src/Model/GServer.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2020, Friendica
   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 Friendica\Core\Protocol;
  27 use Friendica\Core\Worker;
  28 use Friendica\Database\DBA;
  29 use Friendica\DI;
  30 use Friendica\Module\Register;
  31 use Friendica\Network\CurlResult;
  32 use Friendica\Util\Network;
  33 use Friendica\Util\DateTimeFormat;
  34 use Friendica\Util\Strings;
  35 use Friendica\Util\XML;
  36 use Friendica\Core\Logger;
  37 use Friendica\Core\System;
  38 use Friendica\Protocol\PortableContact;
  39 use Friendica\Protocol\Diaspora;
  40 use Friendica\Network\Probe;
  41
  42 /**
  43  * This class handles GServer related functions
  44  */
  45 class GServer
  46 {
  47         // Directory types
  48         const DT_NONE = 0;
  49         const DT_POCO = 1;
  50         const DT_MASTODON = 2;
  51
  52         // Methods to detect server types
  53
  54         // Non endpoint specific methods
  55         const DETECT_MANUAL = 0;
  56         const DETECT_HEADER = 1;
  57         const DETECT_BODY = 2;
  58
  59         // Implementation specific endpoints
  60         const DETECT_FRIENDIKA = 10;
  61         const DETECT_FRIENDICA = 11;
  62         const DETECT_STATUSNET = 12;
  63         const DETECT_GNUSOCIAL = 13;
  64         const DETECT_CONFIG_JSON = 14; // Statusnet, GNU Social, Older Hubzilla/Redmatrix
  65         const DETECT_SITEINFO_JSON = 15; // Newer Hubzilla
  66         const DETECT_MASTODON_API = 16;
  67         const DETECT_STATUS_PHP = 17; // Nextcloud
  68
  69         // Standardized endpoints
  70         const DETECT_STATISTICS_JSON = 100;
  71         const DETECT_NODEINFO_1 = 101;
  72         const DETECT_NODEINFO_2 = 102;
  73
  74         /**
  75          * Get the ID for the given server URL
  76          *
  77          * @param string $url
  78          * @param boolean $no_check Don't check if the server hadn't been found
  79          * @return int gserver id
  80          */
  81         public static function getID(string $url, bool $no_check = false)
  82         {
  83                 if (empty($url)) {
  84                         return null;
  85                 }
  86
  87                 $url = self::cleanURL($url);
  88
  89                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
  90                 if (DBA::isResult($gserver)) {
  91                         Logger::info('Got ID for URL', ['id' => $gserver['id'], 'url' => $url, 'callstack' => System::callstack(20)]);
  92                         return $gserver['id'];
  93                 }
  94
  95                 if ($no_check || !self::check($url)) {
  96                         return null;
  97                 }
  98
  99                 return self::getID($url, true);
 100         }
 101
 102         /**
 103          * Checks if the given server is reachable
 104          *
 105          * @param string  $profile URL of the given profile
 106          * @param string  $server  URL of the given server (If empty, taken from profile)
 107          * @param string  $network Network value that is used, when detection failed
 108          * @param boolean $force   Force an update.
 109          *
 110          * @return boolean 'true' if server seems vital
 111          */
 112         public static function reachable(string $profile, string $server = '', string $network = '', bool $force = false)
 113         {
 114                 if ($server == '') {
 115                         $server = GContact::getBasepath($profile);
 116                 }
 117
 118                 if ($server == '') {
 119                         return true;
 120                 }
 121
 122                 return self::check($server, $network, $force);
 123         }
 124
 125         /**
 126          * Decides if a server needs to be updated, based upon several date fields
 127          *
 128          * @param date $created      Creation date of that server entry
 129          * @param date $updated      When had the server entry be updated
 130          * @param date $last_failure Last failure when contacting that server
 131          * @param date $last_contact Last time the server had been contacted
 132          *
 133          * @return boolean Does the server record needs an update?
 134          */
 135         public static function updateNeeded($created, $updated, $last_failure, $last_contact)
 136         {
 137                 $now = strtotime(DateTimeFormat::utcNow());
 138
 139                 if ($updated > $last_contact) {
 140                         $contact_time = strtotime($updated);
 141                 } else {
 142                         $contact_time = strtotime($last_contact);
 143                 }
 144
 145                 $failure_time = strtotime($last_failure);
 146                 $created_time = strtotime($created);
 147
 148                 // If there is no "created" time then use the current time
 149                 if ($created_time <= 0) {
 150                         $created_time = $now;
 151                 }
 152
 153                 // If the last contact was less than 24 hours then don't update
 154                 if (($now - $contact_time) < (60 * 60 * 24)) {
 155                         return false;
 156                 }
 157
 158                 // If the last failure was less than 24 hours then don't update
 159                 if (($now - $failure_time) < (60 * 60 * 24)) {
 160                         return false;
 161                 }
 162
 163                 // If the last contact was less than a week ago and the last failure is older than a week then don't update
 164                 //if ((($now - $contact_time) < (60 * 60 * 24 * 7)) && ($contact_time > $failure_time))
 165                 //      return false;
 166
 167                 // If the last contact time was more than a week ago and the contact was created more than a week ago, then only try once a week
 168                 if ((($now - $contact_time) > (60 * 60 * 24 * 7)) && (($now - $created_time) > (60 * 60 * 24 * 7)) && (($now - $failure_time) < (60 * 60 * 24 * 7))) {
 169                         return false;
 170                 }
 171
 172                 // If the last contact time was more than a month ago and the contact was created more than a month ago, then only try once a month
 173                 if ((($now - $contact_time) > (60 * 60 * 24 * 30)) && (($now - $created_time) > (60 * 60 * 24 * 30)) && (($now - $failure_time) < (60 * 60 * 24 * 30))) {
 174                         return false;
 175                 }
 176
 177                 return true;
 178         }
 179
 180         /**
 181          * Checks the state of the given server.
 182          *
 183          * @param string  $server_url    URL of the given server
 184          * @param string  $network       Network value that is used, when detection failed
 185          * @param boolean $force         Force an update.
 186          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 187          *
 188          * @return boolean 'true' if server seems vital
 189          */
 190         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false)
 191         {
 192                 $server_url = self::cleanURL($server_url);
 193
 194                 if ($server_url == '') {
 195                         return false;
 196                 }
 197
 198                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
 199                 if (DBA::isResult($gserver)) {
 200                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
 201                                 $fields = ['created' => DateTimeFormat::utcNow()];
 202                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
 203                                 DBA::update('gserver', $fields, $condition);
 204                         }
 205
 206                         $last_contact = $gserver['last_contact'];
 207                         $last_failure = $gserver['last_failure'];
 208
 209                         // See discussion under https://forum.friendi.ca/display/0b6b25a8135aabc37a5a0f5684081633
 210                         // It can happen that a zero date is in the database, but storing it again is forbidden.
 211                         if ($last_contact < DBA::NULL_DATETIME) {
 212                                 $last_contact = DBA::NULL_DATETIME;
 213                         }
 214
 215                         if ($last_failure < DBA::NULL_DATETIME) {
 216                                 $last_failure = DBA::NULL_DATETIME;
 217                         }
 218
 219                         if (!$force && !self::updateNeeded($gserver['created'], '', $last_failure, $last_contact)) {
 220                                 Logger::info('No update needed', ['server' => $server_url]);
 221                                 return ($last_contact >= $last_failure);
 222                         }
 223                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force, 'Created' => $gserver['created'], 'Failure' => $last_failure, 'Contact' => $last_contact]);
 224                 } else {
 225                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
 226                 }
 227
 228                 return self::detect($server_url, $network, $only_nodeinfo);
 229         }
 230
 231         /**
 232          * Set failed server status
 233          *
 234          * @param string $url
 235          */
 236         private static function setFailure(string $url)
 237         {
 238                 if (DBA::exists('gserver', ['nurl' => Strings::normaliseLink($url)])) {
 239                         DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow(), 'detection-method' => null],
 240                         ['nurl' => Strings::normaliseLink($url)]);
 241                         Logger::info('Set failed status for existing server', ['url' => $url]);
 242                         return;
 243                 }
 244                 DBA::insert('gserver', ['url' => $url, 'nurl' => Strings::normaliseLink($url),
 245                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
 246                         'last_failure' => DateTimeFormat::utcNow()]);
 247                 Logger::info('Set failed status for new server', ['url' => $url]);
 248         }
 249
 250         /**
 251          * Remove unwanted content from the given URL
 252          *
 253          * @param string $url
 254          * @return string cleaned URL
 255          */
 256         public static function cleanURL(string $url)
 257         {
 258                 $url = trim($url, '/');
 259                 $url = str_replace('/index.php', '', $url);
 260
 261                 $urlparts = parse_url($url);
 262                 unset($urlparts['user']);
 263                 unset($urlparts['pass']);
 264                 unset($urlparts['query']);
 265                 unset($urlparts['fragment']);
 266                 return Network::unparseURL($urlparts);
 267         }
 268
 269         /**
 270          * Return the base URL
 271          *
 272          * @param string $url
 273          * @return string base URL
 274          */
 275         private static function getBaseURL(string $url)
 276         {
 277                 $urlparts = parse_url(self::cleanURL($url));
 278                 unset($urlparts['path']);
 279                 return Network::unparseURL($urlparts);
 280         }
 281
 282         /**
 283          * Detect server data (type, protocol, version number, ...)
 284          * The detected data is then updated or inserted in the gserver table.
 285          *
 286          * @param string  $url           URL of the given server
 287          * @param string  $network       Network value that is used, when detection failed
 288          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
 289          *
 290          * @return boolean 'true' if server could be detected
 291          */
 292         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false)
 293         {
 294                 Logger::info('Detect server type', ['server' => $url]);
 295                 $serverdata = ['detection-method' => self::DETECT_MANUAL];
 296
 297                 $original_url = $url;
 298
 299                 // Remove URL content that is not supposed to exist for a server url
 300                 $url = self::cleanURL($url);
 301
 302                 // Get base URL
 303                 $baseurl = self::getBaseURL($url);
 304
 305                 // If the URL missmatches, then we mark the old entry as failure
 306                 if ($url != $original_url) {
 307                         DBA::update('gserver', ['last_failure' => DateTimeFormat::utcNow()], ['nurl' => Strings::normaliseLink($original_url)]);
 308                 }
 309
 310                 // When a nodeinfo is present, we don't need to dig further
 311                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
 312                 $curlResult = Network::curl($url . '/.well-known/nodeinfo', false, ['timeout' => $xrd_timeout]);
 313                 if ($curlResult->isTimeout()) {
 314                         self::setFailure($url);
 315                         return false;
 316                 }
 317
 318                 $nodeinfo = self::fetchNodeinfo($url, $curlResult);
 319                 if ($only_nodeinfo && empty($nodeinfo)) {
 320                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
 321                         self::setFailure($url);
 322                         return false;
 323                 }
 324
 325                 // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
 326                 if (empty($nodeinfo)) {
 327                         $nodeinfo = self::fetchStatistics($url);
 328                 }
 329
 330                 // If that didn't work out well, we use some protocol specific endpoints
 331                 // For Friendica and Zot based networks we have to dive deeper to reveal more details
 332                 if (empty($nodeinfo['network']) || in_array($nodeinfo['network'], [Protocol::DFRN, Protocol::ZOT])) {
 333                         if (!empty($nodeinfo['detection-method'])) {
 334                                 $serverdata['detection-method'] = $nodeinfo['detection-method'];
 335                         }
 336
 337                         // Fetch the landing page, possibly it reveals some data
 338                         if (empty($nodeinfo['network'])) {
 339                                 if ($baseurl == $url) {
 340                                         $basedata = $serverdata;
 341                                 } else {
 342                                         $basedata = ['detection-method' => self::DETECT_MANUAL];
 343                                 }
 344
 345                                 $curlResult = Network::curl($baseurl, false, ['timeout' => $xrd_timeout]);
 346                                 if ($curlResult->isSuccess()) {
 347                                         $basedata = self::analyseRootHeader($curlResult, $basedata);
 348                                         $basedata = self::analyseRootBody($curlResult, $basedata, $baseurl);
 349                                 }
 350
 351                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody()) || self::invalidBody($curlResult->getBody())) {
 352                                         self::setFailure($url);
 353                                         return false;
 354                                 }
 355
 356                                 if ($baseurl == $url) {
 357                                         $serverdata = $basedata;
 358                                 } else {
 359                                         // When the base path doesn't seem to contain a social network we try the complete path.
 360                                         // Most detectable system have to be installed in the root directory.
 361                                         // We checked the base to avoid false positives.
 362                                         $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout]);
 363                                         if ($curlResult->isSuccess()) {
 364                                                 $urldata = self::analyseRootHeader($curlResult, $serverdata);
 365                                                 $urldata = self::analyseRootBody($curlResult, $urldata, $url);
 366
 367                                                 $comparebase = $basedata;
 368                                                 unset($comparebase['info']);
 369                                                 unset($comparebase['site_name']);
 370                                                 $compareurl = $urldata;
 371                                                 unset($compareurl['info']);
 372                                                 unset($compareurl['site_name']);
 373
 374                                                 // We assume that no one will install the identical system in the root and a subfolder
 375                                                 if (!empty(array_diff($comparebase, $compareurl))) {
 376                                                         $serverdata = $urldata;
 377                                                 }
 378                                         }
 379                                 }
 380                         }
 381
 382                         if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ACTIVITYPUB)) {
 383                                 $serverdata = self::detectMastodonAlikes($url, $serverdata);
 384                         }
 385
 386                         // All following checks are done for systems that always have got a "host-meta" endpoint.
 387                         // With this check we don't have to waste time and ressources for dead systems.
 388                         // Also this hopefully prevents us from receiving abuse messages.
 389                         if (empty($serverdata['network']) && !self::validHostMeta($url)) {
 390                                 self::setFailure($url);
 391                                 return false;
 392                         }
 393
 394                         if (empty($serverdata['network']) || in_array($serverdata['network'], [Protocol::DFRN, Protocol::ACTIVITYPUB])) {
 395                                 $serverdata = self::detectFriendica($url, $serverdata);
 396                         }
 397
 398                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
 399                         if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ZOT)) {
 400                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
 401                         }
 402
 403                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations
 404                         if (empty($serverdata['network'])) {
 405                                 $serverdata = self::detectHubzilla($url, $serverdata);
 406                         }
 407
 408                         if (empty($serverdata['network'])) {
 409                                 $serverdata = self::detectNextcloud($url, $serverdata);
 410                         }
 411
 412                         if (empty($serverdata['network'])) {
 413                                 $serverdata = self::detectGNUSocial($url, $serverdata);
 414                         }
 415
 416                         $serverdata = array_merge($nodeinfo, $serverdata);
 417                 } else {
 418                         $serverdata = $nodeinfo;
 419                 }
 420
 421                 // Detect the directory type
 422                 $serverdata['directory-type'] = self::DT_NONE;
 423                 $serverdata = self::checkPoCo($url, $serverdata);
 424                 $serverdata = self::checkMastodonDirectory($url, $serverdata);
 425
 426                 // We can't detect the network type. Possibly it is some system that we don't know yet
 427                 if (empty($serverdata['network'])) {
 428                         $serverdata['network'] = Protocol::PHANTOM;
 429                 }
 430
 431                 // When we hadn't been able to detect the network type, we use the hint from the parameter
 432                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
 433                         $serverdata['network'] = $network;
 434                 }
 435
 436                 $serverdata['url'] = $url;
 437                 $serverdata['nurl'] = Strings::normaliseLink($url);
 438
 439                 // We take the highest number that we do find
 440                 $registeredUsers = $serverdata['registered-users'] ?? 0;
 441
 442                 // On an active server there has to be at least a single user
 443                 if (($serverdata['network'] != Protocol::PHANTOM) && ($registeredUsers == 0)) {
 444                         $registeredUsers = 1;
 445                 }
 446
 447                 if ($serverdata['network'] == Protocol::PHANTOM) {
 448                         $serverdata['registered-users'] = $registeredUsers;
 449                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
 450                 }
 451
 452                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
 453
 454                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
 455                 if (!DBA::isResult($gserver)) {
 456                         $serverdata['created'] = DateTimeFormat::utcNow();
 457                         $ret = DBA::insert('gserver', $serverdata);
 458                         $id = DBA::lastInsertId();
 459                 } else {
 460                         // Don't override the network with 'unknown' when there had been a valid entry before
 461                         if (($serverdata['network'] == Protocol::PHANTOM) && !empty($gserver['network'])) {
 462                                 unset($serverdata['network']);
 463                         }
 464
 465                         $ret = DBA::update('gserver', $serverdata, ['nurl' => $serverdata['nurl']]);
 466                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
 467                         if (DBA::isResult($gserver)) {
 468                                 $id = $gserver['id'];
 469                         }
 470                 }
 471
 472                 if (!empty($serverdata['network']) && !empty($id) && ($serverdata['network'] != Protocol::PHANTOM)) {
 473                         $gcontacts = DBA::count('gcontact', ['gsid' => $id]);
 474                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
 475                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id]);
 476                         $max_users = max($gcontacts, $apcontacts, $contacts, $registeredUsers);
 477                         if ($max_users > $registeredUsers) {
 478                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
 479                                 DBA::update('gserver', ['registered-users' => $max_users], ['id' => $id]);
 480                         }
 481                 }
 482
 483                 if (!empty($serverdata['network']) && in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
 484                         self::discoverRelay($url);
 485                 }
 486
 487                 return $ret;
 488         }
 489
 490         /**
 491          * Fetch relay data from a given server url
 492          *
 493          * @param string $server_url address of the server
 494          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 495          */
 496         private static function discoverRelay(string $server_url)
 497         {
 498                 Logger::info('Discover relay data', ['server' => $server_url]);
 499
 500                 $curlResult = Network::curl($server_url . '/.well-known/x-social-relay');
 501                 if (!$curlResult->isSuccess()) {
 502                         return;
 503                 }
 504
 505                 $data = json_decode($curlResult->getBody(), true);
 506                 if (!is_array($data)) {
 507                         return;
 508                 }
 509
 510                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
 511                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
 512
 513                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
 514                         $data['scope'] = '';
 515                         $data['subscribe'] = false;
 516                         $data['tags'] = [];
 517                 }
 518
 519                 $gserver = DBA::selectFirst('gserver', ['id', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
 520                 if (!DBA::isResult($gserver)) {
 521                         return;
 522                 }
 523
 524                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
 525                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
 526                         DBA::update('gserver', $fields, ['id' => $gserver['id']]);
 527                 }
 528
 529                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
 530
 531                 if ($data['scope'] == 'tags') {
 532                         // Avoid duplicates
 533                         $tags = [];
 534                         foreach ($data['tags'] as $tag) {
 535                                 $tag = mb_strtolower($tag);
 536                                 if (strlen($tag) < 100) {
 537                                         $tags[$tag] = $tag;
 538                                 }
 539                         }
 540
 541                         foreach ($tags as $tag) {
 542                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], true);
 543                         }
 544                 }
 545
 546                 // Create or update the relay contact
 547                 $fields = [];
 548                 if (isset($data['protocols'])) {
 549                         if (isset($data['protocols']['diaspora'])) {
 550                                 $fields['network'] = Protocol::DIASPORA;
 551
 552                                 if (isset($data['protocols']['diaspora']['receive'])) {
 553                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
 554                                 } elseif (is_string($data['protocols']['diaspora'])) {
 555                                         $fields['batch'] = $data['protocols']['diaspora'];
 556                                 }
 557                         }
 558
 559                         if (isset($data['protocols']['dfrn'])) {
 560                                 $fields['network'] = Protocol::DFRN;
 561
 562                                 if (isset($data['protocols']['dfrn']['receive'])) {
 563                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
 564                                 } elseif (is_string($data['protocols']['dfrn'])) {
 565                                         $fields['batch'] = $data['protocols']['dfrn'];
 566                                 }
 567                         }
 568                 }
 569                 Diaspora::setRelayContact($server_url, $fields);
 570         }
 571
 572         /**
 573          * Fetch server data from '/statistics.json' on the given server
 574          *
 575          * @param string $url URL of the given server
 576          *
 577          * @return array server data
 578          */
 579         private static function fetchStatistics(string $url)
 580         {
 581                 $curlResult = Network::curl($url . '/statistics.json');
 582                 if (!$curlResult->isSuccess()) {
 583                         return [];
 584                 }
 585
 586                 $data = json_decode($curlResult->getBody(), true);
 587                 if (empty($data)) {
 588                         return [];
 589                 }
 590
 591                 $serverdata = ['detection-method' => self::DETECT_STATISTICS_JSON];
 592
 593                 if (!empty($data['version'])) {
 594                         $serverdata['version'] = $data['version'];
 595                         // Version numbers on statistics.json are presented with additional info, e.g.:
 596                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 597                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
 598                 }
 599
 600                 if (!empty($data['name'])) {
 601                         $serverdata['site_name'] = $data['name'];
 602                 }
 603
 604                 if (!empty($data['network'])) {
 605                         $serverdata['platform'] = strtolower($data['network']);
 606
 607                         if ($serverdata['platform'] == 'diaspora') {
 608                                 $serverdata['network'] = Protocol::DIASPORA;
 609                         } elseif ($serverdata['platform'] == 'friendica') {
 610                                 $serverdata['network'] = Protocol::DFRN;
 611                         } elseif ($serverdata['platform'] == 'hubzilla') {
 612                                 $serverdata['network'] = Protocol::ZOT;
 613                         } elseif ($serverdata['platform'] == 'redmatrix') {
 614                                 $serverdata['network'] = Protocol::ZOT;
 615                         }
 616                 }
 617
 618
 619                 if (!empty($data['registrations_open'])) {
 620                         $serverdata['register_policy'] = Register::OPEN;
 621                 } else {
 622                         $serverdata['register_policy'] = Register::CLOSED;
 623                 }
 624
 625                 return $serverdata;
 626         }
 627
 628         /**
 629          * Detect server type by using the nodeinfo data
 630          *
 631          * @param string     $url        address of the server
 632          * @param CurlResult $curlResult
 633          * @return array Server data
 634          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 635          */
 636         private static function fetchNodeinfo(string $url, CurlResult $curlResult)
 637         {
 638                 if (!$curlResult->isSuccess()) {
 639                         return [];
 640                 }
 641
 642                 $nodeinfo = json_decode($curlResult->getBody(), true);
 643
 644                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
 645                         return [];
 646                 }
 647
 648                 $nodeinfo1_url = '';
 649                 $nodeinfo2_url = '';
 650
 651                 foreach ($nodeinfo['links'] as $link) {
 652                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
 653                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
 654                                 continue;
 655                         }
 656                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
 657                                 $nodeinfo1_url = $link['href'];
 658                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
 659                                 $nodeinfo2_url = $link['href'];
 660                         }
 661                 }
 662
 663                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
 664                         return [];
 665                 }
 666
 667                 $server = [];
 668
 669                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 670                 if (!empty($nodeinfo2_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo2_url, PHP_URL_HOST))) {
 671                         $server = self::parseNodeinfo2($nodeinfo2_url);
 672                 }
 673
 674                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
 675                 if (empty($server) && !empty($nodeinfo1_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo1_url, PHP_URL_HOST))) {
 676                         $server = self::parseNodeinfo1($nodeinfo1_url);
 677                 }
 678
 679                 return $server;
 680         }
 681
 682         /**
 683          * Parses Nodeinfo 1
 684          *
 685          * @param string $nodeinfo_url address of the nodeinfo path
 686          * @return array Server data
 687          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 688          */
 689         private static function parseNodeinfo1(string $nodeinfo_url)
 690         {
 691                 $curlResult = Network::curl($nodeinfo_url);
 692                 if (!$curlResult->isSuccess()) {
 693                         return [];
 694                 }
 695
 696                 $nodeinfo = json_decode($curlResult->getBody(), true);
 697
 698                 if (!is_array($nodeinfo)) {
 699                         return [];
 700                 }
 701
 702                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
 703                         'register_policy' => Register::CLOSED];
 704
 705                 if (!empty($nodeinfo['openRegistrations'])) {
 706                         $server['register_policy'] = Register::OPEN;
 707                 }
 708
 709                 if (is_array($nodeinfo['software'])) {
 710                         if (!empty($nodeinfo['software']['name'])) {
 711                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 712                         }
 713
 714                         if (!empty($nodeinfo['software']['version'])) {
 715                                 $server['version'] = $nodeinfo['software']['version'];
 716                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 717                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 718                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 719                         }
 720                 }
 721
 722                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 723                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 724                 }
 725
 726                 if (!empty($nodeinfo['usage']['users']['total'])) {
 727                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 728                 }
 729
 730                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
 731                         $protocols = [];
 732                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
 733                                 $protocols[$protocol] = true;
 734                         }
 735
 736                         if (!empty($protocols['friendica'])) {
 737                                 $server['network'] = Protocol::DFRN;
 738                         } elseif (!empty($protocols['activitypub'])) {
 739                                 $server['network'] = Protocol::ACTIVITYPUB;
 740                         } elseif (!empty($protocols['diaspora'])) {
 741                                 $server['network'] = Protocol::DIASPORA;
 742                         } elseif (!empty($protocols['ostatus'])) {
 743                                 $server['network'] = Protocol::OSTATUS;
 744                         } elseif (!empty($protocols['gnusocial'])) {
 745                                 $server['network'] = Protocol::OSTATUS;
 746                         } elseif (!empty($protocols['zot'])) {
 747                                 $server['network'] = Protocol::ZOT;
 748                         }
 749                 }
 750
 751                 if (empty($server)) {
 752                         return [];
 753                 }
 754
 755                 return $server;
 756         }
 757
 758         /**
 759          * Parses Nodeinfo 2
 760          *
 761          * @param string $nodeinfo_url address of the nodeinfo path
 762          * @return array Server data
 763          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
 764          */
 765         private static function parseNodeinfo2(string $nodeinfo_url)
 766         {
 767                 $curlResult = Network::curl($nodeinfo_url);
 768                 if (!$curlResult->isSuccess()) {
 769                         return [];
 770                 }
 771
 772                 $nodeinfo = json_decode($curlResult->getBody(), true);
 773
 774                 if (!is_array($nodeinfo)) {
 775                         return [];
 776                 }
 777
 778                 $server = ['detection-method' => self::DETECT_NODEINFO_2,
 779                         'register_policy' => Register::CLOSED];
 780
 781                 if (!empty($nodeinfo['openRegistrations'])) {
 782                         $server['register_policy'] = Register::OPEN;
 783                 }
 784
 785                 if (is_array($nodeinfo['software'])) {
 786                         if (!empty($nodeinfo['software']['name'])) {
 787                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
 788                         }
 789
 790                         if (!empty($nodeinfo['software']['version'])) {
 791                                 $server['version'] = $nodeinfo['software']['version'];
 792                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
 793                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
 794                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
 795                         }
 796                 }
 797
 798                 if (!empty($nodeinfo['metadata']['nodeName'])) {
 799                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
 800                 }
 801
 802                 if (!empty($nodeinfo['usage']['users']['total'])) {
 803                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
 804                 }
 805
 806                 if (!empty($nodeinfo['protocols'])) {
 807                         $protocols = [];
 808                         foreach ($nodeinfo['protocols'] as $protocol) {
 809                                 $protocols[$protocol] = true;
 810                         }
 811
 812                         if (!empty($protocols['dfrn'])) {
 813                                 $server['network'] = Protocol::DFRN;
 814                         } elseif (!empty($protocols['activitypub'])) {
 815                                 $server['network'] = Protocol::ACTIVITYPUB;
 816                         } elseif (!empty($protocols['diaspora'])) {
 817                                 $server['network'] = Protocol::DIASPORA;
 818                         } elseif (!empty($protocols['ostatus'])) {
 819                                 $server['network'] = Protocol::OSTATUS;
 820                         } elseif (!empty($protocols['gnusocial'])) {
 821                                 $server['network'] = Protocol::OSTATUS;
 822                         } elseif (!empty($protocols['zot'])) {
 823                                 $server['network'] = Protocol::ZOT;
 824                         }
 825                 }
 826
 827                 if (empty($server)) {
 828                         return [];
 829                 }
 830
 831                 return $server;
 832         }
 833
 834         /**
 835          * Fetch server information from a 'siteinfo.json' file on the given server
 836          *
 837          * @param string $url        URL of the given server
 838          * @param array  $serverdata array with server data
 839          *
 840          * @return array server data
 841          */
 842         private static function fetchSiteinfo(string $url, array $serverdata)
 843         {
 844                 $curlResult = Network::curl($url . '/siteinfo.json');
 845                 if (!$curlResult->isSuccess()) {
 846                         return $serverdata;
 847                 }
 848
 849                 $data = json_decode($curlResult->getBody(), true);
 850                 if (empty($data)) {
 851                         return $serverdata;
 852                 }
 853
 854                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
 855                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
 856                 }
 857
 858                 if (!empty($data['url'])) {
 859                         $serverdata['platform'] = strtolower($data['platform']);
 860                         $serverdata['version'] = $data['version'];
 861                 }
 862
 863                 if (!empty($data['plugins'])) {
 864                         if (in_array('pubcrawl', $data['plugins'])) {
 865                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
 866                         } elseif (in_array('diaspora', $data['plugins'])) {
 867                                 $serverdata['network'] = Protocol::DIASPORA;
 868                         } elseif (in_array('gnusoc', $data['plugins'])) {
 869                                 $serverdata['network'] = Protocol::OSTATUS;
 870                         } else {
 871                                 $serverdata['network'] = Protocol::ZOT;
 872                         }
 873                 }
 874
 875                 if (!empty($data['site_name'])) {
 876                         $serverdata['site_name'] = $data['site_name'];
 877                 }
 878
 879                 if (!empty($data['channels_total'])) {
 880                         $serverdata['registered-users'] = $data['channels_total'];
 881                 }
 882
 883                 if (!empty($data['register_policy'])) {
 884                         switch ($data['register_policy']) {
 885                                 case 'REGISTER_OPEN':
 886                                         $serverdata['register_policy'] = Register::OPEN;
 887                                         break;
 888
 889                                 case 'REGISTER_APPROVE':
 890                                         $serverdata['register_policy'] = Register::APPROVE;
 891                                         break;
 892
 893                                 case 'REGISTER_CLOSED':
 894                                 default:
 895                                         $serverdata['register_policy'] = Register::CLOSED;
 896                                         break;
 897                         }
 898                 }
 899
 900                 return $serverdata;
 901         }
 902
 903         /**
 904          * Checks if the server contains a valid host meta file
 905          *
 906          * @param string $url URL of the given server
 907          *
 908          * @return boolean 'true' if the server seems to be vital
 909          */
 910         private static function validHostMeta(string $url)
 911         {
 912                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
 913                 $curlResult = Network::curl($url . '/.well-known/host-meta', false, ['timeout' => $xrd_timeout]);
 914                 if (!$curlResult->isSuccess()) {
 915                         return false;
 916                 }
 917
 918                 $xrd = XML::parseString($curlResult->getBody());
 919                 if (!is_object($xrd)) {
 920                         return false;
 921                 }
 922
 923                 $elements = XML::elementToArray($xrd);
 924                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
 925                         return false;
 926                 }
 927
 928                 $valid = false;
 929                 foreach ($elements['xrd']['link'] as $link) {
 930                         // When there is more than a single "link" element, the array looks slightly different
 931                         if (!empty($link['@attributes'])) {
 932                                 $link = $link['@attributes'];
 933                         }
 934
 935                         if (empty($link['rel']) || empty($link['template'])) {
 936                                 continue;
 937                         }
 938
 939                         if ($link['rel'] == 'lrdd') {
 940                                 // When the webfinger host is the same like the system host, it should be ok.
 941                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
 942                         }
 943                 }
 944
 945                 return $valid;
 946         }
 947
 948         /**
 949          * Detect the network of the given server via their known contacts
 950          *
 951          * @param string $url        URL of the given server
 952          * @param array  $serverdata array with server data
 953          *
 954          * @return array server data
 955          */
 956         private static function detectNetworkViaContacts(string $url, array $serverdata)
 957         {
 958                 $contacts = [];
 959
 960                 $gcontacts = DBA::select('gcontact', ['url', 'nurl'], ['server_url' => [$url, $serverdata['nurl']]]);
 961                 while ($gcontact = DBA::fetch($gcontacts)) {
 962                         $contacts[$gcontact['nurl']] = $gcontact['url'];
 963                 }
 964                 DBA::close($gcontacts);
 965
 966                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $serverdata['nurl']]]);
 967                 while ($apcontact = DBA::fetch($apcontacts)) {
 968                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
 969                 }
 970                 DBA::close($apcontacts);
 971
 972                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $serverdata['nurl']]]);
 973                 while ($pcontact = DBA::fetch($pcontacts)) {
 974                         $contacts[$pcontact['nurl']] = $pcontact['url'];
 975                 }
 976                 DBA::close($pcontacts);
 977
 978                 if (empty($contacts)) {
 979                         return $serverdata;
 980                 }
 981
 982                 foreach ($contacts as $contact) {
 983                         $probed = Probe::uri($contact);
 984                         if (in_array($probed['network'], Protocol::FEDERATED)) {
 985                                 $serverdata['network'] = $probed['network'];
 986                                 break;
 987                         }
 988                 }
 989
 990                 $serverdata['registered-users'] = max($serverdata['registered-users'], count($contacts));
 991
 992                 return $serverdata;
 993         }
 994
 995         /**
 996          * Checks if the given server does have a '/poco' endpoint.
 997          * This is used for the 'PortableContact' functionality,
 998          * which is used by both Friendica and Hubzilla.
 999          *
1000          * @param string $url        URL of the given server
1001          * @param array  $serverdata array with server data
1002          *
1003          * @return array server data
1004          */
1005         private static function checkPoCo(string $url, array $serverdata)
1006         {
1007                 $serverdata['poco'] = '';
1008
1009                 $curlResult = Network::curl($url. '/poco');
1010                 if (!$curlResult->isSuccess()) {
1011                         return $serverdata;
1012                 }
1013
1014                 $data = json_decode($curlResult->getBody(), true);
1015                 if (empty($data)) {
1016                         return $serverdata;
1017                 }
1018
1019                 if (!empty($data['totalResults'])) {
1020                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1021                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers);
1022                         $serverdata['directory-type'] = self::DT_POCO;
1023                         $serverdata['poco'] = $url . '/poco';
1024                 }
1025
1026                 return $serverdata;
1027         }
1028
1029         /**
1030          * Checks if the given server does have a Mastodon style directory endpoint.
1031          *
1032          * @param string $url        URL of the given server
1033          * @param array  $serverdata array with server data
1034          *
1035          * @return array server data
1036          */
1037         public static function checkMastodonDirectory(string $url, array $serverdata)
1038         {
1039                 $curlResult = Network::curl($url . '/api/v1/directory?limit=1');
1040                 if (!$curlResult->isSuccess()) {
1041                         return $serverdata;
1042                 }
1043
1044                 $data = json_decode($curlResult->getBody(), true);
1045                 if (empty($data)) {
1046                         return $serverdata;
1047                 }
1048
1049                 if (count($data) == 1) {
1050                         $serverdata['directory-type'] = self::DT_MASTODON;
1051                 }
1052
1053                 return $serverdata;
1054         }
1055
1056         /**
1057          * Detects the version number of a given server when it was a NextCloud installation
1058          *
1059          * @param string $url        URL of the given server
1060          * @param array  $serverdata array with server data
1061          *
1062          * @return array server data
1063          */
1064         private static function detectNextcloud(string $url, array $serverdata)
1065         {
1066                 $curlResult = Network::curl($url . '/status.php');
1067                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1068                         return $serverdata;
1069                 }
1070
1071                 $data = json_decode($curlResult->getBody(), true);
1072                 if (empty($data)) {
1073                         return $serverdata;
1074                 }
1075
1076                 if (!empty($data['version'])) {
1077                         $serverdata['platform'] = 'nextcloud';
1078                         $serverdata['version'] = $data['version'];
1079                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1080
1081                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1082                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1083                         }
1084                 }
1085
1086                 return $serverdata;
1087         }
1088
1089         /**
1090          * Detects data from a given server url if it was a mastodon alike system
1091          *
1092          * @param string $url        URL of the given server
1093          * @param array  $serverdata array with server data
1094          *
1095          * @return array server data
1096          */
1097         private static function detectMastodonAlikes(string $url, array $serverdata)
1098         {
1099                 $curlResult = Network::curl($url . '/api/v1/instance');
1100                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1101                         return $serverdata;
1102                 }
1103
1104                 $data = json_decode($curlResult->getBody(), true);
1105                 if (empty($data)) {
1106                         return $serverdata;
1107                 }
1108
1109                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1110                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1111                 }
1112
1113                 if (!empty($data['version'])) {
1114                         $serverdata['platform'] = 'mastodon';
1115                         $serverdata['version'] = $data['version'] ?? '';
1116                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1117                 }
1118
1119                 if (!empty($data['title'])) {
1120                         $serverdata['site_name'] = $data['title'];
1121                 }
1122
1123                 if (!empty($data['title']) && empty($serverdata['platform']) && empty($serverdata['network'])) {
1124                         $serverdata['platform'] = 'mastodon';
1125                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1126                 }
1127
1128                 if (!empty($data['description'])) {
1129                         $serverdata['info'] = trim($data['description']);
1130                 }
1131
1132                 if (!empty($data['stats']['user_count'])) {
1133                         $serverdata['registered-users'] = $data['stats']['user_count'];
1134                 }
1135
1136                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1137                         $serverdata['platform'] = strtolower($matches[1]);
1138                         $serverdata['version'] = $matches[2];
1139                 }
1140
1141                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1142                         $serverdata['platform'] = 'pleroma';
1143                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1144                 }
1145
1146                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1147                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1148                         $serverdata['platform'] = 'pleroma';
1149                 }
1150
1151                 return $serverdata;
1152         }
1153
1154         /**
1155          * Detects data from typical Hubzilla endpoints
1156          *
1157          * @param string $url        URL of the given server
1158          * @param array  $serverdata array with server data
1159          *
1160          * @return array server data
1161          */
1162         private static function detectHubzilla(string $url, array $serverdata)
1163         {
1164                 $curlResult = Network::curl($url . '/api/statusnet/config.json');
1165                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1166                         return $serverdata;
1167                 }
1168
1169                 $data = json_decode($curlResult->getBody(), true);
1170                 if (empty($data) || empty($data['site'])) {
1171                         return $serverdata;
1172                 }
1173
1174                 if (!empty($data['site']['name'])) {
1175                         $serverdata['site_name'] = $data['site']['name'];
1176                 }
1177
1178                 if (!empty($data['site']['platform'])) {
1179                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1180                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1181                         $serverdata['network'] = Protocol::ZOT;
1182                 }
1183
1184                 if (!empty($data['site']['hubzilla'])) {
1185                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1186                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1187                         $serverdata['network'] = Protocol::ZOT;
1188                 }
1189
1190                 if (!empty($data['site']['redmatrix'])) {
1191                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1192                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1193                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1194                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1195                         }
1196
1197                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1198                         $serverdata['network'] = Protocol::ZOT;
1199                 }
1200
1201                 $private = false;
1202                 $inviteonly = false;
1203                 $closed = false;
1204
1205                 if (!empty($data['site']['closed'])) {
1206                         $closed = self::toBoolean($data['site']['closed']);
1207                 }
1208
1209                 if (!empty($data['site']['private'])) {
1210                         $private = self::toBoolean($data['site']['private']);
1211                 }
1212
1213                 if (!empty($data['site']['inviteonly'])) {
1214                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1215                 }
1216
1217                 if (!$closed && !$private and $inviteonly) {
1218                         $serverdata['register_policy'] = Register::APPROVE;
1219                 } elseif (!$closed && !$private) {
1220                         $serverdata['register_policy'] = Register::OPEN;
1221                 } else {
1222                         $serverdata['register_policy'] = Register::CLOSED;
1223                 }
1224
1225                 if (!empty($serverdata['network']) && in_array($serverdata['detection-method'],
1226                         [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1227                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1228                 }
1229
1230                 return $serverdata;
1231         }
1232
1233         /**
1234          * Converts input value to a boolean value
1235          *
1236          * @param string|integer $val
1237          *
1238          * @return boolean
1239          */
1240         private static function toBoolean($val)
1241         {
1242                 if (($val == 'true') || ($val == 1)) {
1243                         return true;
1244                 } elseif (($val == 'false') || ($val == 0)) {
1245                         return false;
1246                 }
1247
1248                 return $val;
1249         }
1250
1251         /**
1252          * Detect if the URL belongs to a GNU Social server
1253          *
1254          * @param string $url        URL of the given server
1255          * @param array  $serverdata array with server data
1256          *
1257          * @return array server data
1258          */
1259         private static function detectGNUSocial(string $url, array $serverdata)
1260         {
1261                 // Test for GNU Social
1262                 $curlResult = Network::curl($url . '/api/gnusocial/version.json');
1263                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1264                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1265                         $serverdata['platform'] = 'gnusocial';
1266                         // Remove junk that some GNU Social servers return
1267                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1268                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1269                         $serverdata['version'] = trim($serverdata['version'], '"');
1270                         $serverdata['network'] = Protocol::OSTATUS;
1271
1272                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1273                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1274                         }
1275
1276                         return $serverdata;
1277                 }
1278
1279                 // Test for Statusnet
1280                 $curlResult = Network::curl($url . '/api/statusnet/version.json');
1281                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1282                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1283
1284                         // Remove junk that some GNU Social servers return
1285                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1286                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1287                         $serverdata['version'] = trim($serverdata['version'], '"');
1288
1289                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1290                                 $serverdata['platform'] = 'pleroma';
1291                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1292                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1293                         } else {
1294                                 $serverdata['platform'] = 'statusnet';
1295                                 $serverdata['network'] = Protocol::OSTATUS;
1296                         }
1297
1298                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1299                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1300                         }
1301                 }
1302
1303                 return $serverdata;
1304         }
1305
1306         /**
1307          * Detect if the URL belongs to a Friendica server
1308          *
1309          * @param string $url        URL of the given server
1310          * @param array  $serverdata array with server data
1311          *
1312          * @return array server data
1313          */
1314         private static function detectFriendica(string $url, array $serverdata)
1315         {
1316                 $curlResult = Network::curl($url . '/friendica/json');
1317                 if (!$curlResult->isSuccess()) {
1318                         $curlResult = Network::curl($url . '/friendika/json');
1319                         $friendika = true;
1320                         $platform = 'Friendika';
1321                 } else {
1322                         $friendika = false;
1323                         $platform = 'Friendica';
1324                 }
1325
1326                 if (!$curlResult->isSuccess()) {
1327                         return $serverdata;
1328                 }
1329
1330                 $data = json_decode($curlResult->getBody(), true);
1331                 if (empty($data) || empty($data['version'])) {
1332                         return $serverdata;
1333                 }
1334
1335                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1336                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1337                 }
1338
1339                 $serverdata['network'] = Protocol::DFRN;
1340                 $serverdata['version'] = $data['version'];
1341
1342                 if (!empty($data['no_scrape_url'])) {
1343                         $serverdata['noscrape'] = $data['no_scrape_url'];
1344                 }
1345
1346                 if (!empty($data['site_name'])) {
1347                         $serverdata['site_name'] = $data['site_name'];
1348                 }
1349
1350                 if (!empty($data['info'])) {
1351                         $serverdata['info'] = trim($data['info']);
1352                 }
1353
1354                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1355                 switch ($register_policy) {
1356                         case 'REGISTER_OPEN':
1357                                 $serverdata['register_policy'] = Register::OPEN;
1358                                 break;
1359
1360                         case 'REGISTER_APPROVE':
1361                                 $serverdata['register_policy'] = Register::APPROVE;
1362                                 break;
1363
1364                         case 'REGISTER_CLOSED':
1365                         case 'REGISTER_INVITATION':
1366                                 $serverdata['register_policy'] = Register::CLOSED;
1367                                 break;
1368                         default:
1369                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1370                                 $serverdata['register_policy'] = Register::CLOSED;
1371                                 break;
1372                 }
1373
1374                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1375
1376                 return $serverdata;
1377         }
1378
1379         /**
1380          * Analyses the landing page of a given server for hints about type and system of that server
1381          *
1382          * @param object $curlResult result of curl execution
1383          * @param array  $serverdata array with server data
1384          * @param string $url        Server URL
1385          *
1386          * @return array server data
1387          */
1388         private static function analyseRootBody($curlResult, array $serverdata, string $url)
1389         {
1390                 $doc = new DOMDocument();
1391                 @$doc->loadHTML($curlResult->getBody());
1392                 $xpath = new DOMXPath($doc);
1393
1394                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1395                 if (!empty($title)) {
1396                         $serverdata['site_name'] = $title;
1397                 }
1398
1399                 $list = $xpath->query('//meta[@name]');
1400
1401                 foreach ($list as $node) {
1402                         $attr = [];
1403                         if ($node->attributes->length) {
1404                                 foreach ($node->attributes as $attribute) {
1405                                         $value = trim($attribute->value);
1406                                         if (empty($value)) {
1407                                                 continue;
1408                                         }
1409
1410                                         $attr[$attribute->name] = $value;
1411                                 }
1412
1413                                 if (empty($attr['name']) || empty($attr['content'])) {
1414                                         continue;
1415                                 }
1416                         }
1417
1418                         if ($attr['name'] == 'description') {
1419                                 $serverdata['info'] = $attr['content'];
1420                         }
1421
1422                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1423                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad'])) {
1424                                 $serverdata['platform'] = strtolower($attr['content']);
1425                                 if (in_array($attr['content'], ['Misskey', 'Write.as'])) {
1426                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1427                                 }
1428                         }
1429                         if (($attr['name'] == 'generator') && (empty($serverdata['platform']) || (substr(strtolower($attr['content']), 0, 9) == 'wordpress'))) {
1430                                 $serverdata['platform'] = strtolower($attr['content']);
1431                                 $version_part = explode(' ', $attr['content']);
1432
1433                                 if (count($version_part) == 2) {
1434                                         if (in_array($version_part[0], ['WordPress'])) {
1435                                                 $serverdata['platform'] = strtolower($version_part[0]);
1436                                                 $serverdata['version'] = $version_part[1];
1437
1438                                                 // We still do need a reliable test if some AP plugin is activated
1439                                                 if (DBA::exists('apcontact', ['baseurl' => $url])) {
1440                                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1441                                                 } else {
1442                                                         $serverdata['network'] = Protocol::FEED;
1443                                                 }
1444
1445                                                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1446                                                         $serverdata['detection-method'] = self::DETECT_BODY;
1447                                                 }
1448                                         }
1449                                         if (in_array($version_part[0], ['Friendika', 'Friendica'])) {
1450                                                 $serverdata['platform'] = strtolower($version_part[0]);
1451                                                 $serverdata['version'] = $version_part[1];
1452                                                 $serverdata['network'] = Protocol::DFRN;
1453                                         }
1454                                 }
1455                         }
1456                 }
1457
1458                 $list = $xpath->query('//meta[@property]');
1459
1460                 foreach ($list as $node) {
1461                         $attr = [];
1462                         if ($node->attributes->length) {
1463                                 foreach ($node->attributes as $attribute) {
1464                                         $value = trim($attribute->value);
1465                                         if (empty($value)) {
1466                                                 continue;
1467                                         }
1468
1469                                         $attr[$attribute->name] = $value;
1470                                 }
1471
1472                                 if (empty($attr['property']) || empty($attr['content'])) {
1473                                         continue;
1474                                 }
1475                         }
1476
1477                         if ($attr['property'] == 'og:site_name') {
1478                                 $serverdata['site_name'] = $attr['content'];
1479                         }
1480
1481                         if ($attr['property'] == 'og:description') {
1482                                 $serverdata['info'] = $attr['content'];
1483                         }
1484
1485                         if ($attr['property'] == 'og:platform') {
1486                                 $serverdata['platform'] = strtolower($attr['content']);
1487
1488                                 if (in_array($attr['content'], ['PeerTube'])) {
1489                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1490                                 }
1491                         }
1492
1493                         if ($attr['property'] == 'generator') {
1494                                 $serverdata['platform'] = strtolower($attr['content']);
1495
1496                                 if (in_array($attr['content'], ['hubzilla'])) {
1497                                         // We later check which compatible protocol modules are loaded.
1498                                         $serverdata['network'] = Protocol::ZOT;
1499                                 }
1500                         }
1501                 }
1502
1503                 if (!empty($serverdata['network']) && ($serverdata['detection-method'] == self::DETECT_MANUAL)) {
1504                         $serverdata['detection-method'] = self::DETECT_BODY;
1505                 }
1506
1507                 return $serverdata;
1508         }
1509
1510         /**
1511          * Analyses the header data of a given server for hints about type and system of that server
1512          *
1513          * @param object $curlResult result of curl execution
1514          * @param array  $serverdata array with server data
1515          *
1516          * @return array server data
1517          */
1518         private static function analyseRootHeader($curlResult, array $serverdata)
1519         {
1520                 if ($curlResult->getHeader('server') == 'Mastodon') {
1521                         $serverdata['platform'] = 'mastodon';
1522                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1523                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
1524                         $serverdata['platform'] = 'diaspora';
1525                         $serverdata['network'] = Protocol::DIASPORA;
1526                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version');
1527                 } elseif ($curlResult->inHeader('x-friendica-version')) {
1528                         $serverdata['platform'] = 'friendica';
1529                         $serverdata['network'] = Protocol::DFRN;
1530                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version');
1531                 } else {
1532                         return $serverdata;
1533                 }
1534
1535                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1536                         $serverdata['detection-method'] = self::DETECT_HEADER;
1537                 }
1538
1539                 return $serverdata;
1540         }
1541
1542         /**
1543          * Test if the body contains valid content
1544          *
1545          * @param string $body
1546          * @return boolean
1547          */
1548         private static function invalidBody(string $body)
1549         {
1550                 // Currently we only test for a HTML element.
1551                 // Possibly we enhance this in the future.
1552                 return !strpos($body, '>');
1553         }
1554
1555         /**
1556          * Update the user directory of a given gserver record
1557          *
1558          * @param array $gserver gserver record
1559          */
1560         public static function updateDirectory(array $gserver)
1561         {
1562                 /// @todo Add Mastodon API directory
1563
1564                 if (!empty($gserver['poco'])) {
1565                         PortableContact::discoverSingleServer($gserver['id']);
1566                 }
1567         }
1568
1569         /**
1570          * Update GServer entries
1571          */
1572         public static function discover()
1573         {
1574                 // Update the server list
1575                 self::discoverFederation();
1576
1577                 $no_of_queries = 5;
1578
1579                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
1580
1581                 if ($requery_days == 0) {
1582                         $requery_days = 7;
1583                 }
1584
1585                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
1586
1587                 $gservers = DBA::p("SELECT `id`, `url`, `nurl`, `network`, `poco`
1588                         FROM `gserver`
1589                         WHERE `last_contact` >= `last_failure`
1590                         AND `poco` != ''
1591                         AND `last_poco_query` < ?
1592                         ORDER BY RAND()", $last_update
1593                 );
1594
1595                 while ($gserver = DBA::fetch($gservers)) {
1596                         if (!GServer::check($gserver['url'], $gserver['network'])) {
1597                                 // The server is not reachable? Okay, then we will try it later
1598                                 $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1599                                 DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1600                                 continue;
1601                         }
1602
1603                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
1604                         Worker::add(PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
1605
1606                         if (--$no_of_queries == 0) {
1607                                 break;
1608                         }
1609                 }
1610
1611                 DBA::close($gservers);
1612         }
1613
1614         /**
1615          * Discover federated servers
1616          */
1617         private static function discoverFederation()
1618         {
1619                 $last = DI::config()->get('poco', 'last_federation_discovery');
1620
1621                 if ($last) {
1622                         $next = $last + (24 * 60 * 60);
1623
1624                         if ($next > time()) {
1625                                 return;
1626                         }
1627                 }
1628
1629                 // Discover federated servers
1630                 $curlResult = Network::fetchUrl("http://the-federation.info/pods.json");
1631
1632                 if (!empty($curlResult)) {
1633                         $servers = json_decode($curlResult, true);
1634
1635                         if (!empty($servers['pods'])) {
1636                                 foreach ($servers['pods'] as $server) {
1637                                         // Using "only_nodeinfo" since servers that are listed on that page should always have it.
1638                                         Worker::add(PRIORITY_LOW, 'UpdateGServer', 'https://' . $server['host'], true);
1639                                 }
1640                         }
1641                 }
1642
1643                 // Disvover Mastodon servers
1644                 $accesstoken = DI::config()->get('system', 'instances_social_key');
1645
1646                 if (!empty($accesstoken)) {
1647                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
1648                         $header = ['Authorization: Bearer '.$accesstoken];
1649                         $curlResult = Network::curl($api, false, ['headers' => $header]);
1650                         if ($curlResult->isSuccess()) {
1651                                 $servers = json_decode($curlResult->getBody(), true);
1652
1653                                 foreach ($servers['instances'] as $server) {
1654                                         $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
1655                                         Worker::add(PRIORITY_LOW, 'UpdateGServer', $url);
1656                                 }
1657                         }
1658                 }
1659
1660                 DI::config()->set('poco', 'last_federation_discovery', time());
1661         }
1662 }