src/Network/Probe.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2022, the Friendica project
   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 Exception;
  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\Model\User;
  37 use Friendica\Network\HTTPClient\Client\HttpClientAccept;
  38 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  39 use Friendica\Protocol\ActivityNamespace;
  40 use Friendica\Protocol\ActivityPub;
  41 use Friendica\Protocol\Email;
  42 use Friendica\Protocol\Feed;
  43 use Friendica\Protocol\Salmon;
  44 use Friendica\Util\Crypto;
  45 use Friendica\Util\DateTimeFormat;
  46 use Friendica\Util\Network;
  47 use Friendica\Util\Strings;
  48 use Friendica\Util\XML;
  49 use GuzzleHttp\Psr7\Uri;
  50
  51 /**
  52  * This class contain functions for probing URL
  53  */
  54 class Probe
  55 {
  56         const HOST_META = '/.well-known/host-meta';
  57         const WEBFINGER = '/.well-known/webfinger?resource={uri}';
  58
  59         /**
  60          * @var string Base URL
  61          */
  62         private static $baseurl;
  63
  64         /**
  65          * @var boolean Whether a timeout has occured
  66          */
  67         private static $isTimeout;
  68
  69         /**
  70          * Checks if the provided network can be probed
  71          *
  72          * @param string $network
  73          *
  74          * @return boolean
  75          */
  76         public static function isProbable(string $network): bool
  77         {
  78                 return (in_array($network, array_merge(Protocol::FEDERATED, [Protocol::ZOT, Protocol::PHANTOM])));
  79         }
  80
  81         /**
  82          * Remove stuff from an URI that doesn't belong there
  83          *
  84          * @param string $rawUri
  85          * @return string Cleaned URI
  86          */
  87         public static function cleanURI(string $rawUri): string
  88         {
  89                 // At first remove leading and trailing junk
  90                 $rawUri = trim($rawUri, "@#?: \t\n\r\0\x0B");
  91
  92                 $rawUri = Network::convertToIdn($rawUri);
  93
  94                 $uri = new Uri($rawUri);
  95                 if (!$uri->getScheme()) {
  96                         return $uri->__toString();
  97                 }
  98
  99                 // Remove the URL fragment, since these shouldn't be part of any profile URL
 100                 $uri = $uri->withFragment('');
 101
 102                 return $uri->__toString();
 103         }
 104
 105         /**
 106          * Rearrange the array so that it always has the same order
 107          *
 108          * @param array $data Unordered data
 109          * @return array Ordered data
 110          */
 111         private static function rearrangeData(array $data): array
 112         {
 113                 $fields = ['name', 'nick', 'guid', 'url', 'addr', 'alias', 'photo', 'header',
 114                                 'account-type', 'community', 'keywords', 'location', 'about', 'xmpp', 'matrix',
 115                                 'hide', 'batch', 'notify', 'poll', 'request', 'confirm', 'subscribe', 'poco',
 116                                 'following', 'followers', 'inbox', 'outbox', 'sharedinbox',
 117                                 'priority', 'network', 'pubkey', 'manually-approve', 'baseurl', 'gsid'];
 118
 119                 $numeric_fields = ['gsid', 'hide', 'account-type', 'manually-approve'];
 120
 121                 $newdata = [];
 122                 foreach ($fields as $field) {
 123                         if (isset($data[$field])) {
 124                                 if (in_array($field, $numeric_fields)) {
 125                                         $newdata[$field] = (int)$data[$field];
 126                                 } else {
 127                                         $newdata[$field] = $data[$field];
 128                                 }
 129                         } elseif (!in_array($field, $numeric_fields)) {
 130                                 $newdata[$field] = '';
 131                         } else {
 132                                 $newdata[$field] = null;
 133                         }
 134                 }
 135
 136                 // We don't use the "priority" field anymore and replace it with a dummy.
 137                 $newdata['priority'] = 0;
 138
 139                 return $newdata;
 140         }
 141
 142         /**
 143          * Check if the hostname belongs to the own server
 144          *
 145          * @param string $host The hostname that is to be checked
 146          * @return bool Does the testes hostname belongs to the own server?
 147          */
 148         private static function ownHost(string $host): bool
 149         {
 150                 $own_host = DI::baseUrl()->getHostname();
 151
 152                 $parts = parse_url($host);
 153
 154                 if (!isset($parts['scheme'])) {
 155                         $parts = parse_url('http://' . $host);
 156                 }
 157
 158                 if (!isset($parts['host'])) {
 159                         return false;
 160                 }
 161                 return $parts['host'] == $own_host;
 162         }
 163
 164         /**
 165          * Probes for webfinger path via "host-meta"
 166          *
 167          * We have to check if the servers in the future still will offer this.
 168          * It seems as if it was dropped from the standard.
 169          *
 170          * @param string $host The host part of an url
 171          *
 172          * @return array with template and type of the webfinger template for JSON or XML
 173          * @throws HTTPException\InternalServerErrorException
 174          */
 175         private static function hostMeta(string $host): array
 176         {
 177                 // Reset the static variable
 178                 self::$baseurl = '';
 179
 180                 // Handles the case when the hostname contains the scheme
 181                 if (!parse_url($host, PHP_URL_SCHEME)) {
 182                         $ssl_url = 'https://' . $host . self::HOST_META;
 183                         $url = 'http://' . $host . self::HOST_META;
 184                 } else {
 185                         $ssl_url = $host . self::HOST_META;
 186                         $url = '';
 187                 }
 188
 189                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 190
 191                 Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
 192                 $xrd = null;
 193
 194                 $curlResult = DI::httpClient()->get($ssl_url, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 195                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 196                 if ($curlResult->isSuccess()) {
 197                         $xml = $curlResult->getBody();
 198                         $xrd = XML::parseString($xml, true);
 199                         if (!empty($url)) {
 200                                 $host_url = 'https://' . $host;
 201                         } else {
 202                                 $host_url = $host;
 203                         }
 204                 } elseif ($curlResult->isTimeout()) {
 205                         Logger::info('Probing timeout', ['url' => $ssl_url]);
 206                         self::$isTimeout = true;
 207                         return [];
 208                 }
 209
 210                 if (!is_object($xrd) && !empty($url)) {
 211                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 212                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 213                         if ($curlResult->isTimeout()) {
 214                                 Logger::info('Probing timeout', ['url' => $url]);
 215                                 self::$isTimeout = true;
 216                                 return [];
 217                         } elseif ($connection_error && $ssl_connection_error) {
 218                                 self::$isTimeout = true;
 219                                 return [];
 220                         }
 221
 222                         $xml = $curlResult->getBody();
 223                         $xrd = XML::parseString($xml, true);
 224                         $host_url = 'http://'.$host;
 225                 }
 226                 if (!is_object($xrd)) {
 227                         Logger::info('No xrd object found', ['host' => $host]);
 228                         return [];
 229                 }
 230
 231                 $links = XML::elementToArray($xrd);
 232                 if (!isset($links['xrd']['link'])) {
 233                         Logger::info('No xrd data found', ['host' => $host]);
 234                         return [];
 235                 }
 236
 237                 $lrdd = [];
 238
 239                 foreach ($links['xrd']['link'] as $value => $link) {
 240                         if (!empty($link['@attributes'])) {
 241                                 $attributes = $link['@attributes'];
 242                         } elseif ($value == '@attributes') {
 243                                 $attributes = $link;
 244                         } else {
 245                                 continue;
 246                         }
 247
 248                         if (!empty($attributes['rel']) && $attributes['rel'] == 'lrdd' && !empty($attributes['template'])) {
 249                                 $type = (empty($attributes['type']) ? '' : $attributes['type']);
 250
 251                                 $lrdd[$type] = $attributes['template'];
 252                         }
 253                 }
 254
 255                 if (Network::isUrlBlocked($host_url)) {
 256                         Logger::info('Domain is blocked', ['url' => $host]);
 257                         return [];
 258                 }
 259
 260                 self::$baseurl = $host_url;
 261
 262                 Logger::info('Probing successful', ['host' => $host]);
 263
 264                 return $lrdd;
 265         }
 266
 267         /**
 268          * Check an URI for LRDD data
 269          *
 270          * @param string $uri     Address that should be probed
 271          * @return array uri data
 272          * @throws HTTPException\InternalServerErrorException
 273          */
 274         public static function lrdd(string $uri): array
 275         {
 276                 $data = self::getWebfingerArray($uri);
 277                 if (empty($data)) {
 278                         return [];
 279                 }
 280                 $webfinger = $data['webfinger'];
 281
 282                 if (empty($webfinger['links'])) {
 283                         Logger::info('No webfinger links found', ['uri' => $uri]);
 284                         return [];
 285                 }
 286
 287                 $data = [];
 288
 289                 foreach ($webfinger['links'] as $link) {
 290                         $data[] = ['@attributes' => $link];
 291                 }
 292
 293                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 294                         foreach ($webfinger['aliases'] as $alias) {
 295                                 $data[] = [
 296                                         '@attributes' => [
 297                                                 'rel' => 'alias',
 298                                                 'href' => $alias,
 299                                         ]
 300                                 ];
 301                         }
 302                 }
 303
 304                 return $data;
 305         }
 306
 307         /**
 308          * Fetch information (protocol endpoints and user information) about a given uri
 309          *
 310          * @param string  $uri     Address that should be probed
 311          * @param string  $network Test for this specific network
 312          * @param integer $uid     User ID for the probe (only used for mails)
 313          * @param boolean $cache   Use cached values?
 314          *
 315          * @return array uri data
 316          * @throws HTTPException\InternalServerErrorException
 317          * @throws \ImagickException
 318          */
 319         public static function uri(string $uri, string $network = '', int $uid = -1): array
 320         {
 321                 // Local profiles aren't probed via network
 322                 if (empty($network) && Contact::isLocal($uri)) {
 323                         $data = self::localProbe($uri);
 324                         if (!empty($data)) {
 325                                 return $data;
 326                         }
 327                 }
 328
 329                 if ($uid == -1) {
 330                         $uid = DI::userSession()->getLocalUserId();
 331                 }
 332
 333                 if (empty($network) || ($network == Protocol::ACTIVITYPUB)) {
 334                         $ap_profile = ActivityPub::probeProfile($uri);
 335                 } else {
 336                         $ap_profile = [];
 337                 }
 338
 339                 self::$isTimeout = false;
 340
 341                 if ($network != Protocol::ACTIVITYPUB) {
 342                         $data = self::detect($uri, $network, $uid, $ap_profile);
 343                         if (!is_array($data)) {
 344                                 $data = [];
 345                         }
 346                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 347                                 $data = $ap_profile;
 348                         } elseif (!empty($ap_profile)) {
 349                                 $ap_profile['batch'] = '';
 350                                 $data = array_merge($ap_profile, $data);
 351                         }
 352                 } else {
 353                         $data = $ap_profile;
 354                 }
 355
 356                 if (!isset($data['url'])) {
 357                         $data['url'] = $uri;
 358                 }
 359
 360                 if (empty($data['photo'])) {
 361                         $data['photo'] = DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO;
 362                 }
 363
 364                 if (empty($data['name'])) {
 365                         if (!empty($data['nick'])) {
 366                                 $data['name'] = $data['nick'];
 367                         }
 368
 369                         if (empty($data['name'])) {
 370                                 $data['name'] = $data['url'];
 371                         }
 372                 }
 373
 374                 if (empty($data['nick'])) {
 375                         $data['nick'] = strtolower($data['name']);
 376
 377                         if (strpos($data['nick'], ' ')) {
 378                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 379                         }
 380                 }
 381
 382                 if (!empty($data['baseurl']) && empty($data['gsid'])) {
 383                         $data['gsid'] = GServer::getID($data['baseurl']);
 384                 }
 385
 386                 if (empty($data['network'])) {
 387                         $data['network'] = Protocol::PHANTOM;
 388                 }
 389
 390                 // Ensure that local connections always are DFRN
 391                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
 392                         $data['network'] = Protocol::DFRN;
 393                 }
 394
 395                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 396                         $data['hide'] = self::getHideStatus($data['url']);
 397                 }
 398
 399                 return self::rearrangeData($data);
 400         }
 401
 402
 403         /**
 404          * Fetches the "hide" status from the profile
 405          *
 406          * @param string $url URL of the profile
 407          * @return boolean "hide" status
 408          */
 409         private static function getHideStatus(string $url): bool
 410         {
 411                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML, [HttpClientOptions::CONTENT_LENGTH => 1000000]);
 412                 if (!$curlResult->isSuccess()) {
 413                         return false;
 414                 }
 415
 416                 // If it isn't a HTML file then exit
 417                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 418                         return false;
 419                 }
 420
 421                 $body = $curlResult->getBody();
 422                 if (empty($body)) {
 423                         return false;
 424                 }
 425
 426                 $doc = new DOMDocument();
 427                 @$doc->loadHTML($body);
 428
 429                 $xpath = new DOMXPath($doc);
 430
 431                 $list = $xpath->query('//meta[@name]');
 432                 foreach ($list as $node) {
 433                         $meta_tag = [];
 434                         if ($node->attributes->length) {
 435                                 foreach ($node->attributes as $attribute) {
 436                                         $meta_tag[$attribute->name] = $attribute->value;
 437                                 }
 438                         }
 439
 440                         if (empty($meta_tag['content'])) {
 441                                 continue;
 442                         }
 443
 444                         $content = strtolower(trim($meta_tag['content']));
 445
 446                         switch (strtolower(trim($meta_tag['name']))) {
 447                                 case 'dfrn-global-visibility':
 448                                         if ($content == 'false') {
 449                                                 return true;
 450                                         }
 451                                         break;
 452                                 case 'robots':
 453                                         if (strpos($content, 'noindex') !== false) {
 454                                                 return true;
 455                                         }
 456                                         break;
 457                         }
 458                 }
 459
 460                 return false;
 461         }
 462
 463         /**
 464          * Fetch the "subscribe" and add it to the result
 465          *
 466          * @param array $result Result array
 467          * @param array $webfinger Webfinger data
 468          *
 469          * @return array result Altered/unaltered result array
 470          */
 471         private static function getSubscribeLink(array $result, array $webfinger): array
 472         {
 473                 if (empty($webfinger['links'])) {
 474                         return $result;
 475                 }
 476
 477                 foreach ($webfinger['links'] as $link) {
 478                         if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
 479                                 $result['subscribe'] = $link['template'];
 480                         }
 481                 }
 482
 483                 return $result;
 484         }
 485
 486         /**
 487          * Get webfinger data from a given URI
 488          *
 489          * @param string $uri URI
 490          *
 491          * @return array Webfinger data
 492          * @throws HTTPException\InternalServerErrorException
 493          */
 494         private static function getWebfingerArray(string $uri): array
 495         {
 496                 $parts = parse_url($uri);
 497
 498                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 499                         $host = $parts['host'];
 500                         if (!empty($parts['port'])) {
 501                                 $host .= ':' . $parts['port'];
 502                         }
 503
 504                         $baseurl = $parts['scheme'] . '://' . $host;
 505
 506                         $nick = '';
 507                         $addr = '';
 508
 509                         $path_parts = explode('/', trim($parts['path'] ?? '', '/'));
 510                         if (!empty($path_parts)) {
 511                                 $nick = ltrim(end($path_parts), '@');
 512                                 $addr = $nick . '@' . $host;
 513                         }
 514
 515                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 516                         if (empty($webfinger)) {
 517                                 $lrdd = self::hostMeta($host);
 518                         }
 519
 520                         if (empty($webfinger) && empty($lrdd)) {
 521                                 while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
 522                                         $host    .= '/' . array_shift($path_parts);
 523                                         $baseurl = $parts['scheme'] . '://' . $host;
 524
 525                                         if (!empty($nick)) {
 526                                                 $addr = $nick . '@' . $host;
 527                                         }
 528
 529                                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 530                                         if (empty($webfinger)) {
 531                                                 $lrdd = self::hostMeta($host);
 532                                         }
 533                                 }
 534
 535                                 if (empty($lrdd) && empty($webfinger)) {
 536                                         return [];
 537                                 }
 538                         }
 539                 } elseif (strstr($uri, '@')) {
 540                         // Remove "acct:" from the URI
 541                         $uri = str_replace('acct:', '', $uri);
 542
 543                         $host = substr($uri, strpos($uri, '@') + 1);
 544                         $nick = substr($uri, 0, strpos($uri, '@'));
 545                         $addr = $uri;
 546
 547                         $webfinger = self::getWebfinger('https://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 548                         if (self::$isTimeout) {
 549                                 return [];
 550                         }
 551
 552                         if (empty($webfinger)) {
 553                                 $webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 554                                 if (self::$isTimeout) {
 555                                         return [];
 556                                 }
 557                         } else {
 558                                 $baseurl = 'https://' . $host;
 559                         }
 560
 561                         if (empty($webfinger)) {
 562                                 $lrdd = self::hostMeta($host);
 563                                 if (self::$isTimeout) {
 564                                         return [];
 565                                 }
 566                                 $baseurl = self::$baseurl;
 567                         } else {
 568                                 $baseurl = 'http://' . $host;
 569                         }
 570                 } else {
 571                         Logger::info('URI was not detectable', ['uri' => $uri]);
 572                         return [];
 573                 }
 574
 575                 if (empty($webfinger)) {
 576                         foreach ($lrdd as $type => $template) {
 577                                 if ($webfinger) {
 578                                         continue;
 579                                 }
 580
 581                                 $webfinger = self::getWebfinger($template, $type, $uri, $addr);
 582                         }
 583                 }
 584
 585                 if (empty($webfinger)) {
 586                         return [];
 587                 }
 588
 589                 if ($webfinger['detected'] == $addr) {
 590                         $webfinger['nick'] = $nick;
 591                         $webfinger['addr'] = $addr;
 592                 }
 593
 594                 $webfinger['baseurl'] = $baseurl;
 595
 596                 return $webfinger;
 597         }
 598
 599         /**
 600          * Perform network request for webfinger data
 601          *
 602          * @param string $template
 603          * @param string $type
 604          * @param string $uri
 605          * @param string $addr
 606          *
 607          * @return array webfinger results
 608          */
 609         private static function getWebfinger(string $template, string $type, string $uri, string $addr): array
 610         {
 611                 if (Network::isUrlBlocked($template)) {
 612                         Logger::info('Domain is blocked', ['url' => $template]);
 613                         return [];
 614                 }
 615
 616                 // First try the address because this is the primary purpose of webfinger
 617                 if (!empty($addr)) {
 618                         $detected = $addr;
 619                         $path = str_replace('{uri}', urlencode('acct:' . $addr), $template);
 620                         $webfinger = self::webfinger($path, $type);
 621                         if (self::$isTimeout) {
 622                                 return [];
 623                         }
 624                 }
 625
 626                 // Then try the URI
 627                 if (empty($webfinger) && $uri != $addr) {
 628                         $detected = $uri;
 629                         $path = str_replace('{uri}', urlencode($uri), $template);
 630                         $webfinger = self::webfinger($path, $type);
 631                         if (self::$isTimeout) {
 632                                 return [];
 633                         }
 634                 }
 635
 636                 if (empty($webfinger)) {
 637                         return [];
 638                 }
 639
 640                 return ['webfinger' => $webfinger, 'detected' => $detected];
 641         }
 642
 643         /**
 644          * Fetch information (protocol endpoints and user information) about a given uri
 645          *
 646          * This function is only called by the "uri" function that adds caching and rearranging of data.
 647          *
 648          * @param string  $uri        Address that should be probed
 649          * @param string  $network    Test for this specific network
 650          * @param integer $uid        User ID for the probe (only used for mails)
 651          * @param array   $ap_profile Previously probed AP profile
 652          * @return array URI data
 653          * @throws HTTPException\InternalServerErrorException
 654          */
 655         private static function detect(string $uri, string $network, int $uid, array $ap_profile): array
 656         {
 657                 $hookData = [
 658                         'uri'     => $uri,
 659                         'network' => $network,
 660                         'uid'     => $uid,
 661                         'result'  => null,
 662                 ];
 663
 664                 Hook::callAll('probe_detect', $hookData);
 665
 666                 if (isset($hookData['result'])) {
 667                         return is_array($hookData['result']) ? $hookData['result'] : [];
 668                 }
 669
 670                 $parts = parse_url($uri);
 671                 if (empty($parts['scheme']) && empty($parts['host']) && !strstr($parts['path'], '@')) {
 672                         Logger::info('URI was not detectable', ['uri' => $uri]);
 673                         return [];
 674                 }
 675
 676                 // If the URI starts with "mailto:" then jump directly to the mail detection
 677                 if (strpos($uri, 'mailto:') !== false) {
 678                         $uri = str_replace('mailto:', '', $uri);
 679                         return self::mail($uri, $uid);
 680                 }
 681
 682                 if ($network == Protocol::MAIL) {
 683                         return self::mail($uri, $uid);
 684                 }
 685
 686                 Logger::info('Probing start', ['uri' => $uri]);
 687
 688                 if (!empty($ap_profile['addr']) && ($ap_profile['addr'] != $uri)) {
 689                         $data = self::getWebfingerArray($ap_profile['addr']);
 690                 }
 691
 692                 if (empty($data)) {
 693                         $data = self::getWebfingerArray($uri);
 694                 }
 695
 696                 if (empty($data)) {
 697                         if (!empty($parts['scheme'])) {
 698                                 return self::feed($uri);
 699                         } elseif (!empty($uid)) {
 700                                 return self::mail($uri, $uid);
 701                         } else {
 702                                 return [];
 703                         }
 704                 }
 705
 706                 $webfinger = $data['webfinger'];
 707                 $nick = $data['nick'] ?? '';
 708                 $addr = $data['addr'] ?? '';
 709                 $baseurl = $data['baseurl'] ?? '';
 710
 711                 $result = [];
 712
 713                 if (in_array($network, ['', Protocol::DFRN])) {
 714                         $result = self::dfrn($webfinger);
 715                 }
 716                 if ((!$result && ($network == '')) || ($network == Protocol::DIASPORA)) {
 717                         $result = self::diaspora($webfinger);
 718                 }
 719                 if ((!$result && ($network == '')) || ($network == Protocol::OSTATUS)) {
 720                         $result = self::ostatus($webfinger);
 721                 }
 722                 if (in_array($network, ['', Protocol::ZOT])) {
 723                         $result = self::zot($webfinger, $result, $baseurl);
 724                 }
 725                 if ((!$result && ($network == '')) || ($network == Protocol::PUMPIO)) {
 726                         $result = self::pumpio($webfinger, $addr);
 727                 }
 728                 if (empty($result['network']) && empty($ap_profile['network']) || ($network == Protocol::FEED)) {
 729                         $result = self::feed($uri);
 730                 } else {
 731                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 732                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 733                         if ((empty($result['nick']) || (strstr($result['nick'], ' '))) && ($nick != '')) {
 734                                 $result['nick'] = $nick;
 735                         }
 736
 737                         if (empty($result['addr']) && ($addr != '')) {
 738                                 $result['addr'] = $addr;
 739                         }
 740                 }
 741
 742                 $result = self::getSubscribeLink($result, $webfinger);
 743
 744                 if (empty($result['network'])) {
 745                         $result['network'] = Protocol::PHANTOM;
 746                 }
 747
 748                 if (empty($result['baseurl']) && !empty($baseurl)) {
 749                         $result['baseurl'] = $baseurl;
 750                 }
 751
 752                 if (empty($result['url'])) {
 753                         $result['url'] = $uri;
 754                 }
 755
 756                 Logger::info('Probing done', ['uri' => $uri, 'network' => $result['network']]);
 757
 758                 return $result;
 759         }
 760
 761         /**
 762          * Check for Zot contact
 763          *
 764          * @param array  $webfinger Webfinger data
 765          * @param array  $data      previously probed data
 766          * @param string $baseUrl   Base URL
 767          *
 768          * @return array Zot data
 769          * @throws HTTPException\InternalServerErrorException
 770          */
 771         private static function zot(array $webfinger, array $data, string $baseurl): array
 772         {
 773                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 774                         foreach ($webfinger['aliases'] as $alias) {
 775                                 if (substr($alias, 0, 5) == 'acct:') {
 776                                         $data['addr'] = substr($alias, 5);
 777                                 }
 778                         }
 779                 }
 780
 781                 if (!empty($webfinger['subject']) && (substr($webfinger['subject'], 0, 5) == 'acct:')) {
 782                         $data['addr'] = substr($webfinger['subject'], 5);
 783                 }
 784
 785                 $zot_url = '';
 786                 foreach ($webfinger['links'] as $link) {
 787                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 788                                 $zot_url = $link['href'];
 789                         }
 790                 }
 791
 792                 if (empty($zot_url) && !empty($data['addr']) && !empty($baseurl)) {
 793                         $condition = ['nurl' => Strings::normaliseLink($baseurl), 'platform' => ['hubzilla']];
 794                         if (!DBA::exists('gserver', $condition)) {
 795                                 return $data;
 796                         }
 797                         $zot_url = $baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 798                 }
 799
 800                 if (empty($zot_url)) {
 801                         return $data;
 802                 }
 803
 804                 $data = self::pollZot($zot_url, $data);
 805
 806                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 807                         foreach ($webfinger['aliases'] as $alias) {
 808                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
 809                                         $data['alias'] = $alias;
 810                                 }
 811                         }
 812                 }
 813
 814                 return $data;
 815         }
 816
 817         public static function pollZot(string $url, array $data): array
 818         {
 819                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::JSON);
 820                 if ($curlResult->isTimeout()) {
 821                         return $data;
 822                 }
 823                 $content = $curlResult->getBody();
 824                 if (!$content) {
 825                         return $data;
 826                 }
 827
 828                 $json = json_decode($content, true);
 829                 if (!is_array($json)) {
 830                         return $data;
 831                 }
 832
 833                 if (empty($data['network'])) {
 834                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 835                                 $data['network'] = Protocol::ZOT;
 836                         } elseif (!isset($json['protocols'])) {
 837                                 $data['network'] = Protocol::ZOT;
 838                         }
 839                 }
 840
 841                 if (!empty($json['guid']) && empty($data['guid'])) {
 842                         $data['guid'] = $json['guid'];
 843                 }
 844                 if (!empty($json['key']) && empty($data['pubkey'])) {
 845                         $data['pubkey'] = $json['key'];
 846                 }
 847                 if (!empty($json['name'])) {
 848                         $data['name'] = $json['name'];
 849                 }
 850                 if (!empty($json['photo'])) {
 851                         $data['photo'] = $json['photo'];
 852                         if (!empty($json['photo_updated'])) {
 853                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 854                         }
 855                 }
 856                 if (!empty($json['address'])) {
 857                         $data['addr'] = $json['address'];
 858                 }
 859                 if (!empty($json['url'])) {
 860                         $data['url'] = $json['url'];
 861                 }
 862                 if (!empty($json['connections_url'])) {
 863                         $data['poco'] = $json['connections_url'];
 864                 }
 865                 if (isset($json['searchable'])) {
 866                         $data['hide'] = !$json['searchable'];
 867                 }
 868                 if (!empty($json['public_forum'])) {
 869                         $data['community'] = $json['public_forum'];
 870                         $data['account-type'] = User::PAGE_FLAGS_COMMUNITY;
 871                 }
 872
 873                 if (!empty($json['profile'])) {
 874                         $profile = $json['profile'];
 875                         if (!empty($profile['description'])) {
 876                                 $data['about'] = $profile['description'];
 877                         }
 878                         if (!empty($profile['keywords'])) {
 879                                 $keywords = implode(', ', $profile['keywords']);
 880                                 if (!empty($keywords)) {
 881                                         $data['keywords'] = $keywords;
 882                                 }
 883                         }
 884
 885                         $loc = [];
 886                         if (!empty($profile['region'])) {
 887                                 $loc['region'] = $profile['region'];
 888                         }
 889                         if (!empty($profile['country'])) {
 890                                 $loc['country-name'] = $profile['country'];
 891                         }
 892                         $location = Profile::formatLocation($loc);
 893                         if (!empty($location)) {
 894                                 $data['location'] = $location;
 895                         }
 896                 }
 897
 898                 return $data;
 899         }
 900
 901         /**
 902          * Perform a webfinger request.
 903          *
 904          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 905          *
 906          * @param string $url  Address that should be probed
 907          * @param string $type type
 908          *
 909          * @return array webfinger data
 910          * @throws HTTPException\InternalServerErrorException
 911          */
 912         public static function webfinger(string $url, string $type): array
 913         {
 914                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 915
 916                 $curlResult = DI::httpClient()->get($url, $type, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 917                 if ($curlResult->isTimeout()) {
 918                         self::$isTimeout = true;
 919                         return [];
 920                 }
 921                 $data = $curlResult->getBody();
 922
 923                 $webfinger = json_decode($data, true);
 924                 if (!empty($webfinger)) {
 925                         if (!isset($webfinger['links'])) {
 926                                 Logger::info('No json webfinger links', ['url' => $url]);
 927                                 return [];
 928                         }
 929                         return $webfinger;
 930                 }
 931
 932                 // If it is not JSON, maybe it is XML
 933                 $xrd = XML::parseString($data, true);
 934                 if (!is_object($xrd)) {
 935                         Logger::info('No webfinger data retrievable', ['url' => $url]);
 936                         return [];
 937                 }
 938
 939                 $xrd_arr = XML::elementToArray($xrd);
 940                 if (!isset($xrd_arr['xrd']['link'])) {
 941                         Logger::info('No XML webfinger links', ['url' => $url]);
 942                         return [];
 943                 }
 944
 945                 $webfinger = [];
 946
 947                 if (!empty($xrd_arr['xrd']['subject'])) {
 948                         $webfinger['subject'] = $xrd_arr['xrd']['subject'];
 949                 }
 950
 951                 if (!empty($xrd_arr['xrd']['alias'])) {
 952                         $webfinger['aliases'] = $xrd_arr['xrd']['alias'];
 953                 }
 954
 955                 $webfinger['links'] = [];
 956
 957                 foreach ($xrd_arr['xrd']['link'] as $value => $data) {
 958                         if (!empty($data['@attributes'])) {
 959                                 $attributes = $data['@attributes'];
 960                         } elseif ($value == '@attributes') {
 961                                 $attributes = $data;
 962                         } else {
 963                                 continue;
 964                         }
 965
 966                         $webfinger['links'][] = $attributes;
 967                 }
 968                 return $webfinger;
 969         }
 970
 971         /**
 972          * Poll the Friendica specific noscrape page.
 973          *
 974          * "noscrape" is a faster alternative to fetch the data from the hcard.
 975          * This functionality was originally created for the directory.
 976          *
 977          * @param string $noscrape_url Link to the noscrape page
 978          * @param array  $data         The already fetched data
 979          *
 980          * @return array noscrape data
 981          * @throws HTTPException\InternalServerErrorException
 982          */
 983         private static function pollNoscrape(string $noscrape_url, array $data): array
 984         {
 985                 $curlResult = DI::httpClient()->get($noscrape_url, HttpClientAccept::JSON);
 986                 if ($curlResult->isTimeout()) {
 987                         self::$isTimeout = true;
 988                         return $data;
 989                 }
 990                 $content = $curlResult->getBody();
 991                 if (!$content) {
 992                         Logger::info('Empty body', ['url' => $noscrape_url]);
 993                         return $data;
 994                 }
 995
 996                 $json = json_decode($content, true);
 997                 if (!is_array($json)) {
 998                         Logger::info('No json data', ['url' => $noscrape_url]);
 999                         return $data;
1000                 }
1001
1002                 if (!empty($json['fn'])) {
1003                         $data['name'] = $json['fn'];
1004                 }
1005
1006                 if (!empty($json['addr'])) {
1007                         $data['addr'] = $json['addr'];
1008                 }
1009
1010                 if (!empty($json['nick'])) {
1011                         $data['nick'] = $json['nick'];
1012                 }
1013
1014                 if (!empty($json['guid'])) {
1015                         $data['guid'] = $json['guid'];
1016                 }
1017
1018                 if (!empty($json['comm'])) {
1019                         $data['community'] = $json['comm'];
1020                 }
1021
1022                 if (!empty($json['tags'])) {
1023                         $keywords = implode(', ', $json['tags']);
1024                         if ($keywords != '') {
1025                                 $data['keywords'] = $keywords;
1026                         }
1027                 }
1028
1029                 $location = Profile::formatLocation($json);
1030                 if ($location) {
1031                         $data['location'] = $location;
1032                 }
1033
1034                 if (!empty($json['about'])) {
1035                         $data['about'] = $json['about'];
1036                 }
1037
1038                 if (!empty($json['xmpp'])) {
1039                         $data['xmpp'] = $json['xmpp'];
1040                 }
1041
1042                 if (!empty($json['matrix'])) {
1043                         $data['matrix'] = $json['matrix'];
1044                 }
1045
1046                 if (!empty($json['key'])) {
1047                         $data['pubkey'] = $json['key'];
1048                 }
1049
1050                 if (!empty($json['photo'])) {
1051                         $data['photo'] = $json['photo'];
1052                 }
1053
1054                 if (!empty($json['dfrn-request'])) {
1055                         $data['request'] = $json['dfrn-request'];
1056                 }
1057
1058                 if (!empty($json['dfrn-confirm'])) {
1059                         $data['confirm'] = $json['dfrn-confirm'];
1060                 }
1061
1062                 if (!empty($json['dfrn-notify'])) {
1063                         $data['notify'] = $json['dfrn-notify'];
1064                 }
1065
1066                 if (!empty($json['dfrn-poll'])) {
1067                         $data['poll'] = $json['dfrn-poll'];
1068                 }
1069
1070                 if (isset($json['hide'])) {
1071                         $data['hide'] = (bool)$json['hide'];
1072                 } else {
1073                         $data['hide'] = false;
1074                 }
1075
1076                 return $data;
1077         }
1078
1079         /**
1080          * Check for valid DFRN data
1081          *
1082          * @param array $data DFRN data
1083          *
1084          * @return int Number of errors
1085          */
1086         public static function validDfrn(array $data): int
1087         {
1088                 $errors = 0;
1089                 if (!isset($data['key'])) {
1090                         $errors ++;
1091                 }
1092                 if (!isset($data['dfrn-request'])) {
1093                         $errors ++;
1094                 }
1095                 if (!isset($data['dfrn-confirm'])) {
1096                         $errors ++;
1097                 }
1098                 if (!isset($data['dfrn-notify'])) {
1099                         $errors ++;
1100                 }
1101                 if (!isset($data['dfrn-poll'])) {
1102                         $errors ++;
1103                 }
1104                 return $errors;
1105         }
1106
1107         /**
1108          * Fetch data from a DFRN profile page and via "noscrape"
1109          *
1110          * @param string $profile_link Link to the profile page
1111          * @return array profile data
1112          * @throws HTTPException\InternalServerErrorException
1113          * @throws \ImagickException
1114          */
1115         public static function profile(string $profile_link): array
1116         {
1117                 $data = [];
1118
1119                 Logger::info('Check profile', ['link' => $profile_link]);
1120
1121                 // Fetch data via noscrape - this is faster
1122                 $noscrape_url = str_replace(['/hcard/', '/profile/'], '/noscrape/', $profile_link);
1123                 $data = self::pollNoscrape($noscrape_url, $data);
1124
1125                 if (!isset($data['notify'])
1126                         || !isset($data['confirm'])
1127                         || !isset($data['request'])
1128                         || !isset($data['poll'])
1129                         || !isset($data['name'])
1130                         || !isset($data['photo'])
1131                 ) {
1132                         $data = self::pollHcard($profile_link, $data, true);
1133                 }
1134
1135                 $prof_data = [];
1136
1137                 if (empty($data['addr']) || empty($data['nick'])) {
1138                         $probe_data = self::uri($profile_link);
1139                         $data['addr'] = ($data['addr'] ?? '') ?: $probe_data['addr'];
1140                         $data['nick'] = ($data['nick'] ?? '') ?: $probe_data['nick'];
1141                 }
1142
1143                 $prof_data['addr']         = $data['addr'];
1144                 $prof_data['nick']         = $data['nick'];
1145                 $prof_data['dfrn-request'] = $data['request'] ?? null;
1146                 $prof_data['dfrn-confirm'] = $data['confirm'] ?? null;
1147                 $prof_data['dfrn-notify']  = $data['notify']  ?? null;
1148                 $prof_data['dfrn-poll']    = $data['poll']    ?? null;
1149                 $prof_data['photo']        = $data['photo']   ?? null;
1150                 $prof_data['fn']           = $data['name']    ?? null;
1151                 $prof_data['key']          = $data['pubkey']  ?? null;
1152
1153                 Logger::debug('Result', ['link' => $profile_link, 'data' => $prof_data]);
1154
1155                 return $prof_data;
1156         }
1157
1158         /**
1159          * Check for DFRN contact
1160          *
1161          * @param array $webfinger Webfinger data
1162          * @return array DFRN data
1163          * @throws HTTPException\InternalServerErrorException
1164          */
1165         private static function dfrn(array $webfinger): array
1166         {
1167                 $hcard_url = '';
1168                 $data = [];
1169                 // The array is reversed to take into account the order of preference for same-rel links
1170                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1171                 foreach (array_reverse($webfinger['links']) as $link) {
1172                         if (($link['rel'] == ActivityNamespace::DFRN) && !empty($link['href'])) {
1173                                 $data['network'] = Protocol::DFRN;
1174                         } elseif (($link['rel'] == ActivityNamespace::FEED) && !empty($link['href'])) {
1175                                 $data['poll'] = $link['href'];
1176                         } elseif (($link['rel'] == 'http://webfinger.net/rel/profile-page') && (($link['type'] ?? '') == 'text/html') && !empty($link['href'])) {
1177                                 $data['url'] = $link['href'];
1178                         } elseif (($link['rel'] == 'http://microformats.org/profile/hcard') && !empty($link['href'])) {
1179                                 $hcard_url = $link['href'];
1180                         } elseif (($link['rel'] == ActivityNamespace::POCO) && !empty($link['href'])) {
1181                                 $data['poco'] = $link['href'];
1182                         } elseif (($link['rel'] == 'http://webfinger.net/rel/avatar') && !empty($link['href'])) {
1183                                 $data['photo'] = $link['href'];
1184                         } elseif (($link['rel'] == 'http://joindiaspora.com/seed_location') && !empty($link['href'])) {
1185                                 $data['baseurl'] = trim($link['href'], '/');
1186                         } elseif (($link['rel'] == 'http://joindiaspora.com/guid') && !empty($link['href'])) {
1187                                 $data['guid'] = $link['href'];
1188                         } elseif (($link['rel'] == 'diaspora-public-key') && !empty($link['href'])) {
1189                                 $data['pubkey'] = base64_decode($link['href']);
1190
1191                                 if (strstr($data['pubkey'], 'RSA ')) {
1192                                         $data['pubkey'] = Crypto::rsaToPem($data['pubkey']);
1193                                 }
1194                         }
1195                 }
1196
1197                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
1198                         foreach ($webfinger['aliases'] as $alias) {
1199                                 if (empty($data['url']) && !strstr($alias, '@')) {
1200                                         $data['url'] = $alias;
1201                                 } elseif (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
1202                                         $data['alias'] = $alias;
1203                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1204                                         $data['addr'] = substr($alias, 5);
1205                                 }
1206                         }
1207                 }
1208
1209                 if (!empty($webfinger['subject']) && (substr($webfinger['subject'], 0, 5) == 'acct:')) {
1210                         $data['addr'] = substr($webfinger['subject'], 5);
1211                 }
1212
1213                 if (!isset($data['network']) || ($hcard_url == '')) {
1214                         return [];
1215                 }
1216
1217                 // Fetch data via noscrape - this is faster
1218                 $noscrape_url = str_replace('/hcard/', '/noscrape/', $hcard_url);
1219                 $data = self::pollNoscrape($noscrape_url, $data);
1220
1221                 if (isset($data['notify'])
1222                         && isset($data['confirm'])
1223                         && isset($data['request'])
1224                         && isset($data['poll'])
1225                         && isset($data['name'])
1226                         && isset($data['photo'])
1227                 ) {
1228                         return $data;
1229                 }
1230
1231                 $data = self::pollHcard($hcard_url, $data, true);
1232
1233                 return $data;
1234         }
1235
1236         /**
1237          * Poll the hcard page (Diaspora and Friendica specific)
1238          *
1239          * @param string  $hcard_url Link to the hcard page
1240          * @param array   $data      The already fetched data
1241          * @param boolean $dfrn      Poll DFRN specific data
1242          * @return array hcard data
1243          * @throws HTTPException\InternalServerErrorException
1244          */
1245         private static function pollHcard(string $hcard_url, array $data, bool $dfrn = false): array
1246         {
1247                 $curlResult = DI::httpClient()->get($hcard_url, HttpClientAccept::HTML);
1248                 if ($curlResult->isTimeout()) {
1249                         self::$isTimeout = true;
1250                         return [];
1251                 }
1252                 $content = $curlResult->getBody();
1253                 if (empty($content)) {
1254                         return [];
1255                 }
1256
1257                 $doc = new DOMDocument();
1258                 if (!@$doc->loadHTML($content)) {
1259                         return [];
1260                 }
1261
1262                 $xpath = new DomXPath($doc);
1263
1264                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1265                 if (!is_object($vcards)) {
1266                         return [];
1267                 }
1268
1269                 if (!isset($data['baseurl'])) {
1270                         $data['baseurl'] = '';
1271                 }
1272
1273                 if ($vcards->length > 0) {
1274                         $vcard = $vcards->item(0);
1275
1276                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1277                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1278                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1279                         if (($search->length > 0) && empty($data['guid'])) {
1280                                 $data['guid'] = $search->item(0)->nodeValue;
1281                         }
1282
1283                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1284                         if ($search->length > 0) {
1285                                 $data['nick'] = $search->item(0)->nodeValue;
1286                         }
1287
1288                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1289                         if ($search->length > 0) {
1290                                 $data['name'] = $search->item(0)->nodeValue;
1291                         }
1292
1293                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1294                         if ($search->length > 0) {
1295                                 $data['searchable'] = $search->item(0)->nodeValue;
1296                         }
1297
1298                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1299                         if ($search->length > 0) {
1300                                 $data['pubkey'] = $search->item(0)->nodeValue;
1301                                 if (strstr($data['pubkey'], 'RSA ')) {
1302                                         $data['pubkey'] = Crypto::rsaToPem($data['pubkey']);
1303                                 }
1304                         }
1305
1306                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1307                         if ($search->length > 0) {
1308                                 $data['baseurl'] = trim($search->item(0)->nodeValue, '/');
1309                         }
1310                 }
1311
1312                 $avatar = [];
1313                 if (!empty($vcard)) {
1314                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1315                         foreach ($photos as $photo) {
1316                                 $attr = [];
1317                                 foreach ($photo->attributes as $attribute) {
1318                                         $attr[$attribute->name] = trim($attribute->value);
1319                                 }
1320
1321                                 if (isset($attr['src']) && isset($attr['width'])) {
1322                                         $avatar[$attr['width']] = $attr['src'];
1323                                 }
1324
1325                                 // We don't have a width. So we just take everything that we got.
1326                                 // This is a Hubzilla workaround which doesn't send a width.
1327                                 if ((sizeof($avatar) == 0) && !empty($attr['src'])) {
1328                                         $avatar[] = $attr['src'];
1329                                 }
1330                         }
1331                 }
1332
1333                 if (sizeof($avatar)) {
1334                         ksort($avatar);
1335                         $data['photo'] = self::fixAvatar(array_pop($avatar), $data['baseurl']);
1336                 }
1337
1338                 if ($dfrn) {
1339                         // Poll DFRN specific data
1340                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1341                         if ($search->length > 0) {
1342                                 foreach ($search as $link) {
1343                                         //$data['request'] = $search->item(0)->nodeValue;
1344                                         $attr = [];
1345                                         foreach ($link->attributes as $attribute) {
1346                                                 $attr[$attribute->name] = trim($attribute->value);
1347                                         }
1348
1349                                         $data[substr($attr['rel'], 5)] = $attr['href'];
1350                                 }
1351                         }
1352
1353                         // Older Friendica versions had used the "uid" field differently than newer versions
1354                         if (!empty($data['nick']) && !empty($data['guid']) && ($data['nick'] == $data['guid'])) {
1355                                 unset($data['guid']);
1356                         }
1357                 }
1358
1359
1360                 return $data;
1361         }
1362
1363         /**
1364          * Check for Diaspora contact
1365          *
1366          * @param array $webfinger Webfinger data
1367          *
1368          * @return array Diaspora data
1369          * @throws HTTPException\InternalServerErrorException
1370          */
1371         private static function diaspora(array $webfinger): array
1372         {
1373                 $hcard_url = '';
1374                 $data = [];
1375
1376                 // The array is reversed to take into account the order of preference for same-rel links
1377                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1378                 foreach (array_reverse($webfinger['links']) as $link) {
1379                         if (($link['rel'] == 'http://microformats.org/profile/hcard') && !empty($link['href'])) {
1380                                 $hcard_url = $link['href'];
1381                         } elseif (($link['rel'] == 'http://joindiaspora.com/seed_location') && !empty($link['href'])) {
1382                                 $data['baseurl'] = trim($link['href'], '/');
1383                         } elseif (($link['rel'] == 'http://joindiaspora.com/guid') && !empty($link['href'])) {
1384                                 $data['guid'] = $link['href'];
1385                         } elseif (($link['rel'] == 'http://webfinger.net/rel/profile-page') && (($link['type'] ?? '') == 'text/html') && !empty($link['href'])) {
1386                                 $data['url'] = $link['href'];
1387                         } elseif (($link['rel'] == 'http://webfinger.net/rel/profile-page') && empty($link['type']) && !empty($link['href'])) {
1388                                 $profile_url = $link['href'];
1389                         } elseif (($link['rel'] == ActivityNamespace::FEED) && !empty($link['href'])) {
1390                                 $data['poll'] = $link['href'];
1391                         } elseif (($link['rel'] == ActivityNamespace::POCO) && !empty($link['href'])) {
1392                                 $data['poco'] = $link['href'];
1393                         } elseif (($link['rel'] == 'salmon') && !empty($link['href'])) {
1394                                 $data['notify'] = $link['href'];
1395                         } elseif (($link['rel'] == 'diaspora-public-key') && !empty($link['href'])) {
1396                                 $data['pubkey'] = base64_decode($link['href']);
1397
1398                                 if (strstr($data['pubkey'], 'RSA ')) {
1399                                         $data['pubkey'] = Crypto::rsaToPem($data['pubkey']);
1400                                 }
1401                         }
1402                 }
1403
1404                 if (empty($data['url']) && !empty($profile_url)) {
1405                         $data['url'] = $profile_url;
1406                 }
1407
1408                 if (empty($data['url']) || empty($hcard_url)) {
1409                         return [];
1410                 }
1411
1412                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
1413                         foreach ($webfinger['aliases'] as $alias) {
1414                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data['url']) && ! strstr($alias, '@')) {
1415                                         $data['alias'] = $alias;
1416                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1417                                         $data['addr'] = substr($alias, 5);
1418                                 }
1419                         }
1420                 }
1421
1422                 if (!empty($webfinger['subject']) && (substr($webfinger['subject'], 0, 5) == 'acct:')) {
1423                         $data['addr'] = substr($webfinger['subject'], 5);
1424                 }
1425
1426                 // Fetch further information from the hcard
1427                 $data = self::pollHcard($hcard_url, $data);
1428
1429                 if (!$data) {
1430                         return [];
1431                 }
1432
1433                 if (!empty($data['url'])
1434                         && !empty($data['guid'])
1435                         && !empty($data['baseurl'])
1436                         && !empty($data['pubkey'])
1437                         && !empty($hcard_url)
1438                 ) {
1439                         $data['network'] = Protocol::DIASPORA;
1440                         $data['manually-approve'] = false;
1441
1442                         // The Diaspora handle must always be lowercase
1443                         if (!empty($data['addr'])) {
1444                                 $data['addr'] = strtolower($data['addr']);
1445                         }
1446
1447                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1448                         $data['notify'] = $data['baseurl'] . '/receive/users/' . $data['guid'];
1449                         $data['batch']  = $data['baseurl'] . '/receive/public';
1450                 } else {
1451                         return [];
1452                 }
1453
1454                 return $data;
1455         }
1456
1457         /**
1458          * Check for OStatus contact
1459          *
1460          * @param array $webfinger Webfinger data
1461          * @param bool  $short     Short detection mode
1462          *
1463          * @return array|bool OStatus data or "false" on error or "true" on short mode
1464          * @throws HTTPException\InternalServerErrorException
1465          */
1466         private static function ostatus(array $webfinger, bool $short = false)
1467         {
1468                 $data = [];
1469
1470                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
1471                         foreach ($webfinger['aliases'] as $alias) {
1472                                 if (strstr($alias, '@') && !strstr(Strings::normaliseLink($alias), 'http://')) {
1473                                         $data['addr'] = str_replace('acct:', '', $alias);
1474                                 }
1475                         }
1476                 }
1477
1478                 if (!empty($webfinger['subject']) && strstr($webfinger['subject'], '@')
1479                         && !strstr(Strings::normaliseLink($webfinger['subject']), 'http://')
1480                 ) {
1481                         $data['addr'] = str_replace('acct:', '', $webfinger['subject']);
1482                 }
1483
1484                 if (!empty($webfinger['links'])) {
1485                         // The array is reversed to take into account the order of preference for same-rel links
1486                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1487                         foreach (array_reverse($webfinger['links']) as $link) {
1488                                 if (($link['rel'] == 'http://webfinger.net/rel/profile-page')
1489                                         && (($link['type'] ?? '') == 'text/html')
1490                                         && ($link['href'] != '')
1491                                 ) {
1492                                         $data['url'] = $data['alias'] = $link['href'];
1493                                 } elseif (($link['rel'] == 'salmon') && !empty($link['href'])) {
1494                                         $data['notify'] = $link['href'];
1495                                 } elseif (($link['rel'] == ActivityNamespace::FEED) && !empty($link['href'])) {
1496                                         $data['poll'] = $link['href'];
1497                                 } elseif (($link['rel'] == 'magic-public-key') && !empty($link['href'])) {
1498                                         $pubkey = $link['href'];
1499
1500                                         if (substr($pubkey, 0, 5) === 'data:') {
1501                                                 if (strstr($pubkey, ',')) {
1502                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1503                                                 } else {
1504                                                         $pubkey = substr($pubkey, 5);
1505                                                 }
1506                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1507                                                 $curlResult = DI::httpClient()->get($pubkey, HttpClientAccept::MAGIC_KEY);
1508                                                 if ($curlResult->isTimeout()) {
1509                                                         self::$isTimeout = true;
1510                                                         return $short ? false : [];
1511                                                 }
1512                                                 Logger::debug('Fetched public key', ['Content-Type' => $curlResult->getHeader('Content-Type'), 'url' => $pubkey]);
1513                                                 $pubkey = $curlResult->getBody();
1514                                         }
1515
1516                                         try {
1517                                                 $data['pubkey'] = Salmon::magicKeyToPem($pubkey);
1518                                         } catch (\Throwable $e) {
1519
1520                                         }
1521                                 }
1522                         }
1523                 }
1524
1525                 if (isset($data['notify']) && isset($data['pubkey'])
1526                         && isset($data['poll'])
1527                         && isset($data['url'])
1528                 ) {
1529                         $data['network'] = Protocol::OSTATUS;
1530                         $data['manually-approve'] = false;
1531                 } else {
1532                         return $short ? false : [];
1533                 }
1534
1535                 if ($short) {
1536                         return true;
1537                 }
1538
1539                 // Fetch all additional data from the feed
1540                 $curlResult = DI::httpClient()->get($data['poll'], HttpClientAccept::FEED_XML);
1541                 if ($curlResult->isTimeout()) {
1542                         self::$isTimeout = true;
1543                         return [];
1544                 }
1545                 $feed = $curlResult->getBody();
1546                 $feed_data = Feed::import($feed);
1547                 if (!$feed_data) {
1548                         return [];
1549                 }
1550
1551                 if (!empty($feed_data['header']['author-name'])) {
1552                         $data['name'] = $feed_data['header']['author-name'];
1553                 }
1554                 if (!empty($feed_data['header']['author-nick'])) {
1555                         $data['nick'] = $feed_data['header']['author-nick'];
1556                 }
1557                 if (!empty($feed_data['header']['author-avatar'])) {
1558                         $data['photo'] = self::fixAvatar($feed_data['header']['author-avatar'], $data['url']);
1559                 }
1560                 if (!empty($feed_data['header']['author-id'])) {
1561                         $data['alias'] = $feed_data['header']['author-id'];
1562                 }
1563                 if (!empty($feed_data['header']['author-location'])) {
1564                         $data['location'] = $feed_data['header']['author-location'];
1565                 }
1566                 if (!empty($feed_data['header']['author-about'])) {
1567                         $data['about'] = $feed_data['header']['author-about'];
1568                 }
1569                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1570                 // So we take the value that we just fetched, although the other one worked as well
1571                 if (!empty($feed_data['header']['author-link'])) {
1572                         $data['url'] = $feed_data['header']['author-link'];
1573                 }
1574
1575                 if ($data['url'] == $data['alias']) {
1576                         $data['alias'] = '';
1577                 }
1578
1579                 /// @todo Fetch location and "about" from the feed as well
1580                 return $data;
1581         }
1582
1583         /**
1584          * Fetch data from a pump.io profile page
1585          *
1586          * @param string $profile_link Link to the profile page
1587          *
1588          * @return array Profile data
1589          */
1590         private static function pumpioProfileData(string $profile_link): array
1591         {
1592                 $curlResult = DI::httpClient()->get($profile_link, HttpClientAccept::HTML);
1593                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
1594                         return [];
1595                 }
1596
1597                 $doc = new DOMDocument();
1598                 if (!@$doc->loadHTML($curlResult->getBody())) {
1599                         return [];
1600                 }
1601
1602                 $xpath = new DomXPath($doc);
1603
1604                 $data = [];
1605                 $data['name'] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1606
1607                 if ($data['name'] == '') {
1608                         // This is ugly - but pump.io doesn't seem to know a better way for it
1609                         $data['name'] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1610                         $pos = strpos($data['name'], chr(10));
1611                         if ($pos) {
1612                                 $data['name'] = trim(substr($data['name'], 0, $pos));
1613                         }
1614                 }
1615
1616                 $data['location'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1617
1618                 if ($data['location'] == '') {
1619                         $data['location'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1620                 }
1621
1622                 $data['about'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1623
1624                 if ($data['about'] == '') {
1625                         $data['about'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1626                 }
1627
1628                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1629                 if (!$avatar) {
1630                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1631                 }
1632                 if ($avatar) {
1633                         foreach ($avatar->attributes as $attribute) {
1634                                 if ($attribute->name == 'src') {
1635                                         $data['photo'] = trim($attribute->value);
1636                                 }
1637                         }
1638                 }
1639
1640                 return $data;
1641         }
1642
1643         /**
1644          * Check for pump.io contact
1645          *
1646          * @param array  $webfinger Webfinger data
1647          * @param string $addr
1648          *
1649          * @return array pump.io data
1650          */
1651         private static function pumpio(array $webfinger, string $addr): array
1652         {
1653                 $data = [];
1654                 // The array is reversed to take into account the order of preference for same-rel links
1655                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1656                 foreach (array_reverse($webfinger['links']) as $link) {
1657                         if (($link['rel'] == 'http://webfinger.net/rel/profile-page')
1658                                 && (($link['type'] ?? '') == 'text/html')
1659                                 && ($link['href'] != '')
1660                         ) {
1661                                 $data['url'] = $link['href'];
1662                         } elseif (($link['rel'] == 'activity-inbox') && ($link['href'] != '')) {
1663                                 $data['notify'] = $link['href'];
1664                         } elseif (($link['rel'] == 'activity-outbox') && ($link['href'] != '')) {
1665                                 $data['poll'] = $link['href'];
1666                         } elseif (($link['rel'] == 'dialback') && ($link['href'] != '')) {
1667                                 $data['dialback'] = $link['href'];
1668                         }
1669                 }
1670                 if (isset($data['poll']) && isset($data['notify'])
1671                         && isset($data['dialback'])
1672                         && isset($data['url'])
1673                 ) {
1674                         // by now we use these fields only for the network type detection
1675                         // So we unset all data that isn't used at the moment
1676                         unset($data['dialback']);
1677
1678                         $data['network'] = Protocol::PUMPIO;
1679                 } else {
1680                         return [];
1681                 }
1682
1683                 $profile_data = self::pumpioProfileData($data['url']);
1684
1685                 if (!$profile_data) {
1686                         return [];
1687                 }
1688
1689                 $data = array_merge($data, $profile_data);
1690
1691                 if (($addr != '') && ($data['name'] != '')) {
1692                         $name = trim(str_replace($addr, '', $data['name']));
1693                         if ($name != '') {
1694                                 $data['name'] = $name;
1695                         }
1696                 }
1697
1698                 return $data;
1699         }
1700
1701         /**
1702          * Checks HTML page for RSS feed link
1703          *
1704          * @param string $url  Page link
1705          * @param string $body Page body string
1706          *
1707          * @return string|false Feed link or false if body was invalid HTML document
1708          */
1709         public static function getFeedLink(string $url, string $body)
1710         {
1711                 if (empty($body)) {
1712                         return '';
1713                 }
1714
1715                 $doc = new DOMDocument();
1716                 if (!@$doc->loadHTML($body)) {
1717                         return false;
1718                 }
1719
1720                 $xpath = new DOMXPath($doc);
1721
1722                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1723                 $feedUrl = $feedUrl ?: $xpath->evaluate('string(/html/head/link[@type="application/atom+xml" and @rel="alternate"]/@href)');
1724
1725                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1726
1727                 return $feedUrl;
1728         }
1729
1730         /**
1731          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1732          *
1733          * Loosely based on RFC 1808
1734          *
1735          * @see https://tools.ietf.org/html/rfc1808
1736          *
1737          * @param string   $href  The potential relative href found in the HTML document
1738          * @param string   $base  The HTML document URL
1739          * @param DOMXPath $xpath The HTML document XPath
1740          *
1741          * @return string Absolute URL
1742          */
1743         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath): string
1744         {
1745                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1746                         return $href;
1747                 }
1748
1749                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1750
1751                 $baseParts = parse_url($base);
1752                 if (empty($baseParts['host'])) {
1753                         return $href;
1754                 }
1755
1756                 // Naked domain case (scheme://basehost)
1757                 $path = $baseParts['path'] ?? '/';
1758
1759                 // Remove the filename part of the path if it exists (/base/path/file)
1760                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1761
1762                 $hrefParts = parse_url($href);
1763
1764                 if (!empty($hrefParts['path'])) {
1765                         // Root path case (/path) including relative scheme case (//host/path)
1766                         if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1767                                 $path = $hrefParts['path'];
1768                         } else {
1769                                 $path = $path . '/' . $hrefParts['path'];
1770
1771                                 // Resolve arbitrary relative path
1772                                 // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1773                                 $parts = array_filter(explode('/', $path), 'strlen');
1774                                 $absolutes = [];
1775                                 foreach ($parts as $part) {
1776                                         if ('.' == $part) continue;
1777                                         if ('..' == $part) {
1778                                                 array_pop($absolutes);
1779                                         } else {
1780                                                 $absolutes[] = $part;
1781                                         }
1782                                 }
1783
1784                                 $path = '/' . implode('/', $absolutes);
1785                         }
1786                 }
1787
1788                 // Relative scheme case (//host/path)
1789                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1790                 $baseParts['path'] = $path;
1791                 unset($baseParts['query']);
1792                 unset($baseParts['fragment']);
1793
1794                 return Network::unparseURL($baseParts);
1795         }
1796
1797         /**
1798          * Check for feed contact
1799          *
1800          * @param string  $url   Profile link
1801          * @param boolean $probe Do a probe if the page contains a feed link
1802          *
1803          * @return array feed data
1804          * @throws HTTPException\InternalServerErrorException
1805          */
1806         private static function feed(string $url, bool $probe = true): array
1807         {
1808                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::FEED_XML);
1809                 if ($curlResult->isTimeout()) {
1810                         self::$isTimeout = true;
1811                         return [];
1812                 }
1813                 $feed = $curlResult->getBody();
1814                 $feed_data = Feed::import($feed);
1815
1816                 if (!$feed_data) {
1817                         if (!$probe) {
1818                                 return [];
1819                         }
1820
1821                         $feed_url = self::getFeedLink($url, $feed);
1822
1823                         if (!$feed_url) {
1824                                 return [];
1825                         }
1826
1827                         return self::feed($feed_url, false);
1828                 }
1829
1830                 if (!empty($feed_data['header']['author-name'])) {
1831                         $data['name'] = $feed_data['header']['author-name'];
1832                 }
1833
1834                 if (!empty($feed_data['header']['author-nick'])) {
1835                         $data['nick'] = $feed_data['header']['author-nick'];
1836                 }
1837
1838                 if (!empty($feed_data['header']['author-avatar'])) {
1839                         $data['photo'] = $feed_data['header']['author-avatar'];
1840                 }
1841
1842                 if (!empty($feed_data['header']['author-id'])) {
1843                         $data['alias'] = $feed_data['header']['author-id'];
1844                 }
1845
1846                 $data['url'] = $url;
1847                 $data['poll'] = $url;
1848
1849                 $data['network'] = Protocol::FEED;
1850
1851                 return $data;
1852         }
1853
1854         /**
1855          * Check for mail contact
1856          *
1857          * @param string  $uri Profile link
1858          * @param integer $uid User ID
1859          *
1860          * @return array mail data
1861          * @throws \Exception
1862          */
1863         private static function mail(string $uri, int $uid): array
1864         {
1865                 if (!Network::isEmailDomainValid($uri)) {
1866                         return [];
1867                 }
1868
1869                 if ($uid == 0) {
1870                         return [];
1871                 }
1872
1873                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1874
1875                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1876                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1877                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1878
1879                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1880                         return [];
1881                 }
1882
1883                 $mailbox = Email::constructMailboxName($mailacct);
1884                 $password = '';
1885                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1886                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1887                 if (!$mbox) {
1888                         return [];
1889                 }
1890
1891                 $msgs = Email::poll($mbox, $uri);
1892                 Logger::info('Messages found', ['uri' => $uri, 'count' => count($msgs)]);
1893
1894                 if (!count($msgs)) {
1895                         return [];
1896                 }
1897
1898                 $phost = substr($uri, strpos($uri, '@') + 1);
1899
1900                 $data = [
1901                         'addr'    => $uri,
1902                         'network' => Protocol::MAIL,
1903                         'name'    => substr($uri, 0, strpos($uri, '@')),
1904                         'photo'   => Network::lookupAvatarByEmail($uri),
1905                         'url'     => 'mailto:' . $uri,
1906                         'notify'  => 'smtp ' . Strings::getRandomHex(),
1907                         'poll'    => 'email ' . Strings::getRandomHex(),
1908                 ];
1909
1910                 $data['nick']    = $data['name'];
1911
1912                 $x = Email::messageMeta($mbox, $msgs[0]);
1913
1914                 if (stristr($x[0]->from, $uri)) {
1915                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1916                 } elseif (stristr($x[0]->to, $uri)) {
1917                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1918                 }
1919
1920                 if (isset($adr)) {
1921                         foreach ($adr as $feadr) {
1922                                 if ((strcasecmp($feadr->mailbox, $data['name']) == 0)
1923                                         &&(strcasecmp($feadr->host, $phost) == 0)
1924                                         && (strlen($feadr->personal))
1925                                 ) {
1926                                         $personal = imap_mime_header_decode($feadr->personal);
1927                                         $data['name'] = '';
1928                                         foreach ($personal as $perspart) {
1929                                                 if ($perspart->charset != 'default') {
1930                                                         $data['name'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1931                                                 } else {
1932                                                         $data['name'] .= $perspart->text;
1933                                                 }
1934                                         }
1935                                 }
1936                         }
1937                 }
1938
1939                 if (!empty($mbox)) {
1940                         imap_close($mbox);
1941                 }
1942
1943                 return $data;
1944         }
1945
1946         /**
1947          * Mix two paths together to possibly fix missing parts
1948          *
1949          * @param string $avatar Path to the avatar
1950          * @param string $base   Another path that is hopefully complete
1951          *
1952          * @return string fixed avatar path
1953          * @throws \Exception
1954          */
1955         public static function fixAvatar(string $avatar, string $base): string
1956         {
1957                 $base_parts = parse_url($base);
1958
1959                 // Remove all parts that could create a problem
1960                 unset($base_parts['path']);
1961                 unset($base_parts['query']);
1962                 unset($base_parts['fragment']);
1963
1964                 $avatar_parts = parse_url($avatar);
1965
1966                 // Now we mix them
1967                 $parts = array_merge($base_parts, $avatar_parts);
1968
1969                 // And put them together again
1970                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1971                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1972                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1973                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1974                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1975                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1976
1977                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1978
1979                 Logger::debug('Avatar fixed', ['base' => $base, 'avatar' => $avatar, 'fixed' => $fixed]);
1980
1981                 return $fixed;
1982         }
1983
1984         /**
1985          * Fetch the last date that the contact had posted something (publically)
1986          *
1987          * @param array $data  probing result
1988          *
1989          * @return string last activity
1990          */
1991         public static function getLastUpdate(array $data): string
1992         {
1993                 $uid = User::getIdForURL($data['url']);
1994                 if (!empty($uid)) {
1995                         $contact = Contact::selectFirst(['url', 'last-item'], ['self' => true, 'uid' => $uid]);
1996                         if (!empty($contact['last-item'])) {
1997                                 return $contact['last-item'];
1998                         }
1999                 }
2000
2001                 if ($lastUpdate = self::updateFromNoScrape($data)) {
2002                         return $lastUpdate;
2003                 }
2004
2005                 if (!empty($data['outbox'])) {
2006                         return self::updateFromOutbox($data['outbox'], $data);
2007                 } elseif (!empty($data['poll']) && ($data['network'] == Protocol::ACTIVITYPUB)) {
2008                         return self::updateFromOutbox($data['poll'], $data);
2009                 } elseif (!empty($data['poll'])) {
2010                         return self::updateFromFeed($data);
2011                 }
2012
2013                 return '';
2014         }
2015
2016         /**
2017          * Fetch the last activity date from the "noscrape" endpoint
2018          *
2019          * @param array $data Probing result
2020          *
2021          * @return string last activity or true if update was successful or the server was unreachable
2022          */
2023         private static function updateFromNoScrape(array $data): string
2024         {
2025                 if (empty($data['baseurl'])) {
2026                         return '';
2027                 }
2028
2029                 // Check the 'noscrape' endpoint when it is a Friendica server
2030                 $gserver = DBA::selectFirst('gserver', ['noscrape'], ["`nurl` = ? AND `noscrape` != ''",
2031                         Strings::normaliseLink($data['baseurl'])]);
2032                 if (!DBA::isResult($gserver)) {
2033                         return '';
2034                 }
2035
2036                 $curlResult = DI::httpClient()->get($gserver['noscrape'] . '/' . $data['nick'], HttpClientAccept::JSON);
2037
2038                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
2039                         $noscrape = json_decode($curlResult->getBody(), true);
2040                         if (!empty($noscrape) && !empty($noscrape['updated'])) {
2041                                 return DateTimeFormat::utc($noscrape['updated'], DateTimeFormat::MYSQL);
2042                         }
2043                 }
2044
2045                 return '';
2046         }
2047
2048         /**
2049          * Fetch the last activity date from an ActivityPub Outbox
2050          *
2051          * @param string $feed
2052          * @param array  $data Probing result
2053          *
2054          * @return string last activity
2055          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
2056          */
2057         private static function updateFromOutbox(string $feed, array $data): string
2058         {
2059                 $outbox = ActivityPub::fetchContent($feed);
2060                 if (empty($outbox)) {
2061                         return '';
2062                 }
2063
2064                 if (!empty($outbox['orderedItems'])) {
2065                         $items = $outbox['orderedItems'];
2066                 } elseif (!empty($outbox['first']['orderedItems'])) {
2067                         $items = $outbox['first']['orderedItems'];
2068                 } elseif (!empty($outbox['first']['href']) && ($outbox['first']['href'] != $feed)) {
2069                         return self::updateFromOutbox($outbox['first']['href'], $data);
2070                 } elseif (!empty($outbox['first'])) {
2071                         if (is_string($outbox['first']) && ($outbox['first'] != $feed)) {
2072                                 return self::updateFromOutbox($outbox['first'], $data);
2073                         } else {
2074                                 Logger::warning('Unexpected data', ['outbox' => $outbox]);
2075                         }
2076                         return '';
2077                 } else {
2078                         $items = [];
2079                 }
2080
2081                 $last_updated = '';
2082                 foreach ($items as $activity) {
2083                         if (!empty($activity['published'])) {
2084                                 $published =  DateTimeFormat::utc($activity['published']);
2085                         } elseif (!empty($activity['object']['published'])) {
2086                                 $published =  DateTimeFormat::utc($activity['object']['published']);
2087                         } else {
2088                                 continue;
2089                         }
2090
2091                         if ($last_updated < $published) {
2092                                 $last_updated = $published;
2093                         }
2094                 }
2095
2096                 if (!empty($last_updated)) {
2097                         return $last_updated;
2098                 }
2099
2100                 return '';
2101         }
2102
2103         /**
2104          * Fetch the last activity date from an XML feed
2105          *
2106          * @param array $data Probing result
2107          * @return string last activity
2108          */
2109         private static function updateFromFeed(array $data): string
2110         {
2111                 // Search for the newest entry in the feed
2112                 $curlResult = DI::httpClient()->get($data['poll'], HttpClientAccept::ATOM_XML);
2113                 if (!$curlResult->isSuccess() || !$curlResult->getBody()) {
2114                         return '';
2115                 }
2116
2117                 $doc = new DOMDocument();
2118                 @$doc->loadXML($curlResult->getBody());
2119
2120                 $xpath = new DOMXPath($doc);
2121                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
2122
2123                 $entries = $xpath->query('/atom:feed/atom:entry');
2124
2125                 $last_updated = '';
2126
2127                 foreach ($entries as $entry) {
2128                         $published_item = $xpath->query('atom:published/text()', $entry)->item(0);
2129                         $updated_item   = $xpath->query('atom:updated/text()'  , $entry)->item(0);
2130                         $published      = !empty($published_item->nodeValue) ? DateTimeFormat::utc($published_item->nodeValue) : null;
2131                         $updated        = !empty($updated_item->nodeValue) ? DateTimeFormat::utc($updated_item->nodeValue) : null;
2132
2133                         if (empty($published) || empty($updated)) {
2134                                 Logger::notice('Invalid entry for XPath.', ['entry' => $entry, 'url' => $data['url']]);
2135                                 continue;
2136                         }
2137
2138                         if ($last_updated < $published) {
2139                                 $last_updated = $published;
2140                         }
2141
2142                         if ($last_updated < $updated) {
2143                                 $last_updated = $updated;
2144                         }
2145                 }
2146
2147                 if (!empty($last_updated)) {
2148                         return $last_updated;
2149                 }
2150
2151                 return '';
2152         }
2153
2154         /**
2155          * Probe data from local profiles without network traffic
2156          *
2157          * @param string $url
2158          *
2159          * @return array probed data
2160          * @throws HTTPException\InternalServerErrorException
2161          * @throws HTTPException\NotFoundException
2162          */
2163         private static function localProbe(string $url): array
2164         {
2165                 try {
2166                         $uid = User::getIdForURL($url);
2167                         if (!$uid) {
2168                                 throw new HTTPException\NotFoundException('User not found.');
2169                         }
2170
2171                         $owner     = User::getOwnerDataById($uid);
2172                         $approfile = ActivityPub\Transmitter::getProfile($uid);
2173
2174                         if (empty($owner['gsid'])) {
2175                                 $owner['gsid'] = GServer::getID($approfile['generator']['url']);
2176                         }
2177
2178                         $data = [
2179                                 'name'             => $owner['name'], 'nick' => $owner['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
2180                                 'url'              => $owner['url'], 'addr' => $owner['addr'], 'alias' => $owner['alias'],
2181                                 'photo'            => User::getAvatarUrl($owner),
2182                                 'header'           => $owner['header'] ? Contact::getHeaderUrlForId($owner['id'], $owner['updated']) : '',
2183                                 'account-type'     => $owner['contact-type'], 'community' => ($owner['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
2184                                 'keywords'         => $owner['keywords'], 'location' => $owner['location'], 'about' => $owner['about'],
2185                                 'xmpp'             => $owner['xmpp'], 'matrix' => $owner['matrix'],
2186                                 'hide'             => !$owner['net-publish'], 'batch' => '', 'notify' => $owner['notify'],
2187                                 'poll'             => $owner['poll'], 'request' => $owner['request'], 'confirm' => $owner['confirm'],
2188                                 'subscribe'        => $approfile['generator']['url'] . '/contact/follow?url={uri}', 'poco' => $owner['poco'],
2189                                 'following'        => $approfile['following'], 'followers' => $approfile['followers'],
2190                                 'inbox'            => $approfile['inbox'], 'outbox' => $approfile['outbox'],
2191                                 'sharedinbox'      => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
2192                                 'pubkey'           => $owner['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $owner['gsid'],
2193                                 'manually-approve' => in_array($owner['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])
2194                         ];
2195                 } catch (Exception $e) {
2196                         // Default values for non existing targets
2197                         $data = [
2198                                 'name' => $url, 'nick' => $url, 'url' => $url, 'network' => Protocol::PHANTOM,
2199                                 'photo' => DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO
2200                         ];
2201                 }
2202
2203                 return self::rearrangeData($data);
2204         }
2205 }