]> git.mxchange.org Git - friendica.git/blob - src/Network/Probe.php
19fc135d7b4b8f9fbe0ac3dd004be56983edaec9
[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         private static $baseurl;
51         private static $istimeout;
52
53         /**
54          * Remove stuff from an URI that doesn't belong there
55          *
56          * @param string $URI
57          * @return string Cleaned URI
58          */
59         public static function cleanURI(string $URI)
60         {
61                 // At first remove leading and trailing junk
62                 $URI = trim($URI, "@#?:/ \t\n\r\0\x0B");
63
64                 $parts = parse_url($URI);
65
66                 if (empty($parts['scheme'])) {
67                         return $URI;
68                 }
69
70                 // Remove the URL fragment, since these shouldn't be part of any profile URL
71                 unset($parts['fragment']);
72
73                 $URI = Network::unparseURL($parts);
74
75                 return $URI;
76         }
77
78         /**
79          * Rearrange the array so that it always has the same order
80          *
81          * @param array $data Unordered data
82          *
83          * @return array Ordered data
84          */
85         private static function rearrangeData($data)
86         {
87                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "account-type",
88                                 "community", "keywords", "location", "about", "hide",
89                                 "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
90                                 "following", "followers", "inbox", "outbox", "sharedinbox",
91                                 "priority", "network", "pubkey", "baseurl", "gsid"];
92
93                 $newdata = [];
94                 foreach ($fields as $field) {
95                         if (isset($data[$field])) {
96                                 if (in_array($field, ["gsid", "hide", "account-type"])) {
97                                         $newdata[$field] = (int)$data[$field];
98                                 } else {        
99                                         $newdata[$field] = $data[$field];
100                                 }
101                         } elseif ($field != "gsid") {
102                                 $newdata[$field] = "";
103                         } else {
104                                 $newdata[$field] = null;
105                         }
106                 }
107
108                 // We don't use the "priority" field anymore and replace it with a dummy.
109                 $newdata["priority"] = 0;
110
111                 return $newdata;
112         }
113
114         /**
115          * Check if the hostname belongs to the own server
116          *
117          * @param string $host The hostname that is to be checked
118          *
119          * @return bool Does the testes hostname belongs to the own server?
120          */
121         private static function ownHost($host)
122         {
123                 $own_host = DI::baseUrl()->getHostname();
124
125                 $parts = parse_url($host);
126
127                 if (!isset($parts['scheme'])) {
128                         $parts = parse_url('http://'.$host);
129                 }
130
131                 if (!isset($parts['host'])) {
132                         return false;
133                 }
134                 return $parts['host'] == $own_host;
135         }
136
137         /**
138          * Probes for webfinger path via "host-meta"
139          *
140          * We have to check if the servers in the future still will offer this.
141          * It seems as if it was dropped from the standard.
142          *
143          * @param string $host The host part of an url
144          *
145          * @return array with template and type of the webfinger template for JSON or XML
146          * @throws HTTPException\InternalServerErrorException
147          */
148         private static function hostMeta($host)
149         {
150                 // Reset the static variable
151                 self::$baseurl = '';
152
153                 // Handles the case when the hostname contains the scheme
154                 if (!parse_url($host, PHP_URL_SCHEME)) {
155                         $ssl_url = "https://" . $host . "/.well-known/host-meta";
156                         $url = "http://" . $host . "/.well-known/host-meta";
157                 } else {
158                         $ssl_url = $host . "/.well-known/host-meta";
159                         $url = '';
160                 }
161
162                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
163
164                 Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
165                 $xrd = null;
166
167                 $curlResult = Network::curl($ssl_url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
168                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
169                 if ($curlResult->isSuccess()) {
170                         $xml = $curlResult->getBody();
171                         $xrd = XML::parseString($xml, true);
172                         if (!empty($url)) {
173                                 $host_url = 'https://' . $host;
174                         } else {
175                                 $host_url = $host;
176                         }
177                 } elseif ($curlResult->isTimeout()) {
178                         Logger::info('Probing timeout', ['url' => $ssl_url], Logger::DEBUG);
179                         self::$istimeout = true;
180                         return [];
181                 }
182
183                 if (!is_object($xrd) && !empty($url)) {
184                         $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
185                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
186                         if ($curlResult->isTimeout()) {
187                                 Logger::info('Probing timeout', ['url' => $url], Logger::DEBUG);
188                                 self::$istimeout = true;
189                                 return [];
190                         } elseif ($connection_error && $ssl_connection_error) {
191                                 self::$istimeout = true;
192                                 return [];
193                         }
194
195                         $xml = $curlResult->getBody();
196                         $xrd = XML::parseString($xml, true);
197                         $host_url = 'http://'.$host;
198                 }
199                 if (!is_object($xrd)) {
200                         Logger::log("No xrd object found for ".$host, Logger::DEBUG);
201                         return [];
202                 }
203
204                 $links = XML::elementToArray($xrd);
205                 if (!isset($links["xrd"]["link"])) {
206                         Logger::log("No xrd data found for ".$host, Logger::DEBUG);
207                         return [];
208                 }
209
210                 $lrdd = ['application/jrd+json' => $host_url . '/.well-known/webfinger?resource={uri}'];
211
212                 foreach ($links["xrd"]["link"] as $value => $link) {
213                         if (!empty($link["@attributes"])) {
214                                 $attributes = $link["@attributes"];
215                         } elseif ($value == "@attributes") {
216                                 $attributes = $link;
217                         } else {
218                                 continue;
219                         }
220
221                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
222                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
223
224                                 $lrdd[$type] = $attributes["template"];
225                         }
226                 }
227
228                 self::$baseurl = $host_url;
229
230                 Logger::log("Probing successful for ".$host, Logger::DEBUG);
231
232                 return $lrdd;
233         }
234
235         /**
236          * Perform Webfinger lookup and return DFRN data
237          *
238          * Given an email style address, perform webfinger lookup and
239          * return the resulting DFRN profile URL, or if no DFRN profile URL
240          * is located, returns an OStatus subscription template (prefixed
241          * with the string 'stat:' to identify it as on OStatus template).
242          * If this isn't an email style address just return $webbie.
243          * Return an empty string if email-style addresses but webfinger fails,
244          * or if the resultant personal XRD doesn't contain a supported
245          * subscription/friend-request attribute.
246          *
247          * amended 7/9/2011 to return an hcard which could save potentially loading
248          * a lengthy content page to scrape dfrn attributes
249          *
250          * @param string $webbie    Address that should be probed
251          * @param string $hcard_url Link to the hcard - is returned by reference
252          *
253          * @return string profile link
254          * @throws HTTPException\InternalServerErrorException
255          */
256         public static function webfingerDfrn($webbie, &$hcard_url)
257         {
258                 $profile_link = '';
259
260                 $links = self::lrdd($webbie);
261                 Logger::log('webfingerDfrn: '.$webbie.':'.print_r($links, true), Logger::DATA);
262                 if (!empty($links) && is_array($links)) {
263                         foreach ($links as $link) {
264                                 if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) {
265                                         $profile_link = $link['@attributes']['href'];
266                                 }
267                                 if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) {
268                                         $profile_link = 'stat:'.$link['@attributes']['template'];
269                                 }
270                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
271                                         $hcard_url = $link['@attributes']['href'];
272                                 }
273                         }
274                 }
275                 return $profile_link;
276         }
277
278         /**
279          * Check an URI for LRDD data
280          *
281          * @param string $uri     Address that should be probed
282          *
283          * @return array uri data
284          * @throws HTTPException\InternalServerErrorException
285          */
286         public static function lrdd(string $uri)
287         {
288                 $lrdd = self::hostMeta($uri);
289                 $webfinger = null;
290
291                 if (is_bool($lrdd)) {
292                         return [];
293                 }
294
295                 if (!$lrdd) {
296                         $parts = @parse_url($uri);
297                         if (!$parts || empty($parts["host"]) || empty($parts["path"])) {
298                                 return [];
299                         }
300
301                         $host = $parts['scheme'] . '://' . $parts["host"];
302                         if (!empty($parts["port"])) {
303                                 $host .= ':'.$parts["port"];
304                         }
305
306                         $path_parts = explode("/", trim($parts["path"], "/"));
307
308                         $nick = array_pop($path_parts);
309
310                         do {
311                                 $lrdd = self::hostMeta($host);
312                                 $host .= "/".array_shift($path_parts);
313                         } while (!$lrdd && (sizeof($path_parts) > 0));
314                 }
315
316                 if (!$lrdd) {
317                         Logger::log("No lrdd data found for ".$uri, Logger::DEBUG);
318                         return [];
319                 }
320
321                 foreach ($lrdd as $type => $template) {
322                         if ($webfinger) {
323                                 continue;
324                         }
325
326                         $path = str_replace('{uri}', urlencode($uri), $template);
327                         $webfinger = self::webfinger($path, $type);
328
329                         if (!$webfinger && (strstr($uri, "@"))) {
330                                 $path = str_replace('{uri}', urlencode("acct:".$uri), $template);
331                                 $webfinger = self::webfinger($path, $type);
332                         }
333
334                         // Special treatment for Mastodon
335                         // Problem is that Mastodon uses an URL format like http://domain.tld/@nick
336                         // But the webfinger for this format fails.
337                         if (!$webfinger && !empty($nick)) {
338                                 // Mastodon uses a "@" as prefix for usernames in their url format
339                                 $nick = ltrim($nick, '@');
340
341                                 $addr = $nick."@".$host;
342
343                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
344                                 $webfinger = self::webfinger($path, $type);
345                         }
346                 }
347
348                 if (empty($webfinger["links"])) {
349                         Logger::log("No webfinger links found for ".$uri, Logger::DEBUG);
350                         return [];
351                 }
352
353                 $data = [];
354
355                 foreach ($webfinger["links"] as $link) {
356                         $data[] = ["@attributes" => $link];
357                 }
358
359                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
360                         foreach ($webfinger["aliases"] as $alias) {
361                                 $data[] = ["@attributes" =>
362                                                         ["rel" => "alias",
363                                                                 "href" => $alias]];
364                         }
365                 }
366
367                 return $data;
368         }
369
370         /**
371          * Fetch information (protocol endpoints and user information) about a given uri
372          *
373          * @param string  $uri     Address that should be probed
374          * @param string  $network Test for this specific network
375          * @param integer $uid     User ID for the probe (only used for mails)
376          * @param boolean $cache   Use cached values?
377          *
378          * @return array uri data
379          * @throws HTTPException\InternalServerErrorException
380          * @throws \ImagickException
381          */
382         public static function uri($uri, $network = '', $uid = -1, $cache = true)
383         {
384                 if ($cache) {
385                         $result = DI::cache()->get('Probe::uri:' . $network . ':' . $uri);
386                         if (!is_null($result)) {
387                                 return $result;
388                         }
389                 }
390
391                 if ($uid == -1) {
392                         $uid = local_user();
393                 }
394
395                 self::$istimeout = false;
396
397                 if ($network != Protocol::ACTIVITYPUB) {
398                         $data = self::detect($uri, $network, $uid);
399                 } else {
400                         $data = null;
401                 }
402
403                 // When the previous detection process had got a time out
404                 // we could falsely detect a Friendica profile as AP profile.
405                 if (!self::$istimeout) {
406                         $ap_profile = ActivityPub::probeProfile($uri, !$cache);
407
408                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
409                                 $data = $ap_profile;
410                         } elseif (!empty($ap_profile)) {
411                                 $ap_profile['batch'] = '';
412                                 $data = array_merge($ap_profile, $data);
413                         }
414                 } else {
415                         Logger::notice('Time out detected. AP will not be probed.', ['uri' => $uri]);
416                 }
417
418                 if (!isset($data['url'])) {
419                         $data['url'] = $uri;
420                 }
421
422                 if (empty($data['photo'])) {
423                         $data['photo'] = DI::baseUrl() . '/images/person-300.jpg';
424                 }
425
426                 if (empty($data['name'])) {
427                         if (!empty($data['nick'])) {
428                                 $data['name'] = $data['nick'];
429                         }
430
431                         if (empty($data['name'])) {
432                                 $data['name'] = $data['url'];
433                         }
434                 }
435
436                 if (empty($data['nick'])) {
437                         $data['nick'] = strtolower($data['name']);
438
439                         if (strpos($data['nick'], ' ')) {
440                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
441                         }
442                 }
443
444                 if (empty($data['baseurl']) && !empty(self::$baseurl)) {
445                         $data['baseurl'] = self::$baseurl;
446                 }
447
448                 if (!empty($data['baseurl']) && empty($data['gsid'])) {
449                         $data['gsid'] = GServer::getID($data['baseurl']);
450                 }
451
452                 if (empty($data['network'])) {
453                         $data['network'] = Protocol::PHANTOM;
454                 }
455
456                 // Ensure that local connections always are DFRN
457                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
458                         $data['network'] = Protocol::DFRN;
459                 }
460
461                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
462                         $data['hide'] = self::getHideStatus($data['url']);
463                 }
464
465                 $data = self::rearrangeData($data);
466
467                 // Only store into the cache if the value seems to be valid
468                 if (!in_array($data['network'], [Protocol::PHANTOM, Protocol::MAIL])) {
469                         DI::cache()->set('Probe::uri:' . $network . ':' . $uri, $data, Duration::DAY);
470                 }
471
472                 return $data;
473         }
474
475
476         /**
477          * Fetches the "hide" status from the profile
478          *
479          * @param string $url URL of the profile
480          *
481          * @return boolean "hide" status
482          */
483         private static function getHideStatus($url)
484         {
485                 $curlResult = Network::curl($url);
486                 if (!$curlResult->isSuccess()) {
487                         return false;
488                 }
489
490                 // If the file is too large then exit
491                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
492                         return false;
493                 }
494
495                 // If it isn't a HTML file then exit
496                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
497                         return false;
498                 }
499
500                 $body = $curlResult->getBody();
501
502                 $doc = new DOMDocument();
503                 @$doc->loadHTML($body);
504
505                 $xpath = new DOMXPath($doc);
506
507                 $list = $xpath->query('//meta[@name]');
508                 foreach ($list as $node) {
509                         $meta_tag = [];
510                         if ($node->attributes->length) {
511                                 foreach ($node->attributes as $attribute) {
512                                         $meta_tag[$attribute->name] = $attribute->value;
513                                 }
514                         }
515
516                         if (empty($meta_tag['content'])) {
517                                 continue;
518                         }
519
520                         $content = strtolower(trim($meta_tag['content']));
521
522                         switch (strtolower(trim($meta_tag['name']))) {
523                                 case 'dfrn-global-visibility':
524                                         if ($content == 'false') {
525                                                 return true;
526                                         }
527                                         break;
528                                 case 'robots':
529                                         if (strpos($content, 'noindex') !== false) {
530                                                 return true;
531                                         }
532                                         break;
533                         }
534                 }
535
536                 return false;
537         }
538
539         /**
540          * Checks if a profile url should be OStatus but only provides partial information
541          *
542          * @param array  $webfinger Webfinger data
543          * @param string $lrdd      Path template for webfinger request
544          * @param string $type      type
545          *
546          * @return array fixed webfinger data
547          * @throws HTTPException\InternalServerErrorException
548          */
549         private static function fixOStatus($webfinger, $lrdd, $type)
550         {
551                 if (empty($webfinger['links']) || empty($webfinger['subject'])) {
552                         return $webfinger;
553                 }
554
555                 $is_ostatus = false;
556                 $has_key = false;
557
558                 foreach ($webfinger['links'] as $link) {
559                         if ($link['rel'] == ActivityNamespace::OSTATUSSUB) {
560                                 $is_ostatus = true;
561                         }
562                         if ($link['rel'] == 'magic-public-key') {
563                                 $has_key = true;
564                         }
565                 }
566
567                 if (!$is_ostatus || $has_key) {
568                         return $webfinger;
569                 }
570
571                 $url = Network::switchScheme($webfinger['subject']);
572                 $path = str_replace('{uri}', urlencode($url), $lrdd);
573                 $webfinger2 = self::webfinger($path, $type);
574
575                 // Is the new webfinger detectable as OStatus?
576                 if (self::ostatus($webfinger2, true)) {
577                         $webfinger = $webfinger2;
578                 }
579
580                 return $webfinger;
581         }
582
583         /**
584          * Fetch the "subscribe" and add it to the result
585          *
586          * @param array $result
587          * @param array $webfinger
588          * @return array result
589          */
590         private static function getSubscribeLink(array $result, array $webfinger)
591         {
592                 if (empty($webfinger['links'])) {
593                         return $result;
594                 }
595
596                 foreach ($webfinger['links'] as $link) {
597                         if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
598                                 $result['subscribe'] = $link['template'];
599                         }
600                 }
601
602                 return $result;
603         }
604
605         /**
606          * Fetch information (protocol endpoints and user information) about a given uri
607          *
608          * This function is only called by the "uri" function that adds caching and rearranging of data.
609          *
610          * @param string  $uri     Address that should be probed
611          * @param string  $network Test for this specific network
612          * @param integer $uid     User ID for the probe (only used for mails)
613          *
614          * @return array uri data
615          * @throws HTTPException\InternalServerErrorException
616          */
617         private static function detect($uri, $network, $uid)
618         {
619                 $parts = parse_url($uri);
620
621                 $hookData = [
622                         'uri'     => $uri,
623                         'network' => $network,
624                         'uid'     => $uid,
625                         'result'  => [],
626                 ];
627
628                 Hook::callAll('probe_detect', $hookData);
629
630                 if ($hookData['result']) {
631                         return $hookData['result'];
632                 }
633
634                 if (!empty($parts["scheme"]) && !empty($parts["host"])) {
635                         $host = $parts["host"];
636                         if (!empty($parts["port"])) {
637                                 $host .= ':'.$parts["port"];
638                         }
639
640                         if ($host == 'twitter.com') {
641                                 return self::twitter($uri);
642                         }
643                         $lrdd = self::hostMeta($host);
644
645                         if (is_bool($lrdd)) {
646                                 return [];
647                         }
648
649                         $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
650
651                         while (!$lrdd && (sizeof($path_parts) > 1)) {
652                                 $host .= "/".array_shift($path_parts);
653                                 $lrdd = self::hostMeta($host);
654                         }
655                         if (!$lrdd) {
656                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
657                                 return self::feed($uri);
658                         }
659                         $nick = array_pop($path_parts);
660
661                         // Mastodon uses a "@" as prefix for usernames in their url format
662                         $nick = ltrim($nick, '@');
663
664                         $addr = $nick."@".$host;
665                 } elseif (strstr($uri, '@')) {
666                         // If the URI starts with "mailto:" then jump directly to the mail detection
667                         if (strpos($uri, 'mailto:') !== false) {
668                                 $uri = str_replace('mailto:', '', $uri);
669                                 return self::mail($uri, $uid);
670                         }
671
672                         if ($network == Protocol::MAIL) {
673                                 return self::mail($uri, $uid);
674                         }
675                         // Remove "acct:" from the URI
676                         $uri = str_replace('acct:', '', $uri);
677
678                         $host = substr($uri, strpos($uri, '@') + 1);
679                         $nick = substr($uri, 0, strpos($uri, '@'));
680
681                         if (strpos($uri, '@twitter.com')) {
682                                 return self::twitter($uri);
683                         }
684                         $lrdd = self::hostMeta($host);
685
686                         if (is_bool($lrdd)) {
687                                 return [];
688                         }
689
690                         if (!$lrdd) {
691                                 Logger::log('No XRD data was found for '.$uri, Logger::DEBUG);
692                                 return self::mail($uri, $uid);
693                         }
694                         $addr = $uri;
695                 } else {
696                         Logger::log("Uri ".$uri." was not detectable", Logger::DEBUG);
697                         return [];
698                 }
699
700                 $webfinger = false;
701
702                 /// @todo Do we need the prefix "acct:" or "acct://"?
703
704                 foreach ($lrdd as $type => $template) {
705                         if ($webfinger) {
706                                 continue;
707                         }
708
709                         // At first try it with the given uri
710                         $path = str_replace('{uri}', urlencode($uri), $template);
711                         $webfinger = self::webfinger($path, $type);
712
713                         // Fix possible problems with GNU Social probing to wrong scheme
714                         $webfinger = self::fixOStatus($webfinger, $template, $type);
715
716                         // We cannot be sure that the detected address was correct, so we don't use the values
717                         if ($webfinger && ($uri != $addr)) {
718                                 $nick = "";
719                                 $addr = "";
720                         }
721
722                         // Try webfinger with the address (user@domain.tld)
723                         if (!$webfinger) {
724                                 $path = str_replace('{uri}', urlencode($addr), $template);
725                                 $webfinger = self::webfinger($path, $type);
726                         }
727
728                         // Mastodon needs to have it with "acct:"
729                         if (!$webfinger) {
730                                 $path = str_replace('{uri}', urlencode("acct:".$addr), $template);
731                                 $webfinger = self::webfinger($path, $type);
732                         }
733                 }
734
735                 if (!$webfinger) {
736                         return self::feed($uri);
737                 }
738
739                 $result = [];
740
741                 Logger::info("Probing", ['uri' => $uri]);
742
743                 if (in_array($network, ["", Protocol::DFRN])) {
744                         $result = self::dfrn($webfinger);
745                 }
746                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
747                         $result = self::diaspora($webfinger);
748                 }
749                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
750                         $result = self::ostatus($webfinger);
751                 }
752                 if (in_array($network, ['', Protocol::ZOT])) {
753                         $result = self::zot($webfinger, $result);
754                 }
755                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
756                         $result = self::pumpio($webfinger, $addr);
757                 }
758                 if ((!$result && ($network == "")) || ($network == Protocol::FEED)) {
759                         $result = self::feed($uri);
760                 } else {
761                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
762                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
763                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
764                                 $result["nick"] = $nick;
765                         }
766
767                         if (empty($result["addr"]) && ($addr != "")) {
768                                 $result["addr"] = $addr;
769                         }
770                 }
771
772                 $result = self::getSubscribeLink($result, $webfinger);
773
774                 if (empty($result["network"])) {
775                         $result["network"] = Protocol::PHANTOM;
776                 }
777
778                 if (empty($result["url"])) {
779                         $result["url"] = $uri;
780                 }
781
782                 Logger::log($uri." is ".$result["network"], Logger::DEBUG);
783
784                 return $result;
785         }
786
787         /**
788          * Check for Zot contact
789          *
790          * @param array $webfinger Webfinger data
791          * @param array $data      previously probed data
792          *
793          * @return array Zot data
794          * @throws HTTPException\InternalServerErrorException
795          */
796         private static function zot($webfinger, $data)
797         {
798                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
799                         foreach ($webfinger["aliases"] as $alias) {
800                                 if (substr($alias, 0, 5) == 'acct:') {
801                                         $data["addr"] = substr($alias, 5);
802                                 }
803                         }
804                 }
805
806                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
807                         $data["addr"] = substr($webfinger["subject"], 5);
808                 }
809
810                 $zot_url = '';
811                 foreach ($webfinger['links'] as $link) {
812                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
813                                 $zot_url = $link['href'];
814                         }
815                 }
816
817                 if (empty($zot_url) && !empty($data['addr']) && !empty(self::$baseurl)) {
818                         $condition = ['nurl' => Strings::normaliseLink(self::$baseurl), 'platform' => ['hubzilla']];
819                         if (!DBA::exists('gserver', $condition)) {
820                                 return $data;
821                         }
822                         $zot_url = self::$baseurl . '/.well-known/zot-info?address=' . $data['addr'];
823                 }
824
825                 if (empty($zot_url)) {
826                         return $data;
827                 }
828
829                 $data = self::pollZot($zot_url, $data);
830
831                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
832                         foreach ($webfinger['aliases'] as $alias) {
833                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
834                                         $data['alias'] = $alias;
835                                 }
836                         }
837                 }
838
839                 return $data;
840         }
841
842         public static function pollZot($url, $data)
843         {
844                 $curlResult = Network::curl($url);
845                 if ($curlResult->isTimeout()) {
846                         return $data;
847                 }
848                 $content = $curlResult->getBody();
849                 if (!$content) {
850                         return $data;
851                 }
852
853                 $json = json_decode($content, true);
854                 if (!is_array($json)) {
855                         return $data;
856                 }
857
858                 if (empty($data['network'])) {
859                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
860                                 $data['network'] = Protocol::ZOT;
861                         } elseif (!isset($json['protocols'])) {
862                                 $data['network'] = Protocol::ZOT;
863                         }
864                 }
865
866                 if (!empty($json['guid']) && empty($data['guid'])) {
867                         $data['guid'] = $json['guid'];
868                 }
869                 if (!empty($json['key']) && empty($data['pubkey'])) {
870                         $data['pubkey'] = $json['key'];
871                 }
872                 if (!empty($json['name'])) {
873                         $data['name'] = $json['name'];
874                 }
875                 if (!empty($json['photo'])) {
876                         $data['photo'] = $json['photo'];
877                         if (!empty($json['photo_updated'])) {
878                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
879                         }
880                 }
881                 if (!empty($json['address'])) {
882                         $data['addr'] = $json['address'];
883                 }
884                 if (!empty($json['url'])) {
885                         $data['url'] = $json['url'];
886                 }
887                 if (!empty($json['connections_url'])) {
888                         $data['poco'] = $json['connections_url'];
889                 }
890                 if (isset($json['searchable'])) {
891                         $data['hide'] = !$json['searchable'];
892                 }
893                 if (!empty($json['public_forum'])) {
894                         $data['community'] = $json['public_forum'];
895                         $data['account-type'] = Contact::PAGE_COMMUNITY;
896                 }
897
898                 if (!empty($json['profile'])) {
899                         $profile = $json['profile'];
900                         if (!empty($profile['description'])) {
901                                 $data['about'] = $profile['description'];
902                         }
903                         if (!empty($profile['keywords'])) {
904                                 $keywords = implode(', ', $profile['keywords']);
905                                 if (!empty($keywords)) {
906                                         $data['keywords'] = $keywords;
907                                 }
908                         }
909
910                         $loc = [];
911                         if (!empty($profile['region'])) {
912                                 $loc['region'] = $profile['region'];
913                         }
914                         if (!empty($profile['country'])) {
915                                 $loc['country-name'] = $profile['country'];
916                         }
917                         $location = Profile::formatLocation($loc);
918                         if (!empty($location)) {
919                                 $data['location'] = $location;
920                         }
921                 }
922
923                 return $data;
924         }
925
926         /**
927          * Perform a webfinger request.
928          *
929          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
930          *
931          * @param string $url  Address that should be probed
932          * @param string $type type
933          *
934          * @return array webfinger data
935          * @throws HTTPException\InternalServerErrorException
936          */
937         public static function webfinger($url, $type)
938         {
939                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
940
941                 $curlResult = Network::curl($url, false, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
942                 if ($curlResult->isTimeout()) {
943                         self::$istimeout = true;
944                         return [];
945                 }
946                 $data = $curlResult->getBody();
947
948                 $webfinger = json_decode($data, true);
949                 if (!empty($webfinger)) {
950                         if (!isset($webfinger["links"])) {
951                                 Logger::log("No json webfinger links for ".$url, Logger::DEBUG);
952                                 return [];
953                         }
954                         return $webfinger;
955                 }
956
957                 // If it is not JSON, maybe it is XML
958                 $xrd = XML::parseString($data, true);
959                 if (!is_object($xrd)) {
960                         Logger::log("No webfinger data retrievable for ".$url, Logger::DEBUG);
961                         return [];
962                 }
963
964                 $xrd_arr = XML::elementToArray($xrd);
965                 if (!isset($xrd_arr["xrd"]["link"])) {
966                         Logger::log("No XML webfinger links for ".$url, Logger::DEBUG);
967                         return [];
968                 }
969
970                 $webfinger = [];
971
972                 if (!empty($xrd_arr["xrd"]["subject"])) {
973                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
974                 }
975
976                 if (!empty($xrd_arr["xrd"]["alias"])) {
977                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
978                 }
979
980                 $webfinger["links"] = [];
981
982                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
983                         if (!empty($data["@attributes"])) {
984                                 $attributes = $data["@attributes"];
985                         } elseif ($value == "@attributes") {
986                                 $attributes = $data;
987                         } else {
988                                 continue;
989                         }
990
991                         $webfinger["links"][] = $attributes;
992                 }
993                 return $webfinger;
994         }
995
996         /**
997          * Poll the Friendica specific noscrape page.
998          *
999          * "noscrape" is a faster alternative to fetch the data from the hcard.
1000          * This functionality was originally created for the directory.
1001          *
1002          * @param string $noscrape_url Link to the noscrape page
1003          * @param array  $data         The already fetched data
1004          *
1005          * @return array noscrape data
1006          * @throws HTTPException\InternalServerErrorException
1007          */
1008         private static function pollNoscrape($noscrape_url, $data)
1009         {
1010                 $curlResult = Network::curl($noscrape_url);
1011                 if ($curlResult->isTimeout()) {
1012                         self::$istimeout = true;
1013                         return [];
1014                 }
1015                 $content = $curlResult->getBody();
1016                 if (!$content) {
1017                         Logger::log("Empty body for ".$noscrape_url, Logger::DEBUG);
1018                         return [];
1019                 }
1020
1021                 $json = json_decode($content, true);
1022                 if (!is_array($json)) {
1023                         Logger::log("No json data for ".$noscrape_url, Logger::DEBUG);
1024                         return [];
1025                 }
1026
1027                 if (!empty($json["fn"])) {
1028                         $data["name"] = $json["fn"];
1029                 }
1030
1031                 if (!empty($json["addr"])) {
1032                         $data["addr"] = $json["addr"];
1033                 }
1034
1035                 if (!empty($json["nick"])) {
1036                         $data["nick"] = $json["nick"];
1037                 }
1038
1039                 if (!empty($json["guid"])) {
1040                         $data["guid"] = $json["guid"];
1041                 }
1042
1043                 if (!empty($json["comm"])) {
1044                         $data["community"] = $json["comm"];
1045                 }
1046
1047                 if (!empty($json["tags"])) {
1048                         $keywords = implode(", ", $json["tags"]);
1049                         if ($keywords != "") {
1050                                 $data["keywords"] = $keywords;
1051                         }
1052                 }
1053
1054                 $location = Profile::formatLocation($json);
1055                 if ($location) {
1056                         $data["location"] = $location;
1057                 }
1058
1059                 if (!empty($json["about"])) {
1060                         $data["about"] = $json["about"];
1061                 }
1062
1063                 if (!empty($json["key"])) {
1064                         $data["pubkey"] = $json["key"];
1065                 }
1066
1067                 if (!empty($json["photo"])) {
1068                         $data["photo"] = $json["photo"];
1069                 }
1070
1071                 if (!empty($json["dfrn-request"])) {
1072                         $data["request"] = $json["dfrn-request"];
1073                 }
1074
1075                 if (!empty($json["dfrn-confirm"])) {
1076                         $data["confirm"] = $json["dfrn-confirm"];
1077                 }
1078
1079                 if (!empty($json["dfrn-notify"])) {
1080                         $data["notify"] = $json["dfrn-notify"];
1081                 }
1082
1083                 if (!empty($json["dfrn-poll"])) {
1084                         $data["poll"] = $json["dfrn-poll"];
1085                 }
1086
1087                 if (isset($json["hide"])) {
1088                         $data["hide"] = (bool)$json["hide"];
1089                 } else {
1090                         $data["hide"] = false;
1091                 }
1092
1093                 return $data;
1094         }
1095
1096         /**
1097          * Check for valid DFRN data
1098          *
1099          * @param array $data DFRN data
1100          *
1101          * @return int Number of errors
1102          */
1103         public static function validDfrn($data)
1104         {
1105                 $errors = 0;
1106                 if (!isset($data['key'])) {
1107                         $errors ++;
1108                 }
1109                 if (!isset($data['dfrn-request'])) {
1110                         $errors ++;
1111                 }
1112                 if (!isset($data['dfrn-confirm'])) {
1113                         $errors ++;
1114                 }
1115                 if (!isset($data['dfrn-notify'])) {
1116                         $errors ++;
1117                 }
1118                 if (!isset($data['dfrn-poll'])) {
1119                         $errors ++;
1120                 }
1121                 return $errors;
1122         }
1123
1124         /**
1125          * Fetch data from a DFRN profile page and via "noscrape"
1126          *
1127          * @param string $profile_link Link to the profile page
1128          *
1129          * @return array profile data
1130          * @throws HTTPException\InternalServerErrorException
1131          * @throws \ImagickException
1132          */
1133         public static function profile($profile_link)
1134         {
1135                 $data = [];
1136
1137                 Logger::log("Check profile ".$profile_link, Logger::DEBUG);
1138
1139                 // Fetch data via noscrape - this is faster
1140                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1141                 $data = self::pollNoscrape($noscrape_url, $data);
1142
1143                 if (!isset($data["notify"])
1144                         || !isset($data["confirm"])
1145                         || !isset($data["request"])
1146                         || !isset($data["poll"])
1147                         || !isset($data["name"])
1148                         || !isset($data["photo"])
1149                 ) {
1150                         $data = self::pollHcard($profile_link, $data, true);
1151                 }
1152
1153                 $prof_data = [];
1154
1155                 if (empty($data["addr"]) || empty($data["nick"])) {
1156                         $probe_data = self::uri($profile_link);
1157                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1158                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1159                 }
1160
1161                 $prof_data["addr"]         = $data["addr"];
1162                 $prof_data["nick"]         = $data["nick"];
1163                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1164                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1165                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1166                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1167                 $prof_data["photo"]        = $data['photo']   ?? null;
1168                 $prof_data["fn"]           = $data['name']    ?? null;
1169                 $prof_data["key"]          = $data['pubkey']  ?? null;
1170
1171                 Logger::log("Result for profile ".$profile_link.": ".print_r($prof_data, true), Logger::DEBUG);
1172
1173                 return $prof_data;
1174         }
1175
1176         /**
1177          * Check for DFRN contact
1178          *
1179          * @param array $webfinger Webfinger data
1180          *
1181          * @return array DFRN data
1182          * @throws HTTPException\InternalServerErrorException
1183          */
1184         private static function dfrn($webfinger)
1185         {
1186                 $hcard_url = "";
1187                 $data = [];
1188                 // The array is reversed to take into account the order of preference for same-rel links
1189                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1190                 foreach (array_reverse($webfinger["links"]) as $link) {
1191                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1192                                 $data["network"] = Protocol::DFRN;
1193                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1194                                 $data["poll"] = $link["href"];
1195                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1196                                 $data["url"] = $link["href"];
1197                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1198                                 $hcard_url = $link["href"];
1199                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1200                                 $data["poco"] = $link["href"];
1201                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1202                                 $data["photo"] = $link["href"];
1203                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1204                                 $data["baseurl"] = trim($link["href"], '/');
1205                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1206                                 $data["guid"] = $link["href"];
1207                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1208                                 $data["pubkey"] = base64_decode($link["href"]);
1209
1210                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1211                                 if (strstr($data["pubkey"], 'RSA ')) {
1212                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1213                                 }
1214                         }
1215                 }
1216
1217                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1218                         foreach ($webfinger["aliases"] as $alias) {
1219                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1220                                         $data["url"] = $alias;
1221                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1222                                         $data["alias"] = $alias;
1223                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1224                                         $data["addr"] = substr($alias, 5);
1225                                 }
1226                         }
1227                 }
1228
1229                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1230                         $data["addr"] = substr($webfinger["subject"], 5);
1231                 }
1232
1233                 if (!isset($data["network"]) || ($hcard_url == "")) {
1234                         return [];
1235                 }
1236
1237                 // Fetch data via noscrape - this is faster
1238                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1239                 $data = self::pollNoscrape($noscrape_url, $data);
1240
1241                 if (isset($data["notify"])
1242                         && isset($data["confirm"])
1243                         && isset($data["request"])
1244                         && isset($data["poll"])
1245                         && isset($data["name"])
1246                         && isset($data["photo"])
1247                 ) {
1248                         return $data;
1249                 }
1250
1251                 $data = self::pollHcard($hcard_url, $data, true);
1252
1253                 return $data;
1254         }
1255
1256         /**
1257          * Poll the hcard page (Diaspora and Friendica specific)
1258          *
1259          * @param string  $hcard_url Link to the hcard page
1260          * @param array   $data      The already fetched data
1261          * @param boolean $dfrn      Poll DFRN specific data
1262          *
1263          * @return array hcard data
1264          * @throws HTTPException\InternalServerErrorException
1265          */
1266         private static function pollHcard($hcard_url, $data, $dfrn = false)
1267         {
1268                 $curlResult = Network::curl($hcard_url);
1269                 if ($curlResult->isTimeout()) {
1270                         self::$istimeout = true;
1271                         return [];
1272                 }
1273                 $content = $curlResult->getBody();
1274                 if (!$content) {
1275                         return [];
1276                 }
1277
1278                 $doc = new DOMDocument();
1279                 if (!@$doc->loadHTML($content)) {
1280                         return [];
1281                 }
1282
1283                 $xpath = new DomXPath($doc);
1284
1285                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1286                 if (!is_object($vcards)) {
1287                         return [];
1288                 }
1289
1290                 if (!isset($data["baseurl"])) {
1291                         $data["baseurl"] = "";
1292                 }
1293
1294                 if ($vcards->length > 0) {
1295                         $vcard = $vcards->item(0);
1296
1297                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1298                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1299                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1300                         if (($search->length > 0) && empty($data["guid"])) {
1301                                 $data["guid"] = $search->item(0)->nodeValue;
1302                         }
1303
1304                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1305                         if ($search->length > 0) {
1306                                 $data["nick"] = $search->item(0)->nodeValue;
1307                         }
1308
1309                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1310                         if ($search->length > 0) {
1311                                 $data["name"] = $search->item(0)->nodeValue;
1312                         }
1313
1314                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1315                         if ($search->length > 0) {
1316                                 $data["searchable"] = $search->item(0)->nodeValue;
1317                         }
1318
1319                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1320                         if ($search->length > 0) {
1321                                 $data["pubkey"] = $search->item(0)->nodeValue;
1322                                 if (strstr($data["pubkey"], 'RSA ')) {
1323                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1324                                 }
1325                         }
1326
1327                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1328                         if ($search->length > 0) {
1329                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1330                         }
1331                 }
1332
1333                 $avatar = [];
1334                 if (!empty($vcard)) {
1335                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1336                         foreach ($photos as $photo) {
1337                                 $attr = [];
1338                                 foreach ($photo->attributes as $attribute) {
1339                                         $attr[$attribute->name] = trim($attribute->value);
1340                                 }
1341
1342                                 if (isset($attr["src"]) && isset($attr["width"])) {
1343                                         $avatar[$attr["width"]] = $attr["src"];
1344                                 }
1345
1346                                 // We don't have a width. So we just take everything that we got.
1347                                 // This is a Hubzilla workaround which doesn't send a width.
1348                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1349                                         $avatar[] = $attr["src"];
1350                                 }
1351                         }
1352                 }
1353
1354                 if (sizeof($avatar)) {
1355                         ksort($avatar);
1356                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1357                 }
1358
1359                 if ($dfrn) {
1360                         // Poll DFRN specific data
1361                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1362                         if ($search->length > 0) {
1363                                 foreach ($search as $link) {
1364                                         //$data["request"] = $search->item(0)->nodeValue;
1365                                         $attr = [];
1366                                         foreach ($link->attributes as $attribute) {
1367                                                 $attr[$attribute->name] = trim($attribute->value);
1368                                         }
1369
1370                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1371                                 }
1372                         }
1373
1374                         // Older Friendica versions had used the "uid" field differently than newer versions
1375                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1376                                 unset($data["guid"]);
1377                         }
1378                 }
1379
1380
1381                 return $data;
1382         }
1383
1384         /**
1385          * Check for Diaspora contact
1386          *
1387          * @param array $webfinger Webfinger data
1388          *
1389          * @return array Diaspora data
1390          * @throws HTTPException\InternalServerErrorException
1391          */
1392         private static function diaspora($webfinger)
1393         {
1394                 $hcard_url = "";
1395                 $data = [];
1396
1397                 // The array is reversed to take into account the order of preference for same-rel links
1398                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1399                 foreach (array_reverse($webfinger["links"]) as $link) {
1400                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1401                                 $hcard_url = $link["href"];
1402                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1403                                 $data["baseurl"] = trim($link["href"], '/');
1404                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1405                                 $data["guid"] = $link["href"];
1406                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1407                                 $data["url"] = $link["href"];
1408                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1409                                 $data["poll"] = $link["href"];
1410                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1411                                 $data["poco"] = $link["href"];
1412                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1413                                 $data["notify"] = $link["href"];
1414                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1415                                 $data["pubkey"] = base64_decode($link["href"]);
1416
1417                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1418                                 if (strstr($data["pubkey"], 'RSA ')) {
1419                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1420                                 }
1421                         }
1422                 }
1423
1424                 if (empty($data["url"]) || empty($hcard_url)) {
1425                         return [];
1426                 }
1427
1428                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1429                         foreach ($webfinger["aliases"] as $alias) {
1430                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1431                                         $data["alias"] = $alias;
1432                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1433                                         $data["addr"] = substr($alias, 5);
1434                                 }
1435                         }
1436                 }
1437
1438                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1439                         $data["addr"] = substr($webfinger["subject"], 5);
1440                 }
1441
1442                 // Fetch further information from the hcard
1443                 $data = self::pollHcard($hcard_url, $data);
1444
1445                 if (!$data) {
1446                         return [];
1447                 }
1448
1449                 if (!empty($data["url"])
1450                         && !empty($data["guid"])
1451                         && !empty($data["baseurl"])
1452                         && !empty($data["pubkey"])
1453                         && !empty($hcard_url)
1454                 ) {
1455                         $data["network"] = Protocol::DIASPORA;
1456
1457                         // The Diaspora handle must always be lowercase
1458                         if (!empty($data["addr"])) {
1459                                 $data["addr"] = strtolower($data["addr"]);
1460                         }
1461
1462                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1463                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1464                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1465                 } else {
1466                         return [];
1467                 }
1468
1469                 return $data;
1470         }
1471
1472         /**
1473          * Check for OStatus contact
1474          *
1475          * @param array $webfinger Webfinger data
1476          * @param bool  $short     Short detection mode
1477          *
1478          * @return array|bool OStatus data or "false" on error or "true" on short mode
1479          * @throws HTTPException\InternalServerErrorException
1480          */
1481         private static function ostatus($webfinger, $short = false)
1482         {
1483                 $data = [];
1484
1485                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1486                         foreach ($webfinger["aliases"] as $alias) {
1487                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1488                                         $data["addr"] = str_replace('acct:', '', $alias);
1489                                 }
1490                         }
1491                 }
1492
1493                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1494                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1495                 ) {
1496                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1497                 }
1498
1499                 if (!empty($webfinger["links"])) {
1500                         // The array is reversed to take into account the order of preference for same-rel links
1501                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1502                         foreach (array_reverse($webfinger["links"]) as $link) {
1503                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1504                                         && (($link["type"] ?? "") == "text/html")
1505                                         && ($link["href"] != "")
1506                                 ) {
1507                                         $data["url"] = $link["href"];
1508                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1509                                         $data["notify"] = $link["href"];
1510                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1511                                         $data["poll"] = $link["href"];
1512                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1513                                         $pubkey = $link["href"];
1514
1515                                         if (substr($pubkey, 0, 5) === 'data:') {
1516                                                 if (strstr($pubkey, ',')) {
1517                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1518                                                 } else {
1519                                                         $pubkey = substr($pubkey, 5);
1520                                                 }
1521                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1522                                                 $curlResult = Network::curl($pubkey);
1523                                                 if ($curlResult->isTimeout()) {
1524                                                         self::$istimeout = true;
1525                                                         return $short ? false : [];
1526                                                 }
1527                                                 $pubkey = $curlResult->getBody();
1528                                         }
1529
1530                                         $key = explode(".", $pubkey);
1531
1532                                         if (sizeof($key) >= 3) {
1533                                                 $m = Strings::base64UrlDecode($key[1]);
1534                                                 $e = Strings::base64UrlDecode($key[2]);
1535                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1536                                         }
1537                                 }
1538                         }
1539                 }
1540
1541                 if (isset($data["notify"]) && isset($data["pubkey"])
1542                         && isset($data["poll"])
1543                         && isset($data["url"])
1544                 ) {
1545                         $data["network"] = Protocol::OSTATUS;
1546                 } else {
1547                         return $short ? false : [];
1548                 }
1549
1550                 if ($short) {
1551                         return true;
1552                 }
1553
1554                 // Fetch all additional data from the feed
1555                 $curlResult = Network::curl($data["poll"]);
1556                 if ($curlResult->isTimeout()) {
1557                         self::$istimeout = true;
1558                         return [];
1559                 }
1560                 $feed = $curlResult->getBody();
1561                 $feed_data = Feed::import($feed);
1562                 if (!$feed_data) {
1563                         return [];
1564                 }
1565
1566                 if (!empty($feed_data["header"]["author-name"])) {
1567                         $data["name"] = $feed_data["header"]["author-name"];
1568                 }
1569                 if (!empty($feed_data["header"]["author-nick"])) {
1570                         $data["nick"] = $feed_data["header"]["author-nick"];
1571                 }
1572                 if (!empty($feed_data["header"]["author-avatar"])) {
1573                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1574                 }
1575                 if (!empty($feed_data["header"]["author-id"])) {
1576                         $data["alias"] = $feed_data["header"]["author-id"];
1577                 }
1578                 if (!empty($feed_data["header"]["author-location"])) {
1579                         $data["location"] = $feed_data["header"]["author-location"];
1580                 }
1581                 if (!empty($feed_data["header"]["author-about"])) {
1582                         $data["about"] = $feed_data["header"]["author-about"];
1583                 }
1584                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1585                 // So we take the value that we just fetched, although the other one worked as well
1586                 if (!empty($feed_data["header"]["author-link"])) {
1587                         $data["url"] = $feed_data["header"]["author-link"];
1588                 }
1589
1590                 if (($data['poll'] == $data['url']) && ($data["alias"] != '')) {
1591                         $data['url'] = $data["alias"];
1592                         $data["alias"] = '';
1593                 }
1594
1595                 /// @todo Fetch location and "about" from the feed as well
1596                 return $data;
1597         }
1598
1599         /**
1600          * Fetch data from a pump.io profile page
1601          *
1602          * @param string $profile_link Link to the profile page
1603          *
1604          * @return array profile data
1605          */
1606         private static function pumpioProfileData($profile_link)
1607         {
1608                 $curlResult = Network::curl($profile_link);
1609                 if (!$curlResult->isSuccess()) {
1610                         return [];
1611                 }
1612
1613                 $doc = new DOMDocument();
1614                 if (!@$doc->loadHTML($curlResult->getBody())) {
1615                         return [];
1616                 }
1617
1618                 $xpath = new DomXPath($doc);
1619
1620                 $data = [];
1621
1622                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1623
1624                 if ($data["name"] == '') {
1625                         // This is ugly - but pump.io doesn't seem to know a better way for it
1626                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1627                         $pos = strpos($data["name"], chr(10));
1628                         if ($pos) {
1629                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1630                         }
1631                 }
1632
1633                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1634
1635                 if ($data["location"] == '') {
1636                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1637                 }
1638
1639                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1640
1641                 if ($data["about"] == '') {
1642                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1643                 }
1644
1645                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1646                 if (!$avatar) {
1647                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1648                 }
1649                 if ($avatar) {
1650                         foreach ($avatar->attributes as $attribute) {
1651                                 if ($attribute->name == "src") {
1652                                         $data["photo"] = trim($attribute->value);
1653                                 }
1654                         }
1655                 }
1656
1657                 return $data;
1658         }
1659
1660         /**
1661          * Check for pump.io contact
1662          *
1663          * @param array  $webfinger Webfinger data
1664          * @param string $addr
1665          * @return array pump.io data
1666          */
1667         private static function pumpio($webfinger, $addr)
1668         {
1669                 $data = [];
1670                 // The array is reversed to take into account the order of preference for same-rel links
1671                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1672                 foreach (array_reverse($webfinger["links"]) as $link) {
1673                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1674                                 && (($link["type"] ?? "") == "text/html")
1675                                 && ($link["href"] != "")
1676                         ) {
1677                                 $data["url"] = $link["href"];
1678                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1679                                 $data["notify"] = $link["href"];
1680                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1681                                 $data["poll"] = $link["href"];
1682                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1683                                 $data["dialback"] = $link["href"];
1684                         }
1685                 }
1686                 if (isset($data["poll"]) && isset($data["notify"])
1687                         && isset($data["dialback"])
1688                         && isset($data["url"])
1689                 ) {
1690                         // by now we use these fields only for the network type detection
1691                         // So we unset all data that isn't used at the moment
1692                         unset($data["dialback"]);
1693
1694                         $data["network"] = Protocol::PUMPIO;
1695                 } else {
1696                         return [];
1697                 }
1698
1699                 $profile_data = self::pumpioProfileData($data["url"]);
1700
1701                 if (!$profile_data) {
1702                         return [];
1703                 }
1704
1705                 $data = array_merge($data, $profile_data);
1706
1707                 if (($addr != '') && ($data['name'] != '')) {
1708                         $name = trim(str_replace($addr, '', $data['name']));
1709                         if ($name != '') {
1710                                 $data['name'] = $name;
1711                         }
1712                 }
1713
1714                 return $data;
1715         }
1716
1717         /**
1718          * Check for twitter contact
1719          *
1720          * @param string $uri
1721          *
1722          * @return array twitter data
1723          */
1724         private static function twitter($uri)
1725         {
1726                 if (preg_match('=(.*)@twitter.com=i', $uri, $matches)) {
1727                         $nick = $matches[1];
1728                 } elseif (preg_match('=https?://twitter.com/(.*)=i', $uri, $matches)) {
1729                         $nick = $matches[1];
1730                 } else {
1731                         return [];
1732                 }
1733
1734                 $data = [];
1735                 $data['url'] = 'https://twitter.com/' . $nick;
1736                 $data['addr'] = $nick . '@twitter.com';
1737                 $data['nick'] = $data['name'] = $nick;
1738                 $data['network'] = Protocol::TWITTER;
1739                 $data['baseurl'] = 'https://twitter.com';
1740
1741                 return $data;
1742         }
1743
1744         /**
1745          * Checks HTML page for RSS feed link
1746          *
1747          * @param string $url  Page link
1748          * @param string $body Page body string
1749          * @return string|false Feed link or false if body was invalid HTML document
1750          */
1751         public static function getFeedLink(string $url, string $body)
1752         {
1753                 $doc = new DOMDocument();
1754                 if (!@$doc->loadHTML($body)) {
1755                         return false;
1756                 }
1757
1758                 $xpath = new DOMXPath($doc);
1759
1760                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1761
1762                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1763
1764                 return $feedUrl;
1765         }
1766
1767         /**
1768          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1769          *
1770          * Loosely based on RFC 1808
1771          *
1772          * @see https://tools.ietf.org/html/rfc1808
1773          *
1774          * @param string   $href  The potential relative href found in the HTML document
1775          * @param string   $base  The HTML document URL
1776          * @param DOMXPath $xpath The HTML document XPath
1777          * @return string
1778          */
1779         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
1780         {
1781                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1782                         return $href;
1783                 }
1784
1785                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1786
1787                 $baseParts = parse_url($base);
1788
1789                 // Naked domain case (scheme://basehost)
1790                 $path = $baseParts['path'] ?? '/';
1791
1792                 // Remove the filename part of the path if it exists (/base/path/file)
1793                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1794
1795                 $hrefParts = parse_url($href);
1796
1797                 // Root path case (/path) including relative scheme case (//host/path)
1798                 if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1799                         $path = $hrefParts['path'];
1800                 } else {
1801                         $path = $path . '/' . $hrefParts['path'];
1802
1803                         // Resolve arbitrary relative path
1804                         // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1805                         $parts = array_filter(explode('/', $path), 'strlen');
1806                         $absolutes = array();
1807                         foreach ($parts as $part) {
1808                                 if ('.' == $part) continue;
1809                                 if ('..' == $part) {
1810                                         array_pop($absolutes);
1811                                 } else {
1812                                         $absolutes[] = $part;
1813                                 }
1814                         }
1815
1816                         $path = '/' . implode('/', $absolutes);
1817                 }
1818
1819                 // Relative scheme case (//host/path)
1820                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1821                 $baseParts['path'] = $path;
1822                 unset($baseParts['query']);
1823                 unset($baseParts['fragment']);
1824
1825                 return Network::unparseURL($baseParts);
1826         }
1827
1828         /**
1829          * Check for feed contact
1830          *
1831          * @param string  $url   Profile link
1832          * @param boolean $probe Do a probe if the page contains a feed link
1833          *
1834          * @return array feed data
1835          * @throws HTTPException\InternalServerErrorException
1836          */
1837         private static function feed($url, $probe = true)
1838         {
1839                 $curlResult = Network::curl($url);
1840                 if ($curlResult->isTimeout()) {
1841                         self::$istimeout = true;
1842                         return [];
1843                 }
1844                 $feed = $curlResult->getBody();
1845                 $feed_data = Feed::import($feed);
1846
1847                 if (!$feed_data) {
1848                         if (!$probe) {
1849                                 return [];
1850                         }
1851
1852                         $feed_url = self::getFeedLink($url, $feed);
1853
1854                         if (!$feed_url) {
1855                                 return [];
1856                         }
1857
1858                         return self::feed($feed_url, false);
1859                 }
1860
1861                 if (!empty($feed_data["header"]["author-name"])) {
1862                         $data["name"] = $feed_data["header"]["author-name"];
1863                 }
1864
1865                 if (!empty($feed_data["header"]["author-nick"])) {
1866                         $data["nick"] = $feed_data["header"]["author-nick"];
1867                 }
1868
1869                 if (!empty($feed_data["header"]["author-avatar"])) {
1870                         $data["photo"] = $feed_data["header"]["author-avatar"];
1871                 }
1872
1873                 if (!empty($feed_data["header"]["author-id"])) {
1874                         $data["alias"] = $feed_data["header"]["author-id"];
1875                 }
1876
1877                 $data["url"] = $url;
1878                 $data["poll"] = $url;
1879
1880                 if (!empty($feed_data["header"]["author-link"])) {
1881                         $data["baseurl"] = $feed_data["header"]["author-link"];
1882                 } else {
1883                         $data["baseurl"] = $data["url"];
1884                 }
1885
1886                 $data["network"] = Protocol::FEED;
1887
1888                 return $data;
1889         }
1890
1891         /**
1892          * Check for mail contact
1893          *
1894          * @param string  $uri Profile link
1895          * @param integer $uid User ID
1896          *
1897          * @return array mail data
1898          * @throws \Exception
1899          */
1900         private static function mail($uri, $uid)
1901         {
1902                 if (!Network::isEmailDomainValid($uri)) {
1903                         return [];
1904                 }
1905
1906                 if ($uid == 0) {
1907                         return [];
1908                 }
1909
1910                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1911
1912                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1913                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1914                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1915
1916                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1917                         return [];
1918                 }
1919
1920                 $mailbox = Email::constructMailboxName($mailacct);
1921                 $password = '';
1922                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1923                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1924                 if (!$mbox) {
1925                         return [];
1926                 }
1927
1928                 $msgs = Email::poll($mbox, $uri);
1929                 Logger::log('searching '.$uri.', '.count($msgs).' messages found.', Logger::DEBUG);
1930
1931                 if (!count($msgs)) {
1932                         return [];
1933                 }
1934
1935                 $phost = substr($uri, strpos($uri, '@') + 1);
1936
1937                 $data = [];
1938                 $data["addr"]    = $uri;
1939                 $data["network"] = Protocol::MAIL;
1940                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1941                 $data["nick"]    = $data["name"];
1942                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1943                 $data["url"]     = 'mailto:'.$uri;
1944                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1945                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1946
1947                 $x = Email::messageMeta($mbox, $msgs[0]);
1948                 if (stristr($x[0]->from, $uri)) {
1949                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1950                 } elseif (stristr($x[0]->to, $uri)) {
1951                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1952                 }
1953                 if (isset($adr)) {
1954                         foreach ($adr as $feadr) {
1955                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1956                                         &&(strcasecmp($feadr->host, $phost) == 0)
1957                                         && (strlen($feadr->personal))
1958                                 ) {
1959                                         $personal = imap_mime_header_decode($feadr->personal);
1960                                         $data["name"] = "";
1961                                         foreach ($personal as $perspart) {
1962                                                 if ($perspart->charset != "default") {
1963                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1964                                                 } else {
1965                                                         $data["name"] .= $perspart->text;
1966                                                 }
1967                                         }
1968
1969                                         $data["name"] = Strings::escapeTags($data["name"]);
1970                                 }
1971                         }
1972                 }
1973                 if (!empty($mbox)) {
1974                         imap_close($mbox);
1975                 }
1976                 return $data;
1977         }
1978
1979         /**
1980          * Mix two paths together to possibly fix missing parts
1981          *
1982          * @param string $avatar Path to the avatar
1983          * @param string $base   Another path that is hopefully complete
1984          *
1985          * @return string fixed avatar path
1986          * @throws \Exception
1987          */
1988         public static function fixAvatar($avatar, $base)
1989         {
1990                 $base_parts = parse_url($base);
1991
1992                 // Remove all parts that could create a problem
1993                 unset($base_parts['path']);
1994                 unset($base_parts['query']);
1995                 unset($base_parts['fragment']);
1996
1997                 $avatar_parts = parse_url($avatar);
1998
1999                 // Now we mix them
2000                 $parts = array_merge($base_parts, $avatar_parts);
2001
2002                 // And put them together again
2003                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
2004                 $host     = isset($parts['host'])     ? $parts['host']           : '';
2005                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
2006                 $path     = isset($parts['path'])     ? $parts['path']           : '';
2007                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
2008                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
2009
2010                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
2011
2012                 Logger::log('Base: '.$base.' - Avatar: '.$avatar.' - Fixed: '.$fixed, Logger::DATA);
2013
2014                 return $fixed;
2015         }
2016 }