]> git.mxchange.org Git - friendica.git/blob - src/Model/GServer.php
Merge pull request #9405 from nupplaphil/task/psr7_remove_binary
[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\DBA;
31 use Friendica\DI;
32 use Friendica\Module\Register;
33 use Friendica\Network\CurlResult;
34 use Friendica\Protocol\Diaspora;
35 use Friendica\Util\DateTimeFormat;
36 use Friendica\Util\Network;
37 use Friendica\Util\Strings;
38 use Friendica\Util\XML;
39
40 /**
41  * This class handles GServer related functions
42  */
43 class GServer
44 {
45         // Directory types
46         const DT_NONE = 0;
47         const DT_POCO = 1;
48         const DT_MASTODON = 2;
49
50         // Methods to detect server types
51
52         // Non endpoint specific methods
53         const DETECT_MANUAL = 0;
54         const DETECT_HEADER = 1;
55         const DETECT_BODY = 2;
56
57         // Implementation specific endpoints
58         const DETECT_FRIENDIKA = 10;
59         const DETECT_FRIENDICA = 11;
60         const DETECT_STATUSNET = 12;
61         const DETECT_GNUSOCIAL = 13;
62         const DETECT_CONFIG_JSON = 14; // Statusnet, GNU Social, Older Hubzilla/Redmatrix
63         const DETECT_SITEINFO_JSON = 15; // Newer Hubzilla
64         const DETECT_MASTODON_API = 16;
65         const DETECT_STATUS_PHP = 17; // Nextcloud
66
67         // Standardized endpoints
68         const DETECT_STATISTICS_JSON = 100;
69         const DETECT_NODEINFO_1 = 101;
70         const DETECT_NODEINFO_2 = 102;
71
72         /**
73          * Get the ID for the given server URL
74          *
75          * @param string $url
76          * @param boolean $no_check Don't check if the server hadn't been found
77          * @return int gserver id
78          */
79         public static function getID(string $url, bool $no_check = false)
80         {
81                 if (empty($url)) {
82                         return null;
83                 }
84
85                 $url = self::cleanURL($url);
86
87                 $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => Strings::normaliseLink($url)]);
88                 if (DBA::isResult($gserver)) {
89                         Logger::info('Got ID for URL', ['id' => $gserver['id'], 'url' => $url, 'callstack' => System::callstack(20)]);
90                         return $gserver['id'];
91                 }
92
93                 if ($no_check || !self::check($url)) {
94                         return null;
95                 }
96         
97                 return self::getID($url, true);
98         }
99
100         /**
101          * Checks if the given server is reachable
102          *
103          * @param string  $profile URL of the given profile
104          * @param string  $server  URL of the given server (If empty, taken from profile)
105          * @param string  $network Network value that is used, when detection failed
106          * @param boolean $force   Force an update.
107          *
108          * @return boolean 'true' if server seems vital
109          */
110         public static function reachable(string $profile, string $server = '', string $network = '', bool $force = false)
111         {
112                 if ($server == '') {
113                         $contact = Contact::getByURL($profile, null, ['baseurl']);
114                         if (!empty($contact['baseurl'])) {
115                                 $server = $contact['baseurl'];
116                         }
117                 }
118
119                 if ($server == '') {
120                         return true;
121                 }
122
123                 return self::check($server, $network, $force);
124         }
125
126         /**
127          * Decides if a server needs to be updated, based upon several date fields
128          *
129          * @param date $created      Creation date of that server entry
130          * @param date $updated      When had the server entry be updated
131          * @param date $last_failure Last failure when contacting that server
132          * @param date $last_contact Last time the server had been contacted
133          *
134          * @return boolean Does the server record needs an update?
135          */
136         public static function updateNeeded($created, $updated, $last_failure, $last_contact)
137         {
138                 $now = strtotime(DateTimeFormat::utcNow());
139
140                 if ($updated > $last_contact) {
141                         $contact_time = strtotime($updated);
142                 } else {
143                         $contact_time = strtotime($last_contact);
144                 }
145
146                 $failure_time = strtotime($last_failure);
147                 $created_time = strtotime($created);
148
149                 // If there is no "created" time then use the current time
150                 if ($created_time <= 0) {
151                         $created_time = $now;
152                 }
153
154                 // If the last contact was less than 24 hours then don't update
155                 if (($now - $contact_time) < (60 * 60 * 24)) {
156                         return false;
157                 }
158
159                 // If the last failure was less than 24 hours then don't update
160                 if (($now - $failure_time) < (60 * 60 * 24)) {
161                         return false;
162                 }
163
164                 // If the last contact was less than a week ago and the last failure is older than a week then don't update
165                 //if ((($now - $contact_time) < (60 * 60 * 24 * 7)) && ($contact_time > $failure_time))
166                 //      return false;
167
168                 // If the last contact time was more than a week ago and the contact was created more than a week ago, then only try once a week
169                 if ((($now - $contact_time) > (60 * 60 * 24 * 7)) && (($now - $created_time) > (60 * 60 * 24 * 7)) && (($now - $failure_time) < (60 * 60 * 24 * 7))) {
170                         return false;
171                 }
172
173                 // If the last contact time was more than a month ago and the contact was created more than a month ago, then only try once a month
174                 if ((($now - $contact_time) > (60 * 60 * 24 * 30)) && (($now - $created_time) > (60 * 60 * 24 * 30)) && (($now - $failure_time) < (60 * 60 * 24 * 30))) {
175                         return false;
176                 }
177
178                 return true;
179         }
180
181         /**
182          * Checks the state of the given server.
183          *
184          * @param string  $server_url    URL of the given server
185          * @param string  $network       Network value that is used, when detection failed
186          * @param boolean $force         Force an update.
187          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
188          *
189          * @return boolean 'true' if server seems vital
190          */
191         public static function check(string $server_url, string $network = '', bool $force = false, bool $only_nodeinfo = false)
192         {
193                 $server_url = self::cleanURL($server_url);
194
195                 if ($server_url == '') {
196                         return false;
197                 }
198
199                 $gserver = DBA::selectFirst('gserver', [], ['nurl' => Strings::normaliseLink($server_url)]);
200                 if (DBA::isResult($gserver)) {
201                         if ($gserver['created'] <= DBA::NULL_DATETIME) {
202                                 $fields = ['created' => DateTimeFormat::utcNow()];
203                                 $condition = ['nurl' => Strings::normaliseLink($server_url)];
204                                 DBA::update('gserver', $fields, $condition);
205                         }
206
207                         $last_contact = $gserver['last_contact'];
208                         $last_failure = $gserver['last_failure'];
209
210                         // See discussion under https://forum.friendi.ca/display/0b6b25a8135aabc37a5a0f5684081633
211                         // It can happen that a zero date is in the database, but storing it again is forbidden.
212                         if ($last_contact < DBA::NULL_DATETIME) {
213                                 $last_contact = DBA::NULL_DATETIME;
214                         }
215
216                         if ($last_failure < DBA::NULL_DATETIME) {
217                                 $last_failure = DBA::NULL_DATETIME;
218                         }
219
220                         if (!$force && !self::updateNeeded($gserver['created'], '', $last_failure, $last_contact)) {
221                                 Logger::info('No update needed', ['server' => $server_url]);
222                                 return ($last_contact >= $last_failure);
223                         }
224                         Logger::info('Server is outdated. Start discovery.', ['Server' => $server_url, 'Force' => $force, 'Created' => $gserver['created'], 'Failure' => $last_failure, 'Contact' => $last_contact]);
225                 } else {
226                         Logger::info('Server is unknown. Start discovery.', ['Server' => $server_url]);
227                 }
228
229                 return self::detect($server_url, $network, $only_nodeinfo);
230         }
231
232         /**
233          * Set failed server status
234          *
235          * @param string $url
236          */
237         private static function setFailure(string $url)
238         {
239                 if (DBA::exists('gserver', ['nurl' => Strings::normaliseLink($url)])) {
240                         DBA::update('gserver', ['failed' => true, 'last_failure' => DateTimeFormat::utcNow(), 'detection-method' => null],
241                         ['nurl' => Strings::normaliseLink($url)]);
242                         Logger::info('Set failed status for existing server', ['url' => $url]);
243                         return;
244                 }
245                 DBA::insert('gserver', ['url' => $url, 'nurl' => Strings::normaliseLink($url),
246                         'network' => Protocol::PHANTOM, 'created' => DateTimeFormat::utcNow(),
247                         'failed' => true, 'last_failure' => DateTimeFormat::utcNow()]);
248                 Logger::info('Set failed status for new server', ['url' => $url]);
249         }
250
251         /**
252          * Remove unwanted content from the given URL
253          *
254          * @param string $url
255          * @return string cleaned URL
256          */
257         public static function cleanURL(string $url)
258         {
259                 $url = trim($url, '/');
260                 $url = str_replace('/index.php', '', $url);
261
262                 $urlparts = parse_url($url);
263                 unset($urlparts['user']);
264                 unset($urlparts['pass']);
265                 unset($urlparts['query']);
266                 unset($urlparts['fragment']);
267                 return Network::unparseURL($urlparts);
268         }
269
270         /**
271          * Return the base URL
272          *
273          * @param string $url
274          * @return string base URL
275          */
276         private static function getBaseURL(string $url)
277         {
278                 $urlparts = parse_url(self::cleanURL($url));
279                 unset($urlparts['path']);
280                 return Network::unparseURL($urlparts);
281         }
282
283         /**
284          * Detect server data (type, protocol, version number, ...)
285          * The detected data is then updated or inserted in the gserver table.
286          *
287          * @param string  $url           URL of the given server
288          * @param string  $network       Network value that is used, when detection failed
289          * @param boolean $only_nodeinfo Only use nodeinfo for server detection
290          *
291          * @return boolean 'true' if server could be detected
292          */
293         public static function detect(string $url, string $network = '', bool $only_nodeinfo = false)
294         {
295                 Logger::info('Detect server type', ['server' => $url]);
296                 $serverdata = ['detection-method' => self::DETECT_MANUAL];
297
298                 $original_url = $url;
299
300                 // Remove URL content that is not supposed to exist for a server url
301                 $url = self::cleanURL($url);
302
303                 // Get base URL
304                 $baseurl = self::getBaseURL($url);
305
306                 // If the URL missmatches, then we mark the old entry as failure
307                 if ($url != $original_url) {
308                         DBA::update('gserver', ['failed' => true, 'last_failure' => DateTimeFormat::utcNow()],
309                                 ['nurl' => Strings::normaliseLink($original_url)]);
310                 }
311
312                 // When a nodeinfo is present, we don't need to dig further
313                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
314                 $curlResult = DI::httpRequest()->get($url . '/.well-known/nodeinfo', ['timeout' => $xrd_timeout]);
315                 if ($curlResult->isTimeout()) {
316                         self::setFailure($url);
317                         return false;
318                 }
319
320                 $nodeinfo = self::fetchNodeinfo($url, $curlResult);
321                 if ($only_nodeinfo && empty($nodeinfo)) {
322                         Logger::info('Invalid nodeinfo in nodeinfo-mode, server is marked as failure', ['url' => $url]);
323                         self::setFailure($url);
324                         return false;
325                 }
326
327                 // When nodeinfo isn't present, we use the older 'statistics.json' endpoint
328                 if (empty($nodeinfo)) {
329                         $nodeinfo = self::fetchStatistics($url);
330                 }
331
332                 // If that didn't work out well, we use some protocol specific endpoints
333                 // For Friendica and Zot based networks we have to dive deeper to reveal more details
334                 if (empty($nodeinfo['network']) || in_array($nodeinfo['network'], [Protocol::DFRN, Protocol::ZOT])) {
335                         if (!empty($nodeinfo['detection-method'])) {
336                                 $serverdata['detection-method'] = $nodeinfo['detection-method'];
337                         }
338
339                         // Fetch the landing page, possibly it reveals some data
340                         if (empty($nodeinfo['network'])) {
341                                 if ($baseurl == $url) {
342                                         $basedata = $serverdata;
343                                 } else {
344                                         $basedata = ['detection-method' => self::DETECT_MANUAL];
345                                 }
346
347                                 $curlResult = DI::httpRequest()->get($baseurl, ['timeout' => $xrd_timeout]);
348                                 if ($curlResult->isSuccess()) {
349                                         $basedata = self::analyseRootHeader($curlResult, $basedata);
350                                         $basedata = self::analyseRootBody($curlResult, $basedata, $baseurl);
351                                 }
352
353                                 if (!$curlResult->isSuccess() || empty($curlResult->getBody()) || self::invalidBody($curlResult->getBody())) {
354                                         self::setFailure($url);
355                                         return false;
356                                 }
357
358                                 if ($baseurl == $url) {
359                                         $serverdata = $basedata;
360                                 } else {
361                                         // When the base path doesn't seem to contain a social network we try the complete path.
362                                         // Most detectable system have to be installed in the root directory.
363                                         // We checked the base to avoid false positives.
364                                         $curlResult = DI::httpRequest()->get($url, ['timeout' => $xrd_timeout]);
365                                         if ($curlResult->isSuccess()) {
366                                                 $urldata = self::analyseRootHeader($curlResult, $serverdata);
367                                                 $urldata = self::analyseRootBody($curlResult, $urldata, $url);
368
369                                                 $comparebase = $basedata;
370                                                 unset($comparebase['info']);
371                                                 unset($comparebase['site_name']);
372                                                 $compareurl = $urldata;
373                                                 unset($compareurl['info']);
374                                                 unset($compareurl['site_name']);
375
376                                                 // We assume that no one will install the identical system in the root and a subfolder
377                                                 if (!empty(array_diff($comparebase, $compareurl))) {
378                                                         $serverdata = $urldata;
379                                                 }
380                                         }
381                                 }
382                         }
383
384                         if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ACTIVITYPUB)) {
385                                 $serverdata = self::detectMastodonAlikes($url, $serverdata);
386                         }
387
388                         // All following checks are done for systems that always have got a "host-meta" endpoint.
389                         // With this check we don't have to waste time and ressources for dead systems.
390                         // Also this hopefully prevents us from receiving abuse messages.
391                         if (empty($serverdata['network']) && !self::validHostMeta($url)) {
392                                 self::setFailure($url);
393                                 return false;
394                         }
395
396                         if (empty($serverdata['network']) || in_array($serverdata['network'], [Protocol::DFRN, Protocol::ACTIVITYPUB])) {
397                                 $serverdata = self::detectFriendica($url, $serverdata);
398                         }
399
400                         // the 'siteinfo.json' is some specific endpoint of Hubzilla and Red
401                         if (empty($serverdata['network']) || ($serverdata['network'] == Protocol::ZOT)) {
402                                 $serverdata = self::fetchSiteinfo($url, $serverdata);
403                         }
404
405                         // The 'siteinfo.json' doesn't seem to be present on older Hubzilla installations
406                         if (empty($serverdata['network'])) {
407                                 $serverdata = self::detectHubzilla($url, $serverdata);
408                         }
409
410                         if (empty($serverdata['network'])) {
411                                 $serverdata = self::detectNextcloud($url, $serverdata);
412                         }
413
414                         if (empty($serverdata['network'])) {
415                                 $serverdata = self::detectGNUSocial($url, $serverdata);
416                         }
417
418                         $serverdata = array_merge($nodeinfo, $serverdata);
419                 } else {
420                         $serverdata = $nodeinfo;
421                 }
422
423                 // Detect the directory type
424                 $serverdata['directory-type'] = self::DT_NONE;
425                 $serverdata = self::checkPoCo($url, $serverdata);
426                 $serverdata = self::checkMastodonDirectory($url, $serverdata);
427
428                 // We can't detect the network type. Possibly it is some system that we don't know yet
429                 if (empty($serverdata['network'])) {
430                         $serverdata['network'] = Protocol::PHANTOM;
431                 }
432
433                 // When we hadn't been able to detect the network type, we use the hint from the parameter
434                 if (($serverdata['network'] == Protocol::PHANTOM) && !empty($network)) {
435                         $serverdata['network'] = $network;
436                 }
437
438                 $serverdata['url'] = $url;
439                 $serverdata['nurl'] = Strings::normaliseLink($url);
440
441                 // We take the highest number that we do find
442                 $registeredUsers = $serverdata['registered-users'] ?? 0;
443
444                 // On an active server there has to be at least a single user
445                 if (($serverdata['network'] != Protocol::PHANTOM) && ($registeredUsers == 0)) {
446                         $registeredUsers = 1;
447                 }
448
449                 if ($serverdata['network'] == Protocol::PHANTOM) {
450                         $serverdata['registered-users'] = $registeredUsers;
451                         $serverdata = self::detectNetworkViaContacts($url, $serverdata);
452                 }
453
454                 $serverdata['last_contact'] = DateTimeFormat::utcNow();
455                 $serverdata['failed'] = false;
456
457                 $gserver = DBA::selectFirst('gserver', ['network'], ['nurl' => Strings::normaliseLink($url)]);
458                 if (!DBA::isResult($gserver)) {
459                         $serverdata['created'] = DateTimeFormat::utcNow();
460                         $ret = DBA::insert('gserver', $serverdata);
461                         $id = DBA::lastInsertId();
462                 } else {
463                         // Don't override the network with 'unknown' when there had been a valid entry before
464                         if (($serverdata['network'] == Protocol::PHANTOM) && !empty($gserver['network'])) {
465                                 unset($serverdata['network']);
466                         }
467
468                         $ret = DBA::update('gserver', $serverdata, ['nurl' => $serverdata['nurl']]);
469                         $gserver = DBA::selectFirst('gserver', ['id'], ['nurl' => $serverdata['nurl']]);
470                         if (DBA::isResult($gserver)) {
471                                 $id = $gserver['id'];
472                         }
473                 }
474
475                 if (!empty($serverdata['network']) && !empty($id) && ($serverdata['network'] != Protocol::PHANTOM)) {
476                         $apcontacts = DBA::count('apcontact', ['gsid' => $id]);
477                         $contacts = DBA::count('contact', ['uid' => 0, 'gsid' => $id]);
478                         $max_users = max($apcontacts, $contacts, $registeredUsers);
479                         if ($max_users > $registeredUsers) {
480                                 Logger::info('Update registered users', ['id' => $id, 'url' => $serverdata['nurl'], 'registered-users' => $max_users]);
481                                 DBA::update('gserver', ['registered-users' => $max_users], ['id' => $id]);
482                         }
483                 }
484
485                 if (!empty($serverdata['network']) && in_array($serverdata['network'], [Protocol::DFRN, Protocol::DIASPORA])) {
486                         self::discoverRelay($url);
487                 }
488
489                 return $ret;
490         }
491
492         /**
493          * Fetch relay data from a given server url
494          *
495          * @param string $server_url address of the server
496          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
497          */
498         private static function discoverRelay(string $server_url)
499         {
500                 Logger::info('Discover relay data', ['server' => $server_url]);
501
502                 $curlResult = DI::httpRequest()->get($server_url . '/.well-known/x-social-relay');
503                 if (!$curlResult->isSuccess()) {
504                         return;
505                 }
506
507                 $data = json_decode($curlResult->getBody(), true);
508                 if (!is_array($data)) {
509                         return;
510                 }
511
512                 // Sanitize incoming data, see https://github.com/friendica/friendica/issues/8565
513                 $data['subscribe'] = (bool)$data['subscribe'] ?? false;
514
515                 if (!$data['subscribe'] || empty($data['scope']) || !in_array(strtolower($data['scope']), ['all', 'tags'])) {
516                         $data['scope'] = '';
517                         $data['subscribe'] = false;
518                         $data['tags'] = [];
519                 }
520
521                 $gserver = DBA::selectFirst('gserver', ['id', 'relay-subscribe', 'relay-scope'], ['nurl' => Strings::normaliseLink($server_url)]);
522                 if (!DBA::isResult($gserver)) {
523                         return;
524                 }
525
526                 if (($gserver['relay-subscribe'] != $data['subscribe']) || ($gserver['relay-scope'] != $data['scope'])) {
527                         $fields = ['relay-subscribe' => $data['subscribe'], 'relay-scope' => $data['scope']];
528                         DBA::update('gserver', $fields, ['id' => $gserver['id']]);
529                 }
530
531                 DBA::delete('gserver-tag', ['gserver-id' => $gserver['id']]);
532
533                 if ($data['scope'] == 'tags') {
534                         // Avoid duplicates
535                         $tags = [];
536                         foreach ($data['tags'] as $tag) {
537                                 $tag = mb_strtolower($tag);
538                                 if (strlen($tag) < 100) {
539                                         $tags[$tag] = $tag;
540                                 }
541                         }
542
543                         foreach ($tags as $tag) {
544                                 DBA::insert('gserver-tag', ['gserver-id' => $gserver['id'], 'tag' => $tag], true);
545                         }
546                 }
547
548                 // Create or update the relay contact
549                 $fields = [];
550                 if (isset($data['protocols'])) {
551                         if (isset($data['protocols']['diaspora'])) {
552                                 $fields['network'] = Protocol::DIASPORA;
553
554                                 if (isset($data['protocols']['diaspora']['receive'])) {
555                                         $fields['batch'] = $data['protocols']['diaspora']['receive'];
556                                 } elseif (is_string($data['protocols']['diaspora'])) {
557                                         $fields['batch'] = $data['protocols']['diaspora'];
558                                 }
559                         }
560
561                         if (isset($data['protocols']['dfrn'])) {
562                                 $fields['network'] = Protocol::DFRN;
563
564                                 if (isset($data['protocols']['dfrn']['receive'])) {
565                                         $fields['batch'] = $data['protocols']['dfrn']['receive'];
566                                 } elseif (is_string($data['protocols']['dfrn'])) {
567                                         $fields['batch'] = $data['protocols']['dfrn'];
568                                 }
569                         }
570                 }
571                 Diaspora::setRelayContact($server_url, $fields);
572         }
573
574         /**
575          * Fetch server data from '/statistics.json' on the given server
576          *
577          * @param string $url URL of the given server
578          *
579          * @return array server data
580          */
581         private static function fetchStatistics(string $url)
582         {
583                 $curlResult = DI::httpRequest()->get($url . '/statistics.json');
584                 if (!$curlResult->isSuccess()) {
585                         return [];
586                 }
587
588                 $data = json_decode($curlResult->getBody(), true);
589                 if (empty($data)) {
590                         return [];
591                 }
592
593                 $serverdata = ['detection-method' => self::DETECT_STATISTICS_JSON];
594
595                 if (!empty($data['version'])) {
596                         $serverdata['version'] = $data['version'];
597                         // Version numbers on statistics.json are presented with additional info, e.g.:
598                         // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
599                         $serverdata['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $serverdata['version']);
600                 }
601
602                 if (!empty($data['name'])) {
603                         $serverdata['site_name'] = $data['name'];
604                 }
605
606                 if (!empty($data['network'])) {
607                         $serverdata['platform'] = strtolower($data['network']);
608
609                         if ($serverdata['platform'] == 'diaspora') {
610                                 $serverdata['network'] = Protocol::DIASPORA;
611                         } elseif ($serverdata['platform'] == 'friendica') {
612                                 $serverdata['network'] = Protocol::DFRN;
613                         } elseif ($serverdata['platform'] == 'hubzilla') {
614                                 $serverdata['network'] = Protocol::ZOT;
615                         } elseif ($serverdata['platform'] == 'redmatrix') {
616                                 $serverdata['network'] = Protocol::ZOT;
617                         }
618                 }
619
620
621                 if (!empty($data['registrations_open'])) {
622                         $serverdata['register_policy'] = Register::OPEN;
623                 } else {
624                         $serverdata['register_policy'] = Register::CLOSED;
625                 }
626
627                 return $serverdata;
628         }
629
630         /**
631          * Detect server type by using the nodeinfo data
632          *
633          * @param string     $url        address of the server
634          * @param CurlResult $curlResult
635          * @return array Server data
636          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
637          */
638         private static function fetchNodeinfo(string $url, CurlResult $curlResult)
639         {
640                 if (!$curlResult->isSuccess()) {
641                         return [];
642                 }
643
644                 $nodeinfo = json_decode($curlResult->getBody(), true);
645
646                 if (!is_array($nodeinfo) || empty($nodeinfo['links'])) {
647                         return [];
648                 }
649
650                 $nodeinfo1_url = '';
651                 $nodeinfo2_url = '';
652
653                 foreach ($nodeinfo['links'] as $link) {
654                         if (!is_array($link) || empty($link['rel']) || empty($link['href'])) {
655                                 Logger::info('Invalid nodeinfo format', ['url' => $url]);
656                                 continue;
657                         }
658                         if ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/1.0') {
659                                 $nodeinfo1_url = $link['href'];
660                         } elseif ($link['rel'] == 'http://nodeinfo.diaspora.software/ns/schema/2.0') {
661                                 $nodeinfo2_url = $link['href'];
662                         }
663                 }
664
665                 if ($nodeinfo1_url . $nodeinfo2_url == '') {
666                         return [];
667                 }
668
669                 $server = [];
670
671                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
672                 if (!empty($nodeinfo2_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo2_url, PHP_URL_HOST))) {
673                         $server = self::parseNodeinfo2($nodeinfo2_url);
674                 }
675
676                 // When the nodeinfo url isn't on the same host, then there is obviously something wrong
677                 if (empty($server) && !empty($nodeinfo1_url) && (parse_url($url, PHP_URL_HOST) == parse_url($nodeinfo1_url, PHP_URL_HOST))) {
678                         $server = self::parseNodeinfo1($nodeinfo1_url);
679                 }
680
681                 return $server;
682         }
683
684         /**
685          * Parses Nodeinfo 1
686          *
687          * @param string $nodeinfo_url address of the nodeinfo path
688          * @return array Server data
689          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
690          */
691         private static function parseNodeinfo1(string $nodeinfo_url)
692         {
693                 $curlResult = DI::httpRequest()->get($nodeinfo_url);
694
695                 if (!$curlResult->isSuccess()) {
696                         return [];
697                 }
698
699                 $nodeinfo = json_decode($curlResult->getBody(), true);
700
701                 if (!is_array($nodeinfo)) {
702                         return [];
703                 }
704
705                 $server = ['detection-method' => self::DETECT_NODEINFO_1,
706                         'register_policy' => Register::CLOSED];
707
708                 if (!empty($nodeinfo['openRegistrations'])) {
709                         $server['register_policy'] = Register::OPEN;
710                 }
711
712                 if (is_array($nodeinfo['software'])) {
713                         if (!empty($nodeinfo['software']['name'])) {
714                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
715                         }
716
717                         if (!empty($nodeinfo['software']['version'])) {
718                                 $server['version'] = $nodeinfo['software']['version'];
719                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
720                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
721                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
722                         }
723                 }
724
725                 if (!empty($nodeinfo['metadata']['nodeName'])) {
726                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
727                 }
728
729                 if (!empty($nodeinfo['usage']['users']['total'])) {
730                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
731                 }
732
733                 if (!empty($nodeinfo['protocols']['inbound']) && is_array($nodeinfo['protocols']['inbound'])) {
734                         $protocols = [];
735                         foreach ($nodeinfo['protocols']['inbound'] as $protocol) {
736                                 $protocols[$protocol] = true;
737                         }
738
739                         if (!empty($protocols['friendica'])) {
740                                 $server['network'] = Protocol::DFRN;
741                         } elseif (!empty($protocols['activitypub'])) {
742                                 $server['network'] = Protocol::ACTIVITYPUB;
743                         } elseif (!empty($protocols['diaspora'])) {
744                                 $server['network'] = Protocol::DIASPORA;
745                         } elseif (!empty($protocols['ostatus'])) {
746                                 $server['network'] = Protocol::OSTATUS;
747                         } elseif (!empty($protocols['gnusocial'])) {
748                                 $server['network'] = Protocol::OSTATUS;
749                         } elseif (!empty($protocols['zot'])) {
750                                 $server['network'] = Protocol::ZOT;
751                         }
752                 }
753
754                 if (empty($server)) {
755                         return [];
756                 }
757
758                 return $server;
759         }
760
761         /**
762          * Parses Nodeinfo 2
763          *
764          * @param string $nodeinfo_url address of the nodeinfo path
765          * @return array Server data
766          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
767          */
768         private static function parseNodeinfo2(string $nodeinfo_url)
769         {
770                 $curlResult = DI::httpRequest()->get($nodeinfo_url);
771                 if (!$curlResult->isSuccess()) {
772                         return [];
773                 }
774
775                 $nodeinfo = json_decode($curlResult->getBody(), true);
776
777                 if (!is_array($nodeinfo)) {
778                         return [];
779                 }
780
781                 $server = ['detection-method' => self::DETECT_NODEINFO_2,
782                         'register_policy' => Register::CLOSED];
783
784                 if (!empty($nodeinfo['openRegistrations'])) {
785                         $server['register_policy'] = Register::OPEN;
786                 }
787
788                 if (is_array($nodeinfo['software'])) {
789                         if (!empty($nodeinfo['software']['name'])) {
790                                 $server['platform'] = strtolower($nodeinfo['software']['name']);
791                         }
792
793                         if (!empty($nodeinfo['software']['version'])) {
794                                 $server['version'] = $nodeinfo['software']['version'];
795                                 // Version numbers on Nodeinfo are presented with additional info, e.g.:
796                                 // 0.6.3.0-p1702cc1c, 0.6.99.0-p1b9ab160 or 3.4.3-2-1191.
797                                 $server['version'] = preg_replace('=(.+)-(.{4,})=ism', '$1', $server['version']);
798                         }
799                 }
800
801                 if (!empty($nodeinfo['metadata']['nodeName'])) {
802                         $server['site_name'] = $nodeinfo['metadata']['nodeName'];
803                 }
804
805                 if (!empty($nodeinfo['usage']['users']['total'])) {
806                         $server['registered-users'] = $nodeinfo['usage']['users']['total'];
807                 }
808
809                 if (!empty($nodeinfo['protocols'])) {
810                         $protocols = [];
811                         foreach ($nodeinfo['protocols'] as $protocol) {
812                                 $protocols[$protocol] = true;
813                         }
814
815                         if (!empty($protocols['dfrn'])) {
816                                 $server['network'] = Protocol::DFRN;
817                         } elseif (!empty($protocols['activitypub'])) {
818                                 $server['network'] = Protocol::ACTIVITYPUB;
819                         } elseif (!empty($protocols['diaspora'])) {
820                                 $server['network'] = Protocol::DIASPORA;
821                         } elseif (!empty($protocols['ostatus'])) {
822                                 $server['network'] = Protocol::OSTATUS;
823                         } elseif (!empty($protocols['gnusocial'])) {
824                                 $server['network'] = Protocol::OSTATUS;
825                         } elseif (!empty($protocols['zot'])) {
826                                 $server['network'] = Protocol::ZOT;
827                         }
828                 }
829
830                 if (empty($server)) {
831                         return [];
832                 }
833
834                 return $server;
835         }
836
837         /**
838          * Fetch server information from a 'siteinfo.json' file on the given server
839          *
840          * @param string $url        URL of the given server
841          * @param array  $serverdata array with server data
842          *
843          * @return array server data
844          */
845         private static function fetchSiteinfo(string $url, array $serverdata)
846         {
847                 $curlResult = DI::httpRequest()->get($url . '/siteinfo.json');
848                 if (!$curlResult->isSuccess()) {
849                         return $serverdata;
850                 }
851
852                 $data = json_decode($curlResult->getBody(), true);
853                 if (empty($data)) {
854                         return $serverdata;
855                 }
856
857                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
858                         $serverdata['detection-method'] = self::DETECT_SITEINFO_JSON;
859                 }
860
861                 if (!empty($data['url'])) {
862                         $serverdata['platform'] = strtolower($data['platform']);
863                         $serverdata['version'] = $data['version'];
864                 }
865
866                 if (!empty($data['plugins'])) {
867                         if (in_array('pubcrawl', $data['plugins'])) {
868                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
869                         } elseif (in_array('diaspora', $data['plugins'])) {
870                                 $serverdata['network'] = Protocol::DIASPORA;
871                         } elseif (in_array('gnusoc', $data['plugins'])) {
872                                 $serverdata['network'] = Protocol::OSTATUS;
873                         } else {
874                                 $serverdata['network'] = Protocol::ZOT;
875                         }
876                 }
877
878                 if (!empty($data['site_name'])) {
879                         $serverdata['site_name'] = $data['site_name'];
880                 }
881
882                 if (!empty($data['channels_total'])) {
883                         $serverdata['registered-users'] = $data['channels_total'];
884                 }
885
886                 if (!empty($data['register_policy'])) {
887                         switch ($data['register_policy']) {
888                                 case 'REGISTER_OPEN':
889                                         $serverdata['register_policy'] = Register::OPEN;
890                                         break;
891
892                                 case 'REGISTER_APPROVE':
893                                         $serverdata['register_policy'] = Register::APPROVE;
894                                         break;
895
896                                 case 'REGISTER_CLOSED':
897                                 default:
898                                         $serverdata['register_policy'] = Register::CLOSED;
899                                         break;
900                         }
901                 }
902
903                 return $serverdata;
904         }
905
906         /**
907          * Checks if the server contains a valid host meta file
908          *
909          * @param string $url URL of the given server
910          *
911          * @return boolean 'true' if the server seems to be vital
912          */
913         private static function validHostMeta(string $url)
914         {
915                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout');
916                 $curlResult = DI::httpRequest()->get($url . '/.well-known/host-meta', ['timeout' => $xrd_timeout]);
917                 if (!$curlResult->isSuccess()) {
918                         return false;
919                 }
920
921                 $xrd = XML::parseString($curlResult->getBody());
922                 if (!is_object($xrd)) {
923                         return false;
924                 }
925
926                 $elements = XML::elementToArray($xrd);
927                 if (empty($elements) || empty($elements['xrd']) || empty($elements['xrd']['link'])) {
928                         return false;
929                 }
930
931                 $valid = false;
932                 foreach ($elements['xrd']['link'] as $link) {
933                         // When there is more than a single "link" element, the array looks slightly different
934                         if (!empty($link['@attributes'])) {
935                                 $link = $link['@attributes'];
936                         }
937
938                         if (empty($link['rel']) || empty($link['template'])) {
939                                 continue;
940                         }
941
942                         if ($link['rel'] == 'lrdd') {
943                                 // When the webfinger host is the same like the system host, it should be ok.
944                                 $valid = (parse_url($url, PHP_URL_HOST) == parse_url($link['template'], PHP_URL_HOST));
945                         }
946                 }
947
948                 return $valid;
949         }
950
951         /**
952          * Detect the network of the given server via their known contacts
953          *
954          * @param string $url        URL of the given server
955          * @param array  $serverdata array with server data
956          *
957          * @return array server data
958          */
959         private static function detectNetworkViaContacts(string $url, array $serverdata)
960         {
961                 $contacts = [];
962
963                 $apcontacts = DBA::select('apcontact', ['url'], ['baseurl' => [$url, $serverdata['nurl']]]);
964                 while ($apcontact = DBA::fetch($apcontacts)) {
965                         $contacts[Strings::normaliseLink($apcontact['url'])] = $apcontact['url'];
966                 }
967                 DBA::close($apcontacts);
968
969                 $pcontacts = DBA::select('contact', ['url', 'nurl'], ['uid' => 0, 'baseurl' => [$url, $serverdata['nurl']]]);
970                 while ($pcontact = DBA::fetch($pcontacts)) {
971                         $contacts[$pcontact['nurl']] = $pcontact['url'];
972                 }
973                 DBA::close($pcontacts);
974
975                 if (empty($contacts)) {
976                         return $serverdata;
977                 }
978
979                 foreach ($contacts as $contact) {
980                         $probed = Contact::getByURL($contact);
981                         if (!empty($probed) && in_array($probed['network'], Protocol::FEDERATED)) {
982                                 $serverdata['network'] = $probed['network'];
983                                 break;
984                         }
985                 }
986
987                 $serverdata['registered-users'] = max($serverdata['registered-users'], count($contacts));
988
989                 return $serverdata;
990         }
991
992         /**
993          * Checks if the given server does have a '/poco' endpoint.
994          * This is used for the 'PortableContact' functionality,
995          * which is used by both Friendica and Hubzilla.
996          *
997          * @param string $url        URL of the given server
998          * @param array  $serverdata array with server data
999          *
1000          * @return array server data
1001          */
1002         private static function checkPoCo(string $url, array $serverdata)
1003         {
1004                 $serverdata['poco'] = '';
1005
1006                 $curlResult = DI::httpRequest()->get($url . '/poco');
1007                 if (!$curlResult->isSuccess()) {
1008                         return $serverdata;
1009                 }
1010
1011                 $data = json_decode($curlResult->getBody(), true);
1012                 if (empty($data)) {
1013                         return $serverdata;
1014                 }
1015
1016                 if (!empty($data['totalResults'])) {
1017                         $registeredUsers = $serverdata['registered-users'] ?? 0;
1018                         $serverdata['registered-users'] = max($data['totalResults'], $registeredUsers);
1019                         $serverdata['directory-type'] = self::DT_POCO;
1020                         $serverdata['poco'] = $url . '/poco';
1021                 }
1022
1023                 return $serverdata;
1024         }
1025
1026         /**
1027          * Checks if the given server does have a Mastodon style directory endpoint.
1028          *
1029          * @param string $url        URL of the given server
1030          * @param array  $serverdata array with server data
1031          *
1032          * @return array server data
1033          */
1034         public static function checkMastodonDirectory(string $url, array $serverdata)
1035         {
1036                 $curlResult = DI::httpRequest()->get($url . '/api/v1/directory?limit=1');
1037                 if (!$curlResult->isSuccess()) {
1038                         return $serverdata;
1039                 }
1040
1041                 $data = json_decode($curlResult->getBody(), true);
1042                 if (empty($data)) {
1043                         return $serverdata;
1044                 }
1045
1046                 if (count($data) == 1) {
1047                         $serverdata['directory-type'] = self::DT_MASTODON;
1048                 }
1049
1050                 return $serverdata;
1051         }
1052
1053         /**
1054          * Detects the version number of a given server when it was a NextCloud installation
1055          *
1056          * @param string $url        URL of the given server
1057          * @param array  $serverdata array with server data
1058          *
1059          * @return array server data
1060          */
1061         private static function detectNextcloud(string $url, array $serverdata)
1062         {
1063                 $curlResult = DI::httpRequest()->get($url . '/status.php');
1064
1065                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1066                         return $serverdata;
1067                 }
1068
1069                 $data = json_decode($curlResult->getBody(), true);
1070                 if (empty($data)) {
1071                         return $serverdata;
1072                 }
1073
1074                 if (!empty($data['version'])) {
1075                         $serverdata['platform'] = 'nextcloud';
1076                         $serverdata['version'] = $data['version'];
1077                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1078
1079                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1080                                 $serverdata['detection-method'] = self::DETECT_STATUS_PHP;
1081                         }
1082                 }
1083
1084                 return $serverdata;
1085         }
1086
1087         /**
1088          * Detects data from a given server url if it was a mastodon alike system
1089          *
1090          * @param string $url        URL of the given server
1091          * @param array  $serverdata array with server data
1092          *
1093          * @return array server data
1094          */
1095         private static function detectMastodonAlikes(string $url, array $serverdata)
1096         {
1097                 $curlResult = DI::httpRequest()->get($url . '/api/v1/instance');
1098
1099                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1100                         return $serverdata;
1101                 }
1102
1103                 $data = json_decode($curlResult->getBody(), true);
1104                 if (empty($data)) {
1105                         return $serverdata;
1106                 }
1107
1108                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1109                         $serverdata['detection-method'] = self::DETECT_MASTODON_API;
1110                 }
1111
1112                 if (!empty($data['version'])) {
1113                         $serverdata['platform'] = 'mastodon';
1114                         $serverdata['version'] = $data['version'] ?? '';
1115                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1116                 }
1117
1118                 if (!empty($data['title'])) {
1119                         $serverdata['site_name'] = $data['title'];
1120                 }
1121
1122                 if (!empty($data['title']) && empty($serverdata['platform']) && empty($serverdata['network'])) {
1123                         $serverdata['platform'] = 'mastodon';
1124                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1125                 }
1126
1127                 if (!empty($data['description'])) {
1128                         $serverdata['info'] = trim($data['description']);
1129                 }
1130
1131                 if (!empty($data['stats']['user_count'])) {
1132                         $serverdata['registered-users'] = $data['stats']['user_count'];
1133                 }
1134
1135                 if (!empty($serverdata['version']) && preg_match('/.*?\(compatible;\s(.*)\s(.*)\)/ism', $serverdata['version'], $matches)) {
1136                         $serverdata['platform'] = strtolower($matches[1]);
1137                         $serverdata['version'] = $matches[2];
1138                 }
1139
1140                 if (!empty($serverdata['version']) && strstr(strtolower($serverdata['version']), 'pleroma')) {
1141                         $serverdata['platform'] = 'pleroma';
1142                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1143                 }
1144
1145                 if (!empty($serverdata['platform']) && strstr($serverdata['platform'], 'pleroma')) {
1146                         $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['platform']));
1147                         $serverdata['platform'] = 'pleroma';
1148                 }
1149
1150                 return $serverdata;
1151         }
1152
1153         /**
1154          * Detects data from typical Hubzilla endpoints
1155          *
1156          * @param string $url        URL of the given server
1157          * @param array  $serverdata array with server data
1158          *
1159          * @return array server data
1160          */
1161         private static function detectHubzilla(string $url, array $serverdata)
1162         {
1163                 $curlResult = DI::httpRequest()->get($url . '/api/statusnet/config.json');
1164                 if (!$curlResult->isSuccess() || ($curlResult->getBody() == '')) {
1165                         return $serverdata;
1166                 }
1167
1168                 $data = json_decode($curlResult->getBody(), true);
1169                 if (empty($data) || empty($data['site'])) {
1170                         return $serverdata;
1171                 }
1172
1173                 if (!empty($data['site']['name'])) {
1174                         $serverdata['site_name'] = $data['site']['name'];
1175                 }
1176
1177                 if (!empty($data['site']['platform'])) {
1178                         $serverdata['platform'] = strtolower($data['site']['platform']['PLATFORM_NAME']);
1179                         $serverdata['version'] = $data['site']['platform']['STD_VERSION'];
1180                         $serverdata['network'] = Protocol::ZOT;
1181                 }
1182
1183                 if (!empty($data['site']['hubzilla'])) {
1184                         $serverdata['platform'] = strtolower($data['site']['hubzilla']['PLATFORM_NAME']);
1185                         $serverdata['version'] = $data['site']['hubzilla']['RED_VERSION'];
1186                         $serverdata['network'] = Protocol::ZOT;
1187                 }
1188
1189                 if (!empty($data['site']['redmatrix'])) {
1190                         if (!empty($data['site']['redmatrix']['PLATFORM_NAME'])) {
1191                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['PLATFORM_NAME']);
1192                         } elseif (!empty($data['site']['redmatrix']['RED_PLATFORM'])) {
1193                                 $serverdata['platform'] = strtolower($data['site']['redmatrix']['RED_PLATFORM']);
1194                         }
1195
1196                         $serverdata['version'] = $data['site']['redmatrix']['RED_VERSION'];
1197                         $serverdata['network'] = Protocol::ZOT;
1198                 }
1199
1200                 $private = false;
1201                 $inviteonly = false;
1202                 $closed = false;
1203
1204                 if (!empty($data['site']['closed'])) {
1205                         $closed = self::toBoolean($data['site']['closed']);
1206                 }
1207
1208                 if (!empty($data['site']['private'])) {
1209                         $private = self::toBoolean($data['site']['private']);
1210                 }
1211
1212                 if (!empty($data['site']['inviteonly'])) {
1213                         $inviteonly = self::toBoolean($data['site']['inviteonly']);
1214                 }
1215
1216                 if (!$closed && !$private and $inviteonly) {
1217                         $serverdata['register_policy'] = Register::APPROVE;
1218                 } elseif (!$closed && !$private) {
1219                         $serverdata['register_policy'] = Register::OPEN;
1220                 } else {
1221                         $serverdata['register_policy'] = Register::CLOSED;
1222                 }
1223
1224                 if (!empty($serverdata['network']) && in_array($serverdata['detection-method'],
1225                         [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1226                         $serverdata['detection-method'] = self::DETECT_CONFIG_JSON;
1227                 }
1228
1229                 return $serverdata;
1230         }
1231
1232         /**
1233          * Converts input value to a boolean value
1234          *
1235          * @param string|integer $val
1236          *
1237          * @return boolean
1238          */
1239         private static function toBoolean($val)
1240         {
1241                 if (($val == 'true') || ($val == 1)) {
1242                         return true;
1243                 } elseif (($val == 'false') || ($val == 0)) {
1244                         return false;
1245                 }
1246
1247                 return $val;
1248         }
1249
1250         /**
1251          * Detect if the URL belongs to a GNU Social server
1252          *
1253          * @param string $url        URL of the given server
1254          * @param array  $serverdata array with server data
1255          *
1256          * @return array server data
1257          */
1258         private static function detectGNUSocial(string $url, array $serverdata)
1259         {
1260                 // Test for GNU Social
1261                 $curlResult = DI::httpRequest()->get($url . '/api/gnusocial/version.json');
1262                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1263                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1264                         $serverdata['platform'] = 'gnusocial';
1265                         // Remove junk that some GNU Social servers return
1266                         $serverdata['version'] = str_replace(chr(239) . chr(187) . chr(191), '', $curlResult->getBody());
1267                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1268                         $serverdata['version'] = trim($serverdata['version'], '"');
1269                         $serverdata['network'] = Protocol::OSTATUS;
1270
1271                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1272                                 $serverdata['detection-method'] = self::DETECT_GNUSOCIAL;
1273                         }
1274         
1275                         return $serverdata;
1276                 }
1277
1278                 // Test for Statusnet
1279                 $curlResult = DI::httpRequest()->get($url . '/api/statusnet/version.json');
1280                 if ($curlResult->isSuccess() && ($curlResult->getBody() != '{"error":"not implemented"}') &&
1281                         ($curlResult->getBody() != '') && (strlen($curlResult->getBody()) < 30)) {
1282
1283                         // Remove junk that some GNU Social servers return
1284                         $serverdata['version'] = str_replace(chr(239).chr(187).chr(191), '', $curlResult->getBody());
1285                         $serverdata['version'] = str_replace(["\r", "\n", "\t"], '', $serverdata['version']);
1286                         $serverdata['version'] = trim($serverdata['version'], '"');
1287
1288                         if (!empty($serverdata['version']) && strtolower(substr($serverdata['version'], 0, 7)) == 'pleroma') {
1289                                 $serverdata['platform'] = 'pleroma';
1290                                 $serverdata['version'] = trim(str_ireplace('pleroma', '', $serverdata['version']));
1291                                 $serverdata['network'] = Protocol::ACTIVITYPUB;
1292                         } else {
1293                                 $serverdata['platform'] = 'statusnet';
1294                                 $serverdata['network'] = Protocol::OSTATUS;
1295                         }
1296
1297                         if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {
1298                                 $serverdata['detection-method'] = self::DETECT_STATUSNET;
1299                         }
1300                 }
1301
1302                 return $serverdata;
1303         }
1304
1305         /**
1306          * Detect if the URL belongs to a Friendica server
1307          *
1308          * @param string $url        URL of the given server
1309          * @param array  $serverdata array with server data
1310          *
1311          * @return array server data
1312          */
1313         private static function detectFriendica(string $url, array $serverdata)
1314         {
1315                 $curlResult = DI::httpRequest()->get($url . '/friendica/json');
1316                 if (!$curlResult->isSuccess()) {
1317                         $curlResult = DI::httpRequest()->get($url . '/friendika/json');
1318                         $friendika = true;
1319                         $platform = 'Friendika';
1320                 } else {
1321                         $friendika = false;
1322                         $platform = 'Friendica';
1323                 }
1324
1325                 if (!$curlResult->isSuccess()) {
1326                         return $serverdata;
1327                 }
1328
1329                 $data = json_decode($curlResult->getBody(), true);
1330                 if (empty($data) || empty($data['version'])) {
1331                         return $serverdata;
1332                 }
1333
1334                 if (in_array($serverdata['detection-method'], [self::DETECT_HEADER, self::DETECT_BODY, self::DETECT_MANUAL])) {                 
1335                         $serverdata['detection-method'] = $friendika ? self::DETECT_FRIENDIKA : self::DETECT_FRIENDICA;
1336                 }
1337
1338                 $serverdata['network'] = Protocol::DFRN;
1339                 $serverdata['version'] = $data['version'];
1340
1341                 if (!empty($data['no_scrape_url'])) {
1342                         $serverdata['noscrape'] = $data['no_scrape_url'];
1343                 }
1344
1345                 if (!empty($data['site_name'])) {
1346                         $serverdata['site_name'] = $data['site_name'];
1347                 }
1348
1349                 if (!empty($data['info'])) {
1350                         $serverdata['info'] = trim($data['info']);
1351                 }
1352
1353                 $register_policy = ($data['register_policy'] ?? '') ?: 'REGISTER_CLOSED';
1354                 switch ($register_policy) {
1355                         case 'REGISTER_OPEN':
1356                                 $serverdata['register_policy'] = Register::OPEN;
1357                                 break;
1358
1359                         case 'REGISTER_APPROVE':
1360                                 $serverdata['register_policy'] = Register::APPROVE;
1361                                 break;
1362
1363                         case 'REGISTER_CLOSED':
1364                         case 'REGISTER_INVITATION':
1365                                 $serverdata['register_policy'] = Register::CLOSED;
1366                                 break;
1367                         default:
1368                                 Logger::info('Register policy is invalid', ['policy' => $register_policy, 'server' => $url]);
1369                                 $serverdata['register_policy'] = Register::CLOSED;
1370                                 break;
1371                 }
1372
1373                 $serverdata['platform'] = strtolower($data['platform'] ?? $platform);
1374
1375                 return $serverdata;
1376         }
1377
1378         /**
1379          * Analyses the landing page of a given server for hints about type and system of that server
1380          *
1381          * @param object $curlResult result of curl execution
1382          * @param array  $serverdata array with server data
1383          * @param string $url        Server URL
1384          *
1385          * @return array server data
1386          */
1387         private static function analyseRootBody($curlResult, array $serverdata, string $url)
1388         {
1389                 $doc = new DOMDocument();
1390                 @$doc->loadHTML($curlResult->getBody());
1391                 $xpath = new DOMXPath($doc);
1392
1393                 $title = trim(XML::getFirstNodeValue($xpath, '//head/title/text()'));
1394                 if (!empty($title)) {
1395                         $serverdata['site_name'] = $title;
1396                 }
1397
1398                 $list = $xpath->query('//meta[@name]');
1399
1400                 foreach ($list as $node) {
1401                         $attr = [];
1402                         if ($node->attributes->length) {
1403                                 foreach ($node->attributes as $attribute) {
1404                                         $value = trim($attribute->value);
1405                                         if (empty($value)) {
1406                                                 continue;
1407                                         }
1408
1409                                         $attr[$attribute->name] = $value;
1410                                 }
1411
1412                                 if (empty($attr['name']) || empty($attr['content'])) {
1413                                         continue;
1414                                 }
1415                         }
1416
1417                         if ($attr['name'] == 'description') {
1418                                 $serverdata['info'] = $attr['content'];
1419                         }
1420
1421                         if (in_array($attr['name'], ['application-name', 'al:android:app_name', 'al:ios:app_name',
1422                                 'twitter:app:name:googleplay', 'twitter:app:name:iphone', 'twitter:app:name:ipad'])) {
1423                                 $serverdata['platform'] = strtolower($attr['content']);
1424                                 if (in_array($attr['content'], ['Misskey', 'Write.as'])) {
1425                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1426                                 }
1427                         }
1428                         if (($attr['name'] == 'generator') && (empty($serverdata['platform']) || (substr(strtolower($attr['content']), 0, 9) == 'wordpress'))) {
1429                                 $serverdata['platform'] = strtolower($attr['content']);
1430                                 $version_part = explode(' ', $attr['content']);
1431
1432                                 if (count($version_part) == 2) {
1433                                         if (in_array($version_part[0], ['WordPress'])) {
1434                                                 $serverdata['platform'] = strtolower($version_part[0]);
1435                                                 $serverdata['version'] = $version_part[1];
1436
1437                                                 // We still do need a reliable test if some AP plugin is activated
1438                                                 if (DBA::exists('apcontact', ['baseurl' => $url])) {
1439                                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1440                                                 } else {
1441                                                         $serverdata['network'] = Protocol::FEED;
1442                                                 }
1443
1444                                                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1445                                                         $serverdata['detection-method'] = self::DETECT_BODY;
1446                                                 }
1447                                         }
1448                                         if (in_array($version_part[0], ['Friendika', 'Friendica'])) {
1449                                                 $serverdata['platform'] = strtolower($version_part[0]);
1450                                                 $serverdata['version'] = $version_part[1];
1451                                                 $serverdata['network'] = Protocol::DFRN;
1452                                         }
1453                                 }
1454                         }
1455                 }
1456
1457                 $list = $xpath->query('//meta[@property]');
1458
1459                 foreach ($list as $node) {
1460                         $attr = [];
1461                         if ($node->attributes->length) {
1462                                 foreach ($node->attributes as $attribute) {
1463                                         $value = trim($attribute->value);
1464                                         if (empty($value)) {
1465                                                 continue;
1466                                         }
1467
1468                                         $attr[$attribute->name] = $value;
1469                                 }
1470
1471                                 if (empty($attr['property']) || empty($attr['content'])) {
1472                                         continue;
1473                                 }
1474                         }
1475
1476                         if ($attr['property'] == 'og:site_name') {
1477                                 $serverdata['site_name'] = $attr['content'];
1478                         }
1479
1480                         if ($attr['property'] == 'og:description') {
1481                                 $serverdata['info'] = $attr['content'];
1482                         }
1483
1484                         if ($attr['property'] == 'og:platform') {
1485                                 $serverdata['platform'] = strtolower($attr['content']);
1486
1487                                 if (in_array($attr['content'], ['PeerTube'])) {
1488                                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1489                                 }
1490                         }
1491
1492                         if ($attr['property'] == 'generator') {
1493                                 $serverdata['platform'] = strtolower($attr['content']);
1494
1495                                 if (in_array($attr['content'], ['hubzilla'])) {
1496                                         // We later check which compatible protocol modules are loaded.
1497                                         $serverdata['network'] = Protocol::ZOT;
1498                                 }
1499                         }
1500                 }
1501
1502                 if (!empty($serverdata['network']) && ($serverdata['detection-method'] == self::DETECT_MANUAL)) {
1503                         $serverdata['detection-method'] = self::DETECT_BODY;
1504                 }
1505
1506                 return $serverdata;
1507         }
1508
1509         /**
1510          * Analyses the header data of a given server for hints about type and system of that server
1511          *
1512          * @param object $curlResult result of curl execution
1513          * @param array  $serverdata array with server data
1514          *
1515          * @return array server data
1516          */
1517         private static function analyseRootHeader($curlResult, array $serverdata)
1518         {
1519                 if ($curlResult->getHeader('server') == 'Mastodon') {
1520                         $serverdata['platform'] = 'mastodon';
1521                         $serverdata['network'] = Protocol::ACTIVITYPUB;
1522                 } elseif ($curlResult->inHeader('x-diaspora-version')) {
1523                         $serverdata['platform'] = 'diaspora';
1524                         $serverdata['network'] = Protocol::DIASPORA;
1525                         $serverdata['version'] = $curlResult->getHeader('x-diaspora-version');
1526                 } elseif ($curlResult->inHeader('x-friendica-version')) {
1527                         $serverdata['platform'] = 'friendica';
1528                         $serverdata['network'] = Protocol::DFRN;
1529                         $serverdata['version'] = $curlResult->getHeader('x-friendica-version');
1530                 } else {
1531                         return $serverdata;
1532                 }
1533
1534                 if ($serverdata['detection-method'] == self::DETECT_MANUAL) {
1535                         $serverdata['detection-method'] = self::DETECT_HEADER;
1536                 }
1537
1538                 return $serverdata;
1539         }
1540
1541         /**
1542          * Test if the body contains valid content
1543          *
1544          * @param string $body
1545          * @return boolean
1546          */
1547         private static function invalidBody(string $body)
1548         {
1549                 // Currently we only test for a HTML element.
1550                 // Possibly we enhance this in the future.
1551                 return !strpos($body, '>');
1552         }
1553
1554         /**
1555          * Update GServer entries
1556          */
1557         public static function discover()
1558         {
1559                 // Update the server list
1560                 self::discoverFederation();
1561
1562                 $no_of_queries = 5;
1563
1564                 $requery_days = intval(DI::config()->get('system', 'poco_requery_days'));
1565
1566                 if ($requery_days == 0) {
1567                         $requery_days = 7;
1568                 }
1569
1570                 $last_update = date('c', time() - (60 * 60 * 24 * $requery_days));
1571
1572                 $gservers = DBA::p("SELECT `id`, `url`, `nurl`, `network`, `poco`, `directory-type`
1573                         FROM `gserver`
1574                         WHERE NOT `failed`
1575                         AND `directory-type` != ?
1576                         AND `last_poco_query` < ?
1577                         ORDER BY RAND()", self::DT_NONE, $last_update
1578                 );
1579
1580                 while ($gserver = DBA::fetch($gservers)) {
1581                         if (!GServer::check($gserver['url'], $gserver['network'])) {
1582                                 // The server is not reachable? Okay, then we will try it later
1583                                 $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1584                                 DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1585                                 continue;
1586                         }
1587
1588                         Logger::info('Update peer list', ['server' => $gserver['url'], 'id' => $gserver['id']]);
1589                         Worker::add(PRIORITY_LOW, 'UpdateServerPeers', $gserver['url']);
1590
1591                         Logger::info('Update directory', ['server' => $gserver['url'], 'id' => $gserver['id']]);
1592                         Worker::add(PRIORITY_LOW, 'UpdateServerDirectory', $gserver);
1593
1594                         $fields = ['last_poco_query' => DateTimeFormat::utcNow()];
1595                         DBA::update('gserver', $fields, ['nurl' => $gserver['nurl']]);
1596         
1597                         if (--$no_of_queries == 0) {
1598                                 break;
1599                         }
1600                 }
1601
1602                 DBA::close($gservers);
1603         }
1604
1605         /**
1606          * Discover federated servers
1607          */
1608         private static function discoverFederation()
1609         {
1610                 $last = DI::config()->get('poco', 'last_federation_discovery');
1611
1612                 if ($last) {
1613                         $next = $last + (24 * 60 * 60);
1614
1615                         if ($next > time()) {
1616                                 return;
1617                         }
1618                 }
1619
1620                 // Discover federated servers
1621                 $protocols = ['activitypub', 'diaspora', 'dfrn', 'ostatus'];
1622                 foreach ($protocols as $protocol) {
1623                         $query = '{nodes(protocol:"' . $protocol . '"){host}}';
1624                         $curlResult = DI::httpRequest()->fetch('https://the-federation.info/graphql?query=' . urlencode($query));
1625                         if (!empty($curlResult)) {
1626                                 $data = json_decode($curlResult, true);
1627                                 if (!empty($data['data']['nodes'])) {
1628                                         foreach ($data['data']['nodes'] as $server) {
1629                                                 // Using "only_nodeinfo" since servers that are listed on that page should always have it.
1630                                                 Worker::add(PRIORITY_LOW, 'UpdateGServer', 'https://' . $server['host'], true);
1631                                         }
1632                                 }
1633                         }
1634                 }
1635
1636                 // Disvover Mastodon servers
1637                 $accesstoken = DI::config()->get('system', 'instances_social_key');
1638
1639                 if (!empty($accesstoken)) {
1640                         $api = 'https://instances.social/api/1.0/instances/list?count=0';
1641                         $header = ['Authorization: Bearer '.$accesstoken];
1642                         $curlResult = DI::httpRequest()->get($api, ['headers' => $header]);
1643
1644                         if ($curlResult->isSuccess()) {
1645                                 $servers = json_decode($curlResult->getBody(), true);
1646
1647                                 foreach ($servers['instances'] as $server) {
1648                                         $url = (is_null($server['https_score']) ? 'http' : 'https') . '://' . $server['name'];
1649                                         Worker::add(PRIORITY_LOW, 'UpdateGServer', $url);
1650                                 }
1651                         }
1652                 }
1653
1654                 DI::config()->set('poco', 'last_federation_discovery', time());
1655         }
1656 }