src/Network/Probe.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, the Friendica project
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Network;
  23
  24 use DOMDocument;
  25 use DomXPath;
  26 use Exception;
  27 use Friendica\Core\Hook;
  28 use Friendica\Core\Logger;
  29 use Friendica\Core\Protocol;
  30 use Friendica\Core\System;
  31 use Friendica\Database\DBA;
  32 use Friendica\DI;
  33 use Friendica\Model\Contact;
  34 use Friendica\Model\GServer;
  35 use Friendica\Model\Profile;
  36 use Friendica\Model\User;
  37 use Friendica\Network\HTTPClient\Client\HttpClientOptions;
  38 use Friendica\Protocol\ActivityNamespace;
  39 use Friendica\Protocol\ActivityPub;
  40 use Friendica\Protocol\Email;
  41 use Friendica\Protocol\Feed;
  42 use Friendica\Util\Crypto;
  43 use Friendica\Util\DateTimeFormat;
  44 use Friendica\Util\Network;
  45 use Friendica\Util\Strings;
  46 use Friendica\Util\XML;
  47
  48 /**
  49  * This class contain functions for probing URL
  50  */
  51 class Probe
  52 {
  53         const WEBFINGER = '/.well-known/webfinger?resource={uri}';
  54
  55         private static $baseurl;
  56         private static $istimeout;
  57
  58         /**
  59          * Remove stuff from an URI that doesn't belong there
  60          *
  61          * @param string $URI
  62          * @return string Cleaned URI
  63          */
  64         public static function cleanURI(string $URI)
  65         {
  66                 // At first remove leading and trailing junk
  67                 $URI = trim($URI, "@#?:/ \t\n\r\0\x0B");
  68
  69                 $parts = parse_url($URI);
  70
  71                 if (empty($parts['scheme'])) {
  72                         return $URI;
  73                 }
  74
  75                 // Remove the URL fragment, since these shouldn't be part of any profile URL
  76                 unset($parts['fragment']);
  77
  78                 $URI = Network::unparseURL($parts);
  79
  80                 return $URI;
  81         }
  82
  83         /**
  84          * Rearrange the array so that it always has the same order
  85          *
  86          * @param array $data Unordered data
  87          *
  88          * @return array Ordered data
  89          */
  90         private static function rearrangeData($data)
  91         {
  92                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "header",
  93                                 "account-type", "community", "keywords", "location", "about", "xmpp", "matrix",
  94                                 "hide", "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
  95                                 "following", "followers", "inbox", "outbox", "sharedinbox",
  96                                 "priority", "network", "pubkey", "manually-approve", "baseurl", "gsid"];
  97
  98                 $numeric_fields = ["gsid", "hide", "account-type", "manually-approve"];
  99
 100                 $newdata = [];
 101                 foreach ($fields as $field) {
 102                         if (isset($data[$field])) {
 103                                 if (in_array($field, $numeric_fields)) {
 104                                         $newdata[$field] = (int)$data[$field];
 105                                 } else {
 106                                         $newdata[$field] = $data[$field];
 107                                 }
 108                         } elseif (!in_array($field, $numeric_fields)) {
 109                                 $newdata[$field] = "";
 110                         } else {
 111                                 $newdata[$field] = null;
 112                         }
 113                 }
 114
 115                 // We don't use the "priority" field anymore and replace it with a dummy.
 116                 $newdata["priority"] = 0;
 117
 118                 return $newdata;
 119         }
 120
 121         /**
 122          * Check if the hostname belongs to the own server
 123          *
 124          * @param string $host The hostname that is to be checked
 125          *
 126          * @return bool Does the testes hostname belongs to the own server?
 127          */
 128         private static function ownHost($host)
 129         {
 130                 $own_host = DI::baseUrl()->getHostname();
 131
 132                 $parts = parse_url($host);
 133
 134                 if (!isset($parts['scheme'])) {
 135                         $parts = parse_url('http://'.$host);
 136                 }
 137
 138                 if (!isset($parts['host'])) {
 139                         return false;
 140                 }
 141                 return $parts['host'] == $own_host;
 142         }
 143
 144         /**
 145          * Probes for webfinger path via "host-meta"
 146          *
 147          * We have to check if the servers in the future still will offer this.
 148          * It seems as if it was dropped from the standard.
 149          *
 150          * @param string $host The host part of an url
 151          *
 152          * @return array with template and type of the webfinger template for JSON or XML
 153          * @throws HTTPException\InternalServerErrorException
 154          */
 155         private static function hostMeta($host)
 156         {
 157                 // Reset the static variable
 158                 self::$baseurl = '';
 159
 160                 // Handles the case when the hostname contains the scheme
 161                 if (!parse_url($host, PHP_URL_SCHEME)) {
 162                         $ssl_url = "https://" . $host . "/.well-known/host-meta";
 163                         $url = "http://" . $host . "/.well-known/host-meta";
 164                 } else {
 165                         $ssl_url = $host . "/.well-known/host-meta";
 166                         $url = '';
 167                 }
 168
 169                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 170
 171                 Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
 172                 $xrd = null;
 173
 174                 $curlResult = DI::httpClient()->get($ssl_url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => ['application/xrd+xml']]);
 175                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 176                 if ($curlResult->isSuccess()) {
 177                         $xml = $curlResult->getBody();
 178                         $xrd = XML::parseString($xml, true);
 179                         if (!empty($url)) {
 180                                 $host_url = 'https://' . $host;
 181                         } else {
 182                                 $host_url = $host;
 183                         }
 184                 } elseif ($curlResult->isTimeout()) {
 185                         Logger::info('Probing timeout', ['url' => $ssl_url]);
 186                         self::$istimeout = true;
 187                         return [];
 188                 }
 189
 190                 if (!is_object($xrd) && !empty($url)) {
 191                         $curlResult = DI::httpClient()->get($url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => ['application/xrd+xml']]);
 192                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 193                         if ($curlResult->isTimeout()) {
 194                                 Logger::info('Probing timeout', ['url' => $url]);
 195                                 self::$istimeout = true;
 196                                 return [];
 197                         } elseif ($connection_error && $ssl_connection_error) {
 198                                 self::$istimeout = true;
 199                                 return [];
 200                         }
 201
 202                         $xml = $curlResult->getBody();
 203                         $xrd = XML::parseString($xml, true);
 204                         $host_url = 'http://'.$host;
 205                 }
 206                 if (!is_object($xrd)) {
 207                         Logger::info('No xrd object found', ['host' => $host]);
 208                         return [];
 209                 }
 210
 211                 $links = XML::elementToArray($xrd);
 212                 if (!isset($links["xrd"]["link"])) {
 213                         Logger::info('No xrd data found', ['host' => $host]);
 214                         return [];
 215                 }
 216
 217                 $lrdd = [];
 218
 219                 foreach ($links["xrd"]["link"] as $value => $link) {
 220                         if (!empty($link["@attributes"])) {
 221                                 $attributes = $link["@attributes"];
 222                         } elseif ($value == "@attributes") {
 223                                 $attributes = $link;
 224                         } else {
 225                                 continue;
 226                         }
 227
 228                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 229                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 230
 231                                 $lrdd[$type] = $attributes["template"];
 232                         }
 233                 }
 234
 235                 if (Network::isUrlBlocked($host_url)) {
 236                         Logger::info('Domain is blocked', ['url' => $host]);
 237                         return [];
 238                 }
 239
 240                 self::$baseurl = $host_url;
 241
 242                 Logger::info('Probing successful', ['host' => $host]);
 243
 244                 return $lrdd;
 245         }
 246
 247         /**
 248          * Perform Webfinger lookup and return DFRN data
 249          *
 250          * Given an email style address, perform webfinger lookup and
 251          * return the resulting DFRN profile URL, or if no DFRN profile URL
 252          * is located, returns an OStatus subscription template (prefixed
 253          * with the string 'stat:' to identify it as on OStatus template).
 254          * If this isn't an email style address just return $webbie.
 255          * Return an empty string if email-style addresses but webfinger fails,
 256          * or if the resultant personal XRD doesn't contain a supported
 257          * subscription/friend-request attribute.
 258          *
 259          * amended 7/9/2011 to return an hcard which could save potentially loading
 260          * a lengthy content page to scrape dfrn attributes
 261          *
 262          * @param string $webbie    Address that should be probed
 263          * @param string $hcard_url Link to the hcard - is returned by reference
 264          *
 265          * @return string profile link
 266          * @throws HTTPException\InternalServerErrorException
 267          */
 268         public static function webfingerDfrn(string $webbie, string &$hcard_url)
 269         {
 270                 $profile_link = '';
 271
 272                 $links = self::lrdd($webbie);
 273                 Logger::debug('Result', ['url' => $webbie, 'links' => $links]);
 274                 if (!empty($links) && is_array($links)) {
 275                         foreach ($links as $link) {
 276                                 if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) {
 277                                         $profile_link = $link['@attributes']['href'];
 278                                 }
 279                                 if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) {
 280                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 281                                 }
 282                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 283                                         $hcard_url = $link['@attributes']['href'];
 284                                 }
 285                         }
 286                 }
 287                 return $profile_link;
 288         }
 289
 290         /**
 291          * Check an URI for LRDD data
 292          *
 293          * @param string $uri     Address that should be probed
 294          *
 295          * @return array uri data
 296          * @throws HTTPException\InternalServerErrorException
 297          */
 298         public static function lrdd(string $uri)
 299         {
 300                 $data = self::getWebfingerArray($uri);
 301                 if (empty($data)) {
 302                         return [];
 303                 }
 304                 $webfinger = $data['webfinger'];
 305
 306                 if (empty($webfinger["links"])) {
 307                         Logger::info('No webfinger links found', ['uri' => $uri]);
 308                         return [];
 309                 }
 310
 311                 $data = [];
 312
 313                 foreach ($webfinger["links"] as $link) {
 314                         $data[] = ["@attributes" => $link];
 315                 }
 316
 317                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 318                         foreach ($webfinger["aliases"] as $alias) {
 319                                 $data[] = ["@attributes" =>
 320                                                         ["rel" => "alias",
 321                                                                 "href" => $alias]];
 322                         }
 323                 }
 324
 325                 return $data;
 326         }
 327
 328         /**
 329          * Fetch information (protocol endpoints and user information) about a given uri
 330          *
 331          * @param string  $uri     Address that should be probed
 332          * @param string  $network Test for this specific network
 333          * @param integer $uid     User ID for the probe (only used for mails)
 334          * @param boolean $cache   Use cached values?
 335          *
 336          * @return array uri data
 337          * @throws HTTPException\InternalServerErrorException
 338          * @throws \ImagickException
 339          */
 340         public static function uri($uri, $network = '', $uid = -1)
 341         {
 342                 // Local profiles aren't probed via network
 343                 if (empty($network) && Contact::isLocal($uri)) {
 344                         $data = self::localProbe($uri);
 345                         if (!empty($data)) {
 346                                 return $data;
 347                         }
 348                 }
 349
 350                 if ($uid == -1) {
 351                         $uid = local_user();
 352                 }
 353
 354                 if (empty($network) || ($network == Protocol::ACTIVITYPUB)) {
 355                         $ap_profile = ActivityPub::probeProfile($uri);
 356                 } else {
 357                         $ap_profile = [];
 358                 }
 359
 360                 self::$istimeout = false;
 361
 362                 if ($network != Protocol::ACTIVITYPUB) {
 363                         $data = self::detect($uri, $network, $uid, $ap_profile);
 364                         if (!is_array($data)) {
 365                                 $data = [];
 366                         }
 367                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 368                                 $data = $ap_profile;
 369                         } elseif (!empty($ap_profile)) {
 370                                 $ap_profile['batch'] = '';
 371                                 $data = array_merge($ap_profile, $data);
 372                         }
 373                 } else {
 374                         $data = $ap_profile;
 375                 }
 376
 377                 if (!isset($data['url'])) {
 378                         $data['url'] = $uri;
 379                 }
 380
 381                 if (empty($data['photo'])) {
 382                         $data['photo'] = DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO;
 383                 }
 384
 385                 if (empty($data['name'])) {
 386                         if (!empty($data['nick'])) {
 387                                 $data['name'] = $data['nick'];
 388                         }
 389
 390                         if (empty($data['name'])) {
 391                                 $data['name'] = $data['url'];
 392                         }
 393                 }
 394
 395                 if (empty($data['nick'])) {
 396                         $data['nick'] = strtolower($data['name']);
 397
 398                         if (strpos($data['nick'], ' ')) {
 399                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 400                         }
 401                 }
 402
 403                 if (!empty($data['baseurl']) && empty($data['gsid'])) {
 404                         $data['gsid'] = GServer::getID($data['baseurl']);
 405                 }
 406
 407                 if (empty($data['network'])) {
 408                         $data['network'] = Protocol::PHANTOM;
 409                 }
 410
 411                 // Ensure that local connections always are DFRN
 412                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
 413                         $data['network'] = Protocol::DFRN;
 414                 }
 415
 416                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 417                         $data['hide'] = self::getHideStatus($data['url']);
 418                 }
 419
 420                 return self::rearrangeData($data);
 421         }
 422
 423
 424         /**
 425          * Fetches the "hide" status from the profile
 426          *
 427          * @param string $url URL of the profile
 428          *
 429          * @return boolean "hide" status
 430          */
 431         private static function getHideStatus($url)
 432         {
 433                 $curlResult = DI::httpClient()->get($url, [HttpClientOptions::CONTENT_LENGTH => 1000000]);
 434                 if (!$curlResult->isSuccess()) {
 435                         return false;
 436                 }
 437
 438                 // If it isn't a HTML file then exit
 439                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 440                         return false;
 441                 }
 442
 443                 $body = $curlResult->getBody();
 444                 if (empty($body)) {
 445                         return false;
 446                 }
 447
 448                 $doc = new DOMDocument();
 449                 @$doc->loadHTML($body);
 450
 451                 $xpath = new DOMXPath($doc);
 452
 453                 $list = $xpath->query('//meta[@name]');
 454                 foreach ($list as $node) {
 455                         $meta_tag = [];
 456                         if ($node->attributes->length) {
 457                                 foreach ($node->attributes as $attribute) {
 458                                         $meta_tag[$attribute->name] = $attribute->value;
 459                                 }
 460                         }
 461
 462                         if (empty($meta_tag['content'])) {
 463                                 continue;
 464                         }
 465
 466                         $content = strtolower(trim($meta_tag['content']));
 467
 468                         switch (strtolower(trim($meta_tag['name']))) {
 469                                 case 'dfrn-global-visibility':
 470                                         if ($content == 'false') {
 471                                                 return true;
 472                                         }
 473                                         break;
 474                                 case 'robots':
 475                                         if (strpos($content, 'noindex') !== false) {
 476                                                 return true;
 477                                         }
 478                                         break;
 479                         }
 480                 }
 481
 482                 return false;
 483         }
 484
 485         /**
 486          * Fetch the "subscribe" and add it to the result
 487          *
 488          * @param array $result
 489          * @param array $webfinger
 490          * @return array result
 491          */
 492         private static function getSubscribeLink(array $result, array $webfinger)
 493         {
 494                 if (empty($webfinger['links'])) {
 495                         return $result;
 496                 }
 497
 498                 foreach ($webfinger['links'] as $link) {
 499                         if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
 500                                 $result['subscribe'] = $link['template'];
 501                         }
 502                 }
 503
 504                 return $result;
 505         }
 506
 507         /**
 508          * Get webfinger data from a given URI
 509          *
 510          * @param string $uri
 511          * @return array
 512          * @throws HTTPException\InternalServerErrorException
 513          */
 514         private static function getWebfingerArray(string $uri): array
 515         {
 516                 $parts = parse_url($uri);
 517
 518                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 519                         $host = $parts['host'];
 520                         if (!empty($parts['port'])) {
 521                                 $host .= ':' . $parts['port'];
 522                         }
 523
 524                         $baseurl = $parts['scheme'] . '://' . $host;
 525
 526                         $nick = '';
 527                         $addr = '';
 528
 529                         $path_parts = explode('/', trim($parts['path'] ?? '', '/'));
 530                         if (!empty($path_parts)) {
 531                                 $nick = ltrim(end($path_parts), '@');
 532                                 $addr = $nick . '@' . $host;
 533                         }
 534
 535                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 536                         if (empty($webfinger)) {
 537                                 $lrdd = self::hostMeta($host);
 538                         }
 539
 540                         if (empty($webfinger) && empty($lrdd)) {
 541                                 while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
 542                                         $host    .= '/' . array_shift($path_parts);
 543                                         $baseurl = $parts['scheme'] . '://' . $host;
 544
 545                                         if (!empty($nick)) {
 546                                                 $addr = $nick . '@' . $host;
 547                                         }
 548
 549                                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 550                                         if (empty($webfinger)) {
 551                                                 $lrdd = self::hostMeta($host);
 552                                         }
 553                                 }
 554
 555                                 if (empty($lrdd) && empty($webfinger)) {
 556                                         return [];
 557                                 }
 558                         }
 559                 } elseif (strstr($uri, '@')) {
 560                         // Remove "acct:" from the URI
 561                         $uri = str_replace('acct:', '', $uri);
 562
 563                         $host = substr($uri, strpos($uri, '@') + 1);
 564                         $nick = substr($uri, 0, strpos($uri, '@'));
 565                         $addr = $uri;
 566
 567                         $webfinger = self::getWebfinger('https://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 568                         if (self::$istimeout) {
 569                                 return [];
 570                         }
 571
 572                         if (empty($webfinger)) {
 573                                 $webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 574                                 if (self::$istimeout) {
 575                                         return [];
 576                                 }
 577                         } else {
 578                                 $baseurl = 'https://' . $host;
 579                         }
 580
 581                         if (empty($webfinger)) {
 582                                 $lrdd = self::hostMeta($host);
 583                                 if (self::$istimeout) {
 584                                         return [];
 585                                 }
 586                                 $baseurl = self::$baseurl;
 587                         } else {
 588                                 $baseurl = 'http://' . $host;
 589                         }
 590                 } else {
 591                         Logger::info('URI was not detectable', ['uri' => $uri]);
 592                         return [];
 593                 }
 594
 595                 if (empty($webfinger)) {
 596                         foreach ($lrdd as $type => $template) {
 597                                 if ($webfinger) {
 598                                         continue;
 599                                 }
 600
 601                                 $webfinger = self::getWebfinger($template, $type, $uri, $addr);
 602                         }
 603                 }
 604
 605                 if (empty($webfinger)) {
 606                         return [];
 607                 }
 608
 609                 if ($webfinger['detected'] == $addr) {
 610                         $webfinger['nick'] = $nick;
 611                         $webfinger['addr'] = $addr;
 612                 }
 613
 614                 $webfinger['baseurl'] = $baseurl;
 615
 616                 return $webfinger;
 617         }
 618
 619         /**
 620          * Perform network request for webfinger data
 621          *
 622          * @param string $template
 623          * @param string $type
 624          * @param string $uri
 625          * @param string $addr
 626          * @return array webfinger results
 627          */
 628         private static function getWebfinger(string $template, string $type, string $uri, string $addr)
 629         {
 630                 if (Network::isUrlBlocked($template)) {
 631                         Logger::info('Domain is blocked', ['url' => $template]);
 632                         return [];
 633                 }
 634
 635                 // First try the address because this is the primary purpose of webfinger
 636                 if (!empty($addr)) {
 637                         $detected = $addr;
 638                         $path = str_replace('{uri}', urlencode("acct:" . $addr), $template);
 639                         $webfinger = self::webfinger($path, $type);
 640                         if (self::$istimeout) {
 641                                 return [];
 642                         }
 643                 }
 644
 645                 // Then try the URI
 646                 if (empty($webfinger) && $uri != $addr) {
 647                         $detected = $uri;
 648                         $path = str_replace('{uri}', urlencode($uri), $template);
 649                         $webfinger = self::webfinger($path, $type);
 650                         if (self::$istimeout) {
 651                                 return [];
 652                         }
 653                 }
 654
 655                 if (empty($webfinger)) {
 656                         return [];
 657                 }
 658
 659                 return ['webfinger' => $webfinger, 'detected' => $detected];
 660         }
 661
 662         /**
 663          * Fetch information (protocol endpoints and user information) about a given uri
 664          *
 665          * This function is only called by the "uri" function that adds caching and rearranging of data.
 666          *
 667          * @param string  $uri        Address that should be probed
 668          * @param string  $network    Test for this specific network
 669          * @param integer $uid        User ID for the probe (only used for mails)
 670          * @param array   $ap_profile Previously probed AP profile
 671          *
 672          * @return array uri data
 673          * @throws HTTPException\InternalServerErrorException
 674          */
 675         private static function detect(string $uri, string $network, int $uid, array $ap_profile)
 676         {
 677                 $hookData = [
 678                         'uri'     => $uri,
 679                         'network' => $network,
 680                         'uid'     => $uid,
 681                         'result'  => null,
 682                 ];
 683
 684                 Hook::callAll('probe_detect', $hookData);
 685
 686                 if (isset($hookData['result'])) {
 687                         return is_array($hookData['result']) ? $hookData['result'] : [];
 688                 }
 689
 690                 $parts = parse_url($uri);
 691
 692                 if (empty($parts['scheme']) || !empty($parts['host']) && strstr($uri, '@')) {
 693                         // If the URI starts with "mailto:" then jump directly to the mail detection
 694                         if (strpos($uri, 'mailto:') !== false) {
 695                                 $uri = str_replace('mailto:', '', $uri);
 696                                 return self::mail($uri, $uid);
 697                         }
 698
 699                         if ($network == Protocol::MAIL) {
 700                                 return self::mail($uri, $uid);
 701                         }
 702                 } else {
 703                         Logger::info('URI was not detectable', ['uri' => $uri]);
 704                         return [];
 705                 }
 706
 707                 Logger::info('Probing start', ['uri' => $uri]);
 708
 709                 if (!empty($ap_profile['addr']) && ($ap_profile['addr'] != $uri)) {
 710                         $data = self::getWebfingerArray($ap_profile['addr']);
 711                 }
 712
 713                 if (empty($data)) {
 714                         $data = self::getWebfingerArray($uri);
 715                 }
 716
 717                 if (empty($data)) {
 718                         if (!empty($parts['scheme'])) {
 719                                 return self::feed($uri);
 720                         } elseif (!empty($uid)) {
 721                                 return self::mail($uri, $uid);
 722                         } else {
 723                                 return [];
 724                         }
 725                 }
 726
 727                 $webfinger = $data['webfinger'];
 728                 $nick = $data['nick'] ?? '';
 729                 $addr = $data['addr'] ?? '';
 730                 $baseurl = $data['baseurl'] ?? '';
 731
 732                 $result = [];
 733
 734                 if (in_array($network, ["", Protocol::DFRN])) {
 735                         $result = self::dfrn($webfinger);
 736                 }
 737                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 738                         $result = self::diaspora($webfinger);
 739                 }
 740                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 741                         $result = self::ostatus($webfinger);
 742                 }
 743                 if (in_array($network, ['', Protocol::ZOT])) {
 744                         $result = self::zot($webfinger, $result, $baseurl);
 745                 }
 746                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 747                         $result = self::pumpio($webfinger, $addr);
 748                 }
 749                 if (empty($result['network']) && empty($ap_profile['network']) || ($network == Protocol::FEED)) {
 750                         $result = self::feed($uri);
 751                 } else {
 752                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 753                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 754                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 755                                 $result["nick"] = $nick;
 756                         }
 757
 758                         if (empty($result["addr"]) && ($addr != "")) {
 759                                 $result["addr"] = $addr;
 760                         }
 761                 }
 762
 763                 $result = self::getSubscribeLink($result, $webfinger);
 764
 765                 if (empty($result["network"])) {
 766                         $result["network"] = Protocol::PHANTOM;
 767                 }
 768
 769                 if (empty($result['baseurl']) && !empty($baseurl)) {
 770                         $result['baseurl'] = $baseurl;
 771                 }
 772
 773                 if (empty($result["url"])) {
 774                         $result["url"] = $uri;
 775                 }
 776
 777                 Logger::info('Probing done', ['uri' => $uri, 'network' => $result["network"]]);
 778
 779                 return $result;
 780         }
 781
 782         /**
 783          * Check for Zot contact
 784          *
 785          * @param array $webfinger Webfinger data
 786          * @param array $data      previously probed data
 787          *
 788          * @return array Zot data
 789          * @throws HTTPException\InternalServerErrorException
 790          */
 791         private static function zot($webfinger, $data, $baseurl)
 792         {
 793                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 794                         foreach ($webfinger["aliases"] as $alias) {
 795                                 if (substr($alias, 0, 5) == 'acct:') {
 796                                         $data["addr"] = substr($alias, 5);
 797                                 }
 798                         }
 799                 }
 800
 801                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 802                         $data["addr"] = substr($webfinger["subject"], 5);
 803                 }
 804
 805                 $zot_url = '';
 806                 foreach ($webfinger['links'] as $link) {
 807                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 808                                 $zot_url = $link['href'];
 809                         }
 810                 }
 811
 812                 if (empty($zot_url) && !empty($data['addr']) && !empty($baseurl)) {
 813                         $condition = ['nurl' => Strings::normaliseLink($baseurl), 'platform' => ['hubzilla']];
 814                         if (!DBA::exists('gserver', $condition)) {
 815                                 return $data;
 816                         }
 817                         $zot_url = $baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 818                 }
 819
 820                 if (empty($zot_url)) {
 821                         return $data;
 822                 }
 823
 824                 $data = self::pollZot($zot_url, $data);
 825
 826                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 827                         foreach ($webfinger['aliases'] as $alias) {
 828                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
 829                                         $data['alias'] = $alias;
 830                                 }
 831                         }
 832                 }
 833
 834                 return $data;
 835         }
 836
 837         public static function pollZot($url, $data)
 838         {
 839                 $curlResult = DI::httpClient()->get($url);
 840                 if ($curlResult->isTimeout()) {
 841                         return $data;
 842                 }
 843                 $content = $curlResult->getBody();
 844                 if (!$content) {
 845                         return $data;
 846                 }
 847
 848                 $json = json_decode($content, true);
 849                 if (!is_array($json)) {
 850                         return $data;
 851                 }
 852
 853                 if (empty($data['network'])) {
 854                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 855                                 $data['network'] = Protocol::ZOT;
 856                         } elseif (!isset($json['protocols'])) {
 857                                 $data['network'] = Protocol::ZOT;
 858                         }
 859                 }
 860
 861                 if (!empty($json['guid']) && empty($data['guid'])) {
 862                         $data['guid'] = $json['guid'];
 863                 }
 864                 if (!empty($json['key']) && empty($data['pubkey'])) {
 865                         $data['pubkey'] = $json['key'];
 866                 }
 867                 if (!empty($json['name'])) {
 868                         $data['name'] = $json['name'];
 869                 }
 870                 if (!empty($json['photo'])) {
 871                         $data['photo'] = $json['photo'];
 872                         if (!empty($json['photo_updated'])) {
 873                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 874                         }
 875                 }
 876                 if (!empty($json['address'])) {
 877                         $data['addr'] = $json['address'];
 878                 }
 879                 if (!empty($json['url'])) {
 880                         $data['url'] = $json['url'];
 881                 }
 882                 if (!empty($json['connections_url'])) {
 883                         $data['poco'] = $json['connections_url'];
 884                 }
 885                 if (isset($json['searchable'])) {
 886                         $data['hide'] = !$json['searchable'];
 887                 }
 888                 if (!empty($json['public_forum'])) {
 889                         $data['community'] = $json['public_forum'];
 890                         $data['account-type'] = User::PAGE_FLAGS_COMMUNITY;
 891                 }
 892
 893                 if (!empty($json['profile'])) {
 894                         $profile = $json['profile'];
 895                         if (!empty($profile['description'])) {
 896                                 $data['about'] = $profile['description'];
 897                         }
 898                         if (!empty($profile['keywords'])) {
 899                                 $keywords = implode(', ', $profile['keywords']);
 900                                 if (!empty($keywords)) {
 901                                         $data['keywords'] = $keywords;
 902                                 }
 903                         }
 904
 905                         $loc = [];
 906                         if (!empty($profile['region'])) {
 907                                 $loc['region'] = $profile['region'];
 908                         }
 909                         if (!empty($profile['country'])) {
 910                                 $loc['country-name'] = $profile['country'];
 911                         }
 912                         $location = Profile::formatLocation($loc);
 913                         if (!empty($location)) {
 914                                 $data['location'] = $location;
 915                         }
 916                 }
 917
 918                 return $data;
 919         }
 920
 921         /**
 922          * Perform a webfinger request.
 923          *
 924          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 925          *
 926          * @param string $url  Address that should be probed
 927          * @param string $type type
 928          *
 929          * @return array webfinger data
 930          * @throws HTTPException\InternalServerErrorException
 931          */
 932         public static function webfinger($url, $type)
 933         {
 934                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 935
 936                 $curlResult = DI::httpClient()->get($url, [HttpClientOptions::TIMEOUT => $xrd_timeout, HttpClientOptions::ACCEPT_CONTENT => [$type]]);
 937                 if ($curlResult->isTimeout()) {
 938                         self::$istimeout = true;
 939                         return [];
 940                 }
 941                 $data = $curlResult->getBody();
 942
 943                 $webfinger = json_decode($data, true);
 944                 if (!empty($webfinger)) {
 945                         if (!isset($webfinger["links"])) {
 946                                 Logger::info('No json webfinger links', ['url' => $url]);
 947                                 return [];
 948                         }
 949                         return $webfinger;
 950                 }
 951
 952                 // If it is not JSON, maybe it is XML
 953                 $xrd = XML::parseString($data, true);
 954                 if (!is_object($xrd)) {
 955                         Logger::info('No webfinger data retrievable', ['url' => $url]);
 956                         return [];
 957                 }
 958
 959                 $xrd_arr = XML::elementToArray($xrd);
 960                 if (!isset($xrd_arr["xrd"]["link"])) {
 961                         Logger::info('No XML webfinger links', ['url' => $url]);
 962                         return [];
 963                 }
 964
 965                 $webfinger = [];
 966
 967                 if (!empty($xrd_arr["xrd"]["subject"])) {
 968                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 969                 }
 970
 971                 if (!empty($xrd_arr["xrd"]["alias"])) {
 972                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 973                 }
 974
 975                 $webfinger["links"] = [];
 976
 977                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 978                         if (!empty($data["@attributes"])) {
 979                                 $attributes = $data["@attributes"];
 980                         } elseif ($value == "@attributes") {
 981                                 $attributes = $data;
 982                         } else {
 983                                 continue;
 984                         }
 985
 986                         $webfinger["links"][] = $attributes;
 987                 }
 988                 return $webfinger;
 989         }
 990
 991         /**
 992          * Poll the Friendica specific noscrape page.
 993          *
 994          * "noscrape" is a faster alternative to fetch the data from the hcard.
 995          * This functionality was originally created for the directory.
 996          *
 997          * @param string $noscrape_url Link to the noscrape page
 998          * @param array  $data         The already fetched data
 999          *
1000          * @return array noscrape data
1001          * @throws HTTPException\InternalServerErrorException
1002          */
1003         private static function pollNoscrape($noscrape_url, $data)
1004         {
1005                 $curlResult = DI::httpClient()->get($noscrape_url);
1006                 if ($curlResult->isTimeout()) {
1007                         self::$istimeout = true;
1008                         return $data;
1009                 }
1010                 $content = $curlResult->getBody();
1011                 if (!$content) {
1012                         Logger::info('Empty body', ['url' => $noscrape_url]);
1013                         return $data;
1014                 }
1015
1016                 $json = json_decode($content, true);
1017                 if (!is_array($json)) {
1018                         Logger::info('No json data', ['url' => $noscrape_url]);
1019                         return $data;
1020                 }
1021
1022                 if (!empty($json["fn"])) {
1023                         $data["name"] = $json["fn"];
1024                 }
1025
1026                 if (!empty($json["addr"])) {
1027                         $data["addr"] = $json["addr"];
1028                 }
1029
1030                 if (!empty($json["nick"])) {
1031                         $data["nick"] = $json["nick"];
1032                 }
1033
1034                 if (!empty($json["guid"])) {
1035                         $data["guid"] = $json["guid"];
1036                 }
1037
1038                 if (!empty($json["comm"])) {
1039                         $data["community"] = $json["comm"];
1040                 }
1041
1042                 if (!empty($json["tags"])) {
1043                         $keywords = implode(", ", $json["tags"]);
1044                         if ($keywords != "") {
1045                                 $data["keywords"] = $keywords;
1046                         }
1047                 }
1048
1049                 $location = Profile::formatLocation($json);
1050                 if ($location) {
1051                         $data["location"] = $location;
1052                 }
1053
1054                 if (!empty($json["about"])) {
1055                         $data["about"] = $json["about"];
1056                 }
1057
1058                 if (!empty($json["xmpp"])) {
1059                         $data["xmpp"] = $json["xmpp"];
1060                 }
1061
1062                 if (!empty($json["matrix"])) {
1063                         $data["matrix"] = $json["matrix"];
1064                 }
1065
1066                 if (!empty($json["key"])) {
1067                         $data["pubkey"] = $json["key"];
1068                 }
1069
1070                 if (!empty($json["photo"])) {
1071                         $data["photo"] = $json["photo"];
1072                 }
1073
1074                 if (!empty($json["dfrn-request"])) {
1075                         $data["request"] = $json["dfrn-request"];
1076                 }
1077
1078                 if (!empty($json["dfrn-confirm"])) {
1079                         $data["confirm"] = $json["dfrn-confirm"];
1080                 }
1081
1082                 if (!empty($json["dfrn-notify"])) {
1083                         $data["notify"] = $json["dfrn-notify"];
1084                 }
1085
1086                 if (!empty($json["dfrn-poll"])) {
1087                         $data["poll"] = $json["dfrn-poll"];
1088                 }
1089
1090                 if (isset($json["hide"])) {
1091                         $data["hide"] = (bool)$json["hide"];
1092                 } else {
1093                         $data["hide"] = false;
1094                 }
1095
1096                 return $data;
1097         }
1098
1099         /**
1100          * Check for valid DFRN data
1101          *
1102          * @param array $data DFRN data
1103          *
1104          * @return int Number of errors
1105          */
1106         public static function validDfrn($data)
1107         {
1108                 $errors = 0;
1109                 if (!isset($data['key'])) {
1110                         $errors ++;
1111                 }
1112                 if (!isset($data['dfrn-request'])) {
1113                         $errors ++;
1114                 }
1115                 if (!isset($data['dfrn-confirm'])) {
1116                         $errors ++;
1117                 }
1118                 if (!isset($data['dfrn-notify'])) {
1119                         $errors ++;
1120                 }
1121                 if (!isset($data['dfrn-poll'])) {
1122                         $errors ++;
1123                 }
1124                 return $errors;
1125         }
1126
1127         /**
1128          * Fetch data from a DFRN profile page and via "noscrape"
1129          *
1130          * @param string $profile_link Link to the profile page
1131          *
1132          * @return array profile data
1133          * @throws HTTPException\InternalServerErrorException
1134          * @throws \ImagickException
1135          */
1136         public static function profile($profile_link)
1137         {
1138                 $data = [];
1139
1140                 Logger::info('Check profile', ['link' => $profile_link]);
1141
1142                 // Fetch data via noscrape - this is faster
1143                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1144                 $data = self::pollNoscrape($noscrape_url, $data);
1145
1146                 if (!isset($data["notify"])
1147                         || !isset($data["confirm"])
1148                         || !isset($data["request"])
1149                         || !isset($data["poll"])
1150                         || !isset($data["name"])
1151                         || !isset($data["photo"])
1152                 ) {
1153                         $data = self::pollHcard($profile_link, $data, true);
1154                 }
1155
1156                 $prof_data = [];
1157
1158                 if (empty($data["addr"]) || empty($data["nick"])) {
1159                         $probe_data = self::uri($profile_link);
1160                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1161                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1162                 }
1163
1164                 $prof_data["addr"]         = $data["addr"];
1165                 $prof_data["nick"]         = $data["nick"];
1166                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1167                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1168                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1169                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1170                 $prof_data["photo"]        = $data['photo']   ?? null;
1171                 $prof_data["fn"]           = $data['name']    ?? null;
1172                 $prof_data["key"]          = $data['pubkey']  ?? null;
1173
1174                 Logger::debug('Result', ['link' => $profile_link, 'data' => $prof_data]);
1175
1176                 return $prof_data;
1177         }
1178
1179         /**
1180          * Check for DFRN contact
1181          *
1182          * @param array $webfinger Webfinger data
1183          *
1184          * @return array DFRN data
1185          * @throws HTTPException\InternalServerErrorException
1186          */
1187         private static function dfrn($webfinger)
1188         {
1189                 $hcard_url = "";
1190                 $data = [];
1191                 // The array is reversed to take into account the order of preference for same-rel links
1192                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1193                 foreach (array_reverse($webfinger["links"]) as $link) {
1194                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1195                                 $data["network"] = Protocol::DFRN;
1196                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1197                                 $data["poll"] = $link["href"];
1198                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1199                                 $data["url"] = $link["href"];
1200                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1201                                 $hcard_url = $link["href"];
1202                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1203                                 $data["poco"] = $link["href"];
1204                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1205                                 $data["photo"] = $link["href"];
1206                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1207                                 $data["baseurl"] = trim($link["href"], '/');
1208                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1209                                 $data["guid"] = $link["href"];
1210                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1211                                 $data["pubkey"] = base64_decode($link["href"]);
1212
1213                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1214                                 if (strstr($data["pubkey"], 'RSA ')) {
1215                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1216                                 }
1217                         }
1218                 }
1219
1220                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1221                         foreach ($webfinger["aliases"] as $alias) {
1222                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1223                                         $data["url"] = $alias;
1224                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1225                                         $data["alias"] = $alias;
1226                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1227                                         $data["addr"] = substr($alias, 5);
1228                                 }
1229                         }
1230                 }
1231
1232                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1233                         $data["addr"] = substr($webfinger["subject"], 5);
1234                 }
1235
1236                 if (!isset($data["network"]) || ($hcard_url == "")) {
1237                         return [];
1238                 }
1239
1240                 // Fetch data via noscrape - this is faster
1241                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1242                 $data = self::pollNoscrape($noscrape_url, $data);
1243
1244                 if (isset($data["notify"])
1245                         && isset($data["confirm"])
1246                         && isset($data["request"])
1247                         && isset($data["poll"])
1248                         && isset($data["name"])
1249                         && isset($data["photo"])
1250                 ) {
1251                         return $data;
1252                 }
1253
1254                 $data = self::pollHcard($hcard_url, $data, true);
1255
1256                 return $data;
1257         }
1258
1259         /**
1260          * Poll the hcard page (Diaspora and Friendica specific)
1261          *
1262          * @param string  $hcard_url Link to the hcard page
1263          * @param array   $data      The already fetched data
1264          * @param boolean $dfrn      Poll DFRN specific data
1265          *
1266          * @return array hcard data
1267          * @throws HTTPException\InternalServerErrorException
1268          */
1269         private static function pollHcard($hcard_url, $data, $dfrn = false)
1270         {
1271                 $curlResult = DI::httpClient()->get($hcard_url);
1272                 if ($curlResult->isTimeout()) {
1273                         self::$istimeout = true;
1274                         return [];
1275                 }
1276                 $content = $curlResult->getBody();
1277                 if (empty($content)) {
1278                         return [];
1279                 }
1280
1281                 $doc = new DOMDocument();
1282                 if (!@$doc->loadHTML($content)) {
1283                         return [];
1284                 }
1285
1286                 $xpath = new DomXPath($doc);
1287
1288                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1289                 if (!is_object($vcards)) {
1290                         return [];
1291                 }
1292
1293                 if (!isset($data["baseurl"])) {
1294                         $data["baseurl"] = "";
1295                 }
1296
1297                 if ($vcards->length > 0) {
1298                         $vcard = $vcards->item(0);
1299
1300                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1301                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1302                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1303                         if (($search->length > 0) && empty($data["guid"])) {
1304                                 $data["guid"] = $search->item(0)->nodeValue;
1305                         }
1306
1307                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1308                         if ($search->length > 0) {
1309                                 $data["nick"] = $search->item(0)->nodeValue;
1310                         }
1311
1312                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1313                         if ($search->length > 0) {
1314                                 $data["name"] = $search->item(0)->nodeValue;
1315                         }
1316
1317                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1318                         if ($search->length > 0) {
1319                                 $data["searchable"] = $search->item(0)->nodeValue;
1320                         }
1321
1322                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1323                         if ($search->length > 0) {
1324                                 $data["pubkey"] = $search->item(0)->nodeValue;
1325                                 if (strstr($data["pubkey"], 'RSA ')) {
1326                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1327                                 }
1328                         }
1329
1330                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1331                         if ($search->length > 0) {
1332                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1333                         }
1334                 }
1335
1336                 $avatar = [];
1337                 if (!empty($vcard)) {
1338                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1339                         foreach ($photos as $photo) {
1340                                 $attr = [];
1341                                 foreach ($photo->attributes as $attribute) {
1342                                         $attr[$attribute->name] = trim($attribute->value);
1343                                 }
1344
1345                                 if (isset($attr["src"]) && isset($attr["width"])) {
1346                                         $avatar[$attr["width"]] = $attr["src"];
1347                                 }
1348
1349                                 // We don't have a width. So we just take everything that we got.
1350                                 // This is a Hubzilla workaround which doesn't send a width.
1351                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1352                                         $avatar[] = $attr["src"];
1353                                 }
1354                         }
1355                 }
1356
1357                 if (sizeof($avatar)) {
1358                         ksort($avatar);
1359                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1360                 }
1361
1362                 if ($dfrn) {
1363                         // Poll DFRN specific data
1364                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1365                         if ($search->length > 0) {
1366                                 foreach ($search as $link) {
1367                                         //$data["request"] = $search->item(0)->nodeValue;
1368                                         $attr = [];
1369                                         foreach ($link->attributes as $attribute) {
1370                                                 $attr[$attribute->name] = trim($attribute->value);
1371                                         }
1372
1373                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1374                                 }
1375                         }
1376
1377                         // Older Friendica versions had used the "uid" field differently than newer versions
1378                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1379                                 unset($data["guid"]);
1380                         }
1381                 }
1382
1383
1384                 return $data;
1385         }
1386
1387         /**
1388          * Check for Diaspora contact
1389          *
1390          * @param array $webfinger Webfinger data
1391          *
1392          * @return array Diaspora data
1393          * @throws HTTPException\InternalServerErrorException
1394          */
1395         private static function diaspora($webfinger)
1396         {
1397                 $hcard_url = "";
1398                 $data = [];
1399
1400                 // The array is reversed to take into account the order of preference for same-rel links
1401                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1402                 foreach (array_reverse($webfinger["links"]) as $link) {
1403                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1404                                 $hcard_url = $link["href"];
1405                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1406                                 $data["baseurl"] = trim($link["href"], '/');
1407                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1408                                 $data["guid"] = $link["href"];
1409                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1410                                 $data["url"] = $link["href"];
1411                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && empty($link["type"]) && !empty($link["href"])) {
1412                                 $profile_url = $link["href"];
1413                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1414                                 $data["poll"] = $link["href"];
1415                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1416                                 $data["poco"] = $link["href"];
1417                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1418                                 $data["notify"] = $link["href"];
1419                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1420                                 $data["pubkey"] = base64_decode($link["href"]);
1421
1422                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1423                                 if (strstr($data["pubkey"], 'RSA ')) {
1424                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1425                                 }
1426                         }
1427                 }
1428
1429                 if (empty($data["url"]) && !empty($profile_url)) {
1430                         $data["url"] = $profile_url;
1431                 }
1432
1433                 if (empty($data["url"]) || empty($hcard_url)) {
1434                         return [];
1435                 }
1436
1437                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1438                         foreach ($webfinger["aliases"] as $alias) {
1439                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1440                                         $data["alias"] = $alias;
1441                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1442                                         $data["addr"] = substr($alias, 5);
1443                                 }
1444                         }
1445                 }
1446
1447                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1448                         $data["addr"] = substr($webfinger["subject"], 5);
1449                 }
1450
1451                 // Fetch further information from the hcard
1452                 $data = self::pollHcard($hcard_url, $data);
1453
1454                 if (!$data) {
1455                         return [];
1456                 }
1457
1458                 if (!empty($data["url"])
1459                         && !empty($data["guid"])
1460                         && !empty($data["baseurl"])
1461                         && !empty($data["pubkey"])
1462                         && !empty($hcard_url)
1463                 ) {
1464                         $data["network"] = Protocol::DIASPORA;
1465                         $data["manually-approve"] = false;
1466
1467                         // The Diaspora handle must always be lowercase
1468                         if (!empty($data["addr"])) {
1469                                 $data["addr"] = strtolower($data["addr"]);
1470                         }
1471
1472                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1473                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1474                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1475                 } else {
1476                         return [];
1477                 }
1478
1479                 return $data;
1480         }
1481
1482         /**
1483          * Check for OStatus contact
1484          *
1485          * @param array $webfinger Webfinger data
1486          * @param bool  $short     Short detection mode
1487          *
1488          * @return array|bool OStatus data or "false" on error or "true" on short mode
1489          * @throws HTTPException\InternalServerErrorException
1490          */
1491         private static function ostatus($webfinger, $short = false)
1492         {
1493                 $data = [];
1494
1495                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1496                         foreach ($webfinger["aliases"] as $alias) {
1497                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1498                                         $data["addr"] = str_replace('acct:', '', $alias);
1499                                 }
1500                         }
1501                 }
1502
1503                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1504                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1505                 ) {
1506                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1507                 }
1508
1509                 if (!empty($webfinger["links"])) {
1510                         // The array is reversed to take into account the order of preference for same-rel links
1511                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1512                         foreach (array_reverse($webfinger["links"]) as $link) {
1513                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1514                                         && (($link["type"] ?? "") == "text/html")
1515                                         && ($link["href"] != "")
1516                                 ) {
1517                                         $data["url"] = $data["alias"] = $link["href"];
1518                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1519                                         $data["notify"] = $link["href"];
1520                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1521                                         $data["poll"] = $link["href"];
1522                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1523                                         $pubkey = $link["href"];
1524
1525                                         if (substr($pubkey, 0, 5) === 'data:') {
1526                                                 if (strstr($pubkey, ',')) {
1527                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1528                                                 } else {
1529                                                         $pubkey = substr($pubkey, 5);
1530                                                 }
1531                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1532                                                 $curlResult = DI::httpClient()->get($pubkey);
1533                                                 if ($curlResult->isTimeout()) {
1534                                                         self::$istimeout = true;
1535                                                         return $short ? false : [];
1536                                                 }
1537                                                 $pubkey = $curlResult->getBody();
1538                                         }
1539
1540                                         $key = explode(".", $pubkey);
1541
1542                                         if (sizeof($key) >= 3) {
1543                                                 $m = Strings::base64UrlDecode($key[1]);
1544                                                 $e = Strings::base64UrlDecode($key[2]);
1545                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1546                                         }
1547                                 }
1548                         }
1549                 }
1550
1551                 if (isset($data["notify"]) && isset($data["pubkey"])
1552                         && isset($data["poll"])
1553                         && isset($data["url"])
1554                 ) {
1555                         $data["network"] = Protocol::OSTATUS;
1556                         $data["manually-approve"] = false;
1557                 } else {
1558                         return $short ? false : [];
1559                 }
1560
1561                 if ($short) {
1562                         return true;
1563                 }
1564
1565                 // Fetch all additional data from the feed
1566                 $curlResult = DI::httpClient()->get($data["poll"]);
1567                 if ($curlResult->isTimeout()) {
1568                         self::$istimeout = true;
1569                         return [];
1570                 }
1571                 $feed = $curlResult->getBody();
1572                 $feed_data = Feed::import($feed);
1573                 if (!$feed_data) {
1574                         return [];
1575                 }
1576
1577                 if (!empty($feed_data["header"]["author-name"])) {
1578                         $data["name"] = $feed_data["header"]["author-name"];
1579                 }
1580                 if (!empty($feed_data["header"]["author-nick"])) {
1581                         $data["nick"] = $feed_data["header"]["author-nick"];
1582                 }
1583                 if (!empty($feed_data["header"]["author-avatar"])) {
1584                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1585                 }
1586                 if (!empty($feed_data["header"]["author-id"])) {
1587                         $data["alias"] = $feed_data["header"]["author-id"];
1588                 }
1589                 if (!empty($feed_data["header"]["author-location"])) {
1590                         $data["location"] = $feed_data["header"]["author-location"];
1591                 }
1592                 if (!empty($feed_data["header"]["author-about"])) {
1593                         $data["about"] = $feed_data["header"]["author-about"];
1594                 }
1595                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1596                 // So we take the value that we just fetched, although the other one worked as well
1597                 if (!empty($feed_data["header"]["author-link"])) {
1598                         $data["url"] = $feed_data["header"]["author-link"];
1599                 }
1600
1601                 if ($data["url"] == $data["alias"]) {
1602                         $data["alias"] = '';
1603                 }
1604
1605                 /// @todo Fetch location and "about" from the feed as well
1606                 return $data;
1607         }
1608
1609         /**
1610          * Fetch data from a pump.io profile page
1611          *
1612          * @param string $profile_link Link to the profile page
1613          *
1614          * @return array profile data
1615          */
1616         private static function pumpioProfileData($profile_link)
1617         {
1618                 $curlResult = DI::httpClient()->get($profile_link);
1619                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
1620                         return [];
1621                 }
1622
1623                 $doc = new DOMDocument();
1624                 if (!@$doc->loadHTML($curlResult->getBody())) {
1625                         return [];
1626                 }
1627
1628                 $xpath = new DomXPath($doc);
1629
1630                 $data = [];
1631
1632                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1633
1634                 if ($data["name"] == '') {
1635                         // This is ugly - but pump.io doesn't seem to know a better way for it
1636                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1637                         $pos = strpos($data["name"], chr(10));
1638                         if ($pos) {
1639                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1640                         }
1641                 }
1642
1643                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1644
1645                 if ($data["location"] == '') {
1646                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1647                 }
1648
1649                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1650
1651                 if ($data["about"] == '') {
1652                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1653                 }
1654
1655                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1656                 if (!$avatar) {
1657                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1658                 }
1659                 if ($avatar) {
1660                         foreach ($avatar->attributes as $attribute) {
1661                                 if ($attribute->name == "src") {
1662                                         $data["photo"] = trim($attribute->value);
1663                                 }
1664                         }
1665                 }
1666
1667                 return $data;
1668         }
1669
1670         /**
1671          * Check for pump.io contact
1672          *
1673          * @param array  $webfinger Webfinger data
1674          * @param string $addr
1675          * @return array pump.io data
1676          */
1677         private static function pumpio($webfinger, $addr)
1678         {
1679                 $data = [];
1680                 // The array is reversed to take into account the order of preference for same-rel links
1681                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1682                 foreach (array_reverse($webfinger["links"]) as $link) {
1683                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1684                                 && (($link["type"] ?? "") == "text/html")
1685                                 && ($link["href"] != "")
1686                         ) {
1687                                 $data["url"] = $link["href"];
1688                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1689                                 $data["notify"] = $link["href"];
1690                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1691                                 $data["poll"] = $link["href"];
1692                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1693                                 $data["dialback"] = $link["href"];
1694                         }
1695                 }
1696                 if (isset($data["poll"]) && isset($data["notify"])
1697                         && isset($data["dialback"])
1698                         && isset($data["url"])
1699                 ) {
1700                         // by now we use these fields only for the network type detection
1701                         // So we unset all data that isn't used at the moment
1702                         unset($data["dialback"]);
1703
1704                         $data["network"] = Protocol::PUMPIO;
1705                 } else {
1706                         return [];
1707                 }
1708
1709                 $profile_data = self::pumpioProfileData($data["url"]);
1710
1711                 if (!$profile_data) {
1712                         return [];
1713                 }
1714
1715                 $data = array_merge($data, $profile_data);
1716
1717                 if (($addr != '') && ($data['name'] != '')) {
1718                         $name = trim(str_replace($addr, '', $data['name']));
1719                         if ($name != '') {
1720                                 $data['name'] = $name;
1721                         }
1722                 }
1723
1724                 return $data;
1725         }
1726
1727         /**
1728          * Checks HTML page for RSS feed link
1729          *
1730          * @param string $url  Page link
1731          * @param string $body Page body string
1732          * @return string|false Feed link or false if body was invalid HTML document
1733          */
1734         public static function getFeedLink(string $url, string $body)
1735         {
1736                 if (empty($body)) {
1737                         return '';
1738                 }
1739
1740                 $doc = new DOMDocument();
1741                 if (!@$doc->loadHTML($body)) {
1742                         return false;
1743                 }
1744
1745                 $xpath = new DOMXPath($doc);
1746
1747                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1748
1749                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1750
1751                 return $feedUrl;
1752         }
1753
1754         /**
1755          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1756          *
1757          * Loosely based on RFC 1808
1758          *
1759          * @see https://tools.ietf.org/html/rfc1808
1760          *
1761          * @param string   $href  The potential relative href found in the HTML document
1762          * @param string   $base  The HTML document URL
1763          * @param DOMXPath $xpath The HTML document XPath
1764          * @return string
1765          */
1766         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
1767         {
1768                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1769                         return $href;
1770                 }
1771
1772                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1773
1774                 $baseParts = parse_url($base);
1775                 if (empty($baseParts['host'])) {
1776                         return $href;
1777                 }
1778
1779                 // Naked domain case (scheme://basehost)
1780                 $path = $baseParts['path'] ?? '/';
1781
1782                 // Remove the filename part of the path if it exists (/base/path/file)
1783                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1784
1785                 $hrefParts = parse_url($href);
1786
1787                 if (!empty($hrefParts['path'])) {
1788                         // Root path case (/path) including relative scheme case (//host/path)
1789                         if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1790                                 $path = $hrefParts['path'];
1791                         } else {
1792                                 $path = $path . '/' . $hrefParts['path'];
1793
1794                                 // Resolve arbitrary relative path
1795                                 // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1796                                 $parts = array_filter(explode('/', $path), 'strlen');
1797                                 $absolutes = array();
1798                                 foreach ($parts as $part) {
1799                                         if ('.' == $part) continue;
1800                                         if ('..' == $part) {
1801                                                 array_pop($absolutes);
1802                                         } else {
1803                                                 $absolutes[] = $part;
1804                                         }
1805                                 }
1806
1807                                 $path = '/' . implode('/', $absolutes);
1808                         }
1809                 }
1810
1811                 // Relative scheme case (//host/path)
1812                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1813                 $baseParts['path'] = $path;
1814                 unset($baseParts['query']);
1815                 unset($baseParts['fragment']);
1816
1817                 return Network::unparseURL($baseParts);
1818         }
1819
1820         /**
1821          * Check for feed contact
1822          *
1823          * @param string  $url   Profile link
1824          * @param boolean $probe Do a probe if the page contains a feed link
1825          *
1826          * @return array feed data
1827          * @throws HTTPException\InternalServerErrorException
1828          */
1829         private static function feed($url, $probe = true)
1830         {
1831                 $curlResult = DI::httpClient()->get($url);
1832                 if ($curlResult->isTimeout()) {
1833                         self::$istimeout = true;
1834                         return [];
1835                 }
1836                 $feed = $curlResult->getBody();
1837                 $feed_data = Feed::import($feed);
1838
1839                 if (!$feed_data) {
1840                         if (!$probe) {
1841                                 return [];
1842                         }
1843
1844                         $feed_url = self::getFeedLink($url, $feed);
1845
1846                         if (!$feed_url) {
1847                                 return [];
1848                         }
1849
1850                         return self::feed($feed_url, false);
1851                 }
1852
1853                 if (!empty($feed_data["header"]["author-name"])) {
1854                         $data["name"] = $feed_data["header"]["author-name"];
1855                 }
1856
1857                 if (!empty($feed_data["header"]["author-nick"])) {
1858                         $data["nick"] = $feed_data["header"]["author-nick"];
1859                 }
1860
1861                 if (!empty($feed_data["header"]["author-avatar"])) {
1862                         $data["photo"] = $feed_data["header"]["author-avatar"];
1863                 }
1864
1865                 if (!empty($feed_data["header"]["author-id"])) {
1866                         $data["alias"] = $feed_data["header"]["author-id"];
1867                 }
1868
1869                 $data["url"] = $url;
1870                 $data["poll"] = $url;
1871
1872                 $data["network"] = Protocol::FEED;
1873
1874                 return $data;
1875         }
1876
1877         /**
1878          * Check for mail contact
1879          *
1880          * @param string  $uri Profile link
1881          * @param integer $uid User ID
1882          *
1883          * @return array mail data
1884          * @throws \Exception
1885          */
1886         private static function mail($uri, $uid)
1887         {
1888                 if (!Network::isEmailDomainValid($uri)) {
1889                         return [];
1890                 }
1891
1892                 if ($uid == 0) {
1893                         return [];
1894                 }
1895
1896                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1897
1898                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1899                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1900                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1901
1902                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1903                         return [];
1904                 }
1905
1906                 $mailbox = Email::constructMailboxName($mailacct);
1907                 $password = '';
1908                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1909                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1910                 if (!$mbox) {
1911                         return [];
1912                 }
1913
1914                 $msgs = Email::poll($mbox, $uri);
1915                 Logger::info('Messages found', ['uri' => $uri, 'count' => count($msgs)]);
1916
1917                 if (!count($msgs)) {
1918                         return [];
1919                 }
1920
1921                 $phost = substr($uri, strpos($uri, '@') + 1);
1922
1923                 $data = [];
1924                 $data["addr"]    = $uri;
1925                 $data["network"] = Protocol::MAIL;
1926                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1927                 $data["nick"]    = $data["name"];
1928                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1929                 $data["url"]     = 'mailto:'.$uri;
1930                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1931                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1932
1933                 $x = Email::messageMeta($mbox, $msgs[0]);
1934                 if (stristr($x[0]->from, $uri)) {
1935                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1936                 } elseif (stristr($x[0]->to, $uri)) {
1937                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1938                 }
1939                 if (isset($adr)) {
1940                         foreach ($adr as $feadr) {
1941                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1942                                         &&(strcasecmp($feadr->host, $phost) == 0)
1943                                         && (strlen($feadr->personal))
1944                                 ) {
1945                                         $personal = imap_mime_header_decode($feadr->personal);
1946                                         $data["name"] = "";
1947                                         foreach ($personal as $perspart) {
1948                                                 if ($perspart->charset != "default") {
1949                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1950                                                 } else {
1951                                                         $data["name"] .= $perspart->text;
1952                                                 }
1953                                         }
1954                                 }
1955                         }
1956                 }
1957                 if (!empty($mbox)) {
1958                         imap_close($mbox);
1959                 }
1960                 return $data;
1961         }
1962
1963         /**
1964          * Mix two paths together to possibly fix missing parts
1965          *
1966          * @param string $avatar Path to the avatar
1967          * @param string $base   Another path that is hopefully complete
1968          *
1969          * @return string fixed avatar path
1970          * @throws \Exception
1971          */
1972         public static function fixAvatar($avatar, $base)
1973         {
1974                 $base_parts = parse_url($base);
1975
1976                 // Remove all parts that could create a problem
1977                 unset($base_parts['path']);
1978                 unset($base_parts['query']);
1979                 unset($base_parts['fragment']);
1980
1981                 $avatar_parts = parse_url($avatar);
1982
1983                 // Now we mix them
1984                 $parts = array_merge($base_parts, $avatar_parts);
1985
1986                 // And put them together again
1987                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
1988                 $host     = isset($parts['host'])     ? $parts['host']           : '';
1989                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
1990                 $path     = isset($parts['path'])     ? $parts['path']           : '';
1991                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
1992                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
1993
1994                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
1995
1996                 Logger::debug('Avatar fixed', ['base' => $base, 'avatar' => $avatar, 'fixed' => $fixed]);
1997
1998                 return $fixed;
1999         }
2000
2001         /**
2002          * Fetch the last date that the contact had posted something (publically)
2003          *
2004          * @param string $data  probing result
2005          * @return string last activity
2006          */
2007         public static function getLastUpdate(array $data)
2008         {
2009                 $uid = User::getIdForURL($data['url']);
2010                 if (!empty($uid)) {
2011                         $contact = Contact::selectFirst(['url', 'last-item'], ['self' => true, 'uid' => $uid]);
2012                         if (!empty($contact['last-item'])) {
2013                                 return $contact['last-item'];
2014                         }
2015                 }
2016
2017                 if ($lastUpdate = self::updateFromNoScrape($data)) {
2018                         return $lastUpdate;
2019                 }
2020
2021                 if (!empty($data['outbox'])) {
2022                         return self::updateFromOutbox($data['outbox'], $data);
2023                 } elseif (!empty($data['poll']) && ($data['network'] == Protocol::ACTIVITYPUB)) {
2024                         return self::updateFromOutbox($data['poll'], $data);
2025                 } elseif (!empty($data['poll'])) {
2026                         return self::updateFromFeed($data);
2027                 }
2028
2029                 return '';
2030         }
2031
2032         /**
2033          * Fetch the last activity date from the "noscrape" endpoint
2034          *
2035          * @param array $data Probing result
2036          * @return string last activity
2037          *
2038          * @return bool 'true' if update was successful or the server was unreachable
2039          */
2040         private static function updateFromNoScrape(array $data)
2041         {
2042                 if (empty($data['baseurl'])) {
2043                         return '';
2044                 }
2045
2046                 // Check the 'noscrape' endpoint when it is a Friendica server
2047                 $gserver = DBA::selectFirst('gserver', ['noscrape'], ["`nurl` = ? AND `noscrape` != ''",
2048                         Strings::normaliseLink($data['baseurl'])]);
2049                 if (!DBA::isResult($gserver)) {
2050                         return '';
2051                 }
2052
2053                 $curlResult = DI::httpClient()->get($gserver['noscrape'] . '/' . $data['nick']);
2054
2055                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
2056                         $noscrape = json_decode($curlResult->getBody(), true);
2057                         if (!empty($noscrape) && !empty($noscrape['updated'])) {
2058                                 return DateTimeFormat::utc($noscrape['updated'], DateTimeFormat::MYSQL);
2059                         }
2060                 }
2061
2062                 return '';
2063         }
2064
2065         /**
2066          * Fetch the last activity date from an ActivityPub Outbox
2067          *
2068          * @param string $feed
2069          * @param array  $data Probing result
2070          * @return string last activity
2071          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
2072          */
2073         private static function updateFromOutbox(string $feed, array $data)
2074         {
2075                 $outbox = ActivityPub::fetchContent($feed);
2076                 if (empty($outbox)) {
2077                         return '';
2078                 }
2079
2080                 if (!empty($outbox['orderedItems'])) {
2081                         $items = $outbox['orderedItems'];
2082                 } elseif (!empty($outbox['first']['orderedItems'])) {
2083                         $items = $outbox['first']['orderedItems'];
2084                 } elseif (!empty($outbox['first']['href']) && ($outbox['first']['href'] != $feed)) {
2085                         return self::updateFromOutbox($outbox['first']['href'], $data);
2086                 } elseif (!empty($outbox['first'])) {
2087                         if (is_string($outbox['first']) && ($outbox['first'] != $feed)) {
2088                                 return self::updateFromOutbox($outbox['first'], $data);
2089                         } else {
2090                                 Logger::warning('Unexpected data', ['outbox' => $outbox]);
2091                         }
2092                         return '';
2093                 } else {
2094                         $items = [];
2095                 }
2096
2097                 $last_updated = '';
2098                 foreach ($items as $activity) {
2099                         if (!empty($activity['published'])) {
2100                                 $published =  DateTimeFormat::utc($activity['published']);
2101                         } elseif (!empty($activity['object']['published'])) {
2102                                 $published =  DateTimeFormat::utc($activity['object']['published']);
2103                         } else {
2104                                 continue;
2105                         }
2106
2107                         if ($last_updated < $published) {
2108                                 $last_updated = $published;
2109                         }
2110                 }
2111
2112                 if (!empty($last_updated)) {
2113                         return $last_updated;
2114                 }
2115
2116                 return '';
2117         }
2118
2119         /**
2120          * Fetch the last activity date from an XML feed
2121          *
2122          * @param array $data Probing result
2123          * @return string last activity
2124          */
2125         private static function updateFromFeed(array $data)
2126         {
2127                 // Search for the newest entry in the feed
2128                 $curlResult = DI::httpClient()->get($data['poll']);
2129                 if (!$curlResult->isSuccess() || !$curlResult->getBody()) {
2130                         return '';
2131                 }
2132
2133                 $doc = new DOMDocument();
2134                 @$doc->loadXML($curlResult->getBody());
2135
2136                 $xpath = new DOMXPath($doc);
2137                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
2138
2139                 $entries = $xpath->query('/atom:feed/atom:entry');
2140
2141                 $last_updated = '';
2142
2143                 foreach ($entries as $entry) {
2144                         $published_item = $xpath->query('atom:published/text()', $entry)->item(0);
2145                         $updated_item   = $xpath->query('atom:updated/text()'  , $entry)->item(0);
2146                         $published      = !empty($published_item->nodeValue) ? DateTimeFormat::utc($published_item->nodeValue) : null;
2147                         $updated        = !empty($updated_item->nodeValue) ? DateTimeFormat::utc($updated_item->nodeValue) : null;
2148
2149                         if (empty($published) || empty($updated)) {
2150                                 Logger::notice('Invalid entry for XPath.', ['entry' => $entry, 'url' => $data['url']]);
2151                                 continue;
2152                         }
2153
2154                         if ($last_updated < $published) {
2155                                 $last_updated = $published;
2156                         }
2157
2158                         if ($last_updated < $updated) {
2159                                 $last_updated = $updated;
2160                         }
2161                 }
2162
2163                 if (!empty($last_updated)) {
2164                         return $last_updated;
2165                 }
2166
2167                 return '';
2168         }
2169
2170         /**
2171          * Probe data from local profiles without network traffic
2172          *
2173          * @param string $url
2174          * @return array probed data
2175          * @throws HTTPException\InternalServerErrorException
2176          * @throws HTTPException\NotFoundException
2177          */
2178         private static function localProbe(string $url): array
2179         {
2180                 try {
2181                         $uid = User::getIdForURL($url);
2182                         if (!$uid) {
2183                                 throw new HTTPException\NotFoundException('User not found.');
2184                         }
2185
2186                         $owner     = User::getOwnerDataById($uid);
2187                         $approfile = ActivityPub\Transmitter::getProfile($uid);
2188
2189                         if (empty($owner['gsid'])) {
2190                                 $owner['gsid'] = GServer::getID($approfile['generator']['url']);
2191                         }
2192
2193                         $data = [
2194                                 'name' => $owner['name'], 'nick' => $owner['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
2195                                 'url' => $owner['url'], 'addr' => $owner['addr'], 'alias' => $owner['alias'],
2196                                 'photo' => User::getAvatarUrl($owner),
2197                                 'header' => $owner['header'] ? Contact::getHeaderUrlForId($owner['id'], $owner['updated']) : '',
2198                                 'account-type' => $owner['contact-type'], 'community' => ($owner['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
2199                                 'keywords' => $owner['keywords'], 'location' => $owner['location'], 'about' => $owner['about'],
2200                                 'xmpp' => $owner['xmpp'], 'matrix' => $owner['matrix'],
2201                                 'hide' => !$owner['net-publish'], 'batch' => '', 'notify' => $owner['notify'],
2202                                 'poll' => $owner['poll'], 'request' => $owner['request'], 'confirm' => $owner['confirm'],
2203                                 'subscribe' => $approfile['generator']['url'] . '/follow?url={uri}', 'poco' => $owner['poco'],
2204                                 'following' => $approfile['following'], 'followers' => $approfile['followers'],
2205                                 'inbox' => $approfile['inbox'], 'outbox' => $approfile['outbox'],
2206                                 'sharedinbox' => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
2207                                 'pubkey' => $owner['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $owner['gsid'],
2208                                 'manually-approve' => in_array($owner['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])
2209                         ];
2210                 } catch (Exception $e) {
2211                         // Default values for non existing targets
2212                         $data = [
2213                                 'name' => $url, 'nick' => $url, 'url' => $url, 'network' => Protocol::PHANTOM,
2214                                 'photo' => DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO
2215                         ];
2216                 }
2217
2218                 return self::rearrangeData($data);
2219         }
2220 }