httpClient()->get($url . '/.well-known/nodeinfo', HttpClientAccept::JSON);
 420                         $serverdata = self::fetchNodeinfo($url, $curlResult);
 421                 }
 422
 423                 if ($only_nodeinfo && empty($serverdata)) {
 424                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 425                         self::setFailure($url);
 426                         return false;
 427                 } elseif (empty($serverdata)) {
 428                         $serverdata = ['detection-method' => self::DETECT_MANUAL, 'network' => Protocol::PHANTOM, 'platform' => '', 'version' => '', 'site_name' => '', 'info' => ''];
 429                 }
 430
 431                 // When there is no Nodeinfo, then use some protocol specific endpoints
 432                 if ($serverdata['network'] == Protocol::PHANTOM) {
 433                         if ($in_webroot) {
 434                                 // Fetch the landing page, possibly it reveals some data
 435                                 $accept = 'application/activity+json,application/ld+json,application/json,*/*;q=0.9';
 436                                 $curlResult = DI::httpClient()->get($url, $accept);
 437                                 if (!$curlResult->isSuccess() && $curlResult->getReturnCode() == '406') {
 438                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 439                                         $html_fetched = true;
 440                                 } else {
 441                                         $html_fetched = false;
 442                                 }
 443
 444                                 if ($curlResult->isSuccess()) {
 445                                         $json = json_decode($curlResult->getBody(), true);
 446                                         if (!empty($json) && is_array($json)) {
 447                                                 $data = self::fetchDataFromSystemActor($json, $serverdata);
 448                                                 $serverdata = $data['server'];
 449                                                 $systemactor = $data['actor'];
 450                                                 if (!$html_fetched && !in_array($serverdata['detection-method'], [self::DETECT_SYSTEM_ACTOR, self::DETECT_AP_COLLECTION])) {
 451                                                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 452                                                 }
 453                                         } elseif (!$html_fetched && (strlen($curlResult->getBody()) < 1000)) {
 454                                                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML);
 455                                         }
 456
 457                                         if ($serverdata['detection-method'] != self::DETECT_SYSTEM_ACTOR) {
 458                                                 $serverdata = self::analyseRootHeader($curlResult, $serverdata);
 459                                                 $serverdata = self::analyseRootBody($curlResult, $serverdata);
 460                                         }
 461                                 }
 462
 463                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
 464                                         self::setFailure($url);
 465                                         return false;
 466                                 }
 467
 468                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 469                                         $serverdata = self::detectMastodonAlikes($url, $serverdata);
 470                                 }
 471                         }
 472
 473                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 474                         // With this check we don't have to waste time and ressources for dead systems.
 475                         // Also this hopefully prevents us from receiving abuse messages.
 476                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 477                                 $validHostMeta = self::validHostMeta($url);
 478                         } else {
 479                                 $validHostMeta = false;
 480                         }
 481
 482                         if ($validHostMeta) {
 483                                 if (in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY])) {
 484                                         $serverdata['detection-method'] = self::DETECT_HOST_META;
 485                                 }
 486
 487                                 if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 488                                         $serverdata = self::detectFriendica($url, $serverdata);
 489                                 }
 490
 491                                 // The following systems have to be installed in the root directory.
 492                                 if ($in_webroot) {
 493                                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 494                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 495                                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 496                                         }
 497
 498                                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations, so we check other endpoints as well
 499                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 500                                                 $serverdata = self::detectHubzilla($url, $serverdata);
 501                                         }
 502
 503                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 504                                                 $serverdata = self::detectPeertube($url, $serverdata);
 505                                         }
 506
 507                                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 508                                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 509                                         }
 510                                 }
 511                         } elseif (in_array($serverdata['platform'], ['friendica', 'friendika']) && in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_SYSTEM_ACTOR]))) {
 512                                 $serverdata = self::detectFriendica($url, $serverdata);
 513                         }
 514
 515                         if (($serverdata['network'] == Protocol::PHANTOM) || in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
 516                                 $serverdata = self::detectNextcloud($url, $serverdata, $validHostMeta);
 517                         }
 518
 519                         // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 520                         // Since this endpoint is only rarely used, we query it at a later time
 521                         if (in_array($serverdata['detection-method'], array_merge(self::DETECT_UNSPECIFIC, [self::DETECT_FRIENDICA, self::DETECT_CONFIG_JSON]))) {
 522                                 $serverdata = self::fetchStatistics($url, $serverdata);
 523                         }
 524                 }
 525
 526                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 527                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 528                         $serverdata['network'] = $network;
 529                 }
 530
 531                 // Most servers aren't installed in a subdirectory, so we declare this entry as failed
 532                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty(parse_url($url, PHP_URL_PATH)) && in_array($serverdata['detection-method'], [self::DETECT_MANUAL])) {
 533                         self::setFailure($url);
 534                         return false;
 535                 }
 536
 537                 $serverdata['url'] = $url;
 538                 $serverdata['nurl'] = Strings::normaliseLink($url);
 539
 540                 // We have to prevent an endless loop here.
 541                 // When a server is new, then there is no gserver entry yet.
 542                 // But in "detectNetworkViaContacts" it could happen that a contact is updated,
 543                 // and this can call this function here as well.
 544                 if (self::getID($url, true) && (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) ||
 545                         in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_HOST_META]))) {
 546                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 547                 }
 548
 549                 // Detect the directory type
 550                 $serverdata['directory-type'] = self::DT_NONE;
 551
 552                 if (in_array($serverdata['network'], Protocol::FEDERATED)) {
 553                         $serverdata = self::checkMastodonDirectory($url, $serverdata);
 554
 555                         if ($serverdata['directory-type'] == self::DT_NONE) {
 556                                 $serverdata = self::checkPoCo($url, $serverdata);
 557                         }
 558                 }
 559
 560                 if ($serverdata['network'] == Protocol::ACTIVITYPUB) {
 561                         $serverdata = self::fetchWeeklyUsage($url, $serverdata);
 562                 }
 563
 564                 $serverdata['registered-users'] = $serverdata['registered-users'] ?? 0;
 565
 566                 // On an active server there has to be at least a single user
 567                 if (!in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]) && ($serverdata['registered-users'] == 0)) {
 568                         $serverdata['registered-users'] = 1;
 569                 } elseif (in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 570                         $serverdata['registered-users'] = 0;
 571                 }
 572
 573                 $serverdata['next_contact'] = self::getNextUpdateDate(true, '', '', in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED]));
 574
 575                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 576                 $serverdata['failed'] = false;
 577
 578                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 579                 if (!DBA::isResult($gserver)) {
 580                         $serverdata['created'] = DateTimeFormat::utcNow();
 581                         $ret = DBA::insert('gserver', $serverdata);
 582                         $id = DBA::lastInsertId();
 583                 } else {
 584                         $ret = self::update($serverdata, ['nurl' => $serverdata['nurl']]);
 585                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 586                         if (DBA::isResult($gserver)) {
 587                                 $id = $gserver['id'];
 588                         }
 589                 }
 590
 591                 // Count the number of known contacts from this server
 592                 if (!empty($id) && !in_array($serverdata['network'], [Protocol::PHANTOM, Protocol::FEED])) {
 593                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 594                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id, 'failed' => false]);
 595                         $max_users = max($apcontacts, $contacts);
 596                         if ($max_users > $serverdata['registered-users']) {
 597                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 598                                 self::update(['registered-users' => $max_users], ['id' => $id]);
 599                         }
 600
 601                         if (empty($serverdata['active-month-users'])) {
 602                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 30 days')]);
 603                                 if ($contacts > 0) {
 604                                         Logger::info('Update monthly users', ['id' => $id, 'url' => $serverdata['nurl'], 'monthly-users' => $contacts]);
 605                                         self::update(['active-month-users' => $contacts], ['id' => $id]);
 606                                 }
 607                         }
 608
 609                         if (empty($serverdata['active-halfyear-users'])) {
 610                                 $contacts = DBA::count('contact', ["`uid` = ? AND `gsid` = ? AND NOT `failed` AND `last-item` > ?", 0, $id, DateTimeFormat::utc('now - 180 days')]);
 611                                 if ($contacts > 0) {
 612                                         Logger::info('Update halfyear users', ['id' => $id, 'url' => $serverdata['nurl'], 'halfyear-users' => $contacts]);
 613                                         self::update(['active-halfyear-users' => $contacts], ['id' => $id]);
 614                                 }
 615                         }
 616                 }
 617
 618                 if (in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 619                         self::discoverRelay($url);
 620                 }
 621
 622                 if (!empty($systemactor)) {
 623                         $contact = Contact::getByURL($systemactor, true, ['gsid', 'baseurl', 'id', 'network', 'url', 'name']);
 624                         Logger::debug('Fetched system actor',  ['url' => $url, 'gsid' => $id, 'contact' => $contact]);
 625                 }
 626
 627                 return $ret;
 628         }
 629
 630         /**
 631          * Fetch relay data from a given server url
 632          *
 633          * @param string $server_url address of the server
 634          *
 635          * @return void
 636          *
 637          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 638          */
 639         private static function discoverRelay(string $server_url)
 640         {
 641                 Logger::info('Discover relay data', ['server' => $server_url]);
 642
 643                 $curlResult = DI::httpClient()->get($server_url . '/.well-known/x-social-relay', HttpClientAccept::JSON);
 644                 if (!$curlResult->isSuccess()) {
 645                         return;
 646                 }
 647
 648                 $data = json_decode($curlResult->getBody(), true);
 649                 if (!is_array($data)) {
 650                         return;
 651                 }
 652
 653                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 654                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
 655
 656                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 657                         $data['scope'] = '';
 658                         $data['subscribe'] = false;
 659                         $data['tags'] = [];
 660                 }
 661
 662                 $gserver = DBA::selectFirst('gserver', ['id', 'url', 'network', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 663                 if (!DBA::isResult($gserver)) {
 664                         return;
 665                 }
 666
 667                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 668                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 669                         self::update($fields, ['id' => $gserver['id']]);
 670                 }
 671
 672                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 673
 674                 if ($data['scope'] == 'tags') {
 675                         // Avoid duplicates
 676                         $tags = [];
 677                         foreach ($data['tags'] as $tag) {
 678                                 $tag = mb_strtolower($tag);
 679                                 if (strlen($tag) < 100) {
 680                                         $tags[$tag] = $tag;
 681                                 }
 682                         }
 683
 684                         foreach ($tags as $tag) {
 685                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], Database::INSERT_IGNORE);
 686                         }
 687                 }
 688
 689                 // Create or update the relay contact
 690                 $fields = [];
 691                 if (isset($data['protocols'])) {
 692                         if (isset($data['protocols']['diaspora'])) {
 693                                 $fields['network'] = Protocol::DIASPORA;
 694
 695                                 if (isset($data['protocols']['diaspora']['receive'])) {
 696                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 697                                 } elseif (is_string($data['protocols']['diaspora'])) {
 698                                         $fields['batch'] = $data['protocols']['diaspora'];
 699                                 }
 700                         }
 701
 702                         if (isset($data['protocols']['dfrn'])) {
 703                                 $fields['network'] = Protocol::DFRN;
 704
 705                                 if (isset($data['protocols']['dfrn']['receive'])) {
 706                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 707                                 } elseif (is_string($data['protocols']['dfrn'])) {
 708                                         $fields['batch'] = $data['protocols']['dfrn'];
 709                                 }
 710                         }
 711
 712                         if (isset($data['protocols']['activitypub'])) {
 713                                 $fields['network'] = Protocol::ACTIVITYPUB;
 714
 715                                 if (!empty($data['protocols']['activitypub']['actor'])) {
 716                                         $fields['url'] = $data['protocols']['activitypub']['actor'];
 717                                 }
 718                                 if (!empty($data['protocols']['activitypub']['receive'])) {
 719                                         $fields['batch'] = $data['protocols']['activitypub']['receive'];
 720                                 }
 721                         }
 722                 }
 723
 724                 Logger::info('Discovery ended', ['server' => $server_url, 'data' => $fields]);
 725
 726                 Relay::updateContact($gserver, $fields);
 727         }
 728
 729         /**
 730          * Fetch server data from '/statistics.json' on the given server
 731          *
 732          * @param string $url URL of the given server
 733          *
 734          * @return array server data
 735          */
 736         private static function fetchStatistics(string $url, array $serverdata): array
 737         {
 738                 $curlResult = DI::httpClient()->get($url . '/statistics.json', HttpClientAccept::JSON);
 739                 if (!$curlResult->isSuccess()) {
 740                         return $serverdata;
 741                 }
 742
 743                 $data = json_decode($curlResult->getBody(), true);
 744                 if (empty($data)) {
 745                         return $serverdata;
 746                 }
 747
 748                 // Some AP enabled systems return activity data that we don't expect here.
 749                 if (strpos($curlResult->getContentType(), 'application/activity+json') !== false) {
 750                         return $serverdata;
 751                 }
 752
 753                 $valid = false;
 754                 $old_serverdata = $serverdata;
 755
 756                 $serverdata['detection-method'] = self::DETECT_STATISTICS_JSON;
 757
 758                 if (!empty($data['version'])) {
 759                         $valid = true;
 760                         $serverdata['version'] = $data['version'];
 761                         // Version numbers on statistics.json are presented with additional info, e.g.:
 762                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 763                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 764                 }
 765
 766                 if (!empty($data['name'])) {
 767                         $valid = true;
 768                         $serverdata['site_name'] = $data['name'];
 769                 }
 770
 771                 if (!empty($data['network'])) {
 772                         $valid = true;
 773                         $serverdata['platform'] = strtolower($data['network']);
 774
 775                         if ($serverdata['platform'] == 'diaspora') {
 776                                 $serverdata['network'] = Protocol::DIASPORA;
 777                         } elseif ($serverdata['platform'] == 'friendica') {
 778                                 $serverdata['network'] = Protocol::DFRN;
 779                         } elseif ($serverdata['platform'] == 'hubzilla') {
 780                                 $serverdata['network'] = Protocol::ZOT;
 781                         } elseif ($serverdata['platform'] == 'redmatrix') {
 782                                 $serverdata['network'] = Protocol::ZOT;
 783                         }
 784                 }
 785
 786                 if (!empty($data['total_users'])) {
 787                         $valid = true;
 788                         $serverdata['registered-users'] = max($data['total_users'], 1);
 789                 }
 790
 791                 if (!empty($data['active_users_monthly'])) {
 792                         $valid = true;
 793                         $serverdata['active-month-users'] = max($data['active_users_monthly'], 0);
 794                 }
 795
 796                 if (!empty($data['active_users_halfyear'])) {
 797                         $valid = true;
 798                         $serverdata['active-halfyear-users'] = max($data['active_users_halfyear'], 0);
 799                 }
 800
 801                 if (!empty($data['local_posts'])) {
 802                         $valid = true;
 803                         $serverdata['local-posts'] = max($data['local_posts'], 0);
 804                 }
 805
 806                 if (!empty($data['registrations_open'])) {
 807                         $serverdata['register_policy'] = Register::OPEN;
 808                 } else {
 809                         $serverdata['register_policy'] = Register::CLOSED;
 810                 }
 811
 812                 if (!$valid) {
 813                         return $old_serverdata;
 814                 }
 815
 816                 return $serverdata;
 817         }
 818
 819         /**
 820          * Detect server type by using the nodeinfo data
 821          *
 822          * @param string                  $url        address of the server
 823          * @param ICanHandleHttpResponses $httpResult
 824          *
 825          * @return array Server data
 826          *
 827          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 828          */
 829         private static function fetchNodeinfo(string $url, ICanHandleHttpResponses $httpResult): array
 830         {
 831                 if (!$httpResult->isSuccess()) {
 832                         return [];
 833                 }
 834
 835                 $nodeinfo = json_decode($httpResult->getBody(), true);
 836
 837                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
 838                         return [];
 839                 }
 840
 841                 $nodeinfo1_url = '';
 842                 $nodeinfo2_url = '';
 843
 844                 foreach ($nodeinfo['links'] as $link) {
 845                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
 846                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
 847                                 continue;
 848                         }
 849                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 850                                 $nodeinfo1_url = $link['href'];
 851                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 852                                 $nodeinfo2_url = $link['href'];
 853                         }
 854                 }
 855
 856                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 857                         return [];
 858                 }
 859
 860                 $server = [];
 861
 862                 if (!empty($nodeinfo2_url)) {
 863                         $server = self::parseNodeinfo2($nodeinfo2_url);
 864                 }
 865
 866                 if (empty($server) && !empty($nodeinfo1_url)) {
 867                         $server = self::parseNodeinfo1($nodeinfo1_url);
 868                 }
 869
 870                 return $server;
 871         }
 872
 873         /**
 874          * Parses Nodeinfo 1
 875          *
 876          * @param string $nodeinfo_url address of the nodeinfo path
 877          *
 878          * @return array Server data
 879          *
 880          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 881          */
 882         private static function parseNodeinfo1(string $nodeinfo_url): array
 883         {
 884                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 885                 if (!$curlResult->isSuccess()) {
 886                         return [];
 887                 }
 888
 889                 $nodeinfo = json_decode($curlResult->getBody(), true);
 890
 891                 if (!is_array($nodeinfo)) {
 892                         return [];
 893                 }
 894
 895                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
 896                         'register_policy' => Register::CLOSED];
 897
 898                 if (!empty($nodeinfo['openRegistrations'])) {
 899                         $server['register_policy'] = Register::OPEN;
 900                 }
 901
 902                 if (is_array($nodeinfo['software'])) {
 903                         if (!empty($nodeinfo['software']['name'])) {
 904                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 905                         }
 906
 907                         if (!empty($nodeinfo['software']['version'])) {
 908                                 $server['version'] = $nodeinfo['software']['version'];
 909                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 910                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 911                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 912                         }
 913                 }
 914
 915                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 916                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 917                 }
 918
 919                 if (!empty($nodeinfo['usage']['users']['total'])) {
 920                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
 921                 }
 922
 923                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
 924                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
 925                 }
 926
 927                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
 928                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
 929                 }
 930
 931                 if (!empty($nodeinfo['usage']['localPosts'])) {
 932                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
 933                 }
 934
 935                 if (!empty($nodeinfo['usage']['localComments'])) {
 936                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
 937                 }
 938
 939                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
 940                         $protocols = [];
 941                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
 942                                 $protocols[$protocol] = true;
 943                         }
 944
 945                         if (!empty($protocols['friendica'])) {
 946                                 $server['network'] = Protocol::DFRN;
 947                         } elseif (!empty($protocols['activitypub'])) {
 948                                 $server['network'] = Protocol::ACTIVITYPUB;
 949                         } elseif (!empty($protocols['diaspora'])) {
 950                                 $server['network'] = Protocol::DIASPORA;
 951                         } elseif (!empty($protocols['ostatus'])) {
 952                                 $server['network'] = Protocol::OSTATUS;
 953                         } elseif (!empty($protocols['gnusocial'])) {
 954                                 $server['network'] = Protocol::OSTATUS;
 955                         } elseif (!empty($protocols['zot'])) {
 956                                 $server['network'] = Protocol::ZOT;
 957                         }
 958                 }
 959
 960                 if (empty($server)) {
 961                         return [];
 962                 }
 963
 964                 if (empty($server['network'])) {
 965                         $server['network'] = Protocol::PHANTOM;
 966                 }
 967
 968                 return $server;
 969         }
 970
 971         /**
 972          * Parses Nodeinfo 2
 973          *
 974          * @see https://git.feneas.org/jaywink/nodeinfo2
 975          *
 976          * @param string $nodeinfo_url address of the nodeinfo path
 977          *
 978          * @return array Server data
 979          *
 980          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 981          */
 982         private static function parseNodeinfo2(string $nodeinfo_url): array
 983         {
 984                 $curlResult = DI::httpClient()->get($nodeinfo_url, HttpClientAccept::JSON);
 985                 if (!$curlResult->isSuccess()) {
 986                         return [];
 987                 }
 988
 989                 $nodeinfo = json_decode($curlResult->getBody(), true);
 990                 if (!is_array($nodeinfo)) {
 991                         return [];
 992                 }
 993
 994                 $server = [
 995                         'detection-method' => self::DETECT_NODEINFO_2,
 996                         'register_policy' => Register::CLOSED,
 997                         'platform' => 'unknown',
 998                 ];
 999
1000                 if (!empty($nodeinfo['openRegistrations'])) {
1001                         $server['register_policy'] = Register::OPEN;
1002                 }
1003
1004                 if (!empty($nodeinfo['software'])) {
1005                         if (isset($nodeinfo['software']['name'])) {
1006                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
1007                         }
1008
1009                         if (!empty($nodeinfo['software']['version']) && isset($server['platform'])) {
1010                                 $server['version'] = $nodeinfo['software']['version'];
1011                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1012                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1013                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1014
1015                                 // qoto advertises itself as Mastodon
1016                                 if (($server['platform'] == 'mastodon') && substr($nodeinfo['software']['version'], -5) == '-qoto') {
1017                                         $server['platform'] = 'qoto';
1018                                 }
1019                         }
1020                 }
1021
1022                 if (!empty($nodeinfo['metadata']['nodeName'])) {
1023                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
1024                 }
1025
1026                 if (!empty($nodeinfo['usage']['users']['total'])) {
1027                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1028                 }
1029
1030                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1031                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1032                 }
1033
1034                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1035                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1036                 }
1037
1038                 if (!empty($nodeinfo['usage']['localPosts'])) {
1039                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1040                 }
1041
1042                 if (!empty($nodeinfo['usage']['localComments'])) {
1043                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1044                 }
1045
1046                 if (!empty($nodeinfo['protocols'])) {
1047                         $protocols = [];
1048                         foreach ($nodeinfo['protocols'] as $protocol) {
1049                                 if (is_string($protocol)) {
1050                                         $protocols[$protocol] = true;
1051                                 }
1052                         }
1053
1054                         if (!empty($protocols['dfrn'])) {
1055                                 $server['network'] = Protocol::DFRN;
1056                         } elseif (!empty($protocols['activitypub'])) {
1057                                 $server['network'] = Protocol::ACTIVITYPUB;
1058                         } elseif (!empty($protocols['diaspora'])) {
1059                                 $server['network'] = Protocol::DIASPORA;
1060                         } elseif (!empty($protocols['ostatus'])) {
1061                                 $server['network'] = Protocol::OSTATUS;
1062                         } elseif (!empty($protocols['gnusocial'])) {
1063                                 $server['network'] = Protocol::OSTATUS;
1064                         } elseif (!empty($protocols['zot'])) {
1065                                 $server['network'] = Protocol::ZOT;
1066                         }
1067                 }
1068
1069                 if (empty($server)) {
1070                         return [];
1071                 }
1072
1073                 if (empty($server['network'])) {
1074                         $server['network'] = Protocol::PHANTOM;
1075                 }
1076
1077                 return $server;
1078         }
1079
1080         /**
1081          * Parses NodeInfo2 protocol 1.0
1082          *
1083          * @see https://github.com/jaywink/nodeinfo2/blob/master/PROTOCOL.md
1084          *
1085          * @param string $nodeinfo_url address of the nodeinfo path
1086          *
1087          * @return array Server data
1088          *
1089          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
1090          */
1091         private static function parseNodeinfo210(ICanHandleHttpResponses $httpResult): array
1092         {
1093                 if (!$httpResult->isSuccess()) {
1094                         return [];
1095                 }
1096
1097                 $nodeinfo = json_decode($httpResult->getBody(), true);
1098
1099                 if (!is_array($nodeinfo)) {
1100                         return [];
1101                 }
1102
1103                 $server = ['detection-method' => self::DETECT_NODEINFO_210,
1104                         'register_policy' => Register::CLOSED];
1105
1106                 if (!empty($nodeinfo['openRegistrations'])) {
1107                         $server['register_policy'] = Register::OPEN;
1108                 }
1109
1110                 if (!empty($nodeinfo['server'])) {
1111                         if (!empty($nodeinfo['server']['software'])) {
1112                                 $server['platform'] = strtolower($nodeinfo['server']['software']);
1113                         }
1114
1115                         if (!empty($nodeinfo['server']['version'])) {
1116                                 $server['version'] = $nodeinfo['server']['version'];
1117                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
1118                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
1119                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
1120                         }
1121
1122                         if (!empty($nodeinfo['server']['name'])) {
1123                                 $server['site_name'] = $nodeinfo['server']['name'];
1124                         }
1125                 }
1126
1127                 if (!empty($nodeinfo['usage']['users']['total'])) {
1128                         $server['registered-users'] = max($nodeinfo['usage']['users']['total'], 1);
1129                 }
1130
1131                 if (!empty($nodeinfo['usage']['users']['activeMonth'])) {
1132                         $server['active-month-users'] = max($nodeinfo['usage']['users']['activeMonth'], 0);
1133                 }
1134
1135                 if (!empty($nodeinfo['usage']['users']['activeHalfyear'])) {
1136                         $server['active-halfyear-users'] = max($nodeinfo['usage']['users']['activeHalfyear'], 0);
1137                 }
1138
1139                 if (!empty($nodeinfo['usage']['localPosts'])) {
1140                         $server['local-posts'] = max($nodeinfo['usage']['localPosts'], 0);
1141                 }
1142
1143                 if (!empty($nodeinfo['usage']['localComments'])) {
1144                         $server['local-comments'] = max($nodeinfo['usage']['localComments'], 0);
1145                 }
1146
1147                 if (!empty($nodeinfo['protocols'])) {
1148                         $protocols = [];
1149                         foreach ($nodeinfo['protocols'] as $protocol) {
1150                                 if (is_string($protocol)) {
1151                                         $protocols[$protocol] = true;
1152                                 }
1153                         }
1154
1155                         if (!empty($protocols['dfrn'])) {
1156                                 $server['network'] = Protocol::DFRN;
1157                         } elseif (!empty($protocols['activitypub'])) {
1158                                 $server['network'] = Protocol::ACTIVITYPUB;
1159                         } elseif (!empty($protocols['diaspora'])) {
1160                                 $server['network'] = Protocol::DIASPORA;
1161                         } elseif (!empty($protocols['ostatus'])) {
1162                                 $server['network'] = Protocol::OSTATUS;
1163                         } elseif (!empty($protocols['gnusocial'])) {
1164                                 $server['network'] = Protocol::OSTATUS;
1165                         } elseif (!empty($protocols['zot'])) {
1166                                 $server['network'] = Protocol::ZOT;
1167                         }
1168                 }
1169
1170                 if (empty($server) || empty($server['platform'])) {
1171                         return [];
1172                 }
1173
1174                 if (empty($server['network'])) {
1175                         $server['network'] = Protocol::PHANTOM;
1176                 }
1177
1178                 return $server;
1179         }
1180
1181         /**
1182          * Fetch server information from a 'siteinfo.json' file on the given server
1183          *
1184          * @param string $url        URL of the given server
1185          * @param array  $serverdata array with server data
1186          *
1187          * @return array server data
1188          */
1189         private static function fetchSiteinfo(string $url, array $serverdata): array
1190         {
1191                 $curlResult = DI::httpClient()->get($url . '/siteinfo.json', HttpClientAccept::JSON);
1192                 if (!$curlResult->isSuccess()) {
1193                         return $serverdata;
1194                 }
1195
1196                 $data = json_decode($curlResult->getBody(), true);
1197                 if (empty($data)) {
1198                         return $serverdata;
1199                 }
1200
1201                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1202                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
1203                 }
1204
1205                 if (!empty($data['url'])) {
1206                         $serverdata['platform'] = strtolower($data['platform']);
1207                         $serverdata['version'] = $data['version'];
1208                 }
1209
1210                 if (!empty($data['plugins'])) {
1211                         if (in_array('pubcrawl', $data['plugins'])) {
1212                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1213                         } elseif (in_array('diaspora', $data['plugins'])) {
1214                                 $serverdata['network'] = Protocol::DIASPORA;
1215                         } elseif (in_array('gnusoc', $data['plugins'])) {
1216                                 $serverdata['network'] = Protocol::OSTATUS;
1217                         } else {
1218                                 $serverdata['network'] = Protocol::ZOT;
1219                         }
1220                 }
1221
1222                 if (!empty($data['site_name'])) {
1223                         $serverdata['site_name'] = $data['site_name'];
1224                 }
1225
1226                 if (!empty($data['channels_total'])) {
1227                         $serverdata['registered-users'] = max($data['channels_total'], 1);
1228                 }
1229
1230                 if (!empty($data['channels_active_monthly'])) {
1231                         $serverdata['active-month-users'] = max($data['channels_active_monthly'], 0);
1232                 }
1233
1234                 if (!empty($data['channels_active_halfyear'])) {
1235                         $serverdata['active-halfyear-users'] = max($data['channels_active_halfyear'], 0);
1236                 }
1237
1238                 if (!empty($data['local_posts'])) {
1239                         $serverdata['local-posts'] = max($data['local_posts'], 0);
1240                 }
1241
1242                 if (!empty($data['local_comments'])) {
1243                         $serverdata['local-comments'] = max($data['local_comments'], 0);
1244                 }
1245
1246                 if (!empty($data['register_policy'])) {
1247                         switch ($data['register_policy']) {
1248                                 case 'REGISTER_OPEN':
1249                                         $serverdata['register_policy'] = Register::OPEN;
1250                                         break;
1251
1252                                 case 'REGISTER_APPROVE':
1253                                         $serverdata['register_policy'] = Register::APPROVE;
1254                                         break;
1255
1256                                 case 'REGISTER_CLOSED':
1257                                 default:
1258                                         $serverdata['register_policy'] = Register::CLOSED;
1259                                         break;
1260                         }
1261                 }
1262
1263                 return $serverdata;
1264         }
1265
1266         /**
1267          * Fetches server data via an ActivityPub account with url of that server
1268          *
1269          * @param string $url        URL of the given server
1270          * @param array  $serverdata array with server data
1271          *
1272          * @return array server data
1273          *
1274          * @throws Exception
1275          */
1276         private static function fetchDataFromSystemActor(array $data, array $serverdata): array
1277         {
1278                 if (empty($data)) {
1279                         return ['server' => $serverdata, 'actor' => ''];
1280                 }
1281
1282                 $actor = JsonLD::compact($data, false);
1283                 if (in_array(JsonLD::fetchElement($actor, '@type'), ActivityPub\Receiver::ACCOUNT_TYPES)) {
1284                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1285                         $serverdata['site_name'] = JsonLD::fetchElement($actor, 'as:name', '@value');
1286                         $serverdata['info'] = JsonLD::fetchElement($actor, 'as:summary', '@value');
1287                         if (!empty($actor['as:generator'])) {
1288                                 $generator = explode(' ', JsonLD::fetchElement($actor['as:generator'], 'as:name', '@value'));
1289                                 $serverdata['platform'] = strtolower(array_shift($generator));
1290                                 $serverdata['detection-method'] = self::DETECT_SYSTEM_ACTOR;
1291                         } else {
1292                                 $serverdata['detection-method'] = self::DETECT_AP_ACTOR;
1293                         }
1294                         return ['server' => $serverdata, 'actor' => $actor['@id']];
1295                 } elseif ((JsonLD::fetchElement($actor, '@type') == 'as:Collection')) {
1296                         // By now only Ktistec seems to provide collections this way
1297                         $serverdata['platform'] = 'ktistec';
1298                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1299                         $serverdata['detection-method'] = self::DETECT_AP_COLLECTION;
1300
1301                         $actors = JsonLD::fetchElementArray($actor, 'as:items');
1302                         if (!empty($actors) && !empty($actors[0]['@id'])) {
1303                                 $actor_url = $actor['@id'] . $actors[0]['@id'];
1304                         } else {
1305                                 $actor_url = '';
1306                         }
1307
1308                         return ['server' => $serverdata, 'actor' => $actor_url];
1309                 }
1310                 return ['server' => $serverdata, 'actor' => ''];
1311         }
1312
1313         /**
1314          * Checks if the server contains a valid host meta file
1315          *
1316          * @param string $url URL of the given server
1317          *
1318          * @return boolean 'true' if the server seems to be vital
1319          */
1320         private static function validHostMeta(string $url): bool
1321         {
1322                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
1323                 $curlResult = DI::httpClient()->get($url . '/.well-known/host-meta', HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
1324                 if (!$curlResult->isSuccess()) {
1325                         return false;
1326                 }
1327
1328                 $xrd = XML::parseString($curlResult->getBody());
1329                 if (!is_object($xrd)) {
1330                         return false;
1331                 }
1332
1333                 $elements = XML::elementToArray($xrd);
1334                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
1335                         return false;
1336                 }
1337
1338                 $valid = false;
1339                 foreach ($elements['xrd']['link'] as $link) {
1340                         // When there is more than a single "link" element, the array looks slightly different
1341                         if (!empty($link['@attributes'])) {
1342                                 $link = $link['@attributes'];
1343                         }
1344
1345                         if (empty($link['rel']) || empty($link['template'])) {
1346                                 continue;
1347                         }
1348
1349                         if ($link['rel'] == 'lrdd') {
1350                                 // When the webfinger host is the same like the system host, it should be ok.
1351                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
1352                         }
1353                 }
1354
1355                 return $valid;
1356         }
1357
1358         /**
1359          * Detect the network of the given server via their known contacts
1360          *
1361          * @param string $url        URL of the given server
1362          * @param array  $serverdata array with server data
1363          *
1364          * @return array server data
1365          */
1366         private static function detectNetworkViaContacts(string $url, array $serverdata): array
1367         {
1368                 $contacts = [];
1369
1370                 $nurl = Strings::normaliseLink($url);
1371
1372                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $nurl]]);
1373                 while ($apcontact = DBA::fetch($apcontacts)) {
1374                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
1375                 }
1376                 DBA::close($apcontacts);
1377
1378                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $nurl]]);
1379                 while ($pcontact = DBA::fetch($pcontacts)) {
1380                         $contacts[$pcontact['nurl']] = $pcontact['url'];
1381                 }
1382                 DBA::close($pcontacts);
1383
1384                 if (empty($contacts)) {
1385                         return $serverdata;
1386                 }
1387
1388                 $time = time();
1389                 foreach ($contacts as $contact) {
1390                         // Endlosschleife verhindern wegen gsid!
1391                         $data = Probe::uri($contact);
1392                         if (in_array($data['network'], Protocol::FEDERATED)) {
1393                                 $serverdata['network'] = $data['network'];
1394
1395                                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1396                                         $serverdata['detection-method'] = self::DETECT_CONTACTS;
1397                                 }
1398                                 break;
1399                         } elseif ((time() - $time) > 10) {
1400                                 // To reduce the stress on remote systems we probe a maximum of 10 seconds
1401                                 break;
1402                         }
1403                 }
1404
1405                 return $serverdata;
1406         }
1407
1408         /**
1409          * Checks if the given server does have a '/poco' endpoint.
1410          * This is used for the 'PortableContact' functionality,
1411          * which is used by both Friendica and Hubzilla.
1412          *
1413          * @param string $url        URL of the given server
1414          * @param array  $serverdata array with server data
1415          *
1416          * @return array server data
1417          */
1418         private static function checkPoCo(string $url, array $serverdata): array
1419         {
1420                 $serverdata['poco'] = '';
1421
1422                 $curlResult = DI::httpClient()->get($url . '/poco', HttpClientAccept::JSON);
1423                 if (!$curlResult->isSuccess()) {
1424                         return $serverdata;
1425                 }
1426
1427                 $data = json_decode($curlResult->getBody(), true);
1428                 if (empty($data)) {
1429                         return $serverdata;
1430                 }
1431
1432                 if (!empty($data['totalResults'])) {
1433                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1434                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers, 1);
1435                         $serverdata['directory-type'] = self::DT_POCO;
1436                         $serverdata['poco'] = $url . '/poco';
1437                 }
1438
1439                 return $serverdata;
1440         }
1441
1442         /**
1443          * Checks if the given server does have a Mastodon style directory endpoint.
1444          *
1445          * @param string $url        URL of the given server
1446          * @param array  $serverdata array with server data
1447          *
1448          * @return array server data
1449          */
1450         public static function checkMastodonDirectory(string $url, array $serverdata): array
1451         {
1452                 $curlResult = DI::httpClient()->get($url . '/api/v1/directory?limit=1', HttpClientAccept::JSON);
1453                 if (!$curlResult->isSuccess()) {
1454                         return $serverdata;
1455                 }
1456
1457                 $data = json_decode($curlResult->getBody(), true);
1458                 if (empty($data)) {
1459                         return $serverdata;
1460                 }
1461
1462                 if (count($data) == 1) {
1463                         $serverdata['directory-type'] = self::DT_MASTODON;
1464                 }
1465
1466                 return $serverdata;
1467         }
1468
1469         /**
1470          * Detects Peertube via their known endpoint
1471          *
1472          * @param string $url        URL of the given server
1473          * @param array  $serverdata array with server data
1474          *
1475          * @return array server data
1476          */
1477         private static function detectPeertube(string $url, array $serverdata): array
1478         {
1479                 $curlResult = DI::httpClient()->get($url . '/api/v1/config', HttpClientAccept::JSON);
1480                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1481                         return $serverdata;
1482                 }
1483
1484                 $data = json_decode($curlResult->getBody(), true);
1485                 if (empty($data)) {
1486                         return $serverdata;
1487                 }
1488
1489                 if (!empty($data['instance']) && !empty($data['serverVersion'])) {
1490                         $serverdata['platform'] = 'peertube';
1491                         $serverdata['version'] = $data['serverVersion'];
1492                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1493
1494                         if (!empty($data['instance']['name'])) {
1495                                 $serverdata['site_name'] = $data['instance']['name'];
1496                         }
1497
1498                         if (!empty($data['instance']['shortDescription'])) {
1499                                 $serverdata['info'] = $data['instance']['shortDescription'];
1500                         }
1501
1502                         if (!empty($data['signup'])) {
1503                                 if (!empty($data['signup']['allowed'])) {
1504                                         $serverdata['register_policy'] = Register::OPEN;
1505                                 }
1506                         }
1507
1508                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1509                                 $serverdata['detection-method'] = self::DETECT_V1_CONFIG;
1510                         }
1511                 }
1512
1513                 return $serverdata;
1514         }
1515
1516         /**
1517          * Detects the version number of a given server when it was a NextCloud installation
1518          *
1519          * @param string $url        URL of the given server
1520          * @param array  $serverdata array with server data
1521          * @param bool   $validHostMeta
1522          *
1523          * @return array server data
1524          */
1525         private static function detectNextcloud(string $url, array $serverdata, bool $validHostMeta): array
1526         {
1527                 $curlResult = DI::httpClient()->get($url . '/status.php', HttpClientAccept::JSON);
1528                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1529                         return $serverdata;
1530                 }
1531
1532                 $data = json_decode($curlResult->getBody(), true);
1533                 if (empty($data)) {
1534                         return $serverdata;
1535                 }
1536
1537                 if (!empty($data['version'])) {
1538                         $serverdata['platform'] = 'nextcloud';
1539                         $serverdata['version'] = $data['version'];
1540
1541                         if ($validHostMeta) {
1542                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1543                         }
1544
1545                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1546                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1547                         }
1548                 }
1549
1550                 return $serverdata;
1551         }
1552
1553         /**
1554          * Fetches weekly usage data
1555          *
1556          * @param string $url        URL of the given server
1557          * @param array  $serverdata array with server data
1558          *
1559          * @return array server data
1560          */
1561         private static function fetchWeeklyUsage(string $url, array $serverdata): array
1562         {
1563                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance/activity', HttpClientAccept::JSON);
1564                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1565                         return $serverdata;
1566                 }
1567
1568                 $data = json_decode($curlResult->getBody(), true);
1569                 if (empty($data)) {
1570                         return $serverdata;
1571                 }
1572
1573                 $current_week = [];
1574                 foreach ($data as $week) {
1575                         // Use only data from a full week
1576                         if (empty($week['week']) || (time() - $week['week']) < 7 * 24 * 60 * 60) {
1577                                 continue;
1578                         }
1579
1580                         // Most likely the data is sorted correctly. But we better are safe than sorry
1581                         if (empty($current_week['week']) || ($current_week['week'] < $week['week'])) {
1582                                 $current_week = $week;
1583                         }
1584                 }
1585
1586                 if (!empty($current_week['logins'])) {
1587                         $serverdata['active-week-users'] = max($current_week['logins'], 0);
1588                 }
1589
1590                 return $serverdata;
1591         }
1592
1593         /**
1594          * Detects data from a given server url if it was a mastodon alike system
1595          *
1596          * @param string $url        URL of the given server
1597          * @param array  $serverdata array with server data
1598          *
1599          * @return array server data
1600          */
1601         private static function detectMastodonAlikes(string $url, array $serverdata): array
1602         {
1603                 $curlResult = DI::httpClient()->get($url . '/api/v1/instance', HttpClientAccept::JSON);
1604                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1605                         return $serverdata;
1606                 }
1607
1608                 $data = json_decode($curlResult->getBody(), true);
1609                 if (empty($data)) {
1610                         return $serverdata;
1611                 }
1612
1613                 $valid = false;
1614
1615                 if (!empty($data['version'])) {
1616                         $serverdata['platform'] = 'mastodon';
1617                         $serverdata['version'] = $data['version'] ?? '';
1618                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1619                         $valid = true;
1620                 }
1621
1622                 if (!empty($data['title'])) {
1623                         $serverdata['site_name'] = $data['title'];
1624                 }
1625
1626                 if (!empty($data['title']) && empty($serverdata['platform']) && ($serverdata['network'] == Protocol::PHANTOM)) {
1627                         $serverdata['platform'] = 'mastodon';
1628                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1629                         $valid = true;
1630                 }
1631
1632                 if (!empty($data['description'])) {
1633                         $serverdata['info'] = trim($data['description']);
1634                 }
1635
1636                 if (!empty($data['stats']['user_count'])) {
1637                         $serverdata['registered-users'] = max($data['stats']['user_count'], 1);
1638                 }
1639
1640                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1641                         $serverdata['platform'] = strtolower($matches[1]);
1642                         $serverdata['version'] = $matches[2];
1643                         $valid = true;
1644                 }
1645
1646                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1647                         $serverdata['platform'] = 'pleroma';
1648                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1649                         $valid = true;
1650                 }
1651
1652                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1653                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1654                         $serverdata['platform'] = 'pleroma';
1655                         $valid = true;
1656                 }
1657
1658                 if ($valid && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1659                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1660                 }
1661
1662                 return $serverdata;
1663         }
1664
1665         /**
1666          * Detects data from typical Hubzilla endpoints
1667          *
1668          * @param string $url        URL of the given server
1669          * @param array  $serverdata array with server data
1670          *
1671          * @return array server data
1672          */
1673         private static function detectHubzilla(string $url, array $serverdata): array
1674         {
1675                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/config.json', HttpClientAccept::JSON);
1676                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1677                         return $serverdata;
1678                 }
1679
1680                 $data = json_decode($curlResult->getBody(), true);
1681                 if (empty($data) || empty($data['site'])) {
1682                         return $serverdata;
1683                 }
1684
1685                 if (!empty($data['site']['name'])) {
1686                         $serverdata['site_name'] = $data['site']['name'];
1687                 }
1688
1689                 if (!empty($data['site']['platform'])) {
1690                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1691                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1692                         $serverdata['network'] = Protocol::ZOT;
1693                 }
1694
1695                 if (!empty($data['site']['hubzilla'])) {
1696                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1697                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1698                         $serverdata['network'] = Protocol::ZOT;
1699                 }
1700
1701                 if (!empty($data['site']['redmatrix'])) {
1702                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1703                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1704                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1705                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1706                         }
1707
1708                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1709                         $serverdata['network'] = Protocol::ZOT;
1710                 }
1711
1712                 $private = false;
1713                 $inviteonly = false;
1714                 $closed = false;
1715
1716                 if (!empty($data['site']['closed'])) {
1717                         $closed = self::toBoolean($data['site']['closed']);
1718                 }
1719
1720                 if (!empty($data['site']['private'])) {
1721                         $private = self::toBoolean($data['site']['private']);
1722                 }
1723
1724                 if (!empty($data['site']['inviteonly'])) {
1725                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1726                 }
1727
1728                 if (!$closed && !$private and $inviteonly) {
1729                         $serverdata['register_policy'] = Register::APPROVE;
1730                 } elseif (!$closed && !$private) {
1731                         $serverdata['register_policy'] = Register::OPEN;
1732                 } else {
1733                         $serverdata['register_policy'] = Register::CLOSED;
1734                 }
1735
1736                 if (($serverdata['network'] != Protocol::PHANTOM) && in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1737                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1738                 }
1739
1740                 return $serverdata;
1741         }
1742
1743         /**
1744          * Converts input value to a boolean value
1745          *
1746          * @param string|integer $val
1747          *
1748          * @return boolean
1749          */
1750         private static function toBoolean($val): bool
1751         {
1752                 if (($val == 'true') || ($val == 1)) {
1753                         return true;
1754                 } elseif (($val == 'false') || ($val == 0)) {
1755                         return false;
1756                 }
1757
1758                 return $val;
1759         }
1760
1761         /**
1762          * Detect if the URL belongs to a GNU Social server
1763          *
1764          * @param string $url        URL of the given server
1765          * @param array  $serverdata array with server data
1766          *
1767          * @return array server data
1768          */
1769         private static function detectGNUSocial(string $url, array $serverdata): array
1770         {
1771                 // Test for GNU Social
1772                 $curlResult = DI::httpClient()->get($url . '/api/gnusocial/version.json', HttpClientAccept::JSON);
1773                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1774                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1775                         $serverdata['platform'] = 'gnusocial';
1776                         // Remove junk that some GNU Social servers return
1777                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1778                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1779                         $serverdata['version'] = trim($serverdata['version'], '"');
1780                         $serverdata['network'] = Protocol::OSTATUS;
1781
1782                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1783                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1784                         }
1785
1786                         return $serverdata;
1787                 }
1788
1789                 // Test for Statusnet
1790                 $curlResult = DI::httpClient()->get($url . '/api/statusnet/version.json', HttpClientAccept::JSON);
1791                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1792                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1793
1794                         // Remove junk that some GNU Social servers return
1795                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1796                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1797                         $serverdata['version'] = trim($serverdata['version'], '"');
1798
1799                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1800                                 $serverdata['platform'] = 'pleroma';
1801                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1802                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1803                         } else {
1804                                 $serverdata['platform'] = 'statusnet';
1805                                 $serverdata['network'] = Protocol::OSTATUS;
1806                         }
1807
1808                         if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1809                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1810                         }
1811                 }
1812
1813                 return $serverdata;
1814         }
1815
1816         /**
1817          * Detect if the URL belongs to a Friendica server
1818          *
1819          * @param string $url        URL of the given server
1820          * @param array  $serverdata array with server data
1821          *
1822          * @return array server data
1823          */
1824         private static function detectFriendica(string $url, array $serverdata): array
1825         {
1826                 // There is a bug in some versions of Friendica that will return an ActivityStream actor when the content type "application/json" is requested.
1827                 // Because of this me must not use ACCEPT_JSON here.
1828                 $curlResult = DI::httpClient()->get($url . '/friendica/json');
1829                 if (!$curlResult->isSuccess()) {
1830                         $curlResult = DI::httpClient()->get($url . '/friendika/json');
1831                         $friendika = true;
1832                         $platform = 'Friendika';
1833                 } else {
1834                         $friendika = false;
1835                         $platform = 'Friendica';
1836                 }
1837
1838                 if (!$curlResult->isSuccess()) {
1839                         return $serverdata;
1840                 }
1841
1842                 $data = json_decode($curlResult->getBody(), true);
1843                 if (empty($data) || empty($data['version'])) {
1844                         return $serverdata;
1845                 }
1846
1847                 if (in_array($serverdata['detection-method'], self::DETECT_UNSPECIFIC)) {
1848                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1849                 }
1850
1851                 $serverdata['network'] = Protocol::DFRN;
1852                 $serverdata['version'] = $data['version'];
1853
1854                 if (!empty($data['no_scrape_url'])) {
1855                         $serverdata['noscrape'] = $data['no_scrape_url'];
1856                 }
1857
1858                 if (!empty($data['site_name'])) {
1859                         $serverdata['site_name'] = $data['site_name'];
1860                 }
1861
1862                 if (!empty($data['info'])) {
1863                         $serverdata['info'] = trim($data['info']);
1864                 }
1865
1866                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1867                 switch ($register_policy) {
1868                         case 'REGISTER_OPEN':
1869                                 $serverdata['register_policy'] = Register::OPEN;
1870                                 break;
1871
1872                         case 'REGISTER_APPROVE':
1873                                 $serverdata['register_policy'] = Register::APPROVE;
1874                                 break;
1875
1876                         case 'REGISTER_CLOSED':
1877                         case 'REGISTER_INVITATION':
1878                                 $serverdata['register_policy'] = Register::CLOSED;
1879                                 break;
1880                         default:
1881                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1882                                 $serverdata['register_policy'] = Register::CLOSED;
1883                                 break;
1884                 }
1885
1886                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1887
1888                 return $serverdata;
1889         }
1890
1891         /**
1892          * Analyses the landing page of a given server for hints about type and system of that server
1893          *
1894          * @param object $curlResult result of curl execution
1895          * @param array  $serverdata array with server data
1896          *
1897          * @return array server data
1898          */
1899         private static function analyseRootBody($curlResult, array $serverdata): array
1900         {
1901                 if (empty($curlResult->getBody())) {
1902                         return $serverdata;
1903                 }
1904
1905                 if (file_exists(__DIR__ . '/../../static/platforms.config.php')) {
1906                         require __DIR__ . '/../../static/platforms.config.php';
1907                 } else {
1908                         throw new HTTPException\InternalServerErrorException('Invalid platform file');
1909                 }
1910
1911                 $platforms = array_merge($ap_platforms, $dfrn_platforms, $zap_platforms, $platforms);
1912
1913                 $doc = new DOMDocument();
1914                 @$doc->loadHTML($curlResult->getBody());
1915                 $xpath = new DOMXPath($doc);
1916                 $assigned = false;
1917
1918                 // We can only detect honk via some HTML element on their page
1919                 if ($xpath->query('//div[@id="honksonpage"]')->count() == 1) {
1920                         $serverdata['platform'] = 'honk';
1921                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1922                         $assigned = true;
1923                 }
1924
1925                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1926                 if (!empty($title)) {
1927                         $serverdata['site_name'] = $title;
1928                 }
1929
1930                 $list = $xpath->query('//meta[@name]');
1931
1932                 foreach ($list as $node) {
1933                         $attr = [];
1934                         if ($node->attributes->length) {
1935                                 foreach ($node->attributes as $attribute) {
1936                                         $value = trim($attribute->value);
1937                                         if (empty($value)) {
1938                                                 continue;
1939                                         }
1940
1941                                         $attr[$attribute->name] = $value;
1942                                 }
1943
1944                                 if (empty($attr['name']) || empty($attr['content'])) {
1945                                         continue;
1946                                 }
1947                         }
1948
1949                         if ($attr['name'] == 'description') {
1950                                 $serverdata['info'] = $attr['content'];
1951                         }
1952
1953                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1954                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad', 'generator'])) {
1955                                 $platform = str_ireplace(array_keys($platforms), array_values($platforms), $attr['content']);
1956                                 $platform = str_replace('/', ' ', $platform);
1957                                 $platform_parts = explode(' ', $platform);
1958                                 if ((count($platform_parts) >= 2) && in_array(strtolower($platform_parts[0]), array_values($platforms))) {
1959                                         $platform = $platform_parts[0];
1960                                         $serverdata['version'] = $platform_parts[1];
1961                                 }
1962                                 if (in_array($platform, array_values($dfrn_platforms))) {
1963                                         $serverdata['network'] = Protocol::DFRN;
1964                                 } elseif (in_array($platform, array_values($ap_platforms))) {
1965                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1966                                 } elseif (in_array($platform, array_values($zap_platforms))) {
1967                                         $serverdata['network'] = Protocol::ZOT;
1968                                 }
1969                                 if (in_array($platform, array_values($platforms))) {
1970                                         $serverdata['platform'] = $platform;
1971                                         $assigned = true;
1972                                 }
1973                         }
1974                 }
1975
1976                 $list = $xpath->query('//meta[@property]');
1977
1978                 foreach ($list as $node) {
1979                         $attr = [];
1980                         if ($node->attributes->length) {
1981                                 foreach ($node->attributes as $attribute) {
1982                                         $value = trim($attribute->value);
1983                                         if (empty($value)) {
1984                                                 continue;
1985                                         }
1986
1987                                         $attr[$attribute->name] = $value;
1988                                 }
1989
1990                                 if (empty($attr['property']) || empty($attr['content'])) {
1991                                         continue;
1992                                 }
1993                         }
1994
1995                         if ($attr['property'] == 'og:site_name') {
1996                                 $serverdata['site_name'] = $attr['content'];
1997                         }
1998
1999                         if ($attr['property'] == 'og:description') {
2000                                 $serverdata['info'] = $attr['content'];
2001                         }
2002
2003                         if (in_array($attr['property'], ['og:platform', 'generator'])) {
2004                                 if (in_array($attr['content'], array_keys($platforms))) {
2005                                         $serverdata['platform'] = $platforms[$attr['content']];
2006                                         $assigned = true;
2007                                 }
2008
2009                                 if (in_array($attr['content'], array_keys($ap_platforms))) {
2010                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2011                                 } elseif (in_array($attr['content'], array_values($zap_platforms))) {
2012                                         $serverdata['network'] = Protocol::ZOT;
2013                                 }
2014                         }
2015                 }
2016
2017                 $list = $xpath->query('//link[@rel="me"]');
2018                 foreach ($list as $node) {
2019                         foreach ($node->attributes as $attribute) {
2020                                 if (parse_url(trim($attribute->value), PHP_URL_HOST) == 'micro.blog') {
2021                                         $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2022                                         $serverdata['platform'] = 'microblog';
2023                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2024                                         $assigned = true;
2025                                 }
2026                         }
2027                 }
2028
2029                 if ($serverdata['platform'] != 'microblog') {
2030                         $list = $xpath->query('//link[@rel="micropub"]');
2031                         foreach ($list as $node) {
2032                                 foreach ($node->attributes as $attribute) {
2033                                         if (trim($attribute->value) == 'https://micro.blog/micropub') {
2034                                                 $serverdata['version'] = trim($serverdata['platform'] . ' ' . $serverdata['version']);
2035                                                 $serverdata['platform'] = 'microblog';
2036                                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
2037                                                 $assigned = true;
2038                                         }
2039                                 }
2040                         }
2041                 }
2042
2043                 if ($assigned && in_array($serverdata['detection-method'], [self::DETECT_MANUAL, self::DETECT_HEADER])) {
2044                         $serverdata['detection-method'] = self::DETECT_BODY;
2045                 }
2046
2047                 return $serverdata;
2048         }
2049
2050         /**
2051          * Analyses the header data of a given server for hints about type and system of that server
2052          *
2053          * @param object $curlResult result of curl execution
2054          * @param array  $serverdata array with server data
2055          *
2056          * @return array server data
2057          */
2058         private static function analyseRootHeader($curlResult, array $serverdata): array
2059         {
2060                 if ($curlResult->getHeader('server') == 'Mastodon') {
2061                         $serverdata['platform'] = 'mastodon';
2062                         $serverdata['network'] = Protocol::ACTIVITYPUB;
2063                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
2064                         $serverdata['platform'] = 'diaspora';
2065                         $serverdata['network'] = Protocol::DIASPORA;
2066                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version')[0] ?? '';
2067                 } elseif ($curlResult->inHeader('x-friendica-version')) {
2068                         $serverdata['platform'] = 'friendica';
2069                         $serverdata['network'] = Protocol::DFRN;
2070                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version')[0] ?? '';
2071                 } else {
2072                         return $serverdata;
2073                 }
2074
2075                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
2076                         $serverdata['detection-method'] = self::DETECT_HEADER;
2077                 }
2078
2079                 return $serverdata;
2080         }
2081
2082         /**
2083          * Update GServer entries
2084          */
2085         public static function discover()
2086         {
2087                 // Update the server list
2088                 self::discoverFederation();
2089
2090                 $no_of_queries = 5;
2091
2092                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
2093
2094                 if ($requery_days == 0) {
2095                         $requery_days = 7;
2096                 }
2097
2098                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
2099
2100                 $gservers = DBA::select('gserver', ['id', 'url', 'nurl', 'network', 'poco', 'directory-type'],
2101                         ["NOT `failed` AND `directory-type` != ? AND `last_poco_query` < ?", GServer::DT_NONE, $last_update],
2102                         ['order' => ['RAND()']]);
2103
2104                 while ($gserver = DBA::fetch($gservers)) {
2105                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2106                         Worker::add(PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
2107
2108                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
2109                         Worker::add(PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
2110
2111                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
2112                         self::update($fields, ['nurl' => $gserver['nurl']]);
2113
2114                         if (--$no_of_queries == 0) {
2115                                 break;
2116                         }
2117                 }
2118
2119                 DBA::close($gservers);
2120         }
2121
2122         /**
2123          * Discover federated servers
2124          */
2125         private static function discoverFederation()
2126         {
2127                 $last = DI::config()->get('poco', 'last_federation_discovery');
2128
2129                 if ($last) {
2130                         $next = $last + (24 * 60 * 60);
2131
2132                         if ($next > time()) {
2133                                 return;
2134                         }
2135                 }
2136
2137                 // Discover federated servers
2138                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
2139                 foreach ($protocols as $protocol) {
2140                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
2141                         $curlResult = DI::httpClient()->fetch('https://the-federation.info/graphql?query=' . urlencode($query), HttpClientAccept::JSON);
2142                         if (!empty($curlResult)) {
2143                                 $data = json_decode($curlResult, true);
2144                                 if (!empty($data['data']['nodes'])) {
2145                                         foreach ($data['data']['nodes'] as $server) {
2146                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
2147                                                 self::add('https://' . $server['host'], true);
2148                                         }
2149                                 }
2150                         }
2151                 }
2152
2153                 // Disvover Mastodon servers
2154                 $accesstoken = DI::config()->get('system', 'instances_social_key');
2155
2156                 if (!empty($accesstoken)) {
2157                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
2158                         $curlResult = DI::httpClient()->get($api, HttpClientAccept::JSON, [HttpClientOptions::HEADERS => ['Authorization' => ['Bearer ' . $accesstoken]]]);
2159                         if ($curlResult->isSuccess()) {
2160                                 $servers = json_decode($curlResult->getBody(), true);
2161
2162                                 foreach ($servers['instances'] as $server) {
2163                                         $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
2164                                         self::add($url);
2165                                 }
2166                         }
2167                 }
2168
2169                 DI::config()->set('poco', 'last_federation_discovery', time());
2170         }
2171
2172         /**
2173          * Set the protocol for the given server
2174          *
2175          * @param int $gsid     Server id
2176          * @param int $protocol Protocol id
2177          *
2178          * @throws Exception
2179          */
2180         public static function setProtocol(int $gsid, int $protocol)
2181         {
2182                 if (empty($gsid)) {
2183                         return;
2184                 }
2185
2186                 $gserver = DBA::selectFirst('gserver', ['protocol', 'url'], ['id' => $gsid]);
2187                 if (!DBA::isResult($gserver)) {
2188                         return;
2189                 }
2190
2191                 $old = $gserver['protocol'];
2192
2193                 if (!is_null($old)) {
2194                         /*
2195                         The priority for the protocols is:
2196                                 1. ActivityPub
2197                                 2. DFRN via Diaspora
2198                                 3. Legacy DFRN
2199                                 4. Diaspora
2200                                 5. OStatus
2201                         */
2202
2203                         // We don't need to change it when nothing is to be changed
2204                         if ($old == $protocol) {
2205                                 return;
2206                         }
2207
2208                         // We don't want to mark a server as OStatus when it had been marked with any other protocol before
2209                         if ($protocol == Post\DeliveryData::OSTATUS) {
2210                                 return;
2211                         }
2212
2213                         // If the server is marked as ActivityPub then we won't change it to anything different
2214                         if ($old == Post\DeliveryData::ACTIVITYPUB) {
2215                                 return;
2216                         }
2217
2218                         // Don't change it to anything lower than DFRN if the new one wasn't ActivityPub
2219                         if (($old == Post\DeliveryData::DFRN) && ($protocol != Post\DeliveryData::ACTIVITYPUB)) {
2220                                 return;
2221                         }
2222
2223                         // Don't change it to Diaspora when it is a legacy DFRN server
2224                         if (($old == Post\DeliveryData::LEGACY_DFRN) && ($protocol == Post\DeliveryData::DIASPORA)) {
2225                                 return;
2226                         }
2227                 }
2228
2229                 Logger::info('Protocol for server', ['protocol' => $protocol, 'old' => $old, 'id' => $gsid, 'url' => $gserver['url'], 'callstack' => System::callstack(20)]);
2230                 self::update(['protocol' => $protocol], ['id' => $gsid]);
2231         }
2232
2233         /**
2234          * Fetch the protocol of the given server
2235          *
2236          * @param int $gsid Server id
2237          *
2238          * @return ?int One of Post\DeliveryData protocol constants or null if unknown or gserver is missing
2239          *
2240          * @throws Exception
2241          */
2242         public static function getProtocol(int $gsid): ?int
2243         {
2244                 if (empty($gsid)) {
2245                         return null;
2246                 }
2247
2248                 $gserver = DBA::selectFirst('gserver', ['protocol'], ['id' => $gsid]);
2249                 if (DBA::isResult($gserver)) {
2250                         return $gserver['protocol'];
2251                 }
2252
2253                 return null;
2254         }
2255
2256         /**
2257          * Enforces gserver table field maximum sizes to avoid "Data too long" database errors
2258          *
2259          * @param array $fields
2260          * @param array $condition
2261          *
2262          * @return bool
2263          *
2264          * @throws Exception
2265          */
2266         public static function update(array $fields, array $condition): bool
2267         {
2268                 $fields = DI::dbaDefinition()->truncateFieldsForTable('gserver', $fields);
2269
2270                 return DBA::update('gserver', $fields, $condition);
2271         }
2272 }