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