src/Network/Probe.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, 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\HttpClientOptions;
  38 use Friendica\Protocol\ActivityNamespace;
  39 use Friendica\Protocol\ActivityPub;
  40 use Friendica\Protocol\Email;
  41 use Friendica\Protocol\Feed;
  42 use Friendica\Util\Crypto;
  43 use Friendica\Util\DateTimeFormat;
  44 use Friendica\Util\Network;
  45 use Friendica\Util\Strings;
  46 use Friendica\Util\XML;
  47
  48 /**
  49  * This class contain functions for probing URL
  50  */
  51 class Probe
  52 {
  53         const WEBFINGER = '/.well-known/webfinger?resource={uri}';
  54
  55         private static $baseurl;
  56         private static $istimeout;
  57
  58         /**
  59          * Remove stuff from an URI that doesn't belong there
  60          *
  61          * @param string $URI
  62          * @return string Cleaned URI
  63          */
  64         public static function cleanURI(string $URI)
  65         {
  66                 // At first remove leading and trailing junk
  67                 $URI = trim($URI, "@#?:/ \t\n\r\0\x0B");
  68
  69                 $parts = parse_url($URI);
  70
  71                 if (empty($parts['scheme'])) {
  72                         return $URI;
  73                 }
  74
  75                 // Remove the URL fragment, since these shouldn't be part of any profile URL
  76                 unset($parts['fragment']);
  77
  78                 $URI = Network::unparseURL($parts);
  79
  80                 return $URI;
  81         }
  82
  83         /**
  84          * Rearrange the array so that it always has the same order
  85          *
  86          * @param array $data Unordered data
  87          *
  88          * @return array Ordered data
  89          */
  90         private static function rearrangeData($data)
  91         {
  92                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "header",
  93                                 "account-type", "community", "keywords", "location", "about", "xmpp", "matrix",
  94                                 "hide", "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
  95                                 "following", "followers", "inbox", "outbox", "sharedinbox",
  96                                 "priority", "network", "pubkey", "manually-approve", "baseurl", "gsid"];
  97
  98                 $numeric_fields = ["gsid", "hide", "account-type", "manually-approve"];
  99
 100                 $newdata = [];
 101                 foreach ($fields as $field) {
 102                         if (isset($data[$field])) {
 103                                 if (in_array($field, $numeric_fields)) {
 104                                         $newdata[$field] = (int)$data[$field];
 105                                 } else {
 106                                         $newdata[$field] = $data[$field];
 107                                 }
 108                         } elseif (!in_array($field, $numeric_fields)) {
 109                                 $newdata[$field] = "";
 110                         } else {
 111                                 $newdata[$field] = null;
 112                         }
 113                 }
 114
 115                 // We don't use the "priority" field anymore and replace it with a dummy.
 116                 $newdata["priority"] = 0;
 117
 118                 return $newdata;
 119         }
 120
 121         /**
 122          * Check if the hostname belongs to the own server
 123          *
 124          * @param string $host The hostname that is to be checked
 125          *
 126          * @return bool Does the testes hostname belongs to the own server?
 127          */
 128         private static function ownHost($host)
 129         {
 130                 $own_host = DI::baseUrl()->getHostname();
 131
 132                 $parts = parse_url($host);
 133
 134                 if (!isset($parts['scheme'])) {
 135                         $parts = parse_url('http://'.$host);
 136                 }
 137
 138                 if (!isset($parts['host'])) {
 139                         return false;
 140                 }
 141                 return $parts['host'] == $own_host;
 142         }
 143
 144         /**
 145          * Probes for webfinger path via "host-meta"
 146          *
 147          * We have to check if the servers in the future still will offer this.
 148          * It seems as if it was dropped from the standard.
 149          *
 150          * @param string $host The host part of an url
 151          *
 152          * @return array with template and type of the webfinger template for JSON or XML
 153          * @throws HTTPException\InternalServerErrorException
 154          */
 155         private static function hostMeta($host)
 156         {
 157                 // Reset the static variable
 158                 self::$baseurl = '';
 159
 160                 // Handles the case when the hostname contains the scheme
 161                 if (!parse_url($host, PHP_URL_SCHEME)) {
 162                         $ssl_url = "https://" . $host . "/.well-known/host-meta";
 163                         $url = "http://" . $host . "/.well-known/host-meta";
 164                 } else {
 165                         $ssl_url = $host . "/.well-known/host-meta";
 166                         $url = '';
 167                 }
 168
 169                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 170
 171                 Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
 172                 $xrd = null;
 173
 174                 $curlResult = DI::httpClient()->get($ssl_url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => ['application/xrd+xml']]);
 175                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 176                 if ($curlResult->isSuccess()) {
 177                         $xml = $curlResult->getBody();
 178                         $xrd = XML::parseString($xml, true);
 179                         if (!empty($url)) {
 180                                 $host_url = 'https://' . $host;
 181                         } else {
 182                                 $host_url = $host;
 183                         }
 184                 } elseif ($curlResult->isTimeout()) {
 185                         Logger::info('Probing timeout', ['url' => $ssl_url]);
 186                         self::$istimeout = true;
 187                         return [];
 188                 }
 189
 190                 if (!is_object($xrd) && !empty($url)) {
 191                         $curlResult = DI::httpClient()->get($url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => ['application/xrd+xml']]);
 192                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 193                         if ($curlResult->isTimeout()) {
 194                                 Logger::info('Probing timeout', ['url' => $url]);
 195                                 self::$istimeout = true;
 196                                 return [];
 197                         } elseif ($connection_error && $ssl_connection_error) {
 198                                 self::$istimeout = true;
 199                                 return [];
 200                         }
 201
 202                         $xml = $curlResult->getBody();
 203                         $xrd = XML::parseString($xml, true);
 204                         $host_url = 'http://'.$host;
 205                 }
 206                 if (!is_object($xrd)) {
 207                         Logger::info('No xrd object found', ['host' => $host]);
 208                         return [];
 209                 }
 210
 211                 $links = XML::elementToArray($xrd);
 212                 if (!isset($links["xrd"]["link"])) {
 213                         Logger::info('No xrd data found', ['host' => $host]);
 214                         return [];
 215                 }
 216
 217                 $lrdd = [];
 218
 219                 foreach ($links["xrd"]["link"] as $value => $link) {
 220                         if (!empty($link["@attributes"])) {
 221                                 $attributes = $link["@attributes"];
 222                         } elseif ($value == "@attributes") {
 223                                 $attributes = $link;
 224                         } else {
 225                                 continue;
 226                         }
 227
 228                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 229                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 230
 231                                 $lrdd[$type] = $attributes["template"];
 232                         }
 233                 }
 234
 235                 if (Network::isUrlBlocked($host_url)) {
 236                         Logger::info('Domain is blocked', ['url' => $host]);
 237                         return [];
 238                 }
 239
 240                 self::$baseurl = $host_url;
 241
 242                 Logger::info('Probing successful', ['host' => $host]);
 243
 244                 return $lrdd;
 245         }
 246
 247         /**
 248          * Perform Webfinger lookup and return DFRN data
 249          *
 250          * Given an email style address, perform webfinger lookup and
 251          * return the resulting DFRN profile URL, or if no DFRN profile URL
 252          * is located, returns an OStatus subscription template (prefixed
 253          * with the string 'stat:' to identify it as on OStatus template).
 254          * If this isn't an email style address just return $webbie.
 255          * Return an empty string if email-style addresses but webfinger fails,
 256          * or if the resultant personal XRD doesn't contain a supported
 257          * subscription/friend-request attribute.
 258          *
 259          * amended 7/9/2011 to return an hcard which could save potentially loading
 260          * a lengthy content page to scrape dfrn attributes
 261          *
 262          * @param string $webbie    Address that should be probed
 263          * @param string $hcard_url Link to the hcard - is returned by reference
 264          *
 265          * @return string profile link
 266          * @throws HTTPException\InternalServerErrorException
 267          */
 268         public static function webfingerDfrn(string $webbie, string &$hcard_url)
 269         {
 270                 $profile_link = '';
 271
 272                 $links = self::lrdd($webbie);
 273                 Logger::debug('Result', ['url' => $webbie, 'links' => $links]);
 274                 if (!empty($links) && is_array($links)) {
 275                         foreach ($links as $link) {
 276                                 if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) {
 277                                         $profile_link = $link['@attributes']['href'];
 278                                 }
 279                                 if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) {
 280                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 281                                 }
 282                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 283                                         $hcard_url = $link['@attributes']['href'];
 284                                 }
 285                         }
 286                 }
 287                 return $profile_link;
 288         }
 289
 290         /**
 291          * Check an URI for LRDD data
 292          *
 293          * @param string $uri     Address that should be probed
 294          *
 295          * @return array uri data
 296          * @throws HTTPException\InternalServerErrorException
 297          */
 298         public static function lrdd(string $uri)
 299         {
 300                 $data = self::getWebfingerArray($uri);
 301                 if (empty($data)) {
 302                         return [];
 303                 }
 304                 $webfinger = $data['webfinger'];
 305
 306                 if (empty($webfinger["links"])) {
 307                         Logger::info('No webfinger links found', ['uri' => $uri]);
 308                         return [];
 309                 }
 310
 311                 $data = [];
 312
 313                 foreach ($webfinger["links"] as $link) {
 314                         $data[] = ["@attributes" => $link];
 315                 }
 316
 317                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 318                         foreach ($webfinger["aliases"] as $alias) {
 319                                 $data[] = ["@attributes" =>
 320                                                         ["rel" => "alias",
 321                                                                 "href" => $alias]];
 322                         }
 323                 }
 324
 325                 return $data;
 326         }
 327
 328         /**
 329          * Fetch information (protocol endpoints and user information) about a given uri
 330          *
 331          * @param string  $uri     Address that should be probed
 332          * @param string  $network Test for this specific network
 333          * @param integer $uid     User ID for the probe (only used for mails)
 334          * @param boolean $cache   Use cached values?
 335          *
 336          * @return array uri data
 337          * @throws HTTPException\InternalServerErrorException
 338          * @throws \ImagickException
 339          */
 340         public static function uri($uri, $network = '', $uid = -1)
 341         {
 342                 // Local profiles aren't probed via network
 343                 if (empty($network) && Contact::isLocal($uri)) {
 344                         $data = self::localProbe($uri);
 345                         if (!empty($data)) {
 346                                 return $data;
 347                         }
 348                 }
 349
 350                 if ($uid == -1) {
 351                         $uid = local_user();
 352                 }
 353
 354                 if (empty($network) || ($network == Protocol::ACTIVITYPUB)) {
 355                         $ap_profile = ActivityPub::probeProfile($uri);
 356                 } else {
 357                         $ap_profile = [];
 358                 }
 359
 360                 self::$istimeout = false;
 361
 362                 if ($network != Protocol::ACTIVITYPUB) {
 363                         $data = self::detect($uri, $network, $uid, $ap_profile);
 364                         if (!is_array($data)) {
 365                                 $data = [];
 366                         }
 367                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 368                                 $data = $ap_profile;
 369                         } elseif (!empty($ap_profile)) {
 370                                 $ap_profile['batch'] = '';
 371                                 $data = array_merge($ap_profile, $data);
 372                         }
 373                 } else {
 374                         $data = $ap_profile;
 375                 }
 376
 377                 if (!isset($data['url'])) {
 378                         $data['url'] = $uri;
 379                 }
 380
 381                 if (empty($data['photo'])) {
 382                         $data['photo'] = DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO;
 383                 }
 384
 385                 if (empty($data['name'])) {
 386                         if (!empty($data['nick'])) {
 387                                 $data['name'] = $data['nick'];
 388                         }
 389
 390                         if (empty($data['name'])) {
 391                                 $data['name'] = $data['url'];
 392                         }
 393                 }
 394
 395                 if (empty($data['nick'])) {
 396                         $data['nick'] = strtolower($data['name']);
 397
 398                         if (strpos($data['nick'], ' ')) {
 399                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 400                         }
 401                 }
 402
 403                 if (!empty($data['baseurl']) && empty($data['gsid'])) {
 404                         $data['gsid'] = GServer::getID($data['baseurl']);
 405                 }
 406
 407                 if (empty($data['network'])) {
 408                         $data['network'] = Protocol::PHANTOM;
 409                 }
 410
 411                 // Ensure that local connections always are DFRN
 412                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
 413                         $data['network'] = Protocol::DFRN;
 414                 }
 415
 416                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 417                         $data['hide'] = self::getHideStatus($data['url']);
 418                 }
 419
 420                 return self::rearrangeData($data);
 421         }
 422
 423
 424         /**
 425          * Fetches the "hide" status from the profile
 426          *
 427          * @param string $url URL of the profile
 428          *
 429          * @return boolean "hide" status
 430          */
 431         private static function getHideStatus($url)
 432         {
 433                 $curlResult = DI::httpClient()->get($url, [HttpClientOptions::CONTENT_LENGTH => 1000000]);
 434                 if (!$curlResult->isSuccess()) {
 435                         return false;
 436                 }
 437
 438                 // If it isn't a HTML file then exit
 439                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 440                         return false;
 441                 }
 442
 443                 $body = $curlResult->getBody();
 444                 if (empty($body)) {
 445                         return false;
 446                 }
 447
 448                 $doc = new DOMDocument();
 449                 @$doc->loadHTML($body);
 450
 451                 $xpath = new DOMXPath($doc);
 452
 453                 $list = $xpath->query('//meta[@name]');
 454                 foreach ($list as $node) {
 455                         $meta_tag = [];
 456                         if ($node->attributes->length) {
 457                                 foreach ($node->attributes as $attribute) {
 458                                         $meta_tag[$attribute->name] = $attribute->value;
 459                                 }
 460                         }
 461
 462                         if (empty($meta_tag['content'])) {
 463                                 continue;
 464                         }
 465
 466                         $content = strtolower(trim($meta_tag['content']));
 467
 468                         switch (strtolower(trim($meta_tag['name']))) {
 469                                 case 'dfrn-global-visibility':
 470                                         if ($content == 'false') {
 471                                                 return true;
 472                                         }
 473                                         break;
 474                                 case 'robots':
 475                                         if (strpos($content, 'noindex') !== false) {
 476                                                 return true;
 477                                         }
 478                                         break;
 479                         }
 480                 }
 481
 482                 return false;
 483         }
 484
 485         /**
 486          * Fetch the "subscribe" and add it to the result
 487          *
 488          * @param array $result
 489          * @param array $webfinger
 490          * @return array result
 491          */
 492         private static function getSubscribeLink(array $result, array $webfinger)
 493         {
 494                 if (empty($webfinger['links'])) {
 495                         return $result;
 496                 }
 497
 498                 foreach ($webfinger['links'] as $link) {
 499                         if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
 500                                 $result['subscribe'] = $link['template'];
 501                         }
 502                 }
 503
 504                 return $result;
 505         }
 506
 507         /**
 508          * Get webfinger data from a given URI
 509          *
 510          * @param string $uri
 511          * @return array
 512          * @throws HTTPException\InternalServerErrorException
 513          */
 514         private static function getWebfingerArray(string $uri): array
 515         {
 516                 $parts = parse_url($uri);
 517
 518                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 519                         $host = $parts['host'];
 520                         if (!empty($parts['port'])) {
 521                                 $host .= ':' . $parts['port'];
 522                         }
 523
 524                         $baseurl = $parts['scheme'] . '://' . $host;
 525
 526                         $nick = '';
 527                         $addr = '';
 528
 529                         $path_parts = explode('/', trim($parts['path'] ?? '', '/'));
 530                         if (!empty($path_parts)) {
 531                                 $nick = ltrim(end($path_parts), '@');
 532                                 // When the last part of the URI is numeric then it is most likely an ID and not a nick name
 533                                 if (!is_numeric($nick)) {
 534                                         $addr = $nick . '@' . $host;
 535                                 } else {
 536                                         $nick = '';
 537                                 }
 538                         }
 539
 540                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 541                         if (empty($webfinger)) {
 542                                 $lrdd = self::hostMeta($host);
 543                         }
 544
 545                         if (empty($webfinger) && empty($lrdd)) {
 546                                 while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
 547                                         $host    .= '/' . array_shift($path_parts);
 548                                         $baseurl = $parts['scheme'] . '://' . $host;
 549
 550                                         if (!empty($nick)) {
 551                                                 $addr = $nick . '@' . $host;
 552                                         }
 553
 554                                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 555                                         if (empty($webfinger)) {
 556                                                 $lrdd = self::hostMeta($host);
 557                                         }
 558                                 }
 559
 560                                 if (empty($lrdd) && empty($webfinger)) {
 561                                         return [];
 562                                 }
 563                         }
 564                 } elseif (strstr($uri, '@')) {
 565                         // Remove "acct:" from the URI
 566                         $uri = str_replace('acct:', '', $uri);
 567
 568                         $host = substr($uri, strpos($uri, '@') + 1);
 569                         $nick = substr($uri, 0, strpos($uri, '@'));
 570                         $addr = $uri;
 571
 572                         $webfinger = self::getWebfinger('https://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 573                         if (self::$istimeout) {
 574                                 return [];
 575                         }
 576
 577                         if (empty($webfinger)) {
 578                                 $webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 579                                 if (self::$istimeout) {
 580                                         return [];
 581                                 }
 582                         } else {
 583                                 $baseurl = 'https://' . $host;
 584                         }
 585
 586                         if (empty($webfinger)) {
 587                                 $lrdd = self::hostMeta($host);
 588                                 if (self::$istimeout) {
 589                                         return [];
 590                                 }
 591                                 $baseurl = self::$baseurl;
 592                         } else {
 593                                 $baseurl = 'http://' . $host;
 594                         }
 595                 } else {
 596                         Logger::info('URI was not detectable', ['uri' => $uri]);
 597                         return [];
 598                 }
 599
 600                 if (empty($webfinger)) {
 601                         foreach ($lrdd as $type => $template) {
 602                                 if ($webfinger) {
 603                                         continue;
 604                                 }
 605
 606                                 $webfinger = self::getWebfinger($template, $type, $uri, $addr);
 607                         }
 608                 }
 609
 610                 if (empty($webfinger)) {
 611                         return [];
 612                 }
 613
 614                 if ($webfinger['detected'] == $addr) {
 615                         $webfinger['nick'] = $nick;
 616                         $webfinger['addr'] = $addr;
 617                 }
 618
 619                 $webfinger['baseurl'] = $baseurl;
 620
 621                 return $webfinger;
 622         }
 623
 624         /**
 625          * Perform network request for webfinger data
 626          *
 627          * @param string $template
 628          * @param string $type
 629          * @param string $uri
 630          * @param string $addr
 631          * @return array webfinger results
 632          */
 633         private static function getWebfinger(string $template, string $type, string $uri, string $addr)
 634         {
 635                 if (Network::isUrlBlocked($template)) {
 636                         Logger::info('Domain is blocked', ['url' => $template]);
 637                         return [];
 638                 }
 639
 640                 // First try the address because this is the primary purpose of webfinger
 641                 if (!empty($addr)) {
 642                         $detected = $addr;
 643                         $path = str_replace('{uri}', urlencode("acct:" . $addr), $template);
 644                         $webfinger = self::webfinger($path, $type);
 645                         if (self::$istimeout) {
 646                                 return [];
 647                         }
 648                 }
 649
 650                 // Then try the URI
 651                 if (empty($webfinger) && $uri != $addr) {
 652                         $detected = $uri;
 653                         $path = str_replace('{uri}', urlencode($uri), $template);
 654                         $webfinger = self::webfinger($path, $type);
 655                         if (self::$istimeout) {
 656                                 return [];
 657                         }
 658                 }
 659
 660                 if (empty($webfinger)) {
 661                         return [];
 662                 }
 663
 664                 return ['webfinger' => $webfinger, 'detected' => $detected];
 665         }
 666
 667         /**
 668          * Fetch information (protocol endpoints and user information) about a given uri
 669          *
 670          * This function is only called by the "uri" function that adds caching and rearranging of data.
 671          *
 672          * @param string  $uri        Address that should be probed
 673          * @param string  $network    Test for this specific network
 674          * @param integer $uid        User ID for the probe (only used for mails)
 675          * @param array   $ap_profile Previously probed AP profile
 676          *
 677          * @return array uri data
 678          * @throws HTTPException\InternalServerErrorException
 679          */
 680         private static function detect(string $uri, string $network, int $uid, array $ap_profile)
 681         {
 682                 $hookData = [
 683                         'uri'     => $uri,
 684                         'network' => $network,
 685                         'uid'     => $uid,
 686                         'result'  => null,
 687                 ];
 688
 689                 Hook::callAll('probe_detect', $hookData);
 690
 691                 if (isset($hookData['result'])) {
 692                         return is_array($hookData['result']) ? $hookData['result'] : [];
 693                 }
 694
 695                 $parts = parse_url($uri);
 696
 697                 if (empty($parts['scheme']) || !empty($parts['host']) && strstr($uri, '@')) {
 698                         // If the URI starts with "mailto:" then jump directly to the mail detection
 699                         if (strpos($uri, 'mailto:') !== false) {
 700                                 $uri = str_replace('mailto:', '', $uri);
 701                                 return self::mail($uri, $uid);
 702                         }
 703
 704                         if ($network == Protocol::MAIL) {
 705                                 return self::mail($uri, $uid);
 706                         }
 707                 } else {
 708                         Logger::info('URI was not detectable', ['uri' => $uri]);
 709                         return [];
 710                 }
 711
 712                 Logger::info('Probing start', ['uri' => $uri]);
 713
 714                 if (!empty($ap_profile['addr']) && ($ap_profile['addr'] != $uri)) {
 715                         $data = self::getWebfingerArray($ap_profile['addr']);
 716                 }
 717
 718                 if (empty($data)) {
 719                         $data = self::getWebfingerArray($uri);
 720                 }
 721
 722                 if (empty($data)) {
 723                         if (!empty($parts['scheme'])) {
 724                                 return self::feed($uri);
 725                         } elseif (!empty($uid)) {
 726                                 return self::mail($uri, $uid);
 727                         } else {
 728                                 return [];
 729                         }
 730                 }
 731
 732                 $webfinger = $data['webfinger'];
 733                 $nick = $data['nick'] ?? '';
 734                 $addr = $data['addr'] ?? '';
 735                 $baseurl = $data['baseurl'] ?? '';
 736
 737                 $result = [];
 738
 739                 if (in_array($network, ["", Protocol::DFRN])) {
 740                         $result = self::dfrn($webfinger);
 741                 }
 742                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 743                         $result = self::diaspora($webfinger);
 744                 }
 745                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 746                         $result = self::ostatus($webfinger);
 747                 }
 748                 if (in_array($network, ['', Protocol::ZOT])) {
 749                         $result = self::zot($webfinger, $result, $baseurl);
 750                 }
 751                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 752                         $result = self::pumpio($webfinger, $addr);
 753                 }
 754                 if (empty($result['network']) && empty($ap_profile['network']) || ($network == Protocol::FEED)) {
 755                         $result = self::feed($uri);
 756                 } else {
 757                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 758                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 759                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 760                                 $result["nick"] = $nick;
 761                         }
 762
 763                         if (empty($result["addr"]) && ($addr != "")) {
 764                                 $result["addr"] = $addr;
 765                         }
 766                 }
 767
 768                 $result = self::getSubscribeLink($result, $webfinger);
 769
 770                 if (empty($result["network"])) {
 771                         $result["network"] = Protocol::PHANTOM;
 772                 }
 773
 774                 if (empty($result['baseurl']) && !empty($baseurl)) {
 775                         $result['baseurl'] = $baseurl;
 776                 }
 777
 778                 if (empty($result["url"])) {
 779                         $result["url"] = $uri;
 780                 }
 781
 782                 Logger::info('Probing done', ['uri' => $uri, 'network' => $result["network"]]);
 783
 784                 return $result;
 785         }
 786
 787         /**
 788          * Check for Zot contact
 789          *
 790          * @param array $webfinger Webfinger data
 791          * @param array $data      previously probed data
 792          *
 793          * @return array Zot data
 794          * @throws HTTPException\InternalServerErrorException
 795          */
 796         private static function zot($webfinger, $data, $baseurl)
 797         {
 798                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 799                         foreach ($webfinger["aliases"] as $alias) {
 800                                 if (substr($alias, 0, 5) == 'acct:') {
 801                                         $data["addr"] = substr($alias, 5);
 802                                 }
 803                         }
 804                 }
 805
 806                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 807                         $data["addr"] = substr($webfinger["subject"], 5);
 808                 }
 809
 810                 $zot_url = '';
 811                 foreach ($webfinger['links'] as $link) {
 812                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 813                                 $zot_url = $link['href'];
 814                         }
 815                 }
 816
 817                 if (empty($zot_url) && !empty($data['addr']) && !empty($baseurl)) {
 818                         $condition = ['nurl' => Strings::normaliseLink($baseurl), 'platform' => ['hubzilla']];
 819                         if (!DBA::exists('gserver', $condition)) {
 820                                 return $data;
 821                         }
 822                         $zot_url = $baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 823                 }
 824
 825                 if (empty($zot_url)) {
 826                         return $data;
 827                 }
 828
 829                 $data = self::pollZot($zot_url, $data);
 830
 831                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 832                         foreach ($webfinger['aliases'] as $alias) {
 833                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
 834                                         $data['alias'] = $alias;
 835                                 }
 836                         }
 837                 }
 838
 839                 return $data;
 840         }
 841
 842         public static function pollZot($url, $data)
 843         {
 844                 $curlResult = DI::httpClient()->get($url);
 845                 if ($curlResult->isTimeout()) {
 846                         return $data;
 847                 }
 848                 $content = $curlResult->getBody();
 849                 if (!$content) {
 850                         return $data;
 851                 }
 852
 853                 $json = json_decode($content, true);
 854                 if (!is_array($json)) {
 855                         return $data;
 856                 }
 857
 858                 if (empty($data['network'])) {
 859                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 860                                 $data['network'] = Protocol::ZOT;
 861                         } elseif (!isset($json['protocols'])) {
 862                                 $data['network'] = Protocol::ZOT;
 863                         }
 864                 }
 865
 866                 if (!empty($json['guid']) && empty($data['guid'])) {
 867                         $data['guid'] = $json['guid'];
 868                 }
 869                 if (!empty($json['key']) && empty($data['pubkey'])) {
 870                         $data['pubkey'] = $json['key'];
 871                 }
 872                 if (!empty($json['name'])) {
 873                         $data['name'] = $json['name'];
 874                 }
 875                 if (!empty($json['photo'])) {
 876                         $data['photo'] = $json['photo'];
 877                         if (!empty($json['photo_updated'])) {
 878                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 879                         }
 880                 }
 881                 if (!empty($json['address'])) {
 882                         $data['addr'] = $json['address'];
 883                 }
 884                 if (!empty($json['url'])) {
 885                         $data['url'] = $json['url'];
 886                 }
 887                 if (!empty($json['connections_url'])) {
 888                         $data['poco'] = $json['connections_url'];
 889                 }
 890                 if (isset($json['searchable'])) {
 891                         $data['hide'] = !$json['searchable'];
 892                 }
 893                 if (!empty($json['public_forum'])) {
 894                         $data['community'] = $json['public_forum'];
 895                         $data['account-type'] = User::PAGE_FLAGS_COMMUNITY;
 896                 }
 897
 898                 if (!empty($json['profile'])) {
 899                         $profile = $json['profile'];
 900                         if (!empty($profile['description'])) {
 901                                 $data['about'] = $profile['description'];
 902                         }
 903                         if (!empty($profile['keywords'])) {
 904                                 $keywords = implode(', ', $profile['keywords']);
 905                                 if (!empty($keywords)) {
 906                                         $data['keywords'] = $keywords;
 907                                 }
 908                         }
 909
 910                         $loc = [];
 911                         if (!empty($profile['region'])) {
 912                                 $loc['region'] = $profile['region'];
 913                         }
 914                         if (!empty($profile['country'])) {
 915                                 $loc['country-name'] = $profile['country'];
 916                         }
 917                         $location = Profile::formatLocation($loc);
 918                         if (!empty($location)) {
 919                                 $data['location'] = $location;
 920                         }
 921                 }
 922
 923                 return $data;
 924         }
 925
 926         /**
 927          * Perform a webfinger request.
 928          *
 929          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 930          *
 931          * @param string $url  Address that should be probed
 932          * @param string $type type
 933          *
 934          * @return array webfinger data
 935          * @throws HTTPException\InternalServerErrorException
 936          */
 937         public static function webfinger($url, $type)
 938         {
 939                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 940
 941                 $curlResult = DI::httpClient()->get($url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => [$type]]);
 942                 if ($curlResult->isTimeout()) {
 943                         self::$istimeout = true;
 944                         return [];
 945                 }
 946                 $data = $curlResult->getBody();
 947
 948                 $webfinger = json_decode($data, true);
 949                 if (!empty($webfinger)) {
 950                         if (!isset($webfinger["links"])) {
 951                                 Logger::info('No json webfinger links', ['url' => $url]);
 952                                 return [];
 953                         }
 954                         return $webfinger;
 955                 }
 956
 957                 // If it is not JSON, maybe it is XML
 958                 $xrd = XML::parseString($data, true);
 959                 if (!is_object($xrd)) {
 960                         Logger::info('No webfinger data retrievable', ['url' => $url]);
 961                         return [];
 962                 }
 963
 964                 $xrd_arr = XML::elementToArray($xrd);
 965                 if (!isset($xrd_arr["xrd"]["link"])) {
 966                         Logger::info('No XML webfinger links', ['url' => $url]);
 967                         return [];
 968                 }
 969
 970                 $webfinger = [];
 971
 972                 if (!empty($xrd_arr["xrd"]["subject"])) {
 973                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 974                 }
 975
 976                 if (!empty($xrd_arr["xrd"]["alias"])) {
 977                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 978                 }
 979
 980                 $webfinger["links"] = [];
 981
 982                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 983                         if (!empty($data["@attributes"])) {
 984                                 $attributes = $data["@attributes"];
 985                         } elseif ($value == "@attributes") {
 986                                 $attributes = $data;
 987                         } else {
 988                                 continue;
 989                         }
 990
 991                         $webfinger["links"][] = $attributes;
 992                 }
 993                 return $webfinger;
 994         }
 995
 996         /**
 997          * Poll the Friendica specific noscrape page.
 998          *
 999          * "noscrape" is a faster alternative to fetch the data from the hcard.
1000          * This functionality was originally created for the directory.
1001          *
1002          * @param string $noscrape_url Link to the noscrape page
1003          * @param array  $data         The already fetched data
1004          *
1005          * @return array noscrape data
1006          * @throws HTTPException\InternalServerErrorException
1007          */
1008         private static function pollNoscrape($noscrape_url, $data)
1009         {
1010                 $curlResult = DI::httpClient()->get($noscrape_url);
1011                 if ($curlResult->isTimeout()) {
1012                         self::$istimeout = true;
1013                         return $data;
1014                 }
1015                 $content = $curlResult->getBody();
1016                 if (!$content) {
1017                         Logger::info('Empty body', ['url' => $noscrape_url]);
1018                         return $data;
1019                 }
1020
1021                 $json = json_decode($content, true);
1022                 if (!is_array($json)) {
1023                         Logger::info('No json data', ['url' => $noscrape_url]);
1024                         return $data;
1025                 }
1026
1027                 if (!empty($json["fn"])) {
1028                         $data["name"] = $json["fn"];
1029                 }
1030
1031                 if (!empty($json["addr"])) {
1032                         $data["addr"] = $json["addr"];
1033                 }
1034
1035                 if (!empty($json["nick"])) {
1036                         $data["nick"] = $json["nick"];
1037                 }
1038
1039                 if (!empty($json["guid"])) {
1040                         $data["guid"] = $json["guid"];
1041                 }
1042
1043                 if (!empty($json["comm"])) {
1044                         $data["community"] = $json["comm"];
1045                 }
1046
1047                 if (!empty($json["tags"])) {
1048                         $keywords = implode(", ", $json["tags"]);
1049                         if ($keywords != "") {
1050                                 $data["keywords"] = $keywords;
1051                         }
1052                 }
1053
1054                 $location = Profile::formatLocation($json);
1055                 if ($location) {
1056                         $data["location"] = $location;
1057                 }
1058
1059                 if (!empty($json["about"])) {
1060                         $data["about"] = $json["about"];
1061                 }
1062
1063                 if (!empty($json["xmpp"])) {
1064                         $data["xmpp"] = $json["xmpp"];
1065                 }
1066
1067                 if (!empty($json["matrix"])) {
1068                         $data["matrix"] = $json["matrix"];
1069                 }
1070
1071                 if (!empty($json["key"])) {
1072                         $data["pubkey"] = $json["key"];
1073                 }
1074
1075                 if (!empty($json["photo"])) {
1076                         $data["photo"] = $json["photo"];
1077                 }
1078
1079                 if (!empty($json["dfrn-request"])) {
1080                         $data["request"] = $json["dfrn-request"];
1081                 }
1082
1083                 if (!empty($json["dfrn-confirm"])) {
1084                         $data["confirm"] = $json["dfrn-confirm"];
1085                 }
1086
1087                 if (!empty($json["dfrn-notify"])) {
1088                         $data["notify"] = $json["dfrn-notify"];
1089                 }
1090
1091                 if (!empty($json["dfrn-poll"])) {
1092                         $data["poll"] = $json["dfrn-poll"];
1093                 }
1094
1095                 if (isset($json["hide"])) {
1096                         $data["hide"] = (bool)$json["hide"];
1097                 } else {
1098                         $data["hide"] = false;
1099                 }
1100
1101                 return $data;
1102         }
1103
1104         /**
1105          * Check for valid DFRN data
1106          *
1107          * @param array $data DFRN data
1108          *
1109          * @return int Number of errors
1110          */
1111         public static function validDfrn($data)
1112         {
1113                 $errors = 0;
1114                 if (!isset($data['key'])) {
1115                         $errors ++;
1116                 }
1117                 if (!isset($data['dfrn-request'])) {
1118                         $errors ++;
1119                 }
1120                 if (!isset($data['dfrn-confirm'])) {
1121                         $errors ++;
1122                 }
1123                 if (!isset($data['dfrn-notify'])) {
1124                         $errors ++;
1125                 }
1126                 if (!isset($data['dfrn-poll'])) {
1127                         $errors ++;
1128                 }
1129                 return $errors;
1130         }
1131
1132         /**
1133          * Fetch data from a DFRN profile page and via "noscrape"
1134          *
1135          * @param string $profile_link Link to the profile page
1136          *
1137          * @return array profile data
1138          * @throws HTTPException\InternalServerErrorException
1139          * @throws \ImagickException
1140          */
1141         public static function profile($profile_link)
1142         {
1143                 $data = [];
1144
1145                 Logger::info('Check profile', ['link' => $profile_link]);
1146
1147                 // Fetch data via noscrape - this is faster
1148                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1149                 $data = self::pollNoscrape($noscrape_url, $data);
1150
1151                 if (!isset($data["notify"])
1152                         || !isset($data["confirm"])
1153                         || !isset($data["request"])
1154                         || !isset($data["poll"])
1155                         || !isset($data["name"])
1156                         || !isset($data["photo"])
1157                 ) {
1158                         $data = self::pollHcard($profile_link, $data, true);
1159                 }
1160
1161                 $prof_data = [];
1162
1163                 if (empty($data["addr"]) || empty($data["nick"])) {
1164                         $probe_data = self::uri($profile_link);
1165                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1166                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1167                 }
1168
1169                 $prof_data["addr"]         = $data["addr"];
1170                 $prof_data["nick"]         = $data["nick"];
1171                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1172                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1173                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1174                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1175                 $prof_data["photo"]        = $data['photo']   ?? null;
1176                 $prof_data["fn"]           = $data['name']    ?? null;
1177                 $prof_data["key"]          = $data['pubkey']  ?? null;
1178
1179                 Logger::debug('Result', ['link' => $profile_link, 'data' => $prof_data]);
1180
1181                 return $prof_data;
1182         }
1183
1184         /**
1185          * Check for DFRN contact
1186          *
1187          * @param array $webfinger Webfinger data
1188          *
1189          * @return array DFRN data
1190          * @throws HTTPException\InternalServerErrorException
1191          */
1192         private static function dfrn($webfinger)
1193         {
1194                 $hcard_url = "";
1195                 $data = [];
1196                 // The array is reversed to take into account the order of preference for same-rel links
1197                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1198                 foreach (array_reverse($webfinger["links"]) as $link) {
1199                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1200                                 $data["network"] = Protocol::DFRN;
1201                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1202                                 $data["poll"] = $link["href"];
1203                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1204                                 $data["url"] = $link["href"];
1205                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1206                                 $hcard_url = $link["href"];
1207                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1208                                 $data["poco"] = $link["href"];
1209                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1210                                 $data["photo"] = $link["href"];
1211                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1212                                 $data["baseurl"] = trim($link["href"], '/');
1213                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1214                                 $data["guid"] = $link["href"];
1215                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1216                                 $data["pubkey"] = base64_decode($link["href"]);
1217
1218                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1219                                 if (strstr($data["pubkey"], 'RSA ')) {
1220                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1221                                 }
1222                         }
1223                 }
1224
1225                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1226                         foreach ($webfinger["aliases"] as $alias) {
1227                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1228                                         $data["url"] = $alias;
1229                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1230                                         $data["alias"] = $alias;
1231                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1232                                         $data["addr"] = substr($alias, 5);
1233                                 }
1234                         }
1235                 }
1236
1237                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1238                         $data["addr"] = substr($webfinger["subject"], 5);
1239                 }
1240
1241                 if (!isset($data["network"]) || ($hcard_url == "")) {
1242                         return [];
1243                 }
1244
1245                 // Fetch data via noscrape - this is faster
1246                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1247                 $data = self::pollNoscrape($noscrape_url, $data);
1248
1249                 if (isset($data["notify"])
1250                         && isset($data["confirm"])
1251                         && isset($data["request"])
1252                         && isset($data["poll"])
1253                         && isset($data["name"])
1254                         && isset($data["photo"])
1255                 ) {
1256                         return $data;
1257                 }
1258
1259                 $data = self::pollHcard($hcard_url, $data, true);
1260
1261                 return $data;
1262         }
1263
1264         /**
1265          * Poll the hcard page (Diaspora and Friendica specific)
1266          *
1267          * @param string  $hcard_url Link to the hcard page
1268          * @param array   $data      The already fetched data
1269          * @param boolean $dfrn      Poll DFRN specific data
1270          *
1271          * @return array hcard data
1272          * @throws HTTPException\InternalServerErrorException
1273          */
1274         private static function pollHcard($hcard_url, $data, $dfrn = false)
1275         {
1276                 $curlResult = DI::httpClient()->get($hcard_url);
1277                 if ($curlResult->isTimeout()) {
1278                         self::$istimeout = true;
1279                         return [];
1280                 }
1281                 $content = $curlResult->getBody();
1282                 if (empty($content)) {
1283                         return [];
1284                 }
1285
1286                 $doc = new DOMDocument();
1287                 if (!@$doc->loadHTML($content)) {
1288                         return [];
1289                 }
1290
1291                 $xpath = new DomXPath($doc);
1292
1293                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1294                 if (!is_object($vcards)) {
1295                         return [];
1296                 }
1297
1298                 if (!isset($data["baseurl"])) {
1299                         $data["baseurl"] = "";
1300                 }
1301
1302                 if ($vcards->length > 0) {
1303                         $vcard = $vcards->item(0);
1304
1305                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1306                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1307                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1308                         if (($search->length > 0) && empty($data["guid"])) {
1309                                 $data["guid"] = $search->item(0)->nodeValue;
1310                         }
1311
1312                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1313                         if ($search->length > 0) {
1314                                 $data["nick"] = $search->item(0)->nodeValue;
1315                         }
1316
1317                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1318                         if ($search->length > 0) {
1319                                 $data["name"] = $search->item(0)->nodeValue;
1320                         }
1321
1322                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1323                         if ($search->length > 0) {
1324                                 $data["searchable"] = $search->item(0)->nodeValue;
1325                         }
1326
1327                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1328                         if ($search->length > 0) {
1329                                 $data["pubkey"] = $search->item(0)->nodeValue;
1330                                 if (strstr($data["pubkey"], 'RSA ')) {
1331                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1332                                 }
1333                         }
1334
1335                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1336                         if ($search->length > 0) {
1337                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1338                         }
1339                 }
1340
1341                 $avatar = [];
1342                 if (!empty($vcard)) {
1343                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1344                         foreach ($photos as $photo) {
1345                                 $attr = [];
1346                                 foreach ($photo->attributes as $attribute) {
1347                                         $attr[$attribute->name] = trim($attribute->value);
1348                                 }
1349
1350                                 if (isset($attr["src"]) && isset($attr["width"])) {
1351                                         $avatar[$attr["width"]] = $attr["src"];
1352                                 }
1353
1354                                 // We don't have a width. So we just take everything that we got.
1355                                 // This is a Hubzilla workaround which doesn't send a width.
1356                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1357                                         $avatar[] = $attr["src"];
1358                                 }
1359                         }
1360                 }
1361
1362                 if (sizeof($avatar)) {
1363                         ksort($avatar);
1364                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1365                 }
1366
1367                 if ($dfrn) {
1368                         // Poll DFRN specific data
1369                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1370                         if ($search->length > 0) {
1371                                 foreach ($search as $link) {
1372                                         //$data["request"] = $search->item(0)->nodeValue;
1373                                         $attr = [];
1374                                         foreach ($link->attributes as $attribute) {
1375                                                 $attr[$attribute->name] = trim($attribute->value);
1376                                         }
1377
1378                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1379                                 }
1380                         }
1381
1382                         // Older Friendica versions had used the "uid" field differently than newer versions
1383                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1384                                 unset($data["guid"]);
1385                         }
1386                 }
1387
1388
1389                 return $data;
1390         }
1391
1392         /**
1393          * Check for Diaspora contact
1394          *
1395          * @param array $webfinger Webfinger data
1396          *
1397          * @return array Diaspora data
1398          * @throws HTTPException\InternalServerErrorException
1399          */
1400         private static function diaspora($webfinger)
1401         {
1402                 $hcard_url = "";
1403                 $data = [];
1404
1405                 // The array is reversed to take into account the order of preference for same-rel links
1406                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1407                 foreach (array_reverse($webfinger["links"]) as $link) {
1408                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1409                                 $hcard_url = $link["href"];
1410                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1411                                 $data["baseurl"] = trim($link["href"], '/');
1412                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1413                                 $data["guid"] = $link["href"];
1414                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1415                                 $data["url"] = $link["href"];
1416                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && empty($link["type"]) && !empty($link["href"])) {
1417                                 $profile_url = $link["href"];
1418                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1419                                 $data["poll"] = $link["href"];
1420                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1421                                 $data["poco"] = $link["href"];
1422                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1423                                 $data["notify"] = $link["href"];
1424                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1425                                 $data["pubkey"] = base64_decode($link["href"]);
1426
1427                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1428                                 if (strstr($data["pubkey"], 'RSA ')) {
1429                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1430                                 }
1431                         }
1432                 }
1433
1434                 if (empty($data["url"]) && !empty($profile_url)) {
1435                         $data["url"] = $profile_url;
1436                 }
1437
1438                 if (empty($data["url"]) || empty($hcard_url)) {
1439                         return [];
1440                 }
1441
1442                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1443                         foreach ($webfinger["aliases"] as $alias) {
1444                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1445                                         $data["alias"] = $alias;
1446                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1447                                         $data["addr"] = substr($alias, 5);
1448                                 }
1449                         }
1450                 }
1451
1452                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1453                         $data["addr"] = substr($webfinger["subject"], 5);
1454                 }
1455
1456                 // Fetch further information from the hcard
1457                 $data = self::pollHcard($hcard_url, $data);
1458
1459                 if (!$data) {
1460                         return [];
1461                 }
1462
1463                 if (!empty($data["url"])
1464                         && !empty($data["guid"])
1465                         && !empty($data["baseurl"])
1466                         && !empty($data["pubkey"])
1467                         && !empty($hcard_url)
1468                 ) {
1469                         $data["network"] = Protocol::DIASPORA;
1470                         $data["manually-approve"] = false;
1471
1472                         // The Diaspora handle must always be lowercase
1473                         if (!empty($data["addr"])) {
1474                                 $data["addr"] = strtolower($data["addr"]);
1475                         }
1476
1477                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1478                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1479                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1480                 } else {
1481                         return [];
1482                 }
1483
1484                 return $data;
1485         }
1486
1487         /**
1488          * Check for OStatus contact
1489          *
1490          * @param array $webfinger Webfinger data
1491          * @param bool  $short     Short detection mode
1492          *
1493          * @return array|bool OStatus data or "false" on error or "true" on short mode
1494          * @throws HTTPException\InternalServerErrorException
1495          */
1496         private static function ostatus($webfinger, $short = false)
1497         {
1498                 $data = [];
1499
1500                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1501                         foreach ($webfinger["aliases"] as $alias) {
1502                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1503                                         $data["addr"] = str_replace('acct:', '', $alias);
1504                                 }
1505                         }
1506                 }
1507
1508                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1509                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1510                 ) {
1511                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1512                 }
1513
1514                 if (!empty($webfinger["links"])) {
1515                         // The array is reversed to take into account the order of preference for same-rel links
1516                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1517                         foreach (array_reverse($webfinger["links"]) as $link) {
1518                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1519                                         && (($link["type"] ?? "") == "text/html")
1520                                         && ($link["href"] != "")
1521                                 ) {
1522                                         $data["url"] = $data["alias"] = $link["href"];
1523                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1524                                         $data["notify"] = $link["href"];
1525                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1526                                         $data["poll"] = $link["href"];
1527                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1528                                         $pubkey = $link["href"];
1529
1530                                         if (substr($pubkey, 0, 5) === 'data:') {
1531                                                 if (strstr($pubkey, ',')) {
1532                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1533                                                 } else {
1534                                                         $pubkey = substr($pubkey, 5);
1535                                                 }
1536                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1537                                                 $curlResult = DI::httpClient()->get($pubkey);
1538                                                 if ($curlResult->isTimeout()) {
1539                                                         self::$istimeout = true;
1540                                                         return $short ? false : [];
1541                                                 }
1542                                                 $pubkey = $curlResult->getBody();
1543                                         }
1544
1545                                         $key = explode(".", $pubkey);
1546
1547                                         if (sizeof($key) >= 3) {
1548                                                 $m = Strings::base64UrlDecode($key[1]);
1549                                                 $e = Strings::base64UrlDecode($key[2]);
1550                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1551                                         }
1552                                 }
1553                         }
1554                 }
1555
1556                 if (isset($data["notify"]) && isset($data["pubkey"])
1557                         && isset($data["poll"])
1558                         && isset($data["url"])
1559                 ) {
1560                         $data["network"] = Protocol::OSTATUS;
1561                         $data["manually-approve"] = false;
1562                 } else {
1563                         return $short ? false : [];
1564                 }
1565
1566                 if ($short) {
1567                         return true;
1568                 }
1569
1570                 // Fetch all additional data from the feed
1571                 $curlResult = DI::httpClient()->get($data["poll"]);
1572                 if ($curlResult->isTimeout()) {
1573                         self::$istimeout = true;
1574                         return [];
1575                 }
1576                 $feed = $curlResult->getBody();
1577                 $feed_data = Feed::import($feed);
1578                 if (!$feed_data) {
1579                         return [];
1580                 }
1581
1582                 if (!empty($feed_data["header"]["author-name"])) {
1583                         $data["name"] = $feed_data["header"]["author-name"];
1584                 }
1585                 if (!empty($feed_data["header"]["author-nick"])) {
1586                         $data["nick"] = $feed_data["header"]["author-nick"];
1587                 }
1588                 if (!empty($feed_data["header"]["author-avatar"])) {
1589                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1590                 }
1591                 if (!empty($feed_data["header"]["author-id"])) {
1592                         $data["alias"] = $feed_data["header"]["author-id"];
1593                 }
1594                 if (!empty($feed_data["header"]["author-location"])) {
1595                         $data["location"] = $feed_data["header"]["author-location"];
1596                 }
1597                 if (!empty($feed_data["header"]["author-about"])) {
1598                         $data["about"] = $feed_data["header"]["author-about"];
1599                 }
1600                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1601                 // So we take the value that we just fetched, although the other one worked as well
1602                 if (!empty($feed_data["header"]["author-link"])) {
1603                         $data["url"] = $feed_data["header"]["author-link"];
1604                 }
1605
1606                 if ($data["url"] == $data["alias"]) {
1607                         $data["alias"] = '';
1608                 }
1609
1610                 /// @todo Fetch location and "about" from the feed as well
1611                 return $data;
1612         }
1613
1614         /**
1615          * Fetch data from a pump.io profile page
1616          *
1617          * @param string $profile_link Link to the profile page
1618          *
1619          * @return array profile data
1620          */
1621         private static function pumpioProfileData($profile_link)
1622         {
1623                 $curlResult = DI::httpClient()->get($profile_link);
1624                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
1625                         return [];
1626                 }
1627
1628                 $doc = new DOMDocument();
1629                 if (!@$doc->loadHTML($curlResult->getBody())) {
1630                         return [];
1631                 }
1632
1633                 $xpath = new DomXPath($doc);
1634
1635                 $data = [];
1636
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 $addr
1680          * @return array pump.io data
1681          */
1682         private static function pumpio($webfinger, $addr)
1683         {
1684                 $data = [];
1685                 // The array is reversed to take into account the order of preference for same-rel links
1686                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1687                 foreach (array_reverse($webfinger["links"]) as $link) {
1688                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1689                                 && (($link["type"] ?? "") == "text/html")
1690                                 && ($link["href"] != "")
1691                         ) {
1692                                 $data["url"] = $link["href"];
1693                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1694                                 $data["notify"] = $link["href"];
1695                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1696                                 $data["poll"] = $link["href"];
1697                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1698                                 $data["dialback"] = $link["href"];
1699                         }
1700                 }
1701                 if (isset($data["poll"]) && isset($data["notify"])
1702                         && isset($data["dialback"])
1703                         && isset($data["url"])
1704                 ) {
1705                         // by now we use these fields only for the network type detection
1706                         // So we unset all data that isn't used at the moment
1707                         unset($data["dialback"]);
1708
1709                         $data["network"] = Protocol::PUMPIO;
1710                 } else {
1711                         return [];
1712                 }
1713
1714                 $profile_data = self::pumpioProfileData($data["url"]);
1715
1716                 if (!$profile_data) {
1717                         return [];
1718                 }
1719
1720                 $data = array_merge($data, $profile_data);
1721
1722                 if (($addr != '') && ($data['name'] != '')) {
1723                         $name = trim(str_replace($addr, '', $data['name']));
1724                         if ($name != '') {
1725                                 $data['name'] = $name;
1726                         }
1727                 }
1728
1729                 return $data;
1730         }
1731
1732         /**
1733          * Checks HTML page for RSS feed link
1734          *
1735          * @param string $url  Page link
1736          * @param string $body Page body string
1737          * @return string|false Feed link or false if body was invalid HTML document
1738          */
1739         public static function getFeedLink(string $url, string $body)
1740         {
1741                 if (empty($body)) {
1742                         return '';
1743                 }
1744
1745                 $doc = new DOMDocument();
1746                 if (!@$doc->loadHTML($body)) {
1747                         return false;
1748                 }
1749
1750                 $xpath = new DOMXPath($doc);
1751
1752                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1753
1754                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1755
1756                 return $feedUrl;
1757         }
1758
1759         /**
1760          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1761          *
1762          * Loosely based on RFC 1808
1763          *
1764          * @see https://tools.ietf.org/html/rfc1808
1765          *
1766          * @param string   $href  The potential relative href found in the HTML document
1767          * @param string   $base  The HTML document URL
1768          * @param DOMXPath $xpath The HTML document XPath
1769          * @return string
1770          */
1771         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
1772         {
1773                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1774                         return $href;
1775                 }
1776
1777                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1778
1779                 $baseParts = parse_url($base);
1780                 if (empty($baseParts['host'])) {
1781                         return $href;
1782                 }
1783
1784                 // Naked domain case (scheme://basehost)
1785                 $path = $baseParts['path'] ?? '/';
1786
1787                 // Remove the filename part of the path if it exists (/base/path/file)
1788                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1789
1790                 $hrefParts = parse_url($href);
1791
1792                 if (!empty($hrefParts['path'])) {
1793                         // Root path case (/path) including relative scheme case (//host/path)
1794                         if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1795                                 $path = $hrefParts['path'];
1796                         } else {
1797                                 $path = $path . '/' . $hrefParts['path'];
1798
1799                                 // Resolve arbitrary relative path
1800                                 // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1801                                 $parts = array_filter(explode('/', $path), 'strlen');
1802                                 $absolutes = array();
1803                                 foreach ($parts as $part) {
1804                                         if ('.' == $part) continue;
1805                                         if ('..' == $part) {
1806                                                 array_pop($absolutes);
1807                                         } else {
1808                                                 $absolutes[] = $part;
1809                                         }
1810                                 }
1811
1812                                 $path = '/' . implode('/', $absolutes);
1813                         }
1814                 }
1815
1816                 // Relative scheme case (//host/path)
1817                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1818                 $baseParts['path'] = $path;
1819                 unset($baseParts['query']);
1820                 unset($baseParts['fragment']);
1821
1822                 return Network::unparseURL($baseParts);
1823         }
1824
1825         /**
1826          * Check for feed contact
1827          *
1828          * @param string  $url   Profile link
1829          * @param boolean $probe Do a probe if the page contains a feed link
1830          *
1831          * @return array feed data
1832          * @throws HTTPException\InternalServerErrorException
1833          */
1834         private static function feed($url, $probe = true)
1835         {
1836                 $curlResult = DI::httpClient()->get($url);
1837                 if ($curlResult->isTimeout()) {
1838                         self::$istimeout = true;
1839                         return [];
1840                 }
1841                 $feed = $curlResult->getBody();
1842                 $feed_data = Feed::import($feed);
1843
1844                 if (!$feed_data) {
1845                         if (!$probe) {
1846                                 return [];
1847                         }
1848
1849                         $feed_url = self::getFeedLink($url, $feed);
1850
1851                         if (!$feed_url) {
1852                                 return [];
1853                         }
1854
1855                         return self::feed($feed_url, false);
1856                 }
1857
1858                 if (!empty($feed_data["header"]["author-name"])) {
1859                         $data["name"] = $feed_data["header"]["author-name"];
1860                 }
1861
1862                 if (!empty($feed_data["header"]["author-nick"])) {
1863                         $data["nick"] = $feed_data["header"]["author-nick"];
1864                 }
1865
1866                 if (!empty($feed_data["header"]["author-avatar"])) {
1867                         $data["photo"] = $feed_data["header"]["author-avatar"];
1868                 }
1869
1870                 if (!empty($feed_data["header"]["author-id"])) {
1871                         $data["alias"] = $feed_data["header"]["author-id"];
1872                 }
1873
1874                 $data["url"] = $url;
1875                 $data["poll"] = $url;
1876
1877                 $data["network"] = Protocol::FEED;
1878
1879                 return $data;
1880         }
1881
1882         /**
1883          * Check for mail contact
1884          *
1885          * @param string  $uri Profile link
1886          * @param integer $uid User ID
1887          *
1888          * @return array mail data
1889          * @throws \Exception
1890          */
1891         private static function mail($uri, $uid)
1892         {
1893                 if (!Network::isEmailDomainValid($uri)) {
1894                         return [];
1895                 }
1896
1897                 if ($uid == 0) {
1898                         return [];
1899                 }
1900
1901                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1902
1903                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1904                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1905                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1906
1907                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1908                         return [];
1909                 }
1910
1911                 $mailbox = Email::constructMailboxName($mailacct);
1912                 $password = '';
1913                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1914                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1915                 if (!$mbox) {
1916                         return [];
1917                 }
1918
1919                 $msgs = Email::poll($mbox, $uri);
1920                 Logger::info('Messages found', ['uri' => $uri, 'count' => count($msgs)]);
1921
1922                 if (!count($msgs)) {
1923                         return [];
1924                 }
1925
1926                 $phost = substr($uri, strpos($uri, '@') + 1);
1927
1928                 $data = [];
1929                 $data["addr"]    = $uri;
1930                 $data["network"] = Protocol::MAIL;
1931                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1932                 $data["nick"]    = $data["name"];
1933                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1934                 $data["url"]     = 'mailto:'.$uri;
1935                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1936                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1937
1938                 $x = Email::messageMeta($mbox, $msgs[0]);
1939                 if (stristr($x[0]->from, $uri)) {
1940                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1941                 } elseif (stristr($x[0]->to, $uri)) {
1942                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1943                 }
1944                 if (isset($adr)) {
1945                         foreach ($adr as $feadr) {
1946                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1947                                         &&(strcasecmp($feadr->host, $phost) == 0)
1948                                         && (strlen($feadr->personal))
1949                                 ) {
1950                                         $personal = imap_mime_header_decode($feadr->personal);
1951                                         $data["name"] = "";
1952                                         foreach ($personal as $perspart) {
1953                                                 if ($perspart->charset != "default") {
1954                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1955                                                 } else {
1956                                                         $data["name"] .= $perspart->text;
1957                                                 }
1958                                         }
1959                                 }
1960                         }
1961                 }
1962                 if (!empty($mbox)) {
1963                         imap_close($mbox);
1964                 }
1965                 return $data;
1966         }
1967
1968         /**
1969          * Mix two paths together to possibly fix missing parts
1970          *
1971          * @param string $avatar Path to the avatar
1972          * @param string $base   Another path that is hopefully complete
1973          *
1974          * @return string fixed avatar path
1975          * @throws \Exception
1976          */
1977         public static function fixAvatar($avatar, $base)
1978         {
1979                 $base_parts = parse_url($base);
1980
1981                 // Remove all parts that could create a problem
1982                 unset($base_parts['path']);
1983                 unset($base_parts['query']);
1984                 unset($base_parts['fragment']);
1985
1986                 $avatar_parts = parse_url($avatar);
1987
1988                 // Now we mix them
1989                 $parts = array_merge($base_parts, $avatar_parts);
1990
1991                 // And put them together again
1992                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1993                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1994                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1995                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1996                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1997                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1998
1999                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
2000
2001                 Logger::debug('Avatar fixed', ['base' => $base, 'avatar' => $avatar, 'fixed' => $fixed]);
2002
2003                 return $fixed;
2004         }
2005
2006         /**
2007          * Fetch the last date that the contact had posted something (publically)
2008          *
2009          * @param string $data  probing result
2010          * @return string last activity
2011          */
2012         public static function getLastUpdate(array $data)
2013         {
2014                 $uid = User::getIdForURL($data['url']);
2015                 if (!empty($uid)) {
2016                         $contact = Contact::selectFirst(['url', 'last-item'], ['self' => true, 'uid' => $uid]);
2017                         if (!empty($contact['last-item'])) {
2018                                 return $contact['last-item'];
2019                         }
2020                 }
2021
2022                 if ($lastUpdate = self::updateFromNoScrape($data)) {
2023                         return $lastUpdate;
2024                 }
2025
2026                 if (!empty($data['outbox'])) {
2027                         return self::updateFromOutbox($data['outbox'], $data);
2028                 } elseif (!empty($data['poll']) && ($data['network'] == Protocol::ACTIVITYPUB)) {
2029                         return self::updateFromOutbox($data['poll'], $data);
2030                 } elseif (!empty($data['poll'])) {
2031                         return self::updateFromFeed($data);
2032                 }
2033
2034                 return '';
2035         }
2036
2037         /**
2038          * Fetch the last activity date from the "noscrape" endpoint
2039          *
2040          * @param array $data Probing result
2041          * @return string last activity
2042          *
2043          * @return bool 'true' if update was successful or the server was unreachable
2044          */
2045         private static function updateFromNoScrape(array $data)
2046         {
2047                 if (empty($data['baseurl'])) {
2048                         return '';
2049                 }
2050
2051                 // Check the 'noscrape' endpoint when it is a Friendica server
2052                 $gserver = DBA::selectFirst('gserver', ['noscrape'], ["`nurl` = ? AND `noscrape` != ''",
2053                         Strings::normaliseLink($data['baseurl'])]);
2054                 if (!DBA::isResult($gserver)) {
2055                         return '';
2056                 }
2057
2058                 $curlResult = DI::httpClient()->get($gserver['noscrape'] . '/' . $data['nick']);
2059
2060                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
2061                         $noscrape = json_decode($curlResult->getBody(), true);
2062                         if (!empty($noscrape) && !empty($noscrape['updated'])) {
2063                                 return DateTimeFormat::utc($noscrape['updated'], DateTimeFormat::MYSQL);
2064                         }
2065                 }
2066
2067                 return '';
2068         }
2069
2070         /**
2071          * Fetch the last activity date from an ActivityPub Outbox
2072          *
2073          * @param string $feed
2074          * @param array  $data Probing result
2075          * @return string last activity
2076          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
2077          */
2078         private static function updateFromOutbox(string $feed, array $data)
2079         {
2080                 $outbox = ActivityPub::fetchContent($feed);
2081                 if (empty($outbox)) {
2082                         return '';
2083                 }
2084
2085                 if (!empty($outbox['orderedItems'])) {
2086                         $items = $outbox['orderedItems'];
2087                 } elseif (!empty($outbox['first']['orderedItems'])) {
2088                         $items = $outbox['first']['orderedItems'];
2089                 } elseif (!empty($outbox['first']['href']) && ($outbox['first']['href'] != $feed)) {
2090                         return self::updateFromOutbox($outbox['first']['href'], $data);
2091                 } elseif (!empty($outbox['first'])) {
2092                         if (is_string($outbox['first']) && ($outbox['first'] != $feed)) {
2093                                 return self::updateFromOutbox($outbox['first'], $data);
2094                         } else {
2095                                 Logger::warning('Unexpected data', ['outbox' => $outbox]);
2096                         }
2097                         return '';
2098                 } else {
2099                         $items = [];
2100                 }
2101
2102                 $last_updated = '';
2103                 foreach ($items as $activity) {
2104                         if (!empty($activity['published'])) {
2105                                 $published =  DateTimeFormat::utc($activity['published']);
2106                         } elseif (!empty($activity['object']['published'])) {
2107                                 $published =  DateTimeFormat::utc($activity['object']['published']);
2108                         } else {
2109                                 continue;
2110                         }
2111
2112                         if ($last_updated < $published) {
2113                                 $last_updated = $published;
2114                         }
2115                 }
2116
2117                 if (!empty($last_updated)) {
2118                         return $last_updated;
2119                 }
2120
2121                 return '';
2122         }
2123
2124         /**
2125          * Fetch the last activity date from an XML feed
2126          *
2127          * @param array $data Probing result
2128          * @return string last activity
2129          */
2130         private static function updateFromFeed(array $data)
2131         {
2132                 // Search for the newest entry in the feed
2133                 $curlResult = DI::httpClient()->get($data['poll']);
2134                 if (!$curlResult->isSuccess() || !$curlResult->getBody()) {
2135                         return '';
2136                 }
2137
2138                 $doc = new DOMDocument();
2139                 @$doc->loadXML($curlResult->getBody());
2140
2141                 $xpath = new DOMXPath($doc);
2142                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
2143
2144                 $entries = $xpath->query('/atom:feed/atom:entry');
2145
2146                 $last_updated = '';
2147
2148                 foreach ($entries as $entry) {
2149                         $published_item = $xpath->query('atom:published/text()', $entry)->item(0);
2150                         $updated_item   = $xpath->query('atom:updated/text()'  , $entry)->item(0);
2151                         $published      = !empty($published_item->nodeValue) ? DateTimeFormat::utc($published_item->nodeValue) : null;
2152                         $updated        = !empty($updated_item->nodeValue) ? DateTimeFormat::utc($updated_item->nodeValue) : null;
2153
2154                         if (empty($published) || empty($updated)) {
2155                                 Logger::notice('Invalid entry for XPath.', ['entry' => $entry, 'url' => $data['url']]);
2156                                 continue;
2157                         }
2158
2159                         if ($last_updated < $published) {
2160                                 $last_updated = $published;
2161                         }
2162
2163                         if ($last_updated < $updated) {
2164                                 $last_updated = $updated;
2165                         }
2166                 }
2167
2168                 if (!empty($last_updated)) {
2169                         return $last_updated;
2170                 }
2171
2172                 return '';
2173         }
2174
2175         /**
2176          * Probe data from local profiles without network traffic
2177          *
2178          * @param string $url
2179          * @return array probed data
2180          * @throws HTTPException\InternalServerErrorException
2181          * @throws HTTPException\NotFoundException
2182          */
2183         private static function localProbe(string $url): array
2184         {
2185                 try {
2186                         $uid = User::getIdForURL($url);
2187                         if (!$uid) {
2188                                 throw new HTTPException\NotFoundException('User not found.');
2189                         }
2190
2191                         $owner     = User::getOwnerDataById($uid);
2192                         $approfile = ActivityPub\Transmitter::getProfile($uid);
2193
2194                         if (empty($owner['gsid'])) {
2195                                 $owner['gsid'] = GServer::getID($approfile['generator']['url']);
2196                         }
2197
2198                         $data = [
2199                                 'name' => $owner['name'], 'nick' => $owner['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
2200                                 'url' => $owner['url'], 'addr' => $owner['addr'], 'alias' => $owner['alias'],
2201                                 'photo' => User::getAvatarUrl($owner),
2202                                 'header' => $owner['header'] ? Contact::getHeaderUrlForId($owner['id'], $owner['updated']) : '',
2203                                 'account-type' => $owner['contact-type'], 'community' => ($owner['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
2204                                 'keywords' => $owner['keywords'], 'location' => $owner['location'], 'about' => $owner['about'],
2205                                 'xmpp' => $owner['xmpp'], 'matrix' => $owner['matrix'],
2206                                 'hide' => !$owner['net-publish'], 'batch' => '', 'notify' => $owner['notify'],
2207                                 'poll' => $owner['poll'], 'request' => $owner['request'], 'confirm' => $owner['confirm'],
2208                                 'subscribe' => $approfile['generator']['url'] . '/follow?url={uri}', 'poco' => $owner['poco'],
2209                                 'following' => $approfile['following'], 'followers' => $approfile['followers'],
2210                                 'inbox' => $approfile['inbox'], 'outbox' => $approfile['outbox'],
2211                                 'sharedinbox' => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
2212                                 'pubkey' => $owner['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $owner['gsid'],
2213                                 'manually-approve' => in_array($owner['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])
2214                         ];
2215                 } catch (Exception $e) {
2216                         // Default values for non existing targets
2217                         $data = [
2218                                 'name' => $url, 'nick' => $url, 'url' => $url, 'network' => Protocol::PHANTOM,
2219                                 'photo' => DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO
2220                         ];
2221                 }
2222
2223                 return self::rearrangeData($data);
2224         }
2225 }