src/Network/Probe.php

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