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 Webfinger array
 512          */
 513         private static function getWebfingerArray(string $uri)
 514         {
 515                 $parts = parse_url($uri);
 516
 517                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 518                         $host = $parts['host'];
 519                         if (!empty($parts['port'])) {
 520                                 $host .= ':'.$parts['port'];
 521                         }
 522
 523                         $baseurl = $parts['scheme'] . '://' . $host;
 524
 525                         $nick = '';
 526                         $addr = '';
 527
 528                         $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
 529                         if (!empty($path_parts)) {
 530                                 $nick = ltrim(end($path_parts), '@');
 531                                 // When the last part of the URI is numeric then it is most likely an ID and not a nick name
 532                                 if (!is_numeric($nick)) {
 533                                         $addr = $nick."@".$host;
 534                                 } else {
 535                                         $nick = '';
 536                                 }
 537                         }
 538
 539                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 540                         if (empty($webfinger)) {
 541                                 $lrdd = self::hostMeta($host);
 542                         }
 543
 544                         if (empty($webfinger) && empty($lrdd)) {
 545                                 while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
 546                                         $host .= "/".array_shift($path_parts);
 547                                         $baseurl = $parts['scheme'] . '://' . $host;
 548
 549                                         if (!empty($nick)) {
 550                                                 $addr = $nick."@".$host;
 551                                         }
 552
 553                                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 554                                         if (empty($webfinger)) {
 555                                                 $lrdd = self::hostMeta($host);
 556                                         }
 557                                 }
 558
 559                                 if (empty($lrdd) && empty($webfinger)) {
 560                                         return [];
 561                                 }
 562                         }
 563                 } elseif (strstr($uri, '@')) {
 564                         // Remove "acct:" from the URI
 565                         $uri = str_replace('acct:', '', $uri);
 566
 567                         $host = substr($uri, strpos($uri, '@') + 1);
 568                         $nick = substr($uri, 0, strpos($uri, '@'));
 569                         $addr = $uri;
 570
 571                         $webfinger = self::getWebfinger('https://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 572                         if (self::$istimeout) {
 573                                 return [];
 574                         }
 575
 576                         if (empty($webfinger)) {
 577                                 $webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 578                                 if (self::$istimeout) {
 579                                         return [];
 580                                 }
 581                         } else {
 582                                 $baseurl = 'https://' . $host;
 583                         }
 584
 585                         if (empty($webfinger)) {
 586                                 $lrdd = self::hostMeta($host);
 587                                 if (self::$istimeout) {
 588                                         return [];
 589                                 }
 590                                 $baseurl = self::$baseurl;
 591                         } else {
 592                                 $baseurl = 'http://' . $host;
 593                         }
 594                 } else {
 595                         Logger::info('URI was not detectable', ['uri' => $uri]);
 596                         return [];
 597                 }
 598
 599                 if (empty($webfinger)) {
 600                         foreach ($lrdd as $type => $template) {
 601                                 if ($webfinger) {
 602                                         continue;
 603                                 }
 604
 605                                 $webfinger = self::getWebfinger($template, $type, $uri, $addr);
 606                         }
 607                 }
 608
 609                 if (empty($webfinger)) {
 610                         return [];
 611                 }
 612
 613                 if ($webfinger['detected'] == $addr) {
 614                         $webfinger['nick'] = $nick;
 615                         $webfinger['addr'] = $addr;
 616                 }
 617
 618                 $webfinger['baseurl'] = $baseurl;
 619
 620                 return $webfinger;
 621         }
 622
 623         /**
 624          * Perform network request for webfinger data
 625          *
 626          * @param string $template
 627          * @param string $type
 628          * @param string $uri
 629          * @param string $addr
 630          * @return array webfinger results
 631          */
 632         private static function getWebfinger(string $template, string $type, string $uri, string $addr)
 633         {
 634                 if (Network::isUrlBlocked($template)) {
 635                         Logger::info('Domain is blocked', ['url' => $template]);
 636                         return [];
 637                 }
 638
 639                 // First try the address because this is the primary purpose of webfinger
 640                 if (!empty($addr)) {
 641                         $detected = $addr;
 642                         $path = str_replace('{uri}', urlencode("acct:" . $addr), $template);
 643                         $webfinger = self::webfinger($path, $type);
 644                         if (self::$istimeout) {
 645                                 return [];
 646                         }
 647                 }
 648
 649                 // Then try the URI
 650                 if (empty($webfinger) && $uri != $addr) {
 651                         $detected = $uri;
 652                         $path = str_replace('{uri}', urlencode($uri), $template);
 653                         $webfinger = self::webfinger($path, $type);
 654                         if (self::$istimeout) {
 655                                 return [];
 656                         }
 657                 }
 658
 659                 if (empty($webfinger)) {
 660                         return [];
 661                 }
 662
 663                 return ['webfinger' => $webfinger, 'detected' => $detected];
 664         }
 665
 666         /**
 667          * Fetch information (protocol endpoints and user information) about a given uri
 668          *
 669          * This function is only called by the "uri" function that adds caching and rearranging of data.
 670          *
 671          * @param string  $uri        Address that should be probed
 672          * @param string  $network    Test for this specific network
 673          * @param integer $uid        User ID for the probe (only used for mails)
 674          * @param array   $ap_profile Previously probed AP profile
 675          *
 676          * @return array uri data
 677          * @throws HTTPException\InternalServerErrorException
 678          */
 679         private static function detect(string $uri, string $network, int $uid, array $ap_profile)
 680         {
 681                 $hookData = [
 682                         'uri'     => $uri,
 683                         'network' => $network,
 684                         'uid'     => $uid,
 685                         'result'  => [],
 686                 ];
 687
 688                 Hook::callAll('probe_detect', $hookData);
 689
 690                 if ($hookData['result']) {
 691                         if (!is_array($hookData['result'])) {
 692                                 return [];
 693                         } else {
 694                                 return $hookData['result'];
 695                         }
 696                 }
 697
 698                 $parts = parse_url($uri);
 699
 700                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 701                         if (in_array($parts['host'], ['twitter.com', 'mobile.twitter.com'])) {
 702                                 return self::twitter($uri);
 703                         }
 704                 } elseif (strstr($uri, '@')) {
 705                         // If the URI starts with "mailto:" then jump directly to the mail detection
 706                         if (strpos($uri, 'mailto:') !== false) {
 707                                 $uri = str_replace('mailto:', '', $uri);
 708                                 return self::mail($uri, $uid);
 709                         }
 710
 711                         if ($network == Protocol::MAIL) {
 712                                 return self::mail($uri, $uid);
 713                         }
 714
 715                         if (Strings::endsWith($uri, '@twitter.com')
 716                                 || Strings::endsWith($uri, '@mobile.twitter.com')
 717                         ) {
 718                                 return self::twitter($uri);
 719                         }
 720                 } else {
 721                         Logger::info('URI was not detectable', ['uri' => $uri]);
 722                         return [];
 723                 }
 724
 725                 Logger::info('Probing start', ['uri' => $uri]);
 726
 727                 if (!empty($ap_profile['addr']) && ($ap_profile['addr'] != $uri)) {
 728                         $data = self::getWebfingerArray($ap_profile['addr']);
 729                 }
 730
 731                 if (empty($data)) {
 732                         $data = self::getWebfingerArray($uri);
 733                 }
 734
 735                 if (empty($data)) {
 736                         if (!empty($parts['scheme'])) {
 737                                 return self::feed($uri);
 738                         } elseif (!empty($uid)) {
 739                                 return self::mail($uri, $uid);
 740                         } else {
 741                                 return [];
 742                         }
 743                 }
 744
 745                 $webfinger = $data['webfinger'];
 746                 $nick = $data['nick'] ?? '';
 747                 $addr = $data['addr'] ?? '';
 748                 $baseurl = $data['baseurl'] ?? '';
 749
 750                 $result = [];
 751
 752                 if (in_array($network, ["", Protocol::DFRN])) {
 753                         $result = self::dfrn($webfinger);
 754                 }
 755                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 756                         $result = self::diaspora($webfinger);
 757                 }
 758                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 759                         $result = self::ostatus($webfinger);
 760                 }
 761                 if (in_array($network, ['', Protocol::ZOT])) {
 762                         $result = self::zot($webfinger, $result, $baseurl);
 763                 }
 764                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 765                         $result = self::pumpio($webfinger, $addr);
 766                 }
 767                 if (empty($result['network']) && empty($ap_profile['network']) || ($network == Protocol::FEED)) {
 768                         $result = self::feed($uri);
 769                 } else {
 770                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 771                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 772                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 773                                 $result["nick"] = $nick;
 774                         }
 775
 776                         if (empty($result["addr"]) && ($addr != "")) {
 777                                 $result["addr"] = $addr;
 778                         }
 779                 }
 780
 781                 $result = self::getSubscribeLink($result, $webfinger);
 782
 783                 if (empty($result["network"])) {
 784                         $result["network"] = Protocol::PHANTOM;
 785                 }
 786
 787                 if (empty($result['baseurl']) && !empty($baseurl)) {
 788                         $result['baseurl'] = $baseurl;
 789                 }
 790
 791                 if (empty($result["url"])) {
 792                         $result["url"] = $uri;
 793                 }
 794
 795                 Logger::info('Probing done', ['uri' => $uri, 'network' => $result["network"]]);
 796
 797                 return $result;
 798         }
 799
 800         /**
 801          * Check for Zot contact
 802          *
 803          * @param array $webfinger Webfinger data
 804          * @param array $data      previously probed data
 805          *
 806          * @return array Zot data
 807          * @throws HTTPException\InternalServerErrorException
 808          */
 809         private static function zot($webfinger, $data, $baseurl)
 810         {
 811                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 812                         foreach ($webfinger["aliases"] as $alias) {
 813                                 if (substr($alias, 0, 5) == 'acct:') {
 814                                         $data["addr"] = substr($alias, 5);
 815                                 }
 816                         }
 817                 }
 818
 819                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 820                         $data["addr"] = substr($webfinger["subject"], 5);
 821                 }
 822
 823                 $zot_url = '';
 824                 foreach ($webfinger['links'] as $link) {
 825                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 826                                 $zot_url = $link['href'];
 827                         }
 828                 }
 829
 830                 if (empty($zot_url) && !empty($data['addr']) && !empty($baseurl)) {
 831                         $condition = ['nurl' => Strings::normaliseLink($baseurl), 'platform' => ['hubzilla']];
 832                         if (!DBA::exists('gserver', $condition)) {
 833                                 return $data;
 834                         }
 835                         $zot_url = $baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 836                 }
 837
 838                 if (empty($zot_url)) {
 839                         return $data;
 840                 }
 841
 842                 $data = self::pollZot($zot_url, $data);
 843
 844                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 845                         foreach ($webfinger['aliases'] as $alias) {
 846                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
 847                                         $data['alias'] = $alias;
 848                                 }
 849                         }
 850                 }
 851
 852                 return $data;
 853         }
 854
 855         public static function pollZot($url, $data)
 856         {
 857                 $curlResult = DI::httpClient()->get($url);
 858                 if ($curlResult->isTimeout()) {
 859                         return $data;
 860                 }
 861                 $content = $curlResult->getBody();
 862                 if (!$content) {
 863                         return $data;
 864                 }
 865
 866                 $json = json_decode($content, true);
 867                 if (!is_array($json)) {
 868                         return $data;
 869                 }
 870
 871                 if (empty($data['network'])) {
 872                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 873                                 $data['network'] = Protocol::ZOT;
 874                         } elseif (!isset($json['protocols'])) {
 875                                 $data['network'] = Protocol::ZOT;
 876                         }
 877                 }
 878
 879                 if (!empty($json['guid']) && empty($data['guid'])) {
 880                         $data['guid'] = $json['guid'];
 881                 }
 882                 if (!empty($json['key']) && empty($data['pubkey'])) {
 883                         $data['pubkey'] = $json['key'];
 884                 }
 885                 if (!empty($json['name'])) {
 886                         $data['name'] = $json['name'];
 887                 }
 888                 if (!empty($json['photo'])) {
 889                         $data['photo'] = $json['photo'];
 890                         if (!empty($json['photo_updated'])) {
 891                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 892                         }
 893                 }
 894                 if (!empty($json['address'])) {
 895                         $data['addr'] = $json['address'];
 896                 }
 897                 if (!empty($json['url'])) {
 898                         $data['url'] = $json['url'];
 899                 }
 900                 if (!empty($json['connections_url'])) {
 901                         $data['poco'] = $json['connections_url'];
 902                 }
 903                 if (isset($json['searchable'])) {
 904                         $data['hide'] = !$json['searchable'];
 905                 }
 906                 if (!empty($json['public_forum'])) {
 907                         $data['community'] = $json['public_forum'];
 908                         $data['account-type'] = User::PAGE_FLAGS_COMMUNITY;
 909                 }
 910
 911                 if (!empty($json['profile'])) {
 912                         $profile = $json['profile'];
 913                         if (!empty($profile['description'])) {
 914                                 $data['about'] = $profile['description'];
 915                         }
 916                         if (!empty($profile['keywords'])) {
 917                                 $keywords = implode(', ', $profile['keywords']);
 918                                 if (!empty($keywords)) {
 919                                         $data['keywords'] = $keywords;
 920                                 }
 921                         }
 922
 923                         $loc = [];
 924                         if (!empty($profile['region'])) {
 925                                 $loc['region'] = $profile['region'];
 926                         }
 927                         if (!empty($profile['country'])) {
 928                                 $loc['country-name'] = $profile['country'];
 929                         }
 930                         $location = Profile::formatLocation($loc);
 931                         if (!empty($location)) {
 932                                 $data['location'] = $location;
 933                         }
 934                 }
 935
 936                 return $data;
 937         }
 938
 939         /**
 940          * Perform a webfinger request.
 941          *
 942          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 943          *
 944          * @param string $url  Address that should be probed
 945          * @param string $type type
 946          *
 947          * @return array webfinger data
 948          * @throws HTTPException\InternalServerErrorException
 949          */
 950         public static function webfinger($url, $type)
 951         {
 952                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 953
 954                 $curlResult = DI::httpClient()->get($url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => [$type]]);
 955                 if ($curlResult->isTimeout()) {
 956                         self::$istimeout = true;
 957                         return [];
 958                 }
 959                 $data = $curlResult->getBody();
 960
 961                 $webfinger = json_decode($data, true);
 962                 if (!empty($webfinger)) {
 963                         if (!isset($webfinger["links"])) {
 964                                 Logger::info('No json webfinger links', ['url' => $url]);
 965                                 return [];
 966                         }
 967                         return $webfinger;
 968                 }
 969
 970                 // If it is not JSON, maybe it is XML
 971                 $xrd = XML::parseString($data, true);
 972                 if (!is_object($xrd)) {
 973                         Logger::info('No webfinger data retrievable', ['url' => $url]);
 974                         return [];
 975                 }
 976
 977                 $xrd_arr = XML::elementToArray($xrd);
 978                 if (!isset($xrd_arr["xrd"]["link"])) {
 979                         Logger::info('No XML webfinger links', ['url' => $url]);
 980                         return [];
 981                 }
 982
 983                 $webfinger = [];
 984
 985                 if (!empty($xrd_arr["xrd"]["subject"])) {
 986                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 987                 }
 988
 989                 if (!empty($xrd_arr["xrd"]["alias"])) {
 990                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 991                 }
 992
 993                 $webfinger["links"] = [];
 994
 995                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 996                         if (!empty($data["@attributes"])) {
 997                                 $attributes = $data["@attributes"];
 998                         } elseif ($value == "@attributes") {
 999                                 $attributes = $data;
1000                         } else {
1001                                 continue;
1002                         }
1003
1004                         $webfinger["links"][] = $attributes;
1005                 }
1006                 return $webfinger;
1007         }
1008
1009         /**
1010          * Poll the Friendica specific noscrape page.
1011          *
1012          * "noscrape" is a faster alternative to fetch the data from the hcard.
1013          * This functionality was originally created for the directory.
1014          *
1015          * @param string $noscrape_url Link to the noscrape page
1016          * @param array  $data         The already fetched data
1017          *
1018          * @return array noscrape data
1019          * @throws HTTPException\InternalServerErrorException
1020          */
1021         private static function pollNoscrape($noscrape_url, $data)
1022         {
1023                 $curlResult = DI::httpClient()->get($noscrape_url);
1024                 if ($curlResult->isTimeout()) {
1025                         self::$istimeout = true;
1026                         return $data;
1027                 }
1028                 $content = $curlResult->getBody();
1029                 if (!$content) {
1030                         Logger::info('Empty body', ['url' => $noscrape_url]);
1031                         return $data;
1032                 }
1033
1034                 $json = json_decode($content, true);
1035                 if (!is_array($json)) {
1036                         Logger::info('No json data', ['url' => $noscrape_url]);
1037                         return $data;
1038                 }
1039
1040                 if (!empty($json["fn"])) {
1041                         $data["name"] = $json["fn"];
1042                 }
1043
1044                 if (!empty($json["addr"])) {
1045                         $data["addr"] = $json["addr"];
1046                 }
1047
1048                 if (!empty($json["nick"])) {
1049                         $data["nick"] = $json["nick"];
1050                 }
1051
1052                 if (!empty($json["guid"])) {
1053                         $data["guid"] = $json["guid"];
1054                 }
1055
1056                 if (!empty($json["comm"])) {
1057                         $data["community"] = $json["comm"];
1058                 }
1059
1060                 if (!empty($json["tags"])) {
1061                         $keywords = implode(", ", $json["tags"]);
1062                         if ($keywords != "") {
1063                                 $data["keywords"] = $keywords;
1064                         }
1065                 }
1066
1067                 $location = Profile::formatLocation($json);
1068                 if ($location) {
1069                         $data["location"] = $location;
1070                 }
1071
1072                 if (!empty($json["about"])) {
1073                         $data["about"] = $json["about"];
1074                 }
1075
1076                 if (!empty($json["xmpp"])) {
1077                         $data["xmpp"] = $json["xmpp"];
1078                 }
1079
1080                 if (!empty($json["matrix"])) {
1081                         $data["matrix"] = $json["matrix"];
1082                 }
1083
1084                 if (!empty($json["key"])) {
1085                         $data["pubkey"] = $json["key"];
1086                 }
1087
1088                 if (!empty($json["photo"])) {
1089                         $data["photo"] = $json["photo"];
1090                 }
1091
1092                 if (!empty($json["dfrn-request"])) {
1093                         $data["request"] = $json["dfrn-request"];
1094                 }
1095
1096                 if (!empty($json["dfrn-confirm"])) {
1097                         $data["confirm"] = $json["dfrn-confirm"];
1098                 }
1099
1100                 if (!empty($json["dfrn-notify"])) {
1101                         $data["notify"] = $json["dfrn-notify"];
1102                 }
1103
1104                 if (!empty($json["dfrn-poll"])) {
1105                         $data["poll"] = $json["dfrn-poll"];
1106                 }
1107
1108                 if (isset($json["hide"])) {
1109                         $data["hide"] = (bool)$json["hide"];
1110                 } else {
1111                         $data["hide"] = false;
1112                 }
1113
1114                 return $data;
1115         }
1116
1117         /**
1118          * Check for valid DFRN data
1119          *
1120          * @param array $data DFRN data
1121          *
1122          * @return int Number of errors
1123          */
1124         public static function validDfrn($data)
1125         {
1126                 $errors = 0;
1127                 if (!isset($data['key'])) {
1128                         $errors ++;
1129                 }
1130                 if (!isset($data['dfrn-request'])) {
1131                         $errors ++;
1132                 }
1133                 if (!isset($data['dfrn-confirm'])) {
1134                         $errors ++;
1135                 }
1136                 if (!isset($data['dfrn-notify'])) {
1137                         $errors ++;
1138                 }
1139                 if (!isset($data['dfrn-poll'])) {
1140                         $errors ++;
1141                 }
1142                 return $errors;
1143         }
1144
1145         /**
1146          * Fetch data from a DFRN profile page and via "noscrape"
1147          *
1148          * @param string $profile_link Link to the profile page
1149          *
1150          * @return array profile data
1151          * @throws HTTPException\InternalServerErrorException
1152          * @throws \ImagickException
1153          */
1154         public static function profile($profile_link)
1155         {
1156                 $data = [];
1157
1158                 Logger::info('Check profile', ['link' => $profile_link]);
1159
1160                 // Fetch data via noscrape - this is faster
1161                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1162                 $data = self::pollNoscrape($noscrape_url, $data);
1163
1164                 if (!isset($data["notify"])
1165                         || !isset($data["confirm"])
1166                         || !isset($data["request"])
1167                         || !isset($data["poll"])
1168                         || !isset($data["name"])
1169                         || !isset($data["photo"])
1170                 ) {
1171                         $data = self::pollHcard($profile_link, $data, true);
1172                 }
1173
1174                 $prof_data = [];
1175
1176                 if (empty($data["addr"]) || empty($data["nick"])) {
1177                         $probe_data = self::uri($profile_link);
1178                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1179                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1180                 }
1181
1182                 $prof_data["addr"]         = $data["addr"];
1183                 $prof_data["nick"]         = $data["nick"];
1184                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1185                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1186                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1187                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1188                 $prof_data["photo"]        = $data['photo']   ?? null;
1189                 $prof_data["fn"]           = $data['name']    ?? null;
1190                 $prof_data["key"]          = $data['pubkey']  ?? null;
1191
1192                 Logger::debug('Result', ['link' => $profile_link, 'data' => $prof_data]);
1193
1194                 return $prof_data;
1195         }
1196
1197         /**
1198          * Check for DFRN contact
1199          *
1200          * @param array $webfinger Webfinger data
1201          *
1202          * @return array DFRN data
1203          * @throws HTTPException\InternalServerErrorException
1204          */
1205         private static function dfrn($webfinger)
1206         {
1207                 $hcard_url = "";
1208                 $data = [];
1209                 // The array is reversed to take into account the order of preference for same-rel links
1210                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1211                 foreach (array_reverse($webfinger["links"]) as $link) {
1212                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1213                                 $data["network"] = Protocol::DFRN;
1214                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1215                                 $data["poll"] = $link["href"];
1216                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1217                                 $data["url"] = $link["href"];
1218                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1219                                 $hcard_url = $link["href"];
1220                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1221                                 $data["poco"] = $link["href"];
1222                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1223                                 $data["photo"] = $link["href"];
1224                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1225                                 $data["baseurl"] = trim($link["href"], '/');
1226                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1227                                 $data["guid"] = $link["href"];
1228                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1229                                 $data["pubkey"] = base64_decode($link["href"]);
1230
1231                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1232                                 if (strstr($data["pubkey"], 'RSA ')) {
1233                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1234                                 }
1235                         }
1236                 }
1237
1238                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1239                         foreach ($webfinger["aliases"] as $alias) {
1240                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1241                                         $data["url"] = $alias;
1242                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1243                                         $data["alias"] = $alias;
1244                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1245                                         $data["addr"] = substr($alias, 5);
1246                                 }
1247                         }
1248                 }
1249
1250                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1251                         $data["addr"] = substr($webfinger["subject"], 5);
1252                 }
1253
1254                 if (!isset($data["network"]) || ($hcard_url == "")) {
1255                         return [];
1256                 }
1257
1258                 // Fetch data via noscrape - this is faster
1259                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1260                 $data = self::pollNoscrape($noscrape_url, $data);
1261
1262                 if (isset($data["notify"])
1263                         && isset($data["confirm"])
1264                         && isset($data["request"])
1265                         && isset($data["poll"])
1266                         && isset($data["name"])
1267                         && isset($data["photo"])
1268                 ) {
1269                         return $data;
1270                 }
1271
1272                 $data = self::pollHcard($hcard_url, $data, true);
1273
1274                 return $data;
1275         }
1276
1277         /**
1278          * Poll the hcard page (Diaspora and Friendica specific)
1279          *
1280          * @param string  $hcard_url Link to the hcard page
1281          * @param array   $data      The already fetched data
1282          * @param boolean $dfrn      Poll DFRN specific data
1283          *
1284          * @return array hcard data
1285          * @throws HTTPException\InternalServerErrorException
1286          */
1287         private static function pollHcard($hcard_url, $data, $dfrn = false)
1288         {
1289                 $curlResult = DI::httpClient()->get($hcard_url);
1290                 if ($curlResult->isTimeout()) {
1291                         self::$istimeout = true;
1292                         return [];
1293                 }
1294                 $content = $curlResult->getBody();
1295                 if (empty($content)) {
1296                         return [];
1297                 }
1298
1299                 $doc = new DOMDocument();
1300                 if (!@$doc->loadHTML($content)) {
1301                         return [];
1302                 }
1303
1304                 $xpath = new DomXPath($doc);
1305
1306                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1307                 if (!is_object($vcards)) {
1308                         return [];
1309                 }
1310
1311                 if (!isset($data["baseurl"])) {
1312                         $data["baseurl"] = "";
1313                 }
1314
1315                 if ($vcards->length > 0) {
1316                         $vcard = $vcards->item(0);
1317
1318                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1319                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1320                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1321                         if (($search->length > 0) && empty($data["guid"])) {
1322                                 $data["guid"] = $search->item(0)->nodeValue;
1323                         }
1324
1325                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1326                         if ($search->length > 0) {
1327                                 $data["nick"] = $search->item(0)->nodeValue;
1328                         }
1329
1330                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1331                         if ($search->length > 0) {
1332                                 $data["name"] = $search->item(0)->nodeValue;
1333                         }
1334
1335                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1336                         if ($search->length > 0) {
1337                                 $data["searchable"] = $search->item(0)->nodeValue;
1338                         }
1339
1340                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1341                         if ($search->length > 0) {
1342                                 $data["pubkey"] = $search->item(0)->nodeValue;
1343                                 if (strstr($data["pubkey"], 'RSA ')) {
1344                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1345                                 }
1346                         }
1347
1348                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1349                         if ($search->length > 0) {
1350                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1351                         }
1352                 }
1353
1354                 $avatar = [];
1355                 if (!empty($vcard)) {
1356                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1357                         foreach ($photos as $photo) {
1358                                 $attr = [];
1359                                 foreach ($photo->attributes as $attribute) {
1360                                         $attr[$attribute->name] = trim($attribute->value);
1361                                 }
1362
1363                                 if (isset($attr["src"]) && isset($attr["width"])) {
1364                                         $avatar[$attr["width"]] = $attr["src"];
1365                                 }
1366
1367                                 // We don't have a width. So we just take everything that we got.
1368                                 // This is a Hubzilla workaround which doesn't send a width.
1369                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1370                                         $avatar[] = $attr["src"];
1371                                 }
1372                         }
1373                 }
1374
1375                 if (sizeof($avatar)) {
1376                         ksort($avatar);
1377                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1378                 }
1379
1380                 if ($dfrn) {
1381                         // Poll DFRN specific data
1382                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1383                         if ($search->length > 0) {
1384                                 foreach ($search as $link) {
1385                                         //$data["request"] = $search->item(0)->nodeValue;
1386                                         $attr = [];
1387                                         foreach ($link->attributes as $attribute) {
1388                                                 $attr[$attribute->name] = trim($attribute->value);
1389                                         }
1390
1391                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1392                                 }
1393                         }
1394
1395                         // Older Friendica versions had used the "uid" field differently than newer versions
1396                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1397                                 unset($data["guid"]);
1398                         }
1399                 }
1400
1401
1402                 return $data;
1403         }
1404
1405         /**
1406          * Check for Diaspora contact
1407          *
1408          * @param array $webfinger Webfinger data
1409          *
1410          * @return array Diaspora data
1411          * @throws HTTPException\InternalServerErrorException
1412          */
1413         private static function diaspora($webfinger)
1414         {
1415                 $hcard_url = "";
1416                 $data = [];
1417
1418                 // The array is reversed to take into account the order of preference for same-rel links
1419                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1420                 foreach (array_reverse($webfinger["links"]) as $link) {
1421                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1422                                 $hcard_url = $link["href"];
1423                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1424                                 $data["baseurl"] = trim($link["href"], '/');
1425                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1426                                 $data["guid"] = $link["href"];
1427                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1428                                 $data["url"] = $link["href"];
1429                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && empty($link["type"]) && !empty($link["href"])) {
1430                                 $profile_url = $link["href"];
1431                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1432                                 $data["poll"] = $link["href"];
1433                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1434                                 $data["poco"] = $link["href"];
1435                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1436                                 $data["notify"] = $link["href"];
1437                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1438                                 $data["pubkey"] = base64_decode($link["href"]);
1439
1440                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1441                                 if (strstr($data["pubkey"], 'RSA ')) {
1442                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1443                                 }
1444                         }
1445                 }
1446
1447                 if (empty($data["url"]) && !empty($profile_url)) {
1448                         $data["url"] = $profile_url;
1449                 }
1450
1451                 if (empty($data["url"]) || empty($hcard_url)) {
1452                         return [];
1453                 }
1454
1455                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1456                         foreach ($webfinger["aliases"] as $alias) {
1457                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1458                                         $data["alias"] = $alias;
1459                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1460                                         $data["addr"] = substr($alias, 5);
1461                                 }
1462                         }
1463                 }
1464
1465                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1466                         $data["addr"] = substr($webfinger["subject"], 5);
1467                 }
1468
1469                 // Fetch further information from the hcard
1470                 $data = self::pollHcard($hcard_url, $data);
1471
1472                 if (!$data) {
1473                         return [];
1474                 }
1475
1476                 if (!empty($data["url"])
1477                         && !empty($data["guid"])
1478                         && !empty($data["baseurl"])
1479                         && !empty($data["pubkey"])
1480                         && !empty($hcard_url)
1481                 ) {
1482                         $data["network"] = Protocol::DIASPORA;
1483                         $data["manually-approve"] = false;
1484
1485                         // The Diaspora handle must always be lowercase
1486                         if (!empty($data["addr"])) {
1487                                 $data["addr"] = strtolower($data["addr"]);
1488                         }
1489
1490                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1491                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1492                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1493                 } else {
1494                         return [];
1495                 }
1496
1497                 return $data;
1498         }
1499
1500         /**
1501          * Check for OStatus contact
1502          *
1503          * @param array $webfinger Webfinger data
1504          * @param bool  $short     Short detection mode
1505          *
1506          * @return array|bool OStatus data or "false" on error or "true" on short mode
1507          * @throws HTTPException\InternalServerErrorException
1508          */
1509         private static function ostatus($webfinger, $short = false)
1510         {
1511                 $data = [];
1512
1513                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1514                         foreach ($webfinger["aliases"] as $alias) {
1515                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1516                                         $data["addr"] = str_replace('acct:', '', $alias);
1517                                 }
1518                         }
1519                 }
1520
1521                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1522                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1523                 ) {
1524                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1525                 }
1526
1527                 if (!empty($webfinger["links"])) {
1528                         // The array is reversed to take into account the order of preference for same-rel links
1529                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1530                         foreach (array_reverse($webfinger["links"]) as $link) {
1531                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1532                                         && (($link["type"] ?? "") == "text/html")
1533                                         && ($link["href"] != "")
1534                                 ) {
1535                                         $data["url"] = $data["alias"] = $link["href"];
1536                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1537                                         $data["notify"] = $link["href"];
1538                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1539                                         $data["poll"] = $link["href"];
1540                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1541                                         $pubkey = $link["href"];
1542
1543                                         if (substr($pubkey, 0, 5) === 'data:') {
1544                                                 if (strstr($pubkey, ',')) {
1545                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1546                                                 } else {
1547                                                         $pubkey = substr($pubkey, 5);
1548                                                 }
1549                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1550                                                 $curlResult = DI::httpClient()->get($pubkey);
1551                                                 if ($curlResult->isTimeout()) {
1552                                                         self::$istimeout = true;
1553                                                         return $short ? false : [];
1554                                                 }
1555                                                 $pubkey = $curlResult->getBody();
1556                                         }
1557
1558                                         $key = explode(".", $pubkey);
1559
1560                                         if (sizeof($key) >= 3) {
1561                                                 $m = Strings::base64UrlDecode($key[1]);
1562                                                 $e = Strings::base64UrlDecode($key[2]);
1563                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1564                                         }
1565                                 }
1566                         }
1567                 }
1568
1569                 if (isset($data["notify"]) && isset($data["pubkey"])
1570                         && isset($data["poll"])
1571                         && isset($data["url"])
1572                 ) {
1573                         $data["network"] = Protocol::OSTATUS;
1574                         $data["manually-approve"] = false;
1575                 } else {
1576                         return $short ? false : [];
1577                 }
1578
1579                 if ($short) {
1580                         return true;
1581                 }
1582
1583                 // Fetch all additional data from the feed
1584                 $curlResult = DI::httpClient()->get($data["poll"]);
1585                 if ($curlResult->isTimeout()) {
1586                         self::$istimeout = true;
1587                         return [];
1588                 }
1589                 $feed = $curlResult->getBody();
1590                 $feed_data = Feed::import($feed);
1591                 if (!$feed_data) {
1592                         return [];
1593                 }
1594
1595                 if (!empty($feed_data["header"]["author-name"])) {
1596                         $data["name"] = $feed_data["header"]["author-name"];
1597                 }
1598                 if (!empty($feed_data["header"]["author-nick"])) {
1599                         $data["nick"] = $feed_data["header"]["author-nick"];
1600                 }
1601                 if (!empty($feed_data["header"]["author-avatar"])) {
1602                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1603                 }
1604                 if (!empty($feed_data["header"]["author-id"])) {
1605                         $data["alias"] = $feed_data["header"]["author-id"];
1606                 }
1607                 if (!empty($feed_data["header"]["author-location"])) {
1608                         $data["location"] = $feed_data["header"]["author-location"];
1609                 }
1610                 if (!empty($feed_data["header"]["author-about"])) {
1611                         $data["about"] = $feed_data["header"]["author-about"];
1612                 }
1613                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1614                 // So we take the value that we just fetched, although the other one worked as well
1615                 if (!empty($feed_data["header"]["author-link"])) {
1616                         $data["url"] = $feed_data["header"]["author-link"];
1617                 }
1618
1619                 if ($data["url"] == $data["alias"]) {
1620                         $data["alias"] = '';
1621                 }
1622
1623                 /// @todo Fetch location and "about" from the feed as well
1624                 return $data;
1625         }
1626
1627         /**
1628          * Fetch data from a pump.io profile page
1629          *
1630          * @param string $profile_link Link to the profile page
1631          *
1632          * @return array profile data
1633          */
1634         private static function pumpioProfileData($profile_link)
1635         {
1636                 $curlResult = DI::httpClient()->get($profile_link);
1637                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
1638                         return [];
1639                 }
1640
1641                 $doc = new DOMDocument();
1642                 if (!@$doc->loadHTML($curlResult->getBody())) {
1643                         return [];
1644                 }
1645
1646                 $xpath = new DomXPath($doc);
1647
1648                 $data = [];
1649
1650                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1651
1652                 if ($data["name"] == '') {
1653                         // This is ugly - but pump.io doesn't seem to know a better way for it
1654                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1655                         $pos = strpos($data["name"], chr(10));
1656                         if ($pos) {
1657                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1658                         }
1659                 }
1660
1661                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1662
1663                 if ($data["location"] == '') {
1664                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1665                 }
1666
1667                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1668
1669                 if ($data["about"] == '') {
1670                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1671                 }
1672
1673                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1674                 if (!$avatar) {
1675                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1676                 }
1677                 if ($avatar) {
1678                         foreach ($avatar->attributes as $attribute) {
1679                                 if ($attribute->name == "src") {
1680                                         $data["photo"] = trim($attribute->value);
1681                                 }
1682                         }
1683                 }
1684
1685                 return $data;
1686         }
1687
1688         /**
1689          * Check for pump.io contact
1690          *
1691          * @param array  $webfinger Webfinger data
1692          * @param string $addr
1693          * @return array pump.io data
1694          */
1695         private static function pumpio($webfinger, $addr)
1696         {
1697                 $data = [];
1698                 // The array is reversed to take into account the order of preference for same-rel links
1699                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1700                 foreach (array_reverse($webfinger["links"]) as $link) {
1701                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1702                                 && (($link["type"] ?? "") == "text/html")
1703                                 && ($link["href"] != "")
1704                         ) {
1705                                 $data["url"] = $link["href"];
1706                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1707                                 $data["notify"] = $link["href"];
1708                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1709                                 $data["poll"] = $link["href"];
1710                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1711                                 $data["dialback"] = $link["href"];
1712                         }
1713                 }
1714                 if (isset($data["poll"]) && isset($data["notify"])
1715                         && isset($data["dialback"])
1716                         && isset($data["url"])
1717                 ) {
1718                         // by now we use these fields only for the network type detection
1719                         // So we unset all data that isn't used at the moment
1720                         unset($data["dialback"]);
1721
1722                         $data["network"] = Protocol::PUMPIO;
1723                 } else {
1724                         return [];
1725                 }
1726
1727                 $profile_data = self::pumpioProfileData($data["url"]);
1728
1729                 if (!$profile_data) {
1730                         return [];
1731                 }
1732
1733                 $data = array_merge($data, $profile_data);
1734
1735                 if (($addr != '') && ($data['name'] != '')) {
1736                         $name = trim(str_replace($addr, '', $data['name']));
1737                         if ($name != '') {
1738                                 $data['name'] = $name;
1739                         }
1740                 }
1741
1742                 return $data;
1743         }
1744
1745         /**
1746          * Check for twitter contact
1747          *
1748          * @param string $uri
1749          *
1750          * @return array twitter data
1751          */
1752         private static function twitter($uri)
1753         {
1754                 if (preg_match('=([^@]+)@(?:mobile\.)?twitter\.com$=i', $uri, $matches)) {
1755                         $nick = $matches[1];
1756                 } elseif (preg_match('=^https?://(?:mobile\.)?twitter\.com/(.+)=i', $uri, $matches)) {
1757                         $nick = $matches[1];
1758                 } else {
1759                         return [];
1760                 }
1761
1762                 $data = [];
1763                 $data['url'] = 'https://twitter.com/' . $nick;
1764                 $data['addr'] = $nick . '@twitter.com';
1765                 $data['nick'] = $data['name'] = $nick;
1766                 $data['network'] = Protocol::TWITTER;
1767                 $data['baseurl'] = 'https://twitter.com';
1768
1769                 return $data;
1770         }
1771
1772         /**
1773          * Checks HTML page for RSS feed link
1774          *
1775          * @param string $url  Page link
1776          * @param string $body Page body string
1777          * @return string|false Feed link or false if body was invalid HTML document
1778          */
1779         public static function getFeedLink(string $url, string $body)
1780         {
1781                 if (empty($body)) {
1782                         return '';
1783                 }
1784
1785                 $doc = new DOMDocument();
1786                 if (!@$doc->loadHTML($body)) {
1787                         return false;
1788                 }
1789
1790                 $xpath = new DOMXPath($doc);
1791
1792                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1793
1794                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1795
1796                 return $feedUrl;
1797         }
1798
1799         /**
1800          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1801          *
1802          * Loosely based on RFC 1808
1803          *
1804          * @see https://tools.ietf.org/html/rfc1808
1805          *
1806          * @param string   $href  The potential relative href found in the HTML document
1807          * @param string   $base  The HTML document URL
1808          * @param DOMXPath $xpath The HTML document XPath
1809          * @return string
1810          */
1811         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
1812         {
1813                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1814                         return $href;
1815                 }
1816
1817                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1818
1819                 $baseParts = parse_url($base);
1820                 if (empty($baseParts['host'])) {
1821                         return $href;
1822                 }
1823
1824                 // Naked domain case (scheme://basehost)
1825                 $path = $baseParts['path'] ?? '/';
1826
1827                 // Remove the filename part of the path if it exists (/base/path/file)
1828                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1829
1830                 $hrefParts = parse_url($href);
1831
1832                 if (!empty($hrefParts['path'])) {
1833                         // Root path case (/path) including relative scheme case (//host/path)
1834                         if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1835                                 $path = $hrefParts['path'];
1836                         } else {
1837                                 $path = $path . '/' . $hrefParts['path'];
1838
1839                                 // Resolve arbitrary relative path
1840                                 // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1841                                 $parts = array_filter(explode('/', $path), 'strlen');
1842                                 $absolutes = array();
1843                                 foreach ($parts as $part) {
1844                                         if ('.' == $part) continue;
1845                                         if ('..' == $part) {
1846                                                 array_pop($absolutes);
1847                                         } else {
1848                                                 $absolutes[] = $part;
1849                                         }
1850                                 }
1851
1852                                 $path = '/' . implode('/', $absolutes);
1853                         }
1854                 }
1855
1856                 // Relative scheme case (//host/path)
1857                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1858                 $baseParts['path'] = $path;
1859                 unset($baseParts['query']);
1860                 unset($baseParts['fragment']);
1861
1862                 return Network::unparseURL($baseParts);
1863         }
1864
1865         /**
1866          * Check for feed contact
1867          *
1868          * @param string  $url   Profile link
1869          * @param boolean $probe Do a probe if the page contains a feed link
1870          *
1871          * @return array feed data
1872          * @throws HTTPException\InternalServerErrorException
1873          */
1874         private static function feed($url, $probe = true)
1875         {
1876                 $curlResult = DI::httpClient()->get($url);
1877                 if ($curlResult->isTimeout()) {
1878                         self::$istimeout = true;
1879                         return [];
1880                 }
1881                 $feed = $curlResult->getBody();
1882                 $feed_data = Feed::import($feed);
1883
1884                 if (!$feed_data) {
1885                         if (!$probe) {
1886                                 return [];
1887                         }
1888
1889                         $feed_url = self::getFeedLink($url, $feed);
1890
1891                         if (!$feed_url) {
1892                                 return [];
1893                         }
1894
1895                         return self::feed($feed_url, false);
1896                 }
1897
1898                 if (!empty($feed_data["header"]["author-name"])) {
1899                         $data["name"] = $feed_data["header"]["author-name"];
1900                 }
1901
1902                 if (!empty($feed_data["header"]["author-nick"])) {
1903                         $data["nick"] = $feed_data["header"]["author-nick"];
1904                 }
1905
1906                 if (!empty($feed_data["header"]["author-avatar"])) {
1907                         $data["photo"] = $feed_data["header"]["author-avatar"];
1908                 }
1909
1910                 if (!empty($feed_data["header"]["author-id"])) {
1911                         $data["alias"] = $feed_data["header"]["author-id"];
1912                 }
1913
1914                 $data["url"] = $url;
1915                 $data["poll"] = $url;
1916
1917                 $data["network"] = Protocol::FEED;
1918
1919                 return $data;
1920         }
1921
1922         /**
1923          * Check for mail contact
1924          *
1925          * @param string  $uri Profile link
1926          * @param integer $uid User ID
1927          *
1928          * @return array mail data
1929          * @throws \Exception
1930          */
1931         private static function mail($uri, $uid)
1932         {
1933                 if (!Network::isEmailDomainValid($uri)) {
1934                         return [];
1935                 }
1936
1937                 if ($uid == 0) {
1938                         return [];
1939                 }
1940
1941                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1942
1943                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1944                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1945                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1946
1947                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1948                         return [];
1949                 }
1950
1951                 $mailbox = Email::constructMailboxName($mailacct);
1952                 $password = '';
1953                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1954                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1955                 if (!$mbox) {
1956                         return [];
1957                 }
1958
1959                 $msgs = Email::poll($mbox, $uri);
1960                 Logger::info('Messages found', ['uri' => $uri, 'count' => count($msgs)]);
1961
1962                 if (!count($msgs)) {
1963                         return [];
1964                 }
1965
1966                 $phost = substr($uri, strpos($uri, '@') + 1);
1967
1968                 $data = [];
1969                 $data["addr"]    = $uri;
1970                 $data["network"] = Protocol::MAIL;
1971                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1972                 $data["nick"]    = $data["name"];
1973                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1974                 $data["url"]     = 'mailto:'.$uri;
1975                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1976                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1977
1978                 $x = Email::messageMeta($mbox, $msgs[0]);
1979                 if (stristr($x[0]->from, $uri)) {
1980                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1981                 } elseif (stristr($x[0]->to, $uri)) {
1982                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1983                 }
1984                 if (isset($adr)) {
1985                         foreach ($adr as $feadr) {
1986                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1987                                         &&(strcasecmp($feadr->host, $phost) == 0)
1988                                         && (strlen($feadr->personal))
1989                                 ) {
1990                                         $personal = imap_mime_header_decode($feadr->personal);
1991                                         $data["name"] = "";
1992                                         foreach ($personal as $perspart) {
1993                                                 if ($perspart->charset != "default") {
1994                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1995                                                 } else {
1996                                                         $data["name"] .= $perspart->text;
1997                                                 }
1998                                         }
1999
2000                                         $data["name"] = Strings::escapeTags($data["name"]);
2001                                 }
2002                         }
2003                 }
2004                 if (!empty($mbox)) {
2005                         imap_close($mbox);
2006                 }
2007                 return $data;
2008         }
2009
2010         /**
2011          * Mix two paths together to possibly fix missing parts
2012          *
2013          * @param string $avatar Path to the avatar
2014          * @param string $base   Another path that is hopefully complete
2015          *
2016          * @return string fixed avatar path
2017          * @throws \Exception
2018          */
2019         public static function fixAvatar($avatar, $base)
2020         {
2021                 $base_parts = parse_url($base);
2022
2023                 // Remove all parts that could create a problem
2024                 unset($base_parts['path']);
2025                 unset($base_parts['query']);
2026                 unset($base_parts['fragment']);
2027
2028                 $avatar_parts = parse_url($avatar);
2029
2030                 // Now we mix them
2031                 $parts = array_merge($base_parts, $avatar_parts);
2032
2033                 // And put them together again
2034                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
2035                 $host     = isset($parts['host'])     ? $parts['host']           : '';
2036                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
2037                 $path     = isset($parts['path'])     ? $parts['path']           : '';
2038                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
2039                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
2040
2041                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
2042
2043                 Logger::debug('Avatar fixed', ['base' => $base, 'avatar' => $avatar, 'fixed' => $fixed]);
2044
2045                 return $fixed;
2046         }
2047
2048         /**
2049          * Fetch the last date that the contact had posted something (publically)
2050          *
2051          * @param string $data  probing result
2052          * @return string last activity
2053          */
2054         public static function getLastUpdate(array $data)
2055         {
2056                 $uid = User::getIdForURL($data['url']);
2057                 if (!empty($uid)) {
2058                         $contact = Contact::selectFirst(['url', 'last-item'], ['self' => true, 'uid' => $uid]);
2059                         if (!empty($contact['last-item'])) {
2060                                 return $contact['last-item'];
2061                         }
2062                 }
2063
2064                 if ($lastUpdate = self::updateFromNoScrape($data)) {
2065                         return $lastUpdate;
2066                 }
2067
2068                 if (!empty($data['outbox'])) {
2069                         return self::updateFromOutbox($data['outbox'], $data);
2070                 } elseif (!empty($data['poll']) && ($data['network'] == Protocol::ACTIVITYPUB)) {
2071                         return self::updateFromOutbox($data['poll'], $data);
2072                 } elseif (!empty($data['poll'])) {
2073                         return self::updateFromFeed($data);
2074                 }
2075
2076                 return '';
2077         }
2078
2079         /**
2080          * Fetch the last activity date from the "noscrape" endpoint
2081          *
2082          * @param array $data Probing result
2083          * @return string last activity
2084          *
2085          * @return bool 'true' if update was successful or the server was unreachable
2086          */
2087         private static function updateFromNoScrape(array $data)
2088         {
2089                 if (empty($data['baseurl'])) {
2090                         return '';
2091                 }
2092
2093                 // Check the 'noscrape' endpoint when it is a Friendica server
2094                 $gserver = DBA::selectFirst('gserver', ['noscrape'], ["`nurl` = ? AND `noscrape` != ''",
2095                         Strings::normaliseLink($data['baseurl'])]);
2096                 if (!DBA::isResult($gserver)) {
2097                         return '';
2098                 }
2099
2100                 $curlResult = DI::httpClient()->get($gserver['noscrape'] . '/' . $data['nick']);
2101
2102                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
2103                         $noscrape = json_decode($curlResult->getBody(), true);
2104                         if (!empty($noscrape) && !empty($noscrape['updated'])) {
2105                                 return DateTimeFormat::utc($noscrape['updated'], DateTimeFormat::MYSQL);
2106                         }
2107                 }
2108
2109                 return '';
2110         }
2111
2112         /**
2113          * Fetch the last activity date from an ActivityPub Outbox
2114          *
2115          * @param string $feed
2116          * @param array  $data Probing result
2117          * @return string last activity
2118          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
2119          */
2120         private static function updateFromOutbox(string $feed, array $data)
2121         {
2122                 $outbox = ActivityPub::fetchContent($feed);
2123                 if (empty($outbox)) {
2124                         return '';
2125                 }
2126
2127                 if (!empty($outbox['orderedItems'])) {
2128                         $items = $outbox['orderedItems'];
2129                 } elseif (!empty($outbox['first']['orderedItems'])) {
2130                         $items = $outbox['first']['orderedItems'];
2131                 } elseif (!empty($outbox['first']['href']) && ($outbox['first']['href'] != $feed)) {
2132                         return self::updateFromOutbox($outbox['first']['href'], $data);
2133                 } elseif (!empty($outbox['first'])) {
2134                         if (is_string($outbox['first']) && ($outbox['first'] != $feed)) {
2135                                 return self::updateFromOutbox($outbox['first'], $data);
2136                         } else {
2137                                 Logger::warning('Unexpected data', ['outbox' => $outbox]);
2138                         }
2139                         return '';
2140                 } else {
2141                         $items = [];
2142                 }
2143
2144                 $last_updated = '';
2145                 foreach ($items as $activity) {
2146                         if (!empty($activity['published'])) {
2147                                 $published =  DateTimeFormat::utc($activity['published']);
2148                         } elseif (!empty($activity['object']['published'])) {
2149                                 $published =  DateTimeFormat::utc($activity['object']['published']);
2150                         } else {
2151                                 continue;
2152                         }
2153
2154                         if ($last_updated < $published) {
2155                                 $last_updated = $published;
2156                         }
2157                 }
2158
2159                 if (!empty($last_updated)) {
2160                         return $last_updated;
2161                 }
2162
2163                 return '';
2164         }
2165
2166         /**
2167          * Fetch the last activity date from an XML feed
2168          *
2169          * @param array $data Probing result
2170          * @return string last activity
2171          */
2172         private static function updateFromFeed(array $data)
2173         {
2174                 // Search for the newest entry in the feed
2175                 $curlResult = DI::httpClient()->get($data['poll']);
2176                 if (!$curlResult->isSuccess() || !$curlResult->getBody()) {
2177                         return '';
2178                 }
2179
2180                 $doc = new DOMDocument();
2181                 @$doc->loadXML($curlResult->getBody());
2182
2183                 $xpath = new DOMXPath($doc);
2184                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
2185
2186                 $entries = $xpath->query('/atom:feed/atom:entry');
2187
2188                 $last_updated = '';
2189
2190                 foreach ($entries as $entry) {
2191                         $published_item = $xpath->query('atom:published/text()', $entry)->item(0);
2192                         $updated_item   = $xpath->query('atom:updated/text()'  , $entry)->item(0);
2193                         $published      = !empty($published_item->nodeValue) ? DateTimeFormat::utc($published_item->nodeValue) : null;
2194                         $updated        = !empty($updated_item->nodeValue) ? DateTimeFormat::utc($updated_item->nodeValue) : null;
2195
2196                         if (empty($published) || empty($updated)) {
2197                                 Logger::notice('Invalid entry for XPath.', ['entry' => $entry, 'url' => $data['url']]);
2198                                 continue;
2199                         }
2200
2201                         if ($last_updated < $published) {
2202                                 $last_updated = $published;
2203                         }
2204
2205                         if ($last_updated < $updated) {
2206                                 $last_updated = $updated;
2207                         }
2208                 }
2209
2210                 if (!empty($last_updated)) {
2211                         return $last_updated;
2212                 }
2213
2214                 return '';
2215         }
2216
2217         /**
2218          * Probe data from local profiles without network traffic
2219          *
2220          * @param string $url
2221          * @return array probed data
2222          * @throws HTTPException\InternalServerErrorException
2223          * @throws HTTPException\NotFoundException
2224          */
2225         private static function localProbe(string $url): array
2226         {
2227                 try {
2228                         $uid = User::getIdForURL($url);
2229                         if (!$uid) {
2230                                 throw new HTTPException\NotFoundException('User not found.');
2231                         }
2232
2233                         $owner     = User::getOwnerDataById($uid);
2234                         $approfile = ActivityPub\Transmitter::getProfile($uid);
2235
2236                         if (empty($owner['gsid'])) {
2237                                 $owner['gsid'] = GServer::getID($approfile['generator']['url']);
2238                         }
2239
2240                         $data = [
2241                                 'name' => $owner['name'], 'nick' => $owner['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
2242                                 'url' => $owner['url'], 'addr' => $owner['addr'], 'alias' => $owner['alias'],
2243                                 'photo' => User::getAvatarUrl($owner),
2244                                 'header' => $owner['header'] ? Contact::getHeaderUrlForId($owner['id'], $owner['updated']) : '',
2245                                 'account-type' => $owner['contact-type'], 'community' => ($owner['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
2246                                 'keywords' => $owner['keywords'], 'location' => $owner['location'], 'about' => $owner['about'],
2247                                 'xmpp' => $owner['xmpp'], 'matrix' => $owner['matrix'],
2248                                 'hide' => !$owner['net-publish'], 'batch' => '', 'notify' => $owner['notify'],
2249                                 'poll' => $owner['poll'], 'request' => $owner['request'], 'confirm' => $owner['confirm'],
2250                                 'subscribe' => $approfile['generator']['url'] . '/follow?url={uri}', 'poco' => $owner['poco'],
2251                                 'following' => $approfile['following'], 'followers' => $approfile['followers'],
2252                                 'inbox' => $approfile['inbox'], 'outbox' => $approfile['outbox'],
2253                                 'sharedinbox' => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
2254                                 'pubkey' => $owner['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $owner['gsid'],
2255                                 'manually-approve' => in_array($owner['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])
2256                         ];
2257                 } catch (Exception $e) {
2258                         // Default values for non existing targets
2259                         $data = [
2260                                 'name' => $url, 'nick' => $url, 'url' => $url, 'network' => Protocol::PHANTOM,
2261                                 'photo' => DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO
2262                         ];
2263                 }
2264
2265                 return self::rearrangeData($data);
2266         }
2267 }