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', 'given_name', 'family_name', 'nick', 'guid', 'url', 'addr', 'alias',
 114                         'photo', 'photo_medium', 'photo_small', 'header',
 115                                 'account-type', 'community', 'keywords', 'location', 'about', 'xmpp', 'matrix',
 116                                 'hide', 'batch', 'notify', 'poll', 'request', 'confirm', 'subscribe', 'poco',
 117                                 'following', 'followers', 'inbox', 'outbox', 'sharedinbox',
 118                                 'priority', 'network', 'pubkey', 'manually-approve', 'baseurl', 'gsid'];
 119
 120                 $numeric_fields = ['gsid', 'hide', 'account-type', 'manually-approve'];
 121
 122                 $newdata = [];
 123                 foreach ($fields as $field) {
 124                         if (isset($data[$field])) {
 125                                 if (in_array($field, $numeric_fields)) {
 126                                         $newdata[$field] = (int)$data[$field];
 127                                 } else {
 128                                         $newdata[$field] = trim($data[$field]);
 129                                 }
 130                         } elseif (!in_array($field, $numeric_fields)) {
 131                                 $newdata[$field] = '';
 132                         } else {
 133                                 $newdata[$field] = null;
 134                         }
 135                 }
 136
 137                 $newdata['networks'] = $data['networks'] ?? [];
 138                 foreach ([Protocol::DIASPORA, Protocol::OSTATUS] as $network) {
 139                         if (empty($newdata['networks'][$network])) {
 140                                 unset($newdata['networks'][$network]);
 141                         }
 142                 }
 143
 144                 // We don't use the "priority" field anymore and replace it with a dummy.
 145                 $newdata['priority'] = 0;
 146
 147                 return $newdata;
 148         }
 149
 150         /**
 151          * Check if the hostname belongs to the own server
 152          *
 153          * @param string $host The hostname that is to be checked
 154          * @return bool Does the testes hostname belongs to the own server?
 155          */
 156         private static function ownHost(string $host): bool
 157         {
 158                 $own_host = DI::baseUrl()->getHostname();
 159
 160                 $parts = parse_url($host);
 161
 162                 if (!isset($parts['scheme'])) {
 163                         $parts = parse_url('http://' . $host);
 164                 }
 165
 166                 if (!isset($parts['host'])) {
 167                         return false;
 168                 }
 169                 return $parts['host'] == $own_host;
 170         }
 171
 172         /**
 173          * Probes for webfinger path via "host-meta"
 174          *
 175          * We have to check if the servers in the future still will offer this.
 176          * It seems as if it was dropped from the standard.
 177          *
 178          * @param string $host The host part of an url
 179          *
 180          * @return array with template and type of the webfinger template for JSON or XML
 181          * @throws HTTPException\InternalServerErrorException
 182          */
 183         private static function hostMeta(string $host): array
 184         {
 185                 // Reset the static variable
 186                 self::$baseurl = '';
 187
 188                 // Handles the case when the hostname contains the scheme
 189                 if (!parse_url($host, PHP_URL_SCHEME)) {
 190                         $ssl_url = 'https://' . $host . self::HOST_META;
 191                         $url = 'http://' . $host . self::HOST_META;
 192                 } else {
 193                         $ssl_url = $host . self::HOST_META;
 194                         $url = '';
 195                 }
 196
 197                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 198
 199                 Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
 200                 $xrd = null;
 201
 202                 $curlResult = DI::httpClient()->get($ssl_url, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 203                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 204                 if ($curlResult->isSuccess()) {
 205                         $xml = $curlResult->getBody();
 206                         $xrd = XML::parseString($xml, true);
 207                         if (!empty($url)) {
 208                                 $host_url = 'https://' . $host;
 209                         } else {
 210                                 $host_url = $host;
 211                         }
 212                 } elseif ($curlResult->isTimeout()) {
 213                         Logger::info('Probing timeout', ['url' => $ssl_url]);
 214                         self::$isTimeout = true;
 215                         return [];
 216                 }
 217
 218                 if (!is_object($xrd) && !empty($url)) {
 219                         $curlResult = DI::httpClient()->get($url, HttpClientAccept::XRD_XML, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 220                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 221                         if ($curlResult->isTimeout()) {
 222                                 Logger::info('Probing timeout', ['url' => $url]);
 223                                 self::$isTimeout = true;
 224                                 return [];
 225                         } elseif ($connection_error && $ssl_connection_error) {
 226                                 self::$isTimeout = true;
 227                                 return [];
 228                         }
 229
 230                         $xml = $curlResult->getBody();
 231                         $xrd = XML::parseString($xml, true);
 232                         $host_url = 'http://'.$host;
 233                 }
 234                 if (!is_object($xrd)) {
 235                         Logger::info('No xrd object found', ['host' => $host]);
 236                         return [];
 237                 }
 238
 239                 $links = XML::elementToArray($xrd);
 240                 if (!isset($links['xrd']['link'])) {
 241                         Logger::info('No xrd data found', ['host' => $host]);
 242                         return [];
 243                 }
 244
 245                 $lrdd = [];
 246
 247                 foreach ($links['xrd']['link'] as $value => $link) {
 248                         if (!empty($link['@attributes'])) {
 249                                 $attributes = $link['@attributes'];
 250                         } elseif ($value == '@attributes') {
 251                                 $attributes = $link;
 252                         } else {
 253                                 continue;
 254                         }
 255
 256                         if (!empty($attributes['rel']) && $attributes['rel'] == 'lrdd' && !empty($attributes['template'])) {
 257                                 $type = (empty($attributes['type']) ? '' : $attributes['type']);
 258
 259                                 $lrdd[$type] = $attributes['template'];
 260                         }
 261                 }
 262
 263                 if (Network::isUrlBlocked($host_url)) {
 264                         Logger::info('Domain is blocked', ['url' => $host]);
 265                         return [];
 266                 }
 267
 268                 self::$baseurl = $host_url;
 269
 270                 Logger::info('Probing successful', ['host' => $host]);
 271
 272                 return $lrdd;
 273         }
 274
 275         /**
 276          * Check an URI for LRDD data
 277          *
 278          * @param string $uri     Address that should be probed
 279          * @return array uri data
 280          * @throws HTTPException\InternalServerErrorException
 281          */
 282         public static function lrdd(string $uri): array
 283         {
 284                 $data = self::getWebfingerArray($uri);
 285                 if (empty($data)) {
 286                         return [];
 287                 }
 288                 $webfinger = $data['webfinger'];
 289
 290                 if (empty($webfinger['links'])) {
 291                         Logger::info('No webfinger links found', ['uri' => $uri]);
 292                         return [];
 293                 }
 294
 295                 $data = [];
 296
 297                 foreach ($webfinger['links'] as $link) {
 298                         $data[] = ['@attributes' => $link];
 299                 }
 300
 301                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 302                         foreach ($webfinger['aliases'] as $alias) {
 303                                 $data[] = [
 304                                         '@attributes' => [
 305                                                 'rel' => 'alias',
 306                                                 'href' => $alias,
 307                                         ]
 308                                 ];
 309                         }
 310                 }
 311
 312                 return $data;
 313         }
 314
 315         /**
 316          * Fetch information (protocol endpoints and user information) about a given uri
 317          *
 318          * @param string  $uri     Address that should be probed
 319          * @param string  $network Test for this specific network
 320          * @param integer $uid     User ID for the probe (only used for mails)
 321          * @param boolean $cache   Use cached values?
 322          *
 323          * @return array uri data
 324          * @throws HTTPException\InternalServerErrorException
 325          * @throws \ImagickException
 326          */
 327         public static function uri(string $uri, string $network = '', int $uid = -1): array
 328         {
 329                 // Local profiles aren't probed via network
 330                 if (empty($network) && Contact::isLocal($uri)) {
 331                         $data = self::localProbe($uri);
 332                         if (!empty($data)) {
 333                                 return $data;
 334                         }
 335                 }
 336
 337                 if ($uid == -1) {
 338                         $uid = DI::userSession()->getLocalUserId();
 339                 }
 340
 341                 if (empty($network) || ($network == Protocol::ACTIVITYPUB)) {
 342                         $ap_profile = ActivityPub::probeProfile($uri);
 343                 } else {
 344                         $ap_profile = [];
 345                 }
 346
 347                 self::$isTimeout = false;
 348
 349                 if ($network != Protocol::ACTIVITYPUB) {
 350                         $data = self::detect($uri, $network, $uid, $ap_profile);
 351                         if (!is_array($data)) {
 352                                 $data = [];
 353                         }
 354                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 355                                 $networks = $data['networks'] ?? [];
 356                                 unset($data['networks']);
 357                                 $networks[$data['network']] = $data;
 358                                 $data = $ap_profile;
 359                                 $data['networks'] = $networks;
 360                         } elseif (!empty($ap_profile)) {
 361                                 $ap_profile['batch'] = '';
 362                                 $data = array_merge($ap_profile, $data);
 363                         }
 364                 } else {
 365                         $data = $ap_profile;
 366                 }
 367
 368                 if (!isset($data['url'])) {
 369                         $data['url'] = $uri;
 370                 }
 371
 372                 if (empty($data['photo'])) {
 373                         $data['photo'] = DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO;
 374                 }
 375
 376                 if (empty($data['name'])) {
 377                         if (!empty($data['nick'])) {
 378                                 $data['name'] = $data['nick'];
 379                         }
 380
 381                         if (empty($data['name'])) {
 382                                 $data['name'] = $data['url'];
 383                         }
 384                 }
 385
 386                 if (empty($data['nick'])) {
 387                         $data['nick'] = strtolower($data['name']);
 388
 389                         if (strpos($data['nick'], ' ')) {
 390                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 391                         }
 392                 }
 393
 394                 if (!empty($data['baseurl']) && empty($data['gsid'])) {
 395                         $data['gsid'] = GServer::getID($data['baseurl']);
 396                 }
 397
 398                 if (empty($data['network'])) {
 399                         $data['network'] = Protocol::PHANTOM;
 400                 }
 401
 402                 // Ensure that local connections always are DFRN
 403                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
 404                         $data['network'] = Protocol::DFRN;
 405                 }
 406
 407                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 408                         $data['hide'] = self::getHideStatus($data['url']);
 409                 }
 410
 411                 return self::rearrangeData($data);
 412         }
 413
 414
 415         /**
 416          * Fetches the "hide" status from the profile
 417          *
 418          * @param string $url URL of the profile
 419          * @return boolean "hide" status
 420          */
 421         private static function getHideStatus(string $url): bool
 422         {
 423                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::HTML, [HttpClientOptions::CONTENT_LENGTH => 1000000]);
 424                 if (!$curlResult->isSuccess()) {
 425                         return false;
 426                 }
 427
 428                 // If it isn't a HTML file then exit
 429                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 430                         return false;
 431                 }
 432
 433                 $body = $curlResult->getBody();
 434                 if (empty($body)) {
 435                         return false;
 436                 }
 437
 438                 $doc = new DOMDocument();
 439                 @$doc->loadHTML($body);
 440
 441                 $xpath = new DOMXPath($doc);
 442
 443                 $list = $xpath->query('//meta[@name]');
 444                 foreach ($list as $node) {
 445                         $meta_tag = [];
 446                         if ($node->attributes->length) {
 447                                 foreach ($node->attributes as $attribute) {
 448                                         $meta_tag[$attribute->name] = $attribute->value;
 449                                 }
 450                         }
 451
 452                         if (empty($meta_tag['content'])) {
 453                                 continue;
 454                         }
 455
 456                         $content = strtolower(trim($meta_tag['content']));
 457
 458                         switch (strtolower(trim($meta_tag['name']))) {
 459                                 case 'dfrn-global-visibility':
 460                                         if ($content == 'false') {
 461                                                 return true;
 462                                         }
 463                                         break;
 464                                 case 'robots':
 465                                         if (strpos($content, 'noindex') !== false) {
 466                                                 return true;
 467                                         }
 468                                         break;
 469                         }
 470                 }
 471
 472                 return false;
 473         }
 474
 475         /**
 476          * Fetch the "subscribe" and add it to the result
 477          *
 478          * @param array $result Result array
 479          * @param array $webfinger Webfinger data
 480          *
 481          * @return array result Altered/unaltered result array
 482          */
 483         private static function getSubscribeLink(array $result, array $webfinger): array
 484         {
 485                 if (empty($webfinger['links'])) {
 486                         return $result;
 487                 }
 488
 489                 foreach ($webfinger['links'] as $link) {
 490                         if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
 491                                 $result['subscribe'] = $link['template'];
 492                         }
 493                 }
 494
 495                 return $result;
 496         }
 497
 498         /**
 499          * Get webfinger data from a given URI
 500          *
 501          * @param string $uri URI
 502          *
 503          * @return array Webfinger data
 504          * @throws HTTPException\InternalServerErrorException
 505          */
 506         private static function getWebfingerArray(string $uri): array
 507         {
 508                 $parts = parse_url($uri);
 509
 510                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 511                         $host = $parts['host'];
 512                         if (!empty($parts['port'])) {
 513                                 $host .= ':' . $parts['port'];
 514                         }
 515
 516                         $baseurl = $parts['scheme'] . '://' . $host;
 517
 518                         $nick = '';
 519                         $addr = '';
 520
 521                         $path_parts = explode('/', trim($parts['path'] ?? '', '/'));
 522                         if (!empty($path_parts)) {
 523                                 $nick = ltrim(end($path_parts), '@');
 524                                 $addr = $nick . '@' . $host;
 525                         }
 526
 527                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 528                         if (empty($webfinger)) {
 529                                 $lrdd = self::hostMeta($host);
 530                         }
 531
 532                         if (empty($webfinger) && empty($lrdd)) {
 533                                 while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
 534                                         $host    .= '/' . array_shift($path_parts);
 535                                         $baseurl = $parts['scheme'] . '://' . $host;
 536
 537                                         if (!empty($nick)) {
 538                                                 $addr = $nick . '@' . $host;
 539                                         }
 540
 541                                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 542                                         if (empty($webfinger)) {
 543                                                 $lrdd = self::hostMeta($host);
 544                                         }
 545                                 }
 546
 547                                 if (empty($lrdd) && empty($webfinger)) {
 548                                         return [];
 549                                 }
 550                         }
 551                 } elseif (strstr($uri, '@')) {
 552                         // Remove "acct:" from the URI
 553                         $uri = str_replace('acct:', '', $uri);
 554
 555                         $host = substr($uri, strpos($uri, '@') + 1);
 556                         $nick = substr($uri, 0, strpos($uri, '@'));
 557                         $addr = $uri;
 558
 559                         $webfinger = self::getWebfinger('https://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 560                         if (self::$isTimeout) {
 561                                 return [];
 562                         }
 563
 564                         if (empty($webfinger)) {
 565                                 $webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, HttpClientAccept::JRD_JSON, $uri, $addr);
 566                                 if (self::$isTimeout) {
 567                                         return [];
 568                                 }
 569                         } else {
 570                                 $baseurl = 'https://' . $host;
 571                         }
 572
 573                         if (empty($webfinger)) {
 574                                 $lrdd = self::hostMeta($host);
 575                                 if (self::$isTimeout) {
 576                                         return [];
 577                                 }
 578                                 $baseurl = self::$baseurl;
 579                         } else {
 580                                 $baseurl = 'http://' . $host;
 581                         }
 582                 } else {
 583                         Logger::info('URI was not detectable', ['uri' => $uri]);
 584                         return [];
 585                 }
 586
 587                 if (empty($webfinger)) {
 588                         foreach ($lrdd as $type => $template) {
 589                                 if ($webfinger) {
 590                                         continue;
 591                                 }
 592
 593                                 $webfinger = self::getWebfinger($template, $type, $uri, $addr);
 594                         }
 595                 }
 596
 597                 if (empty($webfinger)) {
 598                         return [];
 599                 }
 600
 601                 if ($webfinger['detected'] == $addr) {
 602                         $webfinger['nick'] = $nick;
 603                         $webfinger['addr'] = $addr;
 604                 }
 605
 606                 $webfinger['baseurl'] = $baseurl;
 607
 608                 return $webfinger;
 609         }
 610
 611         /**
 612          * Perform network request for webfinger data
 613          *
 614          * @param string $template
 615          * @param string $type
 616          * @param string $uri
 617          * @param string $addr
 618          *
 619          * @return array webfinger results
 620          */
 621         private static function getWebfinger(string $template, string $type, string $uri, string $addr): array
 622         {
 623                 if (Network::isUrlBlocked($template)) {
 624                         Logger::info('Domain is blocked', ['url' => $template]);
 625                         return [];
 626                 }
 627
 628                 // First try the address because this is the primary purpose of webfinger
 629                 if (!empty($addr)) {
 630                         $detected = $addr;
 631                         $path = str_replace('{uri}', urlencode('acct:' . $addr), $template);
 632                         $webfinger = self::webfinger($path, $type);
 633                         if (self::$isTimeout) {
 634                                 return [];
 635                         }
 636                 }
 637
 638                 // Then try the URI
 639                 if (empty($webfinger) && $uri != $addr) {
 640                         $detected = $uri;
 641                         $path = str_replace('{uri}', urlencode($uri), $template);
 642                         $webfinger = self::webfinger($path, $type);
 643                         if (self::$isTimeout) {
 644                                 return [];
 645                         }
 646                 }
 647
 648                 if (empty($webfinger)) {
 649                         return [];
 650                 }
 651
 652                 return ['webfinger' => $webfinger, 'detected' => $detected];
 653         }
 654
 655         /**
 656          * Fetch information (protocol endpoints and user information) about a given uri
 657          *
 658          * This function is only called by the "uri" function that adds caching and rearranging of data.
 659          *
 660          * @param string  $uri        Address that should be probed
 661          * @param string  $network    Test for this specific network
 662          * @param integer $uid        User ID for the probe (only used for mails)
 663          * @param array   $ap_profile Previously probed AP profile
 664          * @return array URI data
 665          * @throws HTTPException\InternalServerErrorException
 666          */
 667         private static function detect(string $uri, string $network, int $uid, array $ap_profile): array
 668         {
 669                 $hookData = [
 670                         'uri'     => $uri,
 671                         'network' => $network,
 672                         'uid'     => $uid,
 673                         'result'  => null,
 674                 ];
 675
 676                 Hook::callAll('probe_detect', $hookData);
 677
 678                 if (isset($hookData['result'])) {
 679                         return is_array($hookData['result']) ? $hookData['result'] : [];
 680                 }
 681
 682                 $parts = parse_url($uri);
 683                 if (empty($parts['scheme']) && empty($parts['host']) && !strstr($parts['path'], '@')) {
 684                         Logger::info('URI was not detectable', ['uri' => $uri]);
 685                         return [];
 686                 }
 687
 688                 // If the URI starts with "mailto:" then jump directly to the mail detection
 689                 if (strpos($uri, 'mailto:') !== false) {
 690                         $uri = str_replace('mailto:', '', $uri);
 691                         return self::mail($uri, $uid);
 692                 }
 693
 694                 if ($network == Protocol::MAIL) {
 695                         return self::mail($uri, $uid);
 696                 }
 697
 698                 Logger::info('Probing start', ['uri' => $uri]);
 699
 700                 if (!empty($ap_profile['addr']) && ($ap_profile['addr'] != $uri)) {
 701                         $data = self::getWebfingerArray($ap_profile['addr']);
 702                 }
 703
 704                 if (empty($data)) {
 705                         $data = self::getWebfingerArray($uri);
 706                 }
 707
 708                 if (empty($data)) {
 709                         if (!empty($parts['scheme'])) {
 710                                 return self::feed($uri);
 711                         } elseif (!empty($uid)) {
 712                                 return self::mail($uri, $uid);
 713                         } else {
 714                                 return [];
 715                         }
 716                 }
 717
 718                 $webfinger = $data['webfinger'];
 719                 $nick = $data['nick'] ?? '';
 720                 $addr = $data['addr'] ?? '';
 721                 $baseurl = $data['baseurl'] ?? '';
 722
 723                 $result = [];
 724
 725                 if (in_array($network, ['', Protocol::DFRN])) {
 726                         $result = self::dfrn($webfinger);
 727                 }
 728                 if ((!$result && ($network == '')) || ($network == Protocol::DIASPORA)) {
 729                         $result = self::diaspora($webfinger);
 730                 } else {
 731                         $result['networks'][Protocol::DIASPORA] = self::diaspora($webfinger);
 732                 }
 733                 if ((!$result && ($network == '')) || ($network == Protocol::OSTATUS)) {
 734                         $result = self::ostatus($webfinger);
 735                 } else {
 736                         $result['networks'][Protocol::OSTATUS] = self::ostatus($webfinger);
 737                 }
 738                 if (in_array($network, ['', Protocol::ZOT])) {
 739                         $result = self::zot($webfinger, $result, $baseurl);
 740                 }
 741                 if ((!$result && ($network == '')) || ($network == Protocol::PUMPIO)) {
 742                         $result = self::pumpio($webfinger, $addr);
 743                 }
 744                 if (empty($result['network']) && empty($ap_profile['network']) || ($network == Protocol::FEED)) {
 745                         $result = self::feed($uri);
 746                 } else {
 747                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 748                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 749                         if ((empty($result['nick']) || (strstr($result['nick'], ' '))) && ($nick != '')) {
 750                                 $result['nick'] = $nick;
 751                         }
 752
 753                         if (empty($result['addr']) && ($addr != '')) {
 754                                 $result['addr'] = $addr;
 755                         }
 756                 }
 757
 758                 $result = self::getSubscribeLink($result, $webfinger);
 759
 760                 if (empty($result['network'])) {
 761                         $result['network'] = Protocol::PHANTOM;
 762                 }
 763
 764                 if (empty($result['baseurl']) && !empty($baseurl)) {
 765                         $result['baseurl'] = $baseurl;
 766                 }
 767
 768                 if (empty($result['url'])) {
 769                         $result['url'] = $uri;
 770                 }
 771
 772                 Logger::info('Probing done', ['uri' => $uri, 'network' => $result['network']]);
 773
 774                 return $result;
 775         }
 776
 777         /**
 778          * Check for Zot contact
 779          *
 780          * @param array  $webfinger Webfinger data
 781          * @param array  $data      previously probed data
 782          * @param string $baseUrl   Base URL
 783          *
 784          * @return array Zot data
 785          * @throws HTTPException\InternalServerErrorException
 786          */
 787         private static function zot(array $webfinger, array $data, string $baseurl): array
 788         {
 789                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 790                         foreach ($webfinger['aliases'] as $alias) {
 791                                 if (substr($alias, 0, 5) == 'acct:') {
 792                                         $data['addr'] = substr($alias, 5);
 793                                 }
 794                         }
 795                 }
 796
 797                 if (!empty($webfinger['subject']) && (substr($webfinger['subject'], 0, 5) == 'acct:')) {
 798                         $data['addr'] = substr($webfinger['subject'], 5);
 799                 }
 800
 801                 $zot_url = '';
 802                 foreach ($webfinger['links'] as $link) {
 803                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 804                                 $zot_url = $link['href'];
 805                         }
 806                 }
 807
 808                 if (empty($zot_url) && !empty($data['addr']) && !empty($baseurl)) {
 809                         $condition = ['nurl' => Strings::normaliseLink($baseurl), 'platform' => ['hubzilla']];
 810                         if (!DBA::exists('gserver', $condition)) {
 811                                 return $data;
 812                         }
 813                         $zot_url = $baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 814                 }
 815
 816                 if (empty($zot_url)) {
 817                         return $data;
 818                 }
 819
 820                 $data = self::pollZot($zot_url, $data);
 821
 822                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 823                         foreach ($webfinger['aliases'] as $alias) {
 824                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
 825                                         $data['alias'] = $alias;
 826                                 }
 827                         }
 828                 }
 829
 830                 return $data;
 831         }
 832
 833         public static function pollZot(string $url, array $data): array
 834         {
 835                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::JSON);
 836                 if ($curlResult->isTimeout()) {
 837                         return $data;
 838                 }
 839                 $content = $curlResult->getBody();
 840                 if (!$content) {
 841                         return $data;
 842                 }
 843
 844                 $json = json_decode($content, true);
 845                 if (!is_array($json)) {
 846                         return $data;
 847                 }
 848
 849                 if (empty($data['network'])) {
 850                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 851                                 $data['network'] = Protocol::ZOT;
 852                         } elseif (!isset($json['protocols'])) {
 853                                 $data['network'] = Protocol::ZOT;
 854                         }
 855                 }
 856
 857                 if (!empty($json['guid']) && empty($data['guid'])) {
 858                         $data['guid'] = $json['guid'];
 859                 }
 860                 if (!empty($json['key']) && empty($data['pubkey'])) {
 861                         $data['pubkey'] = $json['key'];
 862                 }
 863                 if (!empty($json['name'])) {
 864                         $data['name'] = $json['name'];
 865                 }
 866                 if (!empty($json['photo'])) {
 867                         $data['photo'] = $json['photo'];
 868                         if (!empty($json['photo_updated'])) {
 869                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 870                         }
 871                 }
 872                 if (!empty($json['address'])) {
 873                         $data['addr'] = $json['address'];
 874                 }
 875                 if (!empty($json['url'])) {
 876                         $data['url'] = $json['url'];
 877                 }
 878                 if (!empty($json['connections_url'])) {
 879                         $data['poco'] = $json['connections_url'];
 880                 }
 881                 if (isset($json['searchable'])) {
 882                         $data['hide'] = !$json['searchable'];
 883                 }
 884                 if (!empty($json['public_forum'])) {
 885                         $data['community'] = $json['public_forum'];
 886                         $data['account-type'] = User::PAGE_FLAGS_COMMUNITY;
 887                 }
 888
 889                 if (!empty($json['profile'])) {
 890                         $profile = $json['profile'];
 891                         if (!empty($profile['description'])) {
 892                                 $data['about'] = $profile['description'];
 893                         }
 894                         if (!empty($profile['keywords'])) {
 895                                 $keywords = implode(', ', $profile['keywords']);
 896                                 if (!empty($keywords)) {
 897                                         $data['keywords'] = $keywords;
 898                                 }
 899                         }
 900
 901                         $loc = [];
 902                         if (!empty($profile['region'])) {
 903                                 $loc['region'] = $profile['region'];
 904                         }
 905                         if (!empty($profile['country'])) {
 906                                 $loc['country-name'] = $profile['country'];
 907                         }
 908                         $location = Profile::formatLocation($loc);
 909                         if (!empty($location)) {
 910                                 $data['location'] = $location;
 911                         }
 912                 }
 913
 914                 return $data;
 915         }
 916
 917         /**
 918          * Perform a webfinger request.
 919          *
 920          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 921          *
 922          * @param string $url  Address that should be probed
 923          * @param string $type type
 924          *
 925          * @return array webfinger data
 926          * @throws HTTPException\InternalServerErrorException
 927          */
 928         public static function webfinger(string $url, string $type): array
 929         {
 930                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 931
 932                 $curlResult = DI::httpClient()->get($url, $type, [HttpClientOptions::TIMEOUT => $xrd_timeout]);
 933                 if ($curlResult->isTimeout()) {
 934                         self::$isTimeout = true;
 935                         return [];
 936                 }
 937                 $data = $curlResult->getBody();
 938
 939                 $webfinger = json_decode($data, true);
 940                 if (!empty($webfinger)) {
 941                         if (!isset($webfinger['links'])) {
 942                                 Logger::info('No json webfinger links', ['url' => $url]);
 943                                 return [];
 944                         }
 945                         return $webfinger;
 946                 }
 947
 948                 // If it is not JSON, maybe it is XML
 949                 $xrd = XML::parseString($data, true);
 950                 if (!is_object($xrd)) {
 951                         Logger::info('No webfinger data retrievable', ['url' => $url]);
 952                         return [];
 953                 }
 954
 955                 $xrd_arr = XML::elementToArray($xrd);
 956                 if (!isset($xrd_arr['xrd']['link'])) {
 957                         Logger::info('No XML webfinger links', ['url' => $url]);
 958                         return [];
 959                 }
 960
 961                 $webfinger = [];
 962
 963                 if (!empty($xrd_arr['xrd']['subject'])) {
 964                         $webfinger['subject'] = $xrd_arr['xrd']['subject'];
 965                 }
 966
 967                 if (!empty($xrd_arr['xrd']['alias'])) {
 968                         $webfinger['aliases'] = $xrd_arr['xrd']['alias'];
 969                 }
 970
 971                 $webfinger['links'] = [];
 972
 973                 foreach ($xrd_arr['xrd']['link'] as $value => $data) {
 974                         if (!empty($data['@attributes'])) {
 975                                 $attributes = $data['@attributes'];
 976                         } elseif ($value == '@attributes') {
 977                                 $attributes = $data;
 978                         } else {
 979                                 continue;
 980                         }
 981
 982                         $webfinger['links'][] = $attributes;
 983                 }
 984                 return $webfinger;
 985         }
 986
 987         /**
 988          * Poll the Friendica specific noscrape page.
 989          *
 990          * "noscrape" is a faster alternative to fetch the data from the hcard.
 991          * This functionality was originally created for the directory.
 992          *
 993          * @param string $noscrape_url Link to the noscrape page
 994          * @param array  $data         The already fetched data
 995          *
 996          * @return array noscrape data
 997          * @throws HTTPException\InternalServerErrorException
 998          */
 999         private static function pollNoscrape(string $noscrape_url, array $data): array
1000         {
1001                 $curlResult = DI::httpClient()->get($noscrape_url, HttpClientAccept::JSON);
1002                 if ($curlResult->isTimeout()) {
1003                         self::$isTimeout = true;
1004                         return $data;
1005                 }
1006                 $content = $curlResult->getBody();
1007                 if (!$content) {
1008                         Logger::info('Empty body', ['url' => $noscrape_url]);
1009                         return $data;
1010                 }
1011
1012                 $json = json_decode($content, true);
1013                 if (!is_array($json)) {
1014                         Logger::info('No json data', ['url' => $noscrape_url]);
1015                         return $data;
1016                 }
1017
1018                 if (!empty($json['fn'])) {
1019                         $data['name'] = $json['fn'];
1020                 }
1021
1022                 if (!empty($json['addr'])) {
1023                         $data['addr'] = $json['addr'];
1024                 }
1025
1026                 if (!empty($json['nick'])) {
1027                         $data['nick'] = $json['nick'];
1028                 }
1029
1030                 if (!empty($json['guid'])) {
1031                         $data['guid'] = $json['guid'];
1032                 }
1033
1034                 if (!empty($json['comm'])) {
1035                         $data['community'] = $json['comm'];
1036                 }
1037
1038                 if (!empty($json['tags'])) {
1039                         $keywords = implode(', ', $json['tags']);
1040                         if ($keywords != '') {
1041                                 $data['keywords'] = $keywords;
1042                         }
1043                 }
1044
1045                 $location = Profile::formatLocation($json);
1046                 if ($location) {
1047                         $data['location'] = $location;
1048                 }
1049
1050                 if (!empty($json['about'])) {
1051                         $data['about'] = $json['about'];
1052                 }
1053
1054                 if (!empty($json['xmpp'])) {
1055                         $data['xmpp'] = $json['xmpp'];
1056                 }
1057
1058                 if (!empty($json['matrix'])) {
1059                         $data['matrix'] = $json['matrix'];
1060                 }
1061
1062                 if (!empty($json['key'])) {
1063                         $data['pubkey'] = $json['key'];
1064                 }
1065
1066                 if (!empty($json['photo'])) {
1067                         $data['photo'] = $json['photo'];
1068                 }
1069
1070                 if (!empty($json['dfrn-request'])) {
1071                         $data['request'] = $json['dfrn-request'];
1072                 }
1073
1074                 if (!empty($json['dfrn-confirm'])) {
1075                         $data['confirm'] = $json['dfrn-confirm'];
1076                 }
1077
1078                 if (!empty($json['dfrn-notify'])) {
1079                         $data['notify'] = $json['dfrn-notify'];
1080                 }
1081
1082                 if (!empty($json['dfrn-poll'])) {
1083                         $data['poll'] = $json['dfrn-poll'];
1084                 }
1085
1086                 if (isset($json['hide'])) {
1087                         $data['hide'] = (bool)$json['hide'];
1088                 } else {
1089                         $data['hide'] = false;
1090                 }
1091
1092                 return $data;
1093         }
1094
1095         /**
1096          * Check for valid DFRN data
1097          *
1098          * @param array $data DFRN data
1099          *
1100          * @return int Number of errors
1101          */
1102         public static function validDfrn(array $data): int
1103         {
1104                 $errors = 0;
1105                 if (!isset($data['key'])) {
1106                         $errors ++;
1107                 }
1108                 if (!isset($data['dfrn-request'])) {
1109                         $errors ++;
1110                 }
1111                 if (!isset($data['dfrn-confirm'])) {
1112                         $errors ++;
1113                 }
1114                 if (!isset($data['dfrn-notify'])) {
1115                         $errors ++;
1116                 }
1117                 if (!isset($data['dfrn-poll'])) {
1118                         $errors ++;
1119                 }
1120                 return $errors;
1121         }
1122
1123         /**
1124          * Fetch data from a DFRN profile page and via "noscrape"
1125          *
1126          * @param string $profile_link Link to the profile page
1127          * @return array profile data
1128          * @throws HTTPException\InternalServerErrorException
1129          * @throws \ImagickException
1130          */
1131         public static function profile(string $profile_link): array
1132         {
1133                 $data = [];
1134
1135                 Logger::info('Check profile', ['link' => $profile_link]);
1136
1137                 // Fetch data via noscrape - this is faster
1138                 $noscrape_url = str_replace(['/hcard/', '/profile/'], '/noscrape/', $profile_link);
1139                 $data = self::pollNoscrape($noscrape_url, $data);
1140
1141                 if (!isset($data['notify'])
1142                         || !isset($data['confirm'])
1143                         || !isset($data['request'])
1144                         || !isset($data['poll'])
1145                         || !isset($data['name'])
1146                         || !isset($data['photo'])
1147                 ) {
1148                         $data = self::pollHcard($profile_link, $data, true);
1149                 }
1150
1151                 $prof_data = [];
1152
1153                 if (empty($data['addr']) || empty($data['nick'])) {
1154                         $probe_data = self::uri($profile_link);
1155                         $data['addr'] = ($data['addr'] ?? '') ?: $probe_data['addr'];
1156                         $data['nick'] = ($data['nick'] ?? '') ?: $probe_data['nick'];
1157                 }
1158
1159                 $prof_data['addr']         = $data['addr'];
1160                 $prof_data['nick']         = $data['nick'];
1161                 $prof_data['dfrn-request'] = $data['request'] ?? null;
1162                 $prof_data['dfrn-confirm'] = $data['confirm'] ?? null;
1163                 $prof_data['dfrn-notify']  = $data['notify']  ?? null;
1164                 $prof_data['dfrn-poll']    = $data['poll']    ?? null;
1165                 $prof_data['photo']        = $data['photo']   ?? null;
1166                 $prof_data['fn']           = $data['name']    ?? null;
1167                 $prof_data['key']          = $data['pubkey']  ?? null;
1168
1169                 Logger::debug('Result', ['link' => $profile_link, 'data' => $prof_data]);
1170
1171                 return $prof_data;
1172         }
1173
1174         /**
1175          * Check for DFRN contact
1176          *
1177          * @param array $webfinger Webfinger data
1178          * @return array DFRN data
1179          * @throws HTTPException\InternalServerErrorException
1180          */
1181         private static function dfrn(array $webfinger): array
1182         {
1183                 $hcard_url = '';
1184                 $data = [];
1185                 // The array is reversed to take into account the order of preference for same-rel links
1186                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1187                 foreach (array_reverse($webfinger['links']) as $link) {
1188                         if (($link['rel'] == ActivityNamespace::DFRN) && !empty($link['href'])) {
1189                                 $data['network'] = Protocol::DFRN;
1190                         } elseif (($link['rel'] == ActivityNamespace::FEED) && !empty($link['href'])) {
1191                                 $data['poll'] = $link['href'];
1192                         } elseif (($link['rel'] == 'http://webfinger.net/rel/profile-page') && (($link['type'] ?? '') == 'text/html') && !empty($link['href'])) {
1193                                 $data['url'] = $link['href'];
1194                         } elseif (($link['rel'] == 'http://microformats.org/profile/hcard') && !empty($link['href'])) {
1195                                 $hcard_url = $link['href'];
1196                         } elseif (($link['rel'] == ActivityNamespace::POCO) && !empty($link['href'])) {
1197                                 $data['poco'] = $link['href'];
1198                         } elseif (($link['rel'] == 'http://webfinger.net/rel/avatar') && !empty($link['href'])) {
1199                                 $data['photo'] = $link['href'];
1200                         } elseif (($link['rel'] == 'http://joindiaspora.com/seed_location') && !empty($link['href'])) {
1201                                 $data['baseurl'] = trim($link['href'], '/');
1202                         } elseif (($link['rel'] == 'http://joindiaspora.com/guid') && !empty($link['href'])) {
1203                                 $data['guid'] = $link['href'];
1204                         } elseif (($link['rel'] == 'diaspora-public-key') && !empty($link['href'])) {
1205                                 $data['pubkey'] = base64_decode($link['href']);
1206
1207                                 if (strstr($data['pubkey'], 'RSA ')) {
1208                                         $data['pubkey'] = Crypto::rsaToPem($data['pubkey']);
1209                                 }
1210                         }
1211                 }
1212
1213                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
1214                         foreach ($webfinger['aliases'] as $alias) {
1215                                 if (empty($data['url']) && !strstr($alias, '@')) {
1216                                         $data['url'] = $alias;
1217                                 } elseif (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
1218                                         $data['alias'] = $alias;
1219                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1220                                         $data['addr'] = substr($alias, 5);
1221                                 }
1222                         }
1223                 }
1224
1225                 if (!empty($webfinger['subject']) && (substr($webfinger['subject'], 0, 5) == 'acct:')) {
1226                         $data['addr'] = substr($webfinger['subject'], 5);
1227                 }
1228
1229                 if (!isset($data['network']) || ($hcard_url == '')) {
1230                         return [];
1231                 }
1232
1233                 // Fetch data via noscrape - this is faster
1234                 $noscrape_url = str_replace('/hcard/', '/noscrape/', $hcard_url);
1235                 $data = self::pollNoscrape($noscrape_url, $data);
1236
1237                 if (isset($data['notify'])
1238                         && isset($data['confirm'])
1239                         && isset($data['request'])
1240                         && isset($data['poll'])
1241                         && isset($data['name'])
1242                         && isset($data['photo'])
1243                 ) {
1244                         return $data;
1245                 }
1246
1247                 $data = self::pollHcard($hcard_url, $data, true);
1248
1249                 return $data;
1250         }
1251
1252         /**
1253          * Poll the hcard page (Diaspora and Friendica specific)
1254          *
1255          * @param string  $hcard_url Link to the hcard page
1256          * @param array   $data      The already fetched data
1257          * @param boolean $dfrn      Poll DFRN specific data
1258          * @return array hcard data
1259          * @throws HTTPException\InternalServerErrorException
1260          */
1261         private static function pollHcard(string $hcard_url, array $data, bool $dfrn = false): array
1262         {
1263                 $curlResult = DI::httpClient()->get($hcard_url, HttpClientAccept::HTML);
1264                 if ($curlResult->isTimeout()) {
1265                         self::$isTimeout = true;
1266                         return [];
1267                 }
1268                 $content = $curlResult->getBody();
1269                 if (empty($content)) {
1270                         return [];
1271                 }
1272
1273                 $doc = new DOMDocument();
1274                 if (!@$doc->loadHTML($content)) {
1275                         return [];
1276                 }
1277
1278                 $xpath = new DomXPath($doc);
1279
1280                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1281                 if (!is_object($vcards)) {
1282                         return [];
1283                 }
1284
1285                 if (!isset($data['baseurl'])) {
1286                         $data['baseurl'] = '';
1287                 }
1288
1289                 if ($vcards->length > 0) {
1290                         $vcard = $vcards->item(0);
1291
1292                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1293                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1294                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1295                         if (($search->length > 0) && empty($data['guid'])) {
1296                                 $data['guid'] = $search->item(0)->nodeValue;
1297                         }
1298
1299                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1300                         if ($search->length > 0) {
1301                                 $data['nick'] = $search->item(0)->nodeValue;
1302                         }
1303
1304                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1305                         if ($search->length > 0) {
1306                                 $data['name'] = $search->item(0)->nodeValue;
1307                         }
1308
1309                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' given_name ')]", $vcard); // */
1310                         if ($search->length > 0) {
1311                                 $data["given_name"] = $search->item(0)->nodeValue;
1312                         }
1313
1314                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' family_name ')]", $vcard); // */
1315                         if ($search->length > 0) {
1316                                 $data["family_name"] = $search->item(0)->nodeValue;
1317                         }
1318
1319                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1320                         if ($search->length > 0) {
1321                                 $data['hide'] = (strtolower($search->item(0)->nodeValue) != 'true');
1322                         }
1323
1324                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1325                         if ($search->length > 0) {
1326                                 $data['pubkey'] = $search->item(0)->nodeValue;
1327                                 if (strstr($data['pubkey'], 'RSA ')) {
1328                                         $data['pubkey'] = Crypto::rsaToPem($data['pubkey']);
1329                                 }
1330                         }
1331
1332                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1333                         if ($search->length > 0) {
1334                                 $data['baseurl'] = trim($search->item(0)->nodeValue, '/');
1335                         }
1336                 }
1337
1338                 $avatars = [];
1339                 if (!empty($vcard)) {
1340                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1341                         foreach ($photos as $photo) {
1342                                 $attr = [];
1343                                 foreach ($photo->attributes as $attribute) {
1344                                         $attr[$attribute->name] = trim($attribute->value);
1345                                 }
1346
1347                                 if (isset($attr['src']) && isset($attr['width'])) {
1348                                         $avatars[$attr['width']] = self::fixAvatar($attr['src'], $data['baseurl']);
1349                                 }
1350
1351                                 // We don't have a width. So we just take everything that we got.
1352                                 // This is a Hubzilla workaround which doesn't send a width.
1353                                 if (!$avatars && !empty($attr['src'])) {
1354                                         $avatars[] = self::fixAvatar($attr['src'], $data['baseurl']);
1355                                 }
1356                         }
1357                 }
1358
1359                 if ($avatars) {
1360                         ksort($avatars);
1361                         $data['photo'] = array_pop($avatars);
1362                         if ($avatars) {
1363                                 $data['photo_medium'] = array_pop($avatars);
1364                         }
1365
1366                         if ($avatars) {
1367                                 $data['photo_small'] = array_pop($avatars);
1368                         }
1369                 }
1370
1371                 if ($dfrn) {
1372                         // Poll DFRN specific data
1373                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1374                         if ($search->length > 0) {
1375                                 foreach ($search as $link) {
1376                                         //$data['request'] = $search->item(0)->nodeValue;
1377                                         $attr = [];
1378                                         foreach ($link->attributes as $attribute) {
1379                                                 $attr[$attribute->name] = trim($attribute->value);
1380                                         }
1381
1382                                         $data[substr($attr['rel'], 5)] = $attr['href'];
1383                                 }
1384                         }
1385
1386                         // Older Friendica versions had used the "uid" field differently than newer versions
1387                         if (!empty($data['nick']) && !empty($data['guid']) && ($data['nick'] == $data['guid'])) {
1388                                 unset($data['guid']);
1389                         }
1390                 }
1391
1392                 return $data;
1393         }
1394
1395         /**
1396          * Check for Diaspora contact
1397          *
1398          * @param array $webfinger Webfinger data
1399          *
1400          * @return array Diaspora data
1401          * @throws HTTPException\InternalServerErrorException
1402          */
1403         private static function diaspora(array $webfinger): array
1404         {
1405                 $hcard_url = '';
1406                 $data = [];
1407
1408                 // The array is reversed to take into account the order of preference for same-rel links
1409                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1410                 foreach (array_reverse($webfinger['links']) as $link) {
1411                         if (($link['rel'] == 'http://microformats.org/profile/hcard') && !empty($link['href'])) {
1412                                 $hcard_url = $link['href'];
1413                         } elseif (($link['rel'] == 'http://joindiaspora.com/seed_location') && !empty($link['href'])) {
1414                                 $data['baseurl'] = trim($link['href'], '/');
1415                         } elseif (($link['rel'] == 'http://joindiaspora.com/guid') && !empty($link['href'])) {
1416                                 $data['guid'] = $link['href'];
1417                         } elseif (($link['rel'] == 'http://webfinger.net/rel/profile-page') && (($link['type'] ?? '') == 'text/html') && !empty($link['href'])) {
1418                                 $data['url'] = $link['href'];
1419                         } elseif (($link['rel'] == 'http://webfinger.net/rel/profile-page') && empty($link['type']) && !empty($link['href'])) {
1420                                 $profile_url = $link['href'];
1421                         } elseif (($link['rel'] == ActivityNamespace::FEED) && !empty($link['href'])) {
1422                                 $data['poll'] = $link['href'];
1423                         } elseif (($link['rel'] == ActivityNamespace::POCO) && !empty($link['href'])) {
1424                                 $data['poco'] = $link['href'];
1425                         } elseif (($link['rel'] == 'salmon') && !empty($link['href'])) {
1426                                 $data['notify'] = $link['href'];
1427                         } elseif (($link['rel'] == 'diaspora-public-key') && !empty($link['href'])) {
1428                                 $data['pubkey'] = base64_decode($link['href']);
1429
1430                                 if (strstr($data['pubkey'], 'RSA ')) {
1431                                         $data['pubkey'] = Crypto::rsaToPem($data['pubkey']);
1432                                 }
1433                         }
1434                 }
1435
1436                 if (empty($data['url']) && !empty($profile_url)) {
1437                         $data['url'] = $profile_url;
1438                 }
1439
1440                 if (empty($data['url']) || empty($hcard_url)) {
1441                         return [];
1442                 }
1443
1444                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
1445                         foreach ($webfinger['aliases'] as $alias) {
1446                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data['url']) && ! strstr($alias, '@')) {
1447                                         $data['alias'] = $alias;
1448                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1449                                         $data['addr'] = substr($alias, 5);
1450                                 }
1451                         }
1452                 }
1453
1454                 if (!empty($webfinger['subject']) && (substr($webfinger['subject'], 0, 5) == 'acct:')) {
1455                         $data['addr'] = substr($webfinger['subject'], 5);
1456                 }
1457
1458                 // Fetch further information from the hcard
1459                 $data = self::pollHcard($hcard_url, $data);
1460
1461                 if (!$data) {
1462                         return [];
1463                 }
1464
1465                 if (!empty($data['url'])
1466                         && !empty($data['guid'])
1467                         && !empty($data['baseurl'])
1468                         && !empty($data['pubkey'])
1469                         && !empty($hcard_url)
1470                 ) {
1471                         $data['network'] = Protocol::DIASPORA;
1472                         $data['manually-approve'] = false;
1473
1474                         // The Diaspora handle must always be lowercase
1475                         if (!empty($data['addr'])) {
1476                                 $data['addr'] = strtolower($data['addr']);
1477                         }
1478
1479                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1480                         $data['notify'] = $data['baseurl'] . '/receive/users/' . $data['guid'];
1481                         $data['batch']  = $data['baseurl'] . '/receive/public';
1482                 } else {
1483                         return [];
1484                 }
1485
1486                 return $data;
1487         }
1488
1489         /**
1490          * Check for OStatus contact
1491          *
1492          * @param array $webfinger Webfinger data
1493          * @param bool  $short     Short detection mode
1494          *
1495          * @return array|bool OStatus data or "false" on error or "true" on short mode
1496          * @throws HTTPException\InternalServerErrorException
1497          */
1498         private static function ostatus(array $webfinger, bool $short = false)
1499         {
1500                 $data = [];
1501
1502                 if (!empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
1503                         foreach ($webfinger['aliases'] as $alias) {
1504                                 if (strstr($alias, '@') && !strstr(Strings::normaliseLink($alias), 'http://')) {
1505                                         $data['addr'] = str_replace('acct:', '', $alias);
1506                                 }
1507                         }
1508                 }
1509
1510                 if (!empty($webfinger['subject']) && strstr($webfinger['subject'], '@')
1511                         && !strstr(Strings::normaliseLink($webfinger['subject']), 'http://')
1512                 ) {
1513                         $data['addr'] = str_replace('acct:', '', $webfinger['subject']);
1514                 }
1515
1516                 if (!empty($webfinger['links'])) {
1517                         // The array is reversed to take into account the order of preference for same-rel links
1518                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1519                         foreach (array_reverse($webfinger['links']) as $link) {
1520                                 if (($link['rel'] == 'http://webfinger.net/rel/profile-page')
1521                                         && (($link['type'] ?? '') == 'text/html')
1522                                         && ($link['href'] != '')
1523                                 ) {
1524                                         $data['url'] = $data['alias'] = $link['href'];
1525                                 } elseif (($link['rel'] == 'salmon') && !empty($link['href'])) {
1526                                         $data['notify'] = $link['href'];
1527                                 } elseif (($link['rel'] == ActivityNamespace::FEED) && !empty($link['href'])) {
1528                                         $data['poll'] = $link['href'];
1529                                 } elseif (($link['rel'] == 'magic-public-key') && !empty($link['href'])) {
1530                                         $pubkey = $link['href'];
1531
1532                                         if (substr($pubkey, 0, 5) === 'data:') {
1533                                                 if (strstr($pubkey, ',')) {
1534                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1535                                                 } else {
1536                                                         $pubkey = substr($pubkey, 5);
1537                                                 }
1538                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1539                                                 $curlResult = DI::httpClient()->get($pubkey, HttpClientAccept::MAGIC_KEY);
1540                                                 if ($curlResult->isTimeout()) {
1541                                                         self::$isTimeout = true;
1542                                                         return $short ? false : [];
1543                                                 }
1544                                                 Logger::debug('Fetched public key', ['Content-Type' => $curlResult->getHeader('Content-Type'), 'url' => $pubkey]);
1545                                                 $pubkey = $curlResult->getBody();
1546                                         }
1547
1548                                         try {
1549                                                 $data['pubkey'] = Salmon::magicKeyToPem($pubkey);
1550                                         } catch (\Throwable $e) {
1551
1552                                         }
1553                                 }
1554                         }
1555                 }
1556
1557                 if (isset($data['notify']) && isset($data['pubkey'])
1558                         && isset($data['poll'])
1559                         && isset($data['url'])
1560                 ) {
1561                         $data['network'] = Protocol::OSTATUS;
1562                         $data['manually-approve'] = false;
1563                 } else {
1564                         return $short ? false : [];
1565                 }
1566
1567                 if ($short) {
1568                         return true;
1569                 }
1570
1571                 // Fetch all additional data from the feed
1572                 $curlResult = DI::httpClient()->get($data['poll'], HttpClientAccept::FEED_XML);
1573                 if ($curlResult->isTimeout()) {
1574                         self::$isTimeout = true;
1575                         return [];
1576                 }
1577                 $feed = $curlResult->getBody();
1578                 $feed_data = Feed::import($feed);
1579                 if (!$feed_data) {
1580                         return [];
1581                 }
1582
1583                 if (!empty($feed_data['header']['author-name'])) {
1584                         $data['name'] = $feed_data['header']['author-name'];
1585                 }
1586                 if (!empty($feed_data['header']['author-nick'])) {
1587                         $data['nick'] = $feed_data['header']['author-nick'];
1588                 }
1589                 if (!empty($feed_data['header']['author-avatar'])) {
1590                         $data['photo'] = self::fixAvatar($feed_data['header']['author-avatar'], $data['url']);
1591                 }
1592                 if (!empty($feed_data['header']['author-id'])) {
1593                         $data['alias'] = $feed_data['header']['author-id'];
1594                 }
1595                 if (!empty($feed_data['header']['author-location'])) {
1596                         $data['location'] = $feed_data['header']['author-location'];
1597                 }
1598                 if (!empty($feed_data['header']['author-about'])) {
1599                         $data['about'] = $feed_data['header']['author-about'];
1600                 }
1601                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1602                 // So we take the value that we just fetched, although the other one worked as well
1603                 if (!empty($feed_data['header']['author-link'])) {
1604                         $data['url'] = $feed_data['header']['author-link'];
1605                 }
1606
1607                 if ($data['url'] == $data['alias']) {
1608                         $data['alias'] = '';
1609                 }
1610
1611                 /// @todo Fetch location and "about" from the feed as well
1612                 return $data;
1613         }
1614
1615         /**
1616          * Fetch data from a pump.io profile page
1617          *
1618          * @param string $profile_link Link to the profile page
1619          *
1620          * @return array Profile data
1621          */
1622         private static function pumpioProfileData(string $profile_link): array
1623         {
1624                 $curlResult = DI::httpClient()->get($profile_link, HttpClientAccept::HTML);
1625                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
1626                         return [];
1627                 }
1628
1629                 $doc = new DOMDocument();
1630                 if (!@$doc->loadHTML($curlResult->getBody())) {
1631                         return [];
1632                 }
1633
1634                 $xpath = new DomXPath($doc);
1635
1636                 $data = [];
1637                 $data['name'] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1638
1639                 if ($data['name'] == '') {
1640                         // This is ugly - but pump.io doesn't seem to know a better way for it
1641                         $data['name'] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1642                         $pos = strpos($data['name'], chr(10));
1643                         if ($pos) {
1644                                 $data['name'] = trim(substr($data['name'], 0, $pos));
1645                         }
1646                 }
1647
1648                 $data['location'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1649
1650                 if ($data['location'] == '') {
1651                         $data['location'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1652                 }
1653
1654                 $data['about'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1655
1656                 if ($data['about'] == '') {
1657                         $data['about'] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1658                 }
1659
1660                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1661                 if (!$avatar) {
1662                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1663                 }
1664                 if ($avatar) {
1665                         foreach ($avatar->attributes as $attribute) {
1666                                 if ($attribute->name == 'src') {
1667                                         $data['photo'] = trim($attribute->value);
1668                                 }
1669                         }
1670                 }
1671
1672                 return $data;
1673         }
1674
1675         /**
1676          * Check for pump.io contact
1677          *
1678          * @param array  $webfinger Webfinger data
1679          * @param string $addr
1680          *
1681          * @return array pump.io data
1682          */
1683         private static function pumpio(array $webfinger, string $addr): array
1684         {
1685                 $data = [];
1686                 // The array is reversed to take into account the order of preference for same-rel links
1687                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1688                 foreach (array_reverse($webfinger['links']) as $link) {
1689                         if (($link['rel'] == 'http://webfinger.net/rel/profile-page')
1690                                 && (($link['type'] ?? '') == 'text/html')
1691                                 && ($link['href'] != '')
1692                         ) {
1693                                 $data['url'] = $link['href'];
1694                         } elseif (($link['rel'] == 'activity-inbox') && ($link['href'] != '')) {
1695                                 $data['notify'] = $link['href'];
1696                         } elseif (($link['rel'] == 'activity-outbox') && ($link['href'] != '')) {
1697                                 $data['poll'] = $link['href'];
1698                         } elseif (($link['rel'] == 'dialback') && ($link['href'] != '')) {
1699                                 $data['dialback'] = $link['href'];
1700                         }
1701                 }
1702                 if (isset($data['poll']) && isset($data['notify'])
1703                         && isset($data['dialback'])
1704                         && isset($data['url'])
1705                 ) {
1706                         // by now we use these fields only for the network type detection
1707                         // So we unset all data that isn't used at the moment
1708                         unset($data['dialback']);
1709
1710                         $data['network'] = Protocol::PUMPIO;
1711                 } else {
1712                         return [];
1713                 }
1714
1715                 $profile_data = self::pumpioProfileData($data['url']);
1716
1717                 if (!$profile_data) {
1718                         return [];
1719                 }
1720
1721                 $data = array_merge($data, $profile_data);
1722
1723                 if (($addr != '') && ($data['name'] != '')) {
1724                         $name = trim(str_replace($addr, '', $data['name']));
1725                         if ($name != '') {
1726                                 $data['name'] = $name;
1727                         }
1728                 }
1729
1730                 return $data;
1731         }
1732
1733         /**
1734          * Checks HTML page for RSS feed link
1735          *
1736          * @param string $url  Page link
1737          * @param string $body Page body string
1738          *
1739          * @return string|false Feed link or false if body was invalid HTML document
1740          */
1741         public static function getFeedLink(string $url, string $body)
1742         {
1743                 if (empty($body)) {
1744                         return '';
1745                 }
1746
1747                 $doc = new DOMDocument();
1748                 if (!@$doc->loadHTML($body)) {
1749                         return false;
1750                 }
1751
1752                 $xpath = new DOMXPath($doc);
1753
1754                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1755                 $feedUrl = $feedUrl ?: $xpath->evaluate('string(/html/head/link[@type="application/atom+xml" and @rel="alternate"]/@href)');
1756
1757                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1758
1759                 return $feedUrl;
1760         }
1761
1762         /**
1763          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1764          *
1765          * Loosely based on RFC 1808
1766          *
1767          * @see https://tools.ietf.org/html/rfc1808
1768          *
1769          * @param string   $href  The potential relative href found in the HTML document
1770          * @param string   $base  The HTML document URL
1771          * @param DOMXPath $xpath The HTML document XPath
1772          *
1773          * @return string Absolute URL
1774          */
1775         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath): string
1776         {
1777                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1778                         return $href;
1779                 }
1780
1781                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1782
1783                 $baseParts = parse_url($base);
1784                 if (empty($baseParts['host'])) {
1785                         return $href;
1786                 }
1787
1788                 // Naked domain case (scheme://basehost)
1789                 $path = $baseParts['path'] ?? '/';
1790
1791                 // Remove the filename part of the path if it exists (/base/path/file)
1792                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1793
1794                 $hrefParts = parse_url($href);
1795
1796                 if (!empty($hrefParts['path'])) {
1797                         // Root path case (/path) including relative scheme case (//host/path)
1798                         if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1799                                 $path = $hrefParts['path'];
1800                         } else {
1801                                 $path = $path . '/' . $hrefParts['path'];
1802
1803                                 // Resolve arbitrary relative path
1804                                 // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1805                                 $parts = array_filter(explode('/', $path), 'strlen');
1806                                 $absolutes = [];
1807                                 foreach ($parts as $part) {
1808                                         if ('.' == $part) continue;
1809                                         if ('..' == $part) {
1810                                                 array_pop($absolutes);
1811                                         } else {
1812                                                 $absolutes[] = $part;
1813                                         }
1814                                 }
1815
1816                                 $path = '/' . implode('/', $absolutes);
1817                         }
1818                 }
1819
1820                 // Relative scheme case (//host/path)
1821                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1822                 $baseParts['path'] = $path;
1823                 unset($baseParts['query']);
1824                 unset($baseParts['fragment']);
1825
1826                 return Network::unparseURL($baseParts);
1827         }
1828
1829         /**
1830          * Check for feed contact
1831          *
1832          * @param string  $url   Profile link
1833          * @param boolean $probe Do a probe if the page contains a feed link
1834          *
1835          * @return array feed data
1836          * @throws HTTPException\InternalServerErrorException
1837          */
1838         private static function feed(string $url, bool $probe = true): array
1839         {
1840                 $curlResult = DI::httpClient()->get($url, HttpClientAccept::FEED_XML);
1841                 if ($curlResult->isTimeout()) {
1842                         self::$isTimeout = true;
1843                         return [];
1844                 }
1845                 $feed = $curlResult->getBody();
1846                 $feed_data = Feed::import($feed);
1847
1848                 if (!$feed_data) {
1849                         if (!$probe) {
1850                                 return [];
1851                         }
1852
1853                         $feed_url = self::getFeedLink($url, $feed);
1854
1855                         if (!$feed_url) {
1856                                 return [];
1857                         }
1858
1859                         return self::feed($feed_url, false);
1860                 }
1861
1862                 if (!empty($feed_data['header']['author-name'])) {
1863                         $data['name'] = $feed_data['header']['author-name'];
1864                 }
1865
1866                 if (!empty($feed_data['header']['author-nick'])) {
1867                         $data['nick'] = $feed_data['header']['author-nick'];
1868                 }
1869
1870                 if (!empty($feed_data['header']['author-avatar'])) {
1871                         $data['photo'] = $feed_data['header']['author-avatar'];
1872                 }
1873
1874                 if (!empty($feed_data['header']['author-id'])) {
1875                         $data['alias'] = $feed_data['header']['author-id'];
1876                 }
1877
1878                 $data['url'] = $url;
1879                 $data['poll'] = $url;
1880
1881                 $data['network'] = Protocol::FEED;
1882
1883                 return $data;
1884         }
1885
1886         /**
1887          * Check for mail contact
1888          *
1889          * @param string  $uri Profile link
1890          * @param integer $uid User ID
1891          *
1892          * @return array mail data
1893          * @throws \Exception
1894          */
1895         private static function mail(string $uri, int $uid): array
1896         {
1897                 if (!Network::isEmailDomainValid($uri)) {
1898                         return [];
1899                 }
1900
1901                 if ($uid == 0) {
1902                         return [];
1903                 }
1904
1905                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1906
1907                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1908                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1909                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1910
1911                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1912                         return [];
1913                 }
1914
1915                 $mailbox = Email::constructMailboxName($mailacct);
1916                 $password = '';
1917                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1918                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1919                 if (!$mbox) {
1920                         return [];
1921                 }
1922
1923                 $msgs = Email::poll($mbox, $uri);
1924                 Logger::info('Messages found', ['uri' => $uri, 'count' => count($msgs)]);
1925
1926                 if (!count($msgs)) {
1927                         return [];
1928                 }
1929
1930                 $phost = substr($uri, strpos($uri, '@') + 1);
1931
1932                 $data = [
1933                         'addr'    => $uri,
1934                         'network' => Protocol::MAIL,
1935                         'name'    => substr($uri, 0, strpos($uri, '@')),
1936                         'photo'   => Network::lookupAvatarByEmail($uri),
1937                         'url'     => 'mailto:' . $uri,
1938                         'notify'  => 'smtp ' . Strings::getRandomHex(),
1939                         'poll'    => 'email ' . Strings::getRandomHex(),
1940                 ];
1941
1942                 $data['nick']    = $data['name'];
1943
1944                 $x = Email::messageMeta($mbox, $msgs[0]);
1945
1946                 if (stristr($x[0]->from, $uri)) {
1947                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1948                 } elseif (stristr($x[0]->to, $uri)) {
1949                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1950                 }
1951
1952                 if (isset($adr)) {
1953                         foreach ($adr as $feadr) {
1954                                 if ((strcasecmp($feadr->mailbox, $data['name']) == 0)
1955                                         &&(strcasecmp($feadr->host, $phost) == 0)
1956                                         && (strlen($feadr->personal))
1957                                 ) {
1958                                         $personal = imap_mime_header_decode($feadr->personal);
1959                                         $data['name'] = '';
1960                                         foreach ($personal as $perspart) {
1961                                                 if ($perspart->charset != 'default') {
1962                                                         $data['name'] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1963                                                 } else {
1964                                                         $data['name'] .= $perspart->text;
1965                                                 }
1966                                         }
1967                                 }
1968                         }
1969                 }
1970
1971                 if (!empty($mbox)) {
1972                         imap_close($mbox);
1973                 }
1974
1975                 return $data;
1976         }
1977
1978         /**
1979          * Mix two paths together to possibly fix missing parts
1980          *
1981          * @param string $avatar Path to the avatar
1982          * @param string $base   Another path that is hopefully complete
1983          *
1984          * @return string fixed avatar path
1985          * @throws \Exception
1986          */
1987         public static function fixAvatar(string $avatar, string $base): string
1988         {
1989                 $base_parts = parse_url($base);
1990
1991                 // Remove all parts that could create a problem
1992                 unset($base_parts['path']);
1993                 unset($base_parts['query']);
1994                 unset($base_parts['fragment']);
1995
1996                 $avatar_parts = parse_url($avatar);
1997
1998                 // Now we mix them
1999                 $parts = array_merge($base_parts, $avatar_parts);
2000
2001                 // And put them together again
2002                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
2003                 $host     = isset($parts['host'])     ? $parts['host']           : '';
2004                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
2005                 $path     = isset($parts['path'])     ? $parts['path']           : '';
2006                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
2007                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
2008
2009                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
2010
2011                 Logger::debug('Avatar fixed', ['base' => $base, 'avatar' => $avatar, 'fixed' => $fixed]);
2012
2013                 return $fixed;
2014         }
2015
2016         /**
2017          * Fetch the last date that the contact had posted something (publically)
2018          *
2019          * @param array $data  probing result
2020          *
2021          * @return string last activity
2022          */
2023         public static function getLastUpdate(array $data): string
2024         {
2025                 $uid = User::getIdForURL($data['url']);
2026                 if (!empty($uid)) {
2027                         $contact = Contact::selectFirst(['url', 'last-item'], ['self' => true, 'uid' => $uid]);
2028                         if (!empty($contact['last-item'])) {
2029                                 return $contact['last-item'];
2030                         }
2031                 }
2032
2033                 if ($lastUpdate = self::updateFromNoScrape($data)) {
2034                         return $lastUpdate;
2035                 }
2036
2037                 if (!empty($data['outbox'])) {
2038                         return self::updateFromOutbox($data['outbox'], $data);
2039                 } elseif (!empty($data['poll']) && ($data['network'] == Protocol::ACTIVITYPUB)) {
2040                         return self::updateFromOutbox($data['poll'], $data);
2041                 } elseif (!empty($data['poll'])) {
2042                         return self::updateFromFeed($data);
2043                 }
2044
2045                 return '';
2046         }
2047
2048         /**
2049          * Fetch the last activity date from the "noscrape" endpoint
2050          *
2051          * @param array $data Probing result
2052          *
2053          * @return string last activity or true if update was successful or the server was unreachable
2054          */
2055         private static function updateFromNoScrape(array $data): string
2056         {
2057                 if (empty($data['baseurl'])) {
2058                         return '';
2059                 }
2060
2061                 // Check the 'noscrape' endpoint when it is a Friendica server
2062                 $gserver = DBA::selectFirst('gserver', ['noscrape'], ["`nurl` = ? AND `noscrape` != ''",
2063                         Strings::normaliseLink($data['baseurl'])]);
2064                 if (!DBA::isResult($gserver)) {
2065                         return '';
2066                 }
2067
2068                 $curlResult = DI::httpClient()->get($gserver['noscrape'] . '/' . $data['nick'], HttpClientAccept::JSON);
2069
2070                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
2071                         $noscrape = json_decode($curlResult->getBody(), true);
2072                         if (!empty($noscrape) && !empty($noscrape['updated'])) {
2073                                 return DateTimeFormat::utc($noscrape['updated'], DateTimeFormat::MYSQL);
2074                         }
2075                 }
2076
2077                 return '';
2078         }
2079
2080         /**
2081          * Fetch the last activity date from an ActivityPub Outbox
2082          *
2083          * @param string $feed
2084          * @param array  $data Probing result
2085          *
2086          * @return string last activity
2087          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
2088          */
2089         private static function updateFromOutbox(string $feed, array $data): string
2090         {
2091                 $outbox = ActivityPub::fetchContent($feed);
2092                 if (empty($outbox)) {
2093                         return '';
2094                 }
2095
2096                 if (!empty($outbox['orderedItems'])) {
2097                         $items = $outbox['orderedItems'];
2098                 } elseif (!empty($outbox['first']['orderedItems'])) {
2099                         $items = $outbox['first']['orderedItems'];
2100                 } elseif (!empty($outbox['first']['href']) && ($outbox['first']['href'] != $feed)) {
2101                         return self::updateFromOutbox($outbox['first']['href'], $data);
2102                 } elseif (!empty($outbox['first'])) {
2103                         if (is_string($outbox['first']) && ($outbox['first'] != $feed)) {
2104                                 return self::updateFromOutbox($outbox['first'], $data);
2105                         } else {
2106                                 Logger::warning('Unexpected data', ['outbox' => $outbox]);
2107                         }
2108                         return '';
2109                 } else {
2110                         $items = [];
2111                 }
2112
2113                 $last_updated = '';
2114                 foreach ($items as $activity) {
2115                         if (!empty($activity['published'])) {
2116                                 $published =  DateTimeFormat::utc($activity['published']);
2117                         } elseif (!empty($activity['object']['published'])) {
2118                                 $published =  DateTimeFormat::utc($activity['object']['published']);
2119                         } else {
2120                                 continue;
2121                         }
2122
2123                         if ($last_updated < $published) {
2124                                 $last_updated = $published;
2125                         }
2126                 }
2127
2128                 if (!empty($last_updated)) {
2129                         return $last_updated;
2130                 }
2131
2132                 return '';
2133         }
2134
2135         /**
2136          * Fetch the last activity date from an XML feed
2137          *
2138          * @param array $data Probing result
2139          * @return string last activity
2140          */
2141         private static function updateFromFeed(array $data): string
2142         {
2143                 // Search for the newest entry in the feed
2144                 $curlResult = DI::httpClient()->get($data['poll'], HttpClientAccept::ATOM_XML);
2145                 if (!$curlResult->isSuccess() || !$curlResult->getBody()) {
2146                         return '';
2147                 }
2148
2149                 $doc = new DOMDocument();
2150                 @$doc->loadXML($curlResult->getBody());
2151
2152                 $xpath = new DOMXPath($doc);
2153                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
2154
2155                 $entries = $xpath->query('/atom:feed/atom:entry');
2156
2157                 $last_updated = '';
2158
2159                 foreach ($entries as $entry) {
2160                         $published_item = $xpath->query('atom:published/text()', $entry)->item(0);
2161                         $updated_item   = $xpath->query('atom:updated/text()'  , $entry)->item(0);
2162                         $published      = !empty($published_item->nodeValue) ? DateTimeFormat::utc($published_item->nodeValue) : null;
2163                         $updated        = !empty($updated_item->nodeValue) ? DateTimeFormat::utc($updated_item->nodeValue) : null;
2164
2165                         if (empty($published) || empty($updated)) {
2166                                 Logger::notice('Invalid entry for XPath.', ['entry' => $entry, 'url' => $data['url']]);
2167                                 continue;
2168                         }
2169
2170                         if ($last_updated < $published) {
2171                                 $last_updated = $published;
2172                         }
2173
2174                         if ($last_updated < $updated) {
2175                                 $last_updated = $updated;
2176                         }
2177                 }
2178
2179                 if (!empty($last_updated)) {
2180                         return $last_updated;
2181                 }
2182
2183                 return '';
2184         }
2185
2186         /**
2187          * Probe data from local profiles without network traffic
2188          *
2189          * @param string $url
2190          *
2191          * @return array probed data
2192          * @throws HTTPException\InternalServerErrorException
2193          * @throws HTTPException\NotFoundException
2194          */
2195         private static function localProbe(string $url): array
2196         {
2197                 try {
2198                         $uid = User::getIdForURL($url);
2199                         if (!$uid) {
2200                                 throw new HTTPException\NotFoundException('User not found.');
2201                         }
2202
2203                         $owner     = User::getOwnerDataById($uid);
2204                         $approfile = ActivityPub\Transmitter::getProfile($uid);
2205
2206                         if (empty($owner['gsid'])) {
2207                                 $owner['gsid'] = GServer::getID($approfile['generator']['url']);
2208                         }
2209
2210                         $data = [
2211                                 'name'             => $owner['name'], 'nick' => $owner['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
2212                                 'url'              => $owner['url'], 'addr' => $owner['addr'], 'alias' => $owner['alias'],
2213                                 'photo'            => User::getAvatarUrl($owner),
2214                                 'header'           => $owner['header'] ? Contact::getHeaderUrlForId($owner['id'], $owner['updated']) : '',
2215                                 'account-type'     => $owner['contact-type'], 'community' => ($owner['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
2216                                 'keywords'         => $owner['keywords'], 'location' => $owner['location'], 'about' => $owner['about'],
2217                                 'xmpp'             => $owner['xmpp'], 'matrix' => $owner['matrix'],
2218                                 'hide'             => !$owner['net-publish'], 'batch' => '', 'notify' => $owner['notify'],
2219                                 'poll'             => $owner['poll'], 'request' => $owner['request'], 'confirm' => $owner['confirm'],
2220                                 'subscribe'        => $approfile['generator']['url'] . '/contact/follow?url={uri}', 'poco' => $owner['poco'],
2221                                 'following'        => $approfile['following'], 'followers' => $approfile['followers'],
2222                                 'inbox'            => $approfile['inbox'], 'outbox' => $approfile['outbox'],
2223                                 'sharedinbox'      => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
2224                                 'pubkey'           => $owner['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $owner['gsid'],
2225                                 'manually-approve' => in_array($owner['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])
2226                         ];
2227                 } catch (Exception $e) {
2228                         // Default values for non existing targets
2229                         $data = [
2230                                 'name' => $url, 'nick' => $url, 'url' => $url, 'network' => Protocol::PHANTOM,
2231                                 'photo' => DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO
2232                         ];
2233                 }
2234
2235                 return self::rearrangeData($data);
2236         }
2237 }