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