]> git.mxchange.org Git - friendica.git/blob - src/Network/Probe.php
Abuse mail received because of "unusual traffic"
[friendica.git] / src / Network / Probe.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\Network;
23
24 use DOMDocument;
25 use DomXPath;
26 use Friendica\Core\Cache\Duration;
27 use Friendica\Core\Hook;
28 use Friendica\Core\Logger;
29 use Friendica\Core\Protocol;
30 use Friendica\Core\System;
31 use Friendica\Database\DBA;
32 use Friendica\DI;
33 use Friendica\Model\Contact;
34 use Friendica\Model\GServer;
35 use Friendica\Model\Profile;
36 use Friendica\Protocol\ActivityNamespace;
37 use Friendica\Protocol\ActivityPub;
38 use Friendica\Protocol\Email;
39 use Friendica\Protocol\Feed;
40 use Friendica\Util\Crypto;
41 use Friendica\Util\Network;
42 use Friendica\Util\Strings;
43 use Friendica\Util\XML;
44
45 /**
46  * This class contain functions for probing URL
47  */
48 class Probe
49 {
50         const WEBFINGER = '/.well-known/webfinger?resource={uri}';
51
52         private static $baseurl;
53         private static $istimeout;
54
55         /**
56          * Remove stuff from an URI that doesn't belong there
57          *
58          * @param string $URI
59          * @return string Cleaned URI
60          */
61         public static function cleanURI(string $URI)
62         {
63                 // At first remove leading and trailing junk
64                 $URI = trim($URI, "@#?:/ \t\n\r\0\x0B");
65
66                 $parts = parse_url($URI);
67
68                 if (empty($parts['scheme'])) {
69                         return $URI;
70                 }
71
72                 // Remove the URL fragment, since these shouldn't be part of any profile URL
73                 unset($parts['fragment']);
74
75                 $URI = Network::unparseURL($parts);
76
77                 return $URI;
78         }
79
80         /**
81          * Rearrange the array so that it always has the same order
82          *
83          * @param array $data Unordered data
84          *
85          * @return array Ordered data
86          */
87         private static function rearrangeData($data)
88         {
89                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "account-type",
90                                 "community", "keywords", "location", "about", "hide",
91                                 "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
92                                 "following", "followers", "inbox", "outbox", "sharedinbox",
93                                 "priority", "network", "pubkey", "baseurl", "gsid"];
94
95                 $newdata = [];
96                 foreach ($fields as $field) {
97                         if (isset($data[$field])) {
98                                 if (in_array($field, ["gsid", "hide", "account-type"])) {
99                                         $newdata[$field] = (int)$data[$field];
100                                 } else {        
101                                         $newdata[$field] = $data[$field];
102                                 }
103                         } elseif ($field != "gsid") {
104                                 $newdata[$field] = "";
105                         } else {
106                                 $newdata[$field] = null;
107                         }
108                 }
109
110                 // We don't use the "priority" field anymore and replace it with a dummy.
111                 $newdata["priority"] = 0;
112
113                 return $newdata;
114         }
115
116         /**
117          * Check if the hostname belongs to the own server
118          *
119          * @param string $host The hostname that is to be checked
120          *
121          * @return bool Does the testes hostname belongs to the own server?
122          */
123         private static function ownHost($host)
124         {
125                 $own_host = DI::baseUrl()->getHostname();
126
127                 $parts = parse_url($host);
128
129                 if (!isset($parts['scheme'])) {
130                         $parts = parse_url('http://'.$host);
131                 }
132
133                 if (!isset($parts['host'])) {
134                         return false;
135                 }
136                 return $parts['host'] == $own_host;
137         }
138
139         /**
140          * Probes for webfinger path via "host-meta"
141          *
142          * We have to check if the servers in the future still will offer this.
143          * It seems as if it was dropped from the standard.
144          *
145          * @param string $host The host part of an url
146          *
147          * @return array with template and type of the webfinger template for JSON or XML
148          * @throws HTTPException\InternalServerErrorException
149          */
150         private static function hostMeta($host)
151         {
152                 // Reset the static variable
153                 self::$baseurl = '';
154
155                 // Handles the case when the hostname contains the scheme
156                 if (!parse_url($host, PHP_URL_SCHEME)) {
157                         $ssl_url = "https://" . $host . "/.well-known/host-meta";
158                         $url = "http://" . $host . "/.well-known/host-meta";
159                 } else {
160                         $ssl_url = $host . "/.well-known/host-meta";
161                         $url = '';
162                 }
163
164                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
165
166                 Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
167                 $xrd = null;
168
169                 $curlResult = Network::curl($ssl_url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
170                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
171                 if ($curlResult->isSuccess()) {
172                         $xml = $curlResult->getBody();
173                         $xrd = XML::parseString($xml, true);
174                         if (!empty($url)) {
175                                 $host_url = 'https://' . $host;
176                         } else {
177                                 $host_url = $host;
178                         }
179                 } elseif ($curlResult->isTimeout()) {
180                         Logger::info('Probing timeout', ['url' => $ssl_url], Logger::DEBUG);
181                         self::$istimeout = true;
182                         return [];
183                 }
184
185                 if (!is_object($xrd) && !empty($url)) {
186                         $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
187                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
188                         if ($curlResult->isTimeout()) {
189                                 Logger::info('Probing timeout', ['url' => $url], Logger::DEBUG);
190                                 self::$istimeout = true;
191                                 return [];
192                         } elseif ($connection_error && $ssl_connection_error) {
193                                 self::$istimeout = true;
194                                 return [];
195                         }
196
197                         $xml = $curlResult->getBody();
198                         $xrd = XML::parseString($xml, true);
199                         $host_url = 'http://'.$host;
200                 }
201                 if (!is_object($xrd)) {
202                         Logger::log("No xrd object found for ".$host, Logger::DEBUG);
203                         return [];
204                 }
205
206                 $links = XML::elementToArray($xrd);
207                 if (!isset($links["xrd"]["link"])) {
208                         Logger::log("No xrd data found for ".$host, Logger::DEBUG);
209                         return [];
210                 }
211
212                 $lrdd = [];
213
214                 foreach ($links["xrd"]["link"] as $value => $link) {
215                         if (!empty($link["@attributes"])) {
216                                 $attributes = $link["@attributes"];
217                         } elseif ($value == "@attributes") {
218                                 $attributes = $link;
219                         } else {
220                                 continue;
221                         }
222
223                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
224                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
225
226                                 $lrdd[$type] = $attributes["template"];
227                         }
228                 }
229
230                 self::$baseurl = $host_url;
231
232                 Logger::log("Probing successful for ".$host, Logger::DEBUG);
233
234                 return $lrdd;
235         }
236
237         /**
238          * Perform Webfinger lookup and return DFRN data
239          *
240          * Given an email style address, perform webfinger lookup and
241          * return the resulting DFRN profile URL, or if no DFRN profile URL
242          * is located, returns an OStatus subscription template (prefixed
243          * with the string 'stat:' to identify it as on OStatus template).
244          * If this isn't an email style address just return $webbie.
245          * Return an empty string if email-style addresses but webfinger fails,
246          * or if the resultant personal XRD doesn't contain a supported
247          * subscription/friend-request attribute.
248          *
249          * amended 7/9/2011 to return an hcard which could save potentially loading
250          * a lengthy content page to scrape dfrn attributes
251          *
252          * @param string $webbie    Address that should be probed
253          * @param string $hcard_url Link to the hcard - is returned by reference
254          *
255          * @return string profile link
256          * @throws HTTPException\InternalServerErrorException
257          */
258         public static function webfingerDfrn($webbie, &$hcard_url)
259         {
260                 $profile_link = '';
261
262                 $links = self::lrdd($webbie);
263                 Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA);
264                 if (!empty($links) && is_array($links)) {
265                         foreach ($links as $link) {
266                                 if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) {
267                                         $profile_link = $link['@attributes']['href'];
268                                 }
269                                 if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) {
270                                         $profile_link = 'stat:'.$link['@attributes']['template'];
271                                 }
272                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
273                                         $hcard_url = $link['@attributes']['href'];
274                                 }
275                         }
276                 }
277                 return $profile_link;
278         }
279
280         /**
281          * Check an URI for LRDD data
282          *
283          * @param string $uri     Address that should be probed
284          *
285          * @return array uri data
286          * @throws HTTPException\InternalServerErrorException
287          */
288         public static function lrdd(string $uri)
289         {
290                 $data = self::getWebfingerArray($uri);
291                 if (empty($data)) {
292                         return [];
293                 }
294                 $webfinger = $data['webfinger'];
295
296                 if (empty($webfinger["links"])) {
297                         Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
298                         return [];
299                 }
300
301                 $data = [];
302
303                 foreach ($webfinger["links"] as $link) {
304                         $data[] = ["@attributes" => $link];
305                 }
306
307                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
308                         foreach ($webfinger["aliases"] as $alias) {
309                                 $data[] = ["@attributes" =>
310                                                         ["rel" => "alias",
311                                                                 "href" => $alias]];
312                         }
313                 }
314
315                 return $data;
316         }
317
318         /**
319          * Fetch information (protocol endpoints and user information) about a given uri
320          *
321          * @param string  $uri     Address that should be probed
322          * @param string  $network Test for this specific network
323          * @param integer $uid     User ID for the probe (only used for mails)
324          * @param boolean $cache   Use cached values?
325          *
326          * @return array uri data
327          * @throws HTTPException\InternalServerErrorException
328          * @throws \ImagickException
329          */
330         public static function uri($uri, $network = '', $uid = -1, $cache = true)
331         {
332                 $cachekey = 'Probe::uri:' . $network . ':' . $uri;
333                 if ($cache) {
334                         $result = DI::cache()->get($cachekey);
335                         if (!is_null($result)) {
336                                 return $result;
337                         }
338                 }
339
340                 if ($uid == -1) {
341                         $uid = local_user();
342                 }
343
344                 self::$istimeout = false;
345
346                 if ($network != Protocol::ACTIVITYPUB) {
347                         $data = self::detect($uri, $network, $uid);
348                         if (!is_array($data)) {
349                                 $data = [];
350                         }
351                 } else {
352                         $data = [];
353                 }
354
355                 // When the previous detection process had got a time out
356                 // we could falsely detect a Friendica profile as AP profile.
357                 if (!self::$istimeout && (empty($network) || $network == Protocol::ACTIVITYPUB)) {
358                         $ap_profile = ActivityPub::probeProfile($uri, !$cache);
359
360                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
361                                 $data = $ap_profile;
362                         } elseif (!empty($ap_profile)) {
363                                 $ap_profile['batch'] = '';
364                                 $data = array_merge($ap_profile, $data);
365                         }
366                 }
367
368                 if (!isset($data['url'])) {
369                         $data['url'] = $uri;
370                 }
371
372                 if (empty($data['photo'])) {
373                         $data['photo'] = DI::baseUrl() . '/images/person-300.jpg';
374                 }
375
376                 if (empty($data['name'])) {
377                         if (!empty($data['nick'])) {
378                                 $data['name'] = $data['nick'];
379                         }
380
381                         if (empty($data['name'])) {
382                                 $data['name'] = $data['url'];
383                         }
384                 }
385
386                 if (empty($data['nick'])) {
387                         $data['nick'] = strtolower($data['name']);
388
389                         if (strpos($data['nick'], ' ')) {
390                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
391                         }
392                 }
393
394                 if (!empty($data['baseurl']) && empty($data['gsid'])) {
395                         $data['gsid'] = GServer::getID($data['baseurl']);
396                 }
397
398                 if (empty($data['network'])) {
399                         $data['network'] = Protocol::PHANTOM;
400                 }
401
402                 // Ensure that local connections always are DFRN
403                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
404                         $data['network'] = Protocol::DFRN;
405                 }
406
407                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
408                         $data['hide'] = self::getHideStatus($data['url']);
409                 }
410
411                 $data = self::rearrangeData($data);
412
413                 // Only store into the cache if the value seems to be valid
414                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
415                         DI::cache()->set($cachekey, $data, Duration::DAY);
416                 }
417
418                 return $data;
419         }
420
421
422         /**
423          * Fetches the "hide" status from the profile
424          *
425          * @param string $url URL of the profile
426          *
427          * @return boolean "hide" status
428          */
429         private static function getHideStatus($url)
430         {
431                 $curlResult = Network::curl($url);
432                 if (!$curlResult->isSuccess()) {
433                         return false;
434                 }
435
436                 // If the file is too large then exit
437                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
438                         return false;
439                 }
440
441                 // If it isn't a HTML file then exit
442                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
443                         return false;
444                 }
445
446                 $body = $curlResult->getBody();
447
448                 $doc = new DOMDocument();
449                 @$doc->loadHTML($body);
450
451                 $xpath = new DOMXPath($doc);
452
453                 $list = $xpath->query('//meta[@name]');
454                 foreach ($list as $node) {
455                         $meta_tag = [];
456                         if ($node->attributes->length) {
457                                 foreach ($node->attributes as $attribute) {
458                                         $meta_tag[$attribute->name] = $attribute->value;
459                                 }
460                         }
461
462                         if (empty($meta_tag['content'])) {
463                                 continue;
464                         }
465
466                         $content = strtolower(trim($meta_tag['content']));
467
468                         switch (strtolower(trim($meta_tag['name']))) {
469                                 case 'dfrn-global-visibility':
470                                         if ($content == 'false') {
471                                                 return true;
472                                         }
473                                         break;
474                                 case 'robots':
475                                         if (strpos($content, 'noindex') !== false) {
476                                                 return true;
477                                         }
478                                         break;
479                         }
480                 }
481
482                 return false;
483         }
484
485         /**
486          * Fetch the "subscribe" and add it to the result
487          *
488          * @param array $result
489          * @param array $webfinger
490          * @return array result
491          */
492         private static function getSubscribeLink(array $result, array $webfinger)
493         {
494                 if (empty($webfinger['links'])) {
495                         return $result;
496                 }
497
498                 foreach ($webfinger['links'] as $link) {
499                         if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
500                                 $result['subscribe'] = $link['template'];
501                         }
502                 }
503
504                 return $result;
505         }
506
507         /**
508          * Get webfinger data from a given URI
509          *
510          * @param string $uri
511          * @return array Webfinger array
512          */
513         private static function getWebfingerArray(string $uri)
514         {
515                 $parts = parse_url($uri);
516
517                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
518                         $host = $parts['host'];
519                         if (!empty($parts['port'])) {
520                                 $host .= ':'.$parts['port'];
521                         }
522
523                         $baseurl = $parts['scheme'] . '://' . $host;
524
525                         $nick = '';
526                         $addr = '';
527
528                         $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
529                         if (!empty($path_parts)) {
530                                 $nick = ltrim(end($path_parts), '@');
531                                 // When the last part of the URI is numeric then it is most likely an ID and not a nick name
532                                 if (!is_numeric($nick)) {
533                                         $addr = $nick."@".$host;
534                                 } else {
535                                         $nick = '';
536                                 }
537                         }
538
539                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
540                         if (empty($webfinger)) {
541                                 $lrdd = self::hostMeta($host);
542                         }
543
544                         if (empty($webfinger) && empty($lrdd)) {
545                                 while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
546                                         $host .= "/".array_shift($path_parts);
547                                         $baseurl = $parts['scheme'] . '://' . $host;
548
549                                         if (!empty($nick)) {
550                                                 $addr = $nick."@".$host;
551                                         }
552
553                                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
554                                         if (empty($webfinger)) {
555                                                 $lrdd = self::hostMeta($host);
556                                         }
557                                 }
558
559                                 if (empty($lrdd) && empty($webfinger)) {
560                                         return [];
561                                 }
562                         }
563                 } elseif (strstr($uri, '@')) {
564                         // Remove "acct:" from the URI
565                         $uri = str_replace('acct:', '', $uri);
566
567                         $host = substr($uri, strpos($uri, '@') + 1);
568                         $nick = substr($uri, 0, strpos($uri, '@'));
569                         $addr = $uri;
570
571                         $webfinger = self::getWebfinger('https://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
572                         if (self::$istimeout) {
573                                 return [];
574                         }
575
576                         if (empty($webfinger)) {
577                                 $webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
578                                 if (self::$istimeout) {
579                                         return [];
580                                 }
581                         } else {
582                                 $baseurl = 'https://' . $host;
583                         }
584
585                         if (empty($webfinger)) {
586                                 $lrdd = self::hostMeta($host);
587                                 if (self::$istimeout) {
588                                         return [];
589                                 }
590                                 $baseurl = self::$baseurl;
591                         } else {
592                                 $baseurl = 'http://' . $host;
593                         }
594                 } else {
595                         Logger::info('URI was not detectable', ['uri' => $uri]);
596                         return [];
597                 }
598
599                 if (empty($webfinger)) {
600                         foreach ($lrdd as $type => $template) {
601                                 if ($webfinger) {
602                                         continue;
603                                 }
604
605                                 $webfinger = self::getWebfinger($template, $type, $uri, $addr);
606                         }
607                 }
608
609                 if (empty($webfinger)) {
610                         return [];
611                 }
612
613                 if ($webfinger['detected'] == $addr) {
614                         $webfinger['nick'] = $nick;
615                         $webfinger['addr'] = $addr;
616                 }
617
618                 $webfinger['baseurl'] = $baseurl;
619
620                 return $webfinger;
621         }
622
623         /**
624          * Perform network request for webfinger data
625          *
626          * @param string $template
627          * @param string $type
628          * @param string $uri
629          * @param string $addr
630          * @return array webfinger results
631          */
632         private static function getWebfinger(string $template, string $type, string $uri, string $addr)
633         {
634                 // First try the address because this is the primary purpose of webfinger
635                 if (!empty($addr)) {
636                         $detected = $addr;
637                         $path = str_replace('{uri}', urlencode("acct:" . $addr), $template);
638                         $webfinger = self::webfinger($path, $type);
639                         if (self::$istimeout) {
640                                 return [];
641                         }
642                 }
643
644                 // Then try the URI
645                 if (empty($webfinger) && $uri != $addr) {
646                         $detected = $uri;
647                         $path = str_replace('{uri}', urlencode($uri), $template);
648                         $webfinger = self::webfinger($path, $type);
649                         if (self::$istimeout) {
650                                 return [];
651                         }
652                 }
653
654                 if (empty($webfinger)) {
655                         return [];
656                 }
657
658                 return ['webfinger' => $webfinger, 'detected' => $detected];
659         }
660
661         /**
662          * Fetch information (protocol endpoints and user information) about a given uri
663          *
664          * This function is only called by the "uri" function that adds caching and rearranging of data.
665          *
666          * @param string  $uri     Address that should be probed
667          * @param string  $network Test for this specific network
668          * @param integer $uid     User ID for the probe (only used for mails)
669          *
670          * @return array uri data
671          * @throws HTTPException\InternalServerErrorException
672          */
673         private static function detect($uri, $network, $uid)
674         {
675                 $hookData = [
676                         'uri'     => $uri,
677                         'network' => $network,
678                         'uid'     => $uid,
679                         'result'  => [],
680                 ];
681
682                 Hook::callAll('probe_detect', $hookData);
683
684                 if ($hookData['result']) {
685                         if (!is_array($hookData['result'])) {
686                                 return [];
687                         } else {
688                                 return $hookData['result'];
689                         }
690                 }
691
692                 $parts = parse_url($uri);
693
694                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
695                         if ($parts['host'] == 'twitter.com') {
696                                 return self::twitter($uri);
697                         }
698                 } elseif (strstr($uri, '@')) {
699                         // If the URI starts with "mailto:" then jump directly to the mail detection
700                         if (strpos($uri, 'mailto:') !== false) {
701                                 $uri = str_replace('mailto:', '', $uri);
702                                 return self::mail($uri, $uid);
703                         }
704
705                         if ($network == Protocol::MAIL) {
706                                 return self::mail($uri, $uid);
707                         }
708
709                         if (strpos($uri, '@twitter.com')) {
710                                 return self::twitter($uri);
711                         }
712                 } else {
713                         Logger::info('URI was not detectable', ['uri' => $uri]);
714                         return [];
715                 }
716
717                 Logger::info('Probing start', ['uri' => $uri]);
718
719                 $data = self::getWebfingerArray($uri);
720                 if (empty($data)) {
721                         if (!empty($parts['scheme'])) {
722                                 return self::feed($uri);
723                         } elseif (!empty($uid)) {
724                                 return self::mail($uri, $uid);
725                         } else {
726                                 return [];
727                         }
728                 }
729
730                 $webfinger = $data['webfinger'];
731                 $nick = $data['nick'] ?? '';
732                 $addr = $data['addr'] ?? '';
733                 $baseurl = $data['baseurl'] ?? '';
734
735                 $result = [];
736
737                 if (in_array($network, ["", Protocol::DFRN])) {
738                         $result = self::dfrn($webfinger);
739                 }
740                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
741                         $result = self::diaspora($webfinger);
742                 }
743                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
744                         $result = self::ostatus($webfinger);
745                 }
746                 if (in_array($network, ['', Protocol::ZOT])) {
747                         $result = self::zot($webfinger, $result, $baseurl);
748                 }
749                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
750                         $result = self::pumpio($webfinger, $addr);
751                 }
752                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
753                         $result = self::feed($uri);
754                 } else {
755                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
756                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
757                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
758                                 $result["nick"] = $nick;
759                         }
760
761                         if (empty($result["addr"]) && ($addr != "")) {
762                                 $result["addr"] = $addr;
763                         }
764                 }
765
766                 $result = self::getSubscribeLink($result, $webfinger);
767
768                 if (empty($result["network"])) {
769                         $result["network"] = Protocol::PHANTOM;
770                 }
771
772                 if (empty($result['baseurl']) && !empty($baseurl)) {
773                         $result['baseurl'] = $baseurl;
774                 }
775
776                 if (empty($result["url"])) {
777                         $result["url"] = $uri;
778                 }
779
780                 Logger::info('Probing done', ['uri' => $uri, 'network' => $result["network"]]);
781
782                 return $result;
783         }
784
785         /**
786          * Check for Zot contact
787          *
788          * @param array $webfinger Webfinger data
789          * @param array $data      previously probed data
790          *
791          * @return array Zot data
792          * @throws HTTPException\InternalServerErrorException
793          */
794         private static function zot($webfinger, $data, $baseurl)
795         {
796                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
797                         foreach ($webfinger["aliases"] as $alias) {
798                                 if (substr($alias, 0, 5) == 'acct:') {
799                                         $data["addr"] = substr($alias, 5);
800                                 }
801                         }
802                 }
803
804                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
805                         $data["addr"] = substr($webfinger["subject"], 5);
806                 }
807
808                 $zot_url = '';
809                 foreach ($webfinger['links'] as $link) {
810                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
811                                 $zot_url = $link['href'];
812                         }
813                 }
814
815                 if (empty($zot_url) && !empty($data['addr']) && !empty($baseurl)) {
816                         $condition = ['nurl' => Strings::normaliseLink($baseurl), 'platform' => ['hubzilla']];
817                         if (!DBA::exists('gserver', $condition)) {
818                                 return $data;
819                         }
820                         $zot_url = $baseurl . '/.well-known/zot-info?address=' . $data['addr'];
821                 }
822
823                 if (empty($zot_url)) {
824                         return $data;
825                 }
826
827                 $data = self::pollZot($zot_url, $data);
828
829                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
830                         foreach ($webfinger['aliases'] as $alias) {
831                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
832                                         $data['alias'] = $alias;
833                                 }
834                         }
835                 }
836
837                 return $data;
838         }
839
840         public static function pollZot($url, $data)
841         {
842                 $curlResult = Network::curl($url);
843                 if ($curlResult->isTimeout()) {
844                         return $data;
845                 }
846                 $content = $curlResult->getBody();
847                 if (!$content) {
848                         return $data;
849                 }
850
851                 $json = json_decode($content, true);
852                 if (!is_array($json)) {
853                         return $data;
854                 }
855
856                 if (empty($data['network'])) {
857                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
858                                 $data['network'] = Protocol::ZOT;
859                         } elseif (!isset($json['protocols'])) {
860                                 $data['network'] = Protocol::ZOT;
861                         }
862                 }
863
864                 if (!empty($json['guid']) && empty($data['guid'])) {
865                         $data['guid'] = $json['guid'];
866                 }
867                 if (!empty($json['key']) && empty($data['pubkey'])) {
868                         $data['pubkey'] = $json['key'];
869                 }
870                 if (!empty($json['name'])) {
871                         $data['name'] = $json['name'];
872                 }
873                 if (!empty($json['photo'])) {
874                         $data['photo'] = $json['photo'];
875                         if (!empty($json['photo_updated'])) {
876                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
877                         }
878                 }
879                 if (!empty($json['address'])) {
880                         $data['addr'] = $json['address'];
881                 }
882                 if (!empty($json['url'])) {
883                         $data['url'] = $json['url'];
884                 }
885                 if (!empty($json['connections_url'])) {
886                         $data['poco'] = $json['connections_url'];
887                 }
888                 if (isset($json['searchable'])) {
889                         $data['hide'] = !$json['searchable'];
890                 }
891                 if (!empty($json['public_forum'])) {
892                         $data['community'] = $json['public_forum'];
893                         $data['account-type'] = Contact::PAGE_COMMUNITY;
894                 }
895
896                 if (!empty($json['profile'])) {
897                         $profile = $json['profile'];
898                         if (!empty($profile['description'])) {
899                                 $data['about'] = $profile['description'];
900                         }
901                         if (!empty($profile['keywords'])) {
902                                 $keywords = implode(', ', $profile['keywords']);
903                                 if (!empty($keywords)) {
904                                         $data['keywords'] = $keywords;
905                                 }
906                         }
907
908                         $loc = [];
909                         if (!empty($profile['region'])) {
910                                 $loc['region'] = $profile['region'];
911                         }
912                         if (!empty($profile['country'])) {
913                                 $loc['country-name'] = $profile['country'];
914                         }
915                         $location = Profile::formatLocation($loc);
916                         if (!empty($location)) {
917                                 $data['location'] = $location;
918                         }
919                 }
920
921                 return $data;
922         }
923
924         /**
925          * Perform a webfinger request.
926          *
927          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
928          *
929          * @param string $url  Address that should be probed
930          * @param string $type type
931          *
932          * @return array webfinger data
933          * @throws HTTPException\InternalServerErrorException
934          */
935         public static function webfinger($url, $type)
936         {
937                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
938
939                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
940                 if ($curlResult->isTimeout()) {
941                         self::$istimeout = true;
942                         return [];
943                 }
944                 $data = $curlResult->getBody();
945
946                 $webfinger = json_decode($data, true);
947                 if (!empty($webfinger)) {
948                         if (!isset($webfinger["links"])) {
949                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
950                                 return [];
951                         }
952                         return $webfinger;
953                 }
954
955                 // If it is not JSON, maybe it is XML
956                 $xrd = XML::parseString($data, true);
957                 if (!is_object($xrd)) {
958                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
959                         return [];
960                 }
961
962                 $xrd_arr = XML::elementToArray($xrd);
963                 if (!isset($xrd_arr["xrd"]["link"])) {
964                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
965                         return [];
966                 }
967
968                 $webfinger = [];
969
970                 if (!empty($xrd_arr["xrd"]["subject"])) {
971                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
972                 }
973
974                 if (!empty($xrd_arr["xrd"]["alias"])) {
975                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
976                 }
977
978                 $webfinger["links"] = [];
979
980                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
981                         if (!empty($data["@attributes"])) {
982                                 $attributes = $data["@attributes"];
983                         } elseif ($value == "@attributes") {
984                                 $attributes = $data;
985                         } else {
986                                 continue;
987                         }
988
989                         $webfinger["links"][] = $attributes;
990                 }
991                 return $webfinger;
992         }
993
994         /**
995          * Poll the Friendica specific noscrape page.
996          *
997          * "noscrape" is a faster alternative to fetch the data from the hcard.
998          * This functionality was originally created for the directory.
999          *
1000          * @param string $noscrape_url Link to the noscrape page
1001          * @param array  $data         The already fetched data
1002          *
1003          * @return array noscrape data
1004          * @throws HTTPException\InternalServerErrorException
1005          */
1006         private static function pollNoscrape($noscrape_url, $data)
1007         {
1008                 $curlResult = Network::curl($noscrape_url);
1009                 if ($curlResult->isTimeout()) {
1010                         self::$istimeout = true;
1011                         return [];
1012                 }
1013                 $content = $curlResult->getBody();
1014                 if (!$content) {
1015                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
1016                         return [];
1017                 }
1018
1019                 $json = json_decode($content, true);
1020                 if (!is_array($json)) {
1021                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
1022                         return [];
1023                 }
1024
1025                 if (!empty($json["fn"])) {
1026                         $data["name"] = $json["fn"];
1027                 }
1028
1029                 if (!empty($json["addr"])) {
1030                         $data["addr"] = $json["addr"];
1031                 }
1032
1033                 if (!empty($json["nick"])) {
1034                         $data["nick"] = $json["nick"];
1035                 }
1036
1037                 if (!empty($json["guid"])) {
1038                         $data["guid"] = $json["guid"];
1039                 }
1040
1041                 if (!empty($json["comm"])) {
1042                         $data["community"] = $json["comm"];
1043                 }
1044
1045                 if (!empty($json["tags"])) {
1046                         $keywords = implode(", ", $json["tags"]);
1047                         if ($keywords != "") {
1048                                 $data["keywords"] = $keywords;
1049                         }
1050                 }
1051
1052                 $location = Profile::formatLocation($json);
1053                 if ($location) {
1054                         $data["location"] = $location;
1055                 }
1056
1057                 if (!empty($json["about"])) {
1058                         $data["about"] = $json["about"];
1059                 }
1060
1061                 if (!empty($json["key"])) {
1062                         $data["pubkey"] = $json["key"];
1063                 }
1064
1065                 if (!empty($json["photo"])) {
1066                         $data["photo"] = $json["photo"];
1067                 }
1068
1069                 if (!empty($json["dfrn-request"])) {
1070                         $data["request"] = $json["dfrn-request"];
1071                 }
1072
1073                 if (!empty($json["dfrn-confirm"])) {
1074                         $data["confirm"] = $json["dfrn-confirm"];
1075                 }
1076
1077                 if (!empty($json["dfrn-notify"])) {
1078                         $data["notify"] = $json["dfrn-notify"];
1079                 }
1080
1081                 if (!empty($json["dfrn-poll"])) {
1082                         $data["poll"] = $json["dfrn-poll"];
1083                 }
1084
1085                 if (isset($json["hide"])) {
1086                         $data["hide"] = (bool)$json["hide"];
1087                 } else {
1088                         $data["hide"] = false;
1089                 }
1090
1091                 return $data;
1092         }
1093
1094         /**
1095          * Check for valid DFRN data
1096          *
1097          * @param array $data DFRN data
1098          *
1099          * @return int Number of errors
1100          */
1101         public static function validDfrn($data)
1102         {
1103                 $errors = 0;
1104                 if (!isset($data['key'])) {
1105                         $errors ++;
1106                 }
1107                 if (!isset($data['dfrn-request'])) {
1108                         $errors ++;
1109                 }
1110                 if (!isset($data['dfrn-confirm'])) {
1111                         $errors ++;
1112                 }
1113                 if (!isset($data['dfrn-notify'])) {
1114                         $errors ++;
1115                 }
1116                 if (!isset($data['dfrn-poll'])) {
1117                         $errors ++;
1118                 }
1119                 return $errors;
1120         }
1121
1122         /**
1123          * Fetch data from a DFRN profile page and via "noscrape"
1124          *
1125          * @param string $profile_link Link to the profile page
1126          *
1127          * @return array profile data
1128          * @throws HTTPException\InternalServerErrorException
1129          * @throws \ImagickException
1130          */
1131         public static function profile($profile_link)
1132         {
1133                 $data = [];
1134
1135                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
1136
1137                 // Fetch data via noscrape - this is faster
1138                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1139                 $data = self::pollNoscrape($noscrape_url, $data);
1140
1141                 if (!isset($data["notify"])
1142                         || !isset($data["confirm"])
1143                         || !isset($data["request"])
1144                         || !isset($data["poll"])
1145                         || !isset($data["name"])
1146                         || !isset($data["photo"])
1147                 ) {
1148                         $data = self::pollHcard($profile_link, $data, true);
1149                 }
1150
1151                 $prof_data = [];
1152
1153                 if (empty($data["addr"]) || empty($data["nick"])) {
1154                         $probe_data = self::uri($profile_link);
1155                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1156                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1157                 }
1158
1159                 $prof_data["addr"]         = $data["addr"];
1160                 $prof_data["nick"]         = $data["nick"];
1161                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1162                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1163                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1164                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1165                 $prof_data["photo"]        = $data['photo']   ?? null;
1166                 $prof_data["fn"]           = $data['name']    ?? null;
1167                 $prof_data["key"]          = $data['pubkey']  ?? null;
1168
1169                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
1170
1171                 return $prof_data;
1172         }
1173
1174         /**
1175          * Check for DFRN contact
1176          *
1177          * @param array $webfinger Webfinger data
1178          *
1179          * @return array DFRN data
1180          * @throws HTTPException\InternalServerErrorException
1181          */
1182         private static function dfrn($webfinger)
1183         {
1184                 $hcard_url = "";
1185                 $data = [];
1186                 // The array is reversed to take into account the order of preference for same-rel links
1187                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1188                 foreach (array_reverse($webfinger["links"]) as $link) {
1189                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1190                                 $data["network"] = Protocol::DFRN;
1191                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1192                                 $data["poll"] = $link["href"];
1193                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1194                                 $data["url"] = $link["href"];
1195                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1196                                 $hcard_url = $link["href"];
1197                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1198                                 $data["poco"] = $link["href"];
1199                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1200                                 $data["photo"] = $link["href"];
1201                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1202                                 $data["baseurl"] = trim($link["href"], '/');
1203                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1204                                 $data["guid"] = $link["href"];
1205                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1206                                 $data["pubkey"] = base64_decode($link["href"]);
1207
1208                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1209                                 if (strstr($data["pubkey"], 'RSA ')) {
1210                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1211                                 }
1212                         }
1213                 }
1214
1215                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1216                         foreach ($webfinger["aliases"] as $alias) {
1217                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1218                                         $data["url"] = $alias;
1219                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1220                                         $data["alias"] = $alias;
1221                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1222                                         $data["addr"] = substr($alias, 5);
1223                                 }
1224                         }
1225                 }
1226
1227                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1228                         $data["addr"] = substr($webfinger["subject"], 5);
1229                 }
1230
1231                 if (!isset($data["network"]) || ($hcard_url == "")) {
1232                         return [];
1233                 }
1234
1235                 // Fetch data via noscrape - this is faster
1236                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1237                 $data = self::pollNoscrape($noscrape_url, $data);
1238
1239                 if (isset($data["notify"])
1240                         && isset($data["confirm"])
1241                         && isset($data["request"])
1242                         && isset($data["poll"])
1243                         && isset($data["name"])
1244                         && isset($data["photo"])
1245                 ) {
1246                         return $data;
1247                 }
1248
1249                 $data = self::pollHcard($hcard_url, $data, true);
1250
1251                 return $data;
1252         }
1253
1254         /**
1255          * Poll the hcard page (Diaspora and Friendica specific)
1256          *
1257          * @param string  $hcard_url Link to the hcard page
1258          * @param array   $data      The already fetched data
1259          * @param boolean $dfrn      Poll DFRN specific data
1260          *
1261          * @return array hcard data
1262          * @throws HTTPException\InternalServerErrorException
1263          */
1264         private static function pollHcard($hcard_url, $data, $dfrn = false)
1265         {
1266                 $curlResult = Network::curl($hcard_url);
1267                 if ($curlResult->isTimeout()) {
1268                         self::$istimeout = true;
1269                         return [];
1270                 }
1271                 $content = $curlResult->getBody();
1272                 if (!$content) {
1273                         return [];
1274                 }
1275
1276                 $doc = new DOMDocument();
1277                 if (!@$doc->loadHTML($content)) {
1278                         return [];
1279                 }
1280
1281                 $xpath = new DomXPath($doc);
1282
1283                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1284                 if (!is_object($vcards)) {
1285                         return [];
1286                 }
1287
1288                 if (!isset($data["baseurl"])) {
1289                         $data["baseurl"] = "";
1290                 }
1291
1292                 if ($vcards->length > 0) {
1293                         $vcard = $vcards->item(0);
1294
1295                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1296                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1297                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1298                         if (($search->length > 0) && empty($data["guid"])) {
1299                                 $data["guid"] = $search->item(0)->nodeValue;
1300                         }
1301
1302                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1303                         if ($search->length > 0) {
1304                                 $data["nick"] = $search->item(0)->nodeValue;
1305                         }
1306
1307                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1308                         if ($search->length > 0) {
1309                                 $data["name"] = $search->item(0)->nodeValue;
1310                         }
1311
1312                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1313                         if ($search->length > 0) {
1314                                 $data["searchable"] = $search->item(0)->nodeValue;
1315                         }
1316
1317                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1318                         if ($search->length > 0) {
1319                                 $data["pubkey"] = $search->item(0)->nodeValue;
1320                                 if (strstr($data["pubkey"], 'RSA ')) {
1321                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1322                                 }
1323                         }
1324
1325                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1326                         if ($search->length > 0) {
1327                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1328                         }
1329                 }
1330
1331                 $avatar = [];
1332                 if (!empty($vcard)) {
1333                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1334                         foreach ($photos as $photo) {
1335                                 $attr = [];
1336                                 foreach ($photo->attributes as $attribute) {
1337                                         $attr[$attribute->name] = trim($attribute->value);
1338                                 }
1339
1340                                 if (isset($attr["src"]) && isset($attr["width"])) {
1341                                         $avatar[$attr["width"]] = $attr["src"];
1342                                 }
1343
1344                                 // We don't have a width. So we just take everything that we got.
1345                                 // This is a Hubzilla workaround which doesn't send a width.
1346                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1347                                         $avatar[] = $attr["src"];
1348                                 }
1349                         }
1350                 }
1351
1352                 if (sizeof($avatar)) {
1353                         ksort($avatar);
1354                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1355                 }
1356
1357                 if ($dfrn) {
1358                         // Poll DFRN specific data
1359                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1360                         if ($search->length > 0) {
1361                                 foreach ($search as $link) {
1362                                         //$data["request"] = $search->item(0)->nodeValue;
1363                                         $attr = [];
1364                                         foreach ($link->attributes as $attribute) {
1365                                                 $attr[$attribute->name] = trim($attribute->value);
1366                                         }
1367
1368                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1369                                 }
1370                         }
1371
1372                         // Older Friendica versions had used the "uid" field differently than newer versions
1373                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1374                                 unset($data["guid"]);
1375                         }
1376                 }
1377
1378
1379                 return $data;
1380         }
1381
1382         /**
1383          * Check for Diaspora contact
1384          *
1385          * @param array $webfinger Webfinger data
1386          *
1387          * @return array Diaspora data
1388          * @throws HTTPException\InternalServerErrorException
1389          */
1390         private static function diaspora($webfinger)
1391         {
1392                 $hcard_url = "";
1393                 $data = [];
1394
1395                 // The array is reversed to take into account the order of preference for same-rel links
1396                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1397                 foreach (array_reverse($webfinger["links"]) as $link) {
1398                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1399                                 $hcard_url = $link["href"];
1400                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1401                                 $data["baseurl"] = trim($link["href"], '/');
1402                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1403                                 $data["guid"] = $link["href"];
1404                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1405                                 $data["url"] = $link["href"];
1406                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1407                                 $data["poll"] = $link["href"];
1408                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1409                                 $data["poco"] = $link["href"];
1410                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1411                                 $data["notify"] = $link["href"];
1412                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1413                                 $data["pubkey"] = base64_decode($link["href"]);
1414
1415                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1416                                 if (strstr($data["pubkey"], 'RSA ')) {
1417                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1418                                 }
1419                         }
1420                 }
1421
1422                 if (empty($data["url"]) || empty($hcard_url)) {
1423                         return [];
1424                 }
1425
1426                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1427                         foreach ($webfinger["aliases"] as $alias) {
1428                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1429                                         $data["alias"] = $alias;
1430                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1431                                         $data["addr"] = substr($alias, 5);
1432                                 }
1433                         }
1434                 }
1435
1436                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1437                         $data["addr"] = substr($webfinger["subject"], 5);
1438                 }
1439
1440                 // Fetch further information from the hcard
1441                 $data = self::pollHcard($hcard_url, $data);
1442
1443                 if (!$data) {
1444                         return [];
1445                 }
1446
1447                 if (!empty($data["url"])
1448                         && !empty($data["guid"])
1449                         && !empty($data["baseurl"])
1450                         && !empty($data["pubkey"])
1451                         && !empty($hcard_url)
1452                 ) {
1453                         $data["network"] = Protocol::DIASPORA;
1454
1455                         // The Diaspora handle must always be lowercase
1456                         if (!empty($data["addr"])) {
1457                                 $data["addr"] = strtolower($data["addr"]);
1458                         }
1459
1460                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1461                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1462                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1463                 } else {
1464                         return [];
1465                 }
1466
1467                 return $data;
1468         }
1469
1470         /**
1471          * Check for OStatus contact
1472          *
1473          * @param array $webfinger Webfinger data
1474          * @param bool  $short     Short detection mode
1475          *
1476          * @return array|bool OStatus data or "false" on error or "true" on short mode
1477          * @throws HTTPException\InternalServerErrorException
1478          */
1479         private static function ostatus($webfinger, $short = false)
1480         {
1481                 $data = [];
1482
1483                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1484                         foreach ($webfinger["aliases"] as $alias) {
1485                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1486                                         $data["addr"] = str_replace('acct:', '', $alias);
1487                                 }
1488                         }
1489                 }
1490
1491                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1492                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1493                 ) {
1494                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1495                 }
1496
1497                 if (!empty($webfinger["links"])) {
1498                         // The array is reversed to take into account the order of preference for same-rel links
1499                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1500                         foreach (array_reverse($webfinger["links"]) as $link) {
1501                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1502                                         && (($link["type"] ?? "") == "text/html")
1503                                         && ($link["href"] != "")
1504                                 ) {
1505                                         $data["url"] = $data["alias"] = $link["href"];
1506                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1507                                         $data["notify"] = $link["href"];
1508                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1509                                         $data["poll"] = $link["href"];
1510                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1511                                         $pubkey = $link["href"];
1512
1513                                         if (substr($pubkey, 0, 5) === 'data:') {
1514                                                 if (strstr($pubkey, ',')) {
1515                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1516                                                 } else {
1517                                                         $pubkey = substr($pubkey, 5);
1518                                                 }
1519                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1520                                                 $curlResult = Network::curl($pubkey);
1521                                                 if ($curlResult->isTimeout()) {
1522                                                         self::$istimeout = true;
1523                                                         return $short ? false : [];
1524                                                 }
1525                                                 $pubkey = $curlResult->getBody();
1526                                         }
1527
1528                                         $key = explode(".", $pubkey);
1529
1530                                         if (sizeof($key) >= 3) {
1531                                                 $m = Strings::base64UrlDecode($key[1]);
1532                                                 $e = Strings::base64UrlDecode($key[2]);
1533                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1534                                         }
1535                                 }
1536                         }
1537                 }
1538
1539                 if (isset($data["notify"]) && isset($data["pubkey"])
1540                         && isset($data["poll"])
1541                         && isset($data["url"])
1542                 ) {
1543                         $data["network"] = Protocol::OSTATUS;
1544                 } else {
1545                         return $short ? false : [];
1546                 }
1547
1548                 if ($short) {
1549                         return true;
1550                 }
1551
1552                 // Fetch all additional data from the feed
1553                 $curlResult = Network::curl($data["poll"]);
1554                 if ($curlResult->isTimeout()) {
1555                         self::$istimeout = true;
1556                         return [];
1557                 }
1558                 $feed = $curlResult->getBody();
1559                 $feed_data = Feed::import($feed);
1560                 if (!$feed_data) {
1561                         return [];
1562                 }
1563
1564                 if (!empty($feed_data["header"]["author-name"])) {
1565                         $data["name"] = $feed_data["header"]["author-name"];
1566                 }
1567                 if (!empty($feed_data["header"]["author-nick"])) {
1568                         $data["nick"] = $feed_data["header"]["author-nick"];
1569                 }
1570                 if (!empty($feed_data["header"]["author-avatar"])) {
1571                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1572                 }
1573                 if (!empty($feed_data["header"]["author-id"])) {
1574                         $data["alias"] = $feed_data["header"]["author-id"];
1575                 }
1576                 if (!empty($feed_data["header"]["author-location"])) {
1577                         $data["location"] = $feed_data["header"]["author-location"];
1578                 }
1579                 if (!empty($feed_data["header"]["author-about"])) {
1580                         $data["about"] = $feed_data["header"]["author-about"];
1581                 }
1582                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1583                 // So we take the value that we just fetched, although the other one worked as well
1584                 if (!empty($feed_data["header"]["author-link"])) {
1585                         $data["url"] = $feed_data["header"]["author-link"];
1586                 }
1587
1588                 if ($data["url"] == $data["alias"]) {
1589                         $data["alias"] = '';
1590                 }
1591
1592                 /// @todo Fetch location and "about" from the feed as well
1593                 return $data;
1594         }
1595
1596         /**
1597          * Fetch data from a pump.io profile page
1598          *
1599          * @param string $profile_link Link to the profile page
1600          *
1601          * @return array profile data
1602          */
1603         private static function pumpioProfileData($profile_link)
1604         {
1605                 $curlResult = Network::curl($profile_link);
1606                 if (!$curlResult->isSuccess()) {
1607                         return [];
1608                 }
1609
1610                 $doc = new DOMDocument();
1611                 if (!@$doc->loadHTML($curlResult->getBody())) {
1612                         return [];
1613                 }
1614
1615                 $xpath = new DomXPath($doc);
1616
1617                 $data = [];
1618
1619                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1620
1621                 if ($data["name"] == '') {
1622                         // This is ugly - but pump.io doesn't seem to know a better way for it
1623                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1624                         $pos = strpos($data["name"], chr(10));
1625                         if ($pos) {
1626                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1627                         }
1628                 }
1629
1630                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1631
1632                 if ($data["location"] == '') {
1633                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1634                 }
1635
1636                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1637
1638                 if ($data["about"] == '') {
1639                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1640                 }
1641
1642                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1643                 if (!$avatar) {
1644                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1645                 }
1646                 if ($avatar) {
1647                         foreach ($avatar->attributes as $attribute) {
1648                                 if ($attribute->name == "src") {
1649                                         $data["photo"] = trim($attribute->value);
1650                                 }
1651                         }
1652                 }
1653
1654                 return $data;
1655         }
1656
1657         /**
1658          * Check for pump.io contact
1659          *
1660          * @param array  $webfinger Webfinger data
1661          * @param string $addr
1662          * @return array pump.io data
1663          */
1664         private static function pumpio($webfinger, $addr)
1665         {
1666                 $data = [];
1667                 // The array is reversed to take into account the order of preference for same-rel links
1668                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1669                 foreach (array_reverse($webfinger["links"]) as $link) {
1670                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1671                                 && (($link["type"] ?? "") == "text/html")
1672                                 && ($link["href"] != "")
1673                         ) {
1674                                 $data["url"] = $link["href"];
1675                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1676                                 $data["notify"] = $link["href"];
1677                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1678                                 $data["poll"] = $link["href"];
1679                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1680                                 $data["dialback"] = $link["href"];
1681                         }
1682                 }
1683                 if (isset($data["poll"]) && isset($data["notify"])
1684                         && isset($data["dialback"])
1685                         && isset($data["url"])
1686                 ) {
1687                         // by now we use these fields only for the network type detection
1688                         // So we unset all data that isn't used at the moment
1689                         unset($data["dialback"]);
1690
1691                         $data["network"] = Protocol::PUMPIO;
1692                 } else {
1693                         return [];
1694                 }
1695
1696                 $profile_data = self::pumpioProfileData($data["url"]);
1697
1698                 if (!$profile_data) {
1699                         return [];
1700                 }
1701
1702                 $data = array_merge($data, $profile_data);
1703
1704                 if (($addr != '') && ($data['name'] != '')) {
1705                         $name = trim(str_replace($addr, '', $data['name']));
1706                         if ($name != '') {
1707                                 $data['name'] = $name;
1708                         }
1709                 }
1710
1711                 return $data;
1712         }
1713
1714         /**
1715          * Check for twitter contact
1716          *
1717          * @param string $uri
1718          *
1719          * @return array twitter data
1720          */
1721         private static function twitter($uri)
1722         {
1723                 if (preg_match('=(.*)@twitter.com=i', $uri, $matches)) {
1724                         $nick = $matches[1];
1725                 } elseif (preg_match('=https?://twitter.com/(.*)=i', $uri, $matches)) {
1726                         $nick = $matches[1];
1727                 } else {
1728                         return [];
1729                 }
1730
1731                 $data = [];
1732                 $data['url'] = 'https://twitter.com/' . $nick;
1733                 $data['addr'] = $nick . '@twitter.com';
1734                 $data['nick'] = $data['name'] = $nick;
1735                 $data['network'] = Protocol::TWITTER;
1736                 $data['baseurl'] = 'https://twitter.com';
1737
1738                 return $data;
1739         }
1740
1741         /**
1742          * Checks HTML page for RSS feed link
1743          *
1744          * @param string $url  Page link
1745          * @param string $body Page body string
1746          * @return string|false Feed link or false if body was invalid HTML document
1747          */
1748         public static function getFeedLink(string $url, string $body)
1749         {
1750                 $doc = new DOMDocument();
1751                 if (!@$doc->loadHTML($body)) {
1752                         return false;
1753                 }
1754
1755                 $xpath = new DOMXPath($doc);
1756
1757                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1758
1759                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1760
1761                 return $feedUrl;
1762         }
1763
1764         /**
1765          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1766          *
1767          * Loosely based on RFC 1808
1768          *
1769          * @see https://tools.ietf.org/html/rfc1808
1770          *
1771          * @param string   $href  The potential relative href found in the HTML document
1772          * @param string   $base  The HTML document URL
1773          * @param DOMXPath $xpath The HTML document XPath
1774          * @return string
1775          */
1776         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
1777         {
1778                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1779                         return $href;
1780                 }
1781
1782                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1783
1784                 $baseParts = parse_url($base);
1785
1786                 // Naked domain case (scheme://basehost)
1787                 $path = $baseParts['path'] ?? '/';
1788
1789                 // Remove the filename part of the path if it exists (/base/path/file)
1790                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1791
1792                 $hrefParts = parse_url($href);
1793
1794                 // Root path case (/path) including relative scheme case (//host/path)
1795                 if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1796                         $path = $hrefParts['path'];
1797                 } else {
1798                         $path = $path . '/' . $hrefParts['path'];
1799
1800                         // Resolve arbitrary relative path
1801                         // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1802                         $parts = array_filter(explode('/', $path), 'strlen');
1803                         $absolutes = array();
1804                         foreach ($parts as $part) {
1805                                 if ('.' == $part) continue;
1806                                 if ('..' == $part) {
1807                                         array_pop($absolutes);
1808                                 } else {
1809                                         $absolutes[] = $part;
1810                                 }
1811                         }
1812
1813                         $path = '/' . implode('/', $absolutes);
1814                 }
1815
1816                 // Relative scheme case (//host/path)
1817                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1818                 $baseParts['path'] = $path;
1819                 unset($baseParts['query']);
1820                 unset($baseParts['fragment']);
1821
1822                 return Network::unparseURL($baseParts);
1823         }
1824
1825         /**
1826          * Check for feed contact
1827          *
1828          * @param string  $url   Profile link
1829          * @param boolean $probe Do a probe if the page contains a feed link
1830          *
1831          * @return array feed data
1832          * @throws HTTPException\InternalServerErrorException
1833          */
1834         private static function feed($url, $probe = true)
1835         {
1836                 $curlResult = Network::curl($url);
1837                 if ($curlResult->isTimeout()) {
1838                         self::$istimeout = true;
1839                         return [];
1840                 }
1841                 $feed = $curlResult->getBody();
1842                 $feed_data = Feed::import($feed);
1843
1844                 if (!$feed_data) {
1845                         if (!$probe) {
1846                                 return [];
1847                         }
1848
1849                         $feed_url = self::getFeedLink($url, $feed);
1850
1851                         if (!$feed_url) {
1852                                 return [];
1853                         }
1854
1855                         return self::feed($feed_url, false);
1856                 }
1857
1858                 if (!empty($feed_data["header"]["author-name"])) {
1859                         $data["name"] = $feed_data["header"]["author-name"];
1860                 }
1861
1862                 if (!empty($feed_data["header"]["author-nick"])) {
1863                         $data["nick"] = $feed_data["header"]["author-nick"];
1864                 }
1865
1866                 if (!empty($feed_data["header"]["author-avatar"])) {
1867                         $data["photo"] = $feed_data["header"]["author-avatar"];
1868                 }
1869
1870                 if (!empty($feed_data["header"]["author-id"])) {
1871                         $data["alias"] = $feed_data["header"]["author-id"];
1872                 }
1873
1874                 $data["url"] = $url;
1875                 $data["poll"] = $url;
1876
1877                 if (!empty($feed_data["header"]["author-link"])) {
1878                         $data["baseurl"] = $feed_data["header"]["author-link"];
1879                 } else {
1880                         $data["baseurl"] = $data["url"];
1881                 }
1882
1883                 $data["network"] = Protocol::FEED;
1884
1885                 return $data;
1886         }
1887
1888         /**
1889          * Check for mail contact
1890          *
1891          * @param string  $uri Profile link
1892          * @param integer $uid User ID
1893          *
1894          * @return array mail data
1895          * @throws \Exception
1896          */
1897         private static function mail($uri, $uid)
1898         {
1899                 if (!Network::isEmailDomainValid($uri)) {
1900                         return [];
1901                 }
1902
1903                 if ($uid == 0) {
1904                         return [];
1905                 }
1906
1907                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1908
1909                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1910                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1911                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1912
1913                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1914                         return [];
1915                 }
1916
1917                 $mailbox = Email::constructMailboxName($mailacct);
1918                 $password = '';
1919                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1920                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1921                 if (!$mbox) {
1922                         return [];
1923                 }
1924
1925                 $msgs = Email::poll($mbox, $uri);
1926                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1927
1928                 if (!count($msgs)) {
1929                         return [];
1930                 }
1931
1932                 $phost = substr($uri, strpos($uri, '@') + 1);
1933
1934                 $data = [];
1935                 $data["addr"]    = $uri;
1936                 $data["network"] = Protocol::MAIL;
1937                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1938                 $data["nick"]    = $data["name"];
1939                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1940                 $data["url"]     = 'mailto:'.$uri;
1941                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1942                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1943
1944                 $x = Email::messageMeta($mbox, $msgs[0]);
1945                 if (stristr($x[0]->from, $uri)) {
1946                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1947                 } elseif (stristr($x[0]->to, $uri)) {
1948                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1949                 }
1950                 if (isset($adr)) {
1951                         foreach ($adr as $feadr) {
1952                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1953                                         &&(strcasecmp($feadr->host, $phost) == 0)
1954                                         && (strlen($feadr->personal))
1955                                 ) {
1956                                         $personal = imap_mime_header_decode($feadr->personal);
1957                                         $data["name"] = "";
1958                                         foreach ($personal as $perspart) {
1959                                                 if ($perspart->charset != "default") {
1960                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1961                                                 } else {
1962                                                         $data["name"] .= $perspart->text;
1963                                                 }
1964                                         }
1965
1966                                         $data["name"] = Strings::escapeTags($data["name"]);
1967                                 }
1968                         }
1969                 }
1970                 if (!empty($mbox)) {
1971                         imap_close($mbox);
1972                 }
1973                 return $data;
1974         }
1975
1976         /**
1977          * Mix two paths together to possibly fix missing parts
1978          *
1979          * @param string $avatar Path to the avatar
1980          * @param string $base   Another path that is hopefully complete
1981          *
1982          * @return string fixed avatar path
1983          * @throws \Exception
1984          */
1985         public static function fixAvatar($avatar, $base)
1986         {
1987                 $base_parts = parse_url($base);
1988
1989                 // Remove all parts that could create a problem
1990                 unset($base_parts['path']);
1991                 unset($base_parts['query']);
1992                 unset($base_parts['fragment']);
1993
1994                 $avatar_parts = parse_url($avatar);
1995
1996                 // Now we mix them
1997                 $parts = array_merge($base_parts, $avatar_parts);
1998
1999                 // And put them together again
2000                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
2001                 $host     = isset($parts['host'])     ? $parts['host']           : '';
2002                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
2003                 $path     = isset($parts['path'])     ? $parts['path']           : '';
2004                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
2005                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
2006
2007                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
2008
2009                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
2010
2011                 return $fixed;
2012         }
2013 }