src/Network/Probe.php

   1 <?php
   2 /**
   3  * @copyright Copyright (C) 2010-2021, the Friendica project
   4  *
   5  * @license GNU AGPL version 3 or any later version
   6  *
   7  * This program is free software: you can redistribute it and/or modify
   8  * it under the terms of the GNU Affero General Public License as
   9  * published by the Free Software Foundation, either version 3 of the
  10  * License, or (at your option) any later version.
  11  *
  12  * This program is distributed in the hope that it will be useful,
  13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  15  * GNU Affero General Public License for more details.
  16  *
  17  * You should have received a copy of the GNU Affero General Public License
  18  * along with this program.  If not, see <https://www.gnu.org/licenses/>.
  19  *
  20  */
  21
  22 namespace Friendica\Network;
  23
  24 use DOMDocument;
  25 use DomXPath;
  26 use Exception;
  27 use Friendica\Core\Hook;
  28 use Friendica\Core\Logger;
  29 use Friendica\Core\Protocol;
  30 use Friendica\Core\System;
  31 use Friendica\Database\DBA;
  32 use Friendica\DI;
  33 use Friendica\Model\Contact;
  34 use Friendica\Model\GServer;
  35 use Friendica\Model\Profile;
  36 use Friendica\Model\User;
  37 use Friendica\Protocol\ActivityNamespace;
  38 use Friendica\Protocol\ActivityPub;
  39 use Friendica\Protocol\Email;
  40 use Friendica\Protocol\Feed;
  41 use Friendica\Util\Crypto;
  42 use Friendica\Util\DateTimeFormat;
  43 use Friendica\Util\Network;
  44 use Friendica\Util\Strings;
  45 use Friendica\Util\XML;
  46
  47 /**
  48  * This class contain functions for probing URL
  49  */
  50 class Probe
  51 {
  52         const WEBFINGER = '/.well-known/webfinger?resource={uri}';
  53
  54         private static $baseurl;
  55         private static $istimeout;
  56
  57         /**
  58          * Remove stuff from an URI that doesn't belong there
  59          *
  60          * @param string $URI
  61          * @return string Cleaned URI
  62          */
  63         public static function cleanURI(string $URI)
  64         {
  65                 // At first remove leading and trailing junk
  66                 $URI = trim($URI, "@#?:/ \t\n\r\0\x0B");
  67
  68                 $parts = parse_url($URI);
  69
  70                 if (empty($parts['scheme'])) {
  71                         return $URI;
  72                 }
  73
  74                 // Remove the URL fragment, since these shouldn't be part of any profile URL
  75                 unset($parts['fragment']);
  76
  77                 $URI = Network::unparseURL($parts);
  78
  79                 return $URI;
  80         }
  81
  82         /**
  83          * Rearrange the array so that it always has the same order
  84          *
  85          * @param array $data Unordered data
  86          *
  87          * @return array Ordered data
  88          */
  89         private static function rearrangeData($data)
  90         {
  91                 $fields = ["name", "nick", "guid", "url", "addr", "alias", "photo", "header",
  92                                 "account-type", "community", "keywords", "location", "about", "hide",
  93                                 "batch", "notify", "poll", "request", "confirm", "subscribe", "poco",
  94                                 "following", "followers", "inbox", "outbox", "sharedinbox",
  95                                 "priority", "network", "pubkey", "manually-approve", "baseurl", "gsid"];
  96
  97                 $numeric_fields = ["gsid", "hide", "account-type", "manually-approve"];
  98
  99                 $newdata = [];
 100                 foreach ($fields as $field) {
 101                         if (isset($data[$field])) {
 102                                 if (in_array($field, $numeric_fields)) {
 103                                         $newdata[$field] = (int)$data[$field];
 104                                 } else {
 105                                         $newdata[$field] = $data[$field];
 106                                 }
 107                         } elseif (!in_array($field, $numeric_fields)) {
 108                                 $newdata[$field] = "";
 109                         } else {
 110                                 $newdata[$field] = null;
 111                         }
 112                 }
 113
 114                 // We don't use the "priority" field anymore and replace it with a dummy.
 115                 $newdata["priority"] = 0;
 116
 117                 return $newdata;
 118         }
 119
 120         /**
 121          * Check if the hostname belongs to the own server
 122          *
 123          * @param string $host The hostname that is to be checked
 124          *
 125          * @return bool Does the testes hostname belongs to the own server?
 126          */
 127         private static function ownHost($host)
 128         {
 129                 $own_host = DI::baseUrl()->getHostname();
 130
 131                 $parts = parse_url($host);
 132
 133                 if (!isset($parts['scheme'])) {
 134                         $parts = parse_url('http://'.$host);
 135                 }
 136
 137                 if (!isset($parts['host'])) {
 138                         return false;
 139                 }
 140                 return $parts['host'] == $own_host;
 141         }
 142
 143         /**
 144          * Probes for webfinger path via "host-meta"
 145          *
 146          * We have to check if the servers in the future still will offer this.
 147          * It seems as if it was dropped from the standard.
 148          *
 149          * @param string $host The host part of an url
 150          *
 151          * @return array with template and type of the webfinger template for JSON or XML
 152          * @throws HTTPException\InternalServerErrorException
 153          */
 154         private static function hostMeta($host)
 155         {
 156                 // Reset the static variable
 157                 self::$baseurl = '';
 158
 159                 // Handles the case when the hostname contains the scheme
 160                 if (!parse_url($host, PHP_URL_SCHEME)) {
 161                         $ssl_url = "https://" . $host . "/.well-known/host-meta";
 162                         $url = "http://" . $host . "/.well-known/host-meta";
 163                 } else {
 164                         $ssl_url = $host . "/.well-known/host-meta";
 165                         $url = '';
 166                 }
 167
 168                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 169
 170                 Logger::info('Probing', ['host' => $host, 'ssl_url' => $ssl_url, 'url' => $url, 'callstack' => System::callstack(20)]);
 171                 $xrd = null;
 172
 173                 $curlResult = DI::httpRequest()->get($ssl_url, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 174                 $ssl_connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 175                 if ($curlResult->isSuccess()) {
 176                         $xml = $curlResult->getBody();
 177                         $xrd = XML::parseString($xml, true);
 178                         if (!empty($url)) {
 179                                 $host_url = 'https://' . $host;
 180                         } else {
 181                                 $host_url = $host;
 182                         }
 183                 } elseif ($curlResult->isTimeout()) {
 184                         Logger::info('Probing timeout', ['url' => $ssl_url]);
 185                         self::$istimeout = true;
 186                         return [];
 187                 }
 188
 189                 if (!is_object($xrd) && !empty($url)) {
 190                         $curlResult = DI::httpRequest()->get($url, ['timeout' => $xrd_timeout, 'accept_content' => 'application/xrd+xml']);
 191                         $connection_error = ($curlResult->getErrorNumber() == CURLE_COULDNT_CONNECT) || ($curlResult->getReturnCode() == 0);
 192                         if ($curlResult->isTimeout()) {
 193                                 Logger::info('Probing timeout', ['url' => $url]);
 194                                 self::$istimeout = true;
 195                                 return [];
 196                         } elseif ($connection_error && $ssl_connection_error) {
 197                                 self::$istimeout = true;
 198                                 return [];
 199                         }
 200
 201                         $xml = $curlResult->getBody();
 202                         $xrd = XML::parseString($xml, true);
 203                         $host_url = 'http://'.$host;
 204                 }
 205                 if (!is_object($xrd)) {
 206                         Logger::info('No xrd object found', ['host' => $host]);
 207                         return [];
 208                 }
 209
 210                 $links = XML::elementToArray($xrd);
 211                 if (!isset($links["xrd"]["link"])) {
 212                         Logger::info('No xrd data found', ['host' => $host]);
 213                         return [];
 214                 }
 215
 216                 $lrdd = [];
 217
 218                 foreach ($links["xrd"]["link"] as $value => $link) {
 219                         if (!empty($link["@attributes"])) {
 220                                 $attributes = $link["@attributes"];
 221                         } elseif ($value == "@attributes") {
 222                                 $attributes = $link;
 223                         } else {
 224                                 continue;
 225                         }
 226
 227                         if (!empty($attributes["rel"]) && $attributes["rel"] == "lrdd" && !empty($attributes["template"])) {
 228                                 $type = (empty($attributes["type"]) ? '' : $attributes["type"]);
 229
 230                                 $lrdd[$type] = $attributes["template"];
 231                         }
 232                 }
 233
 234                 self::$baseurl = $host_url;
 235
 236                 Logger::info('Probing successful', ['host' => $host]);
 237
 238                 return $lrdd;
 239         }
 240
 241         /**
 242          * Perform Webfinger lookup and return DFRN data
 243          *
 244          * Given an email style address, perform webfinger lookup and
 245          * return the resulting DFRN profile URL, or if no DFRN profile URL
 246          * is located, returns an OStatus subscription template (prefixed
 247          * with the string 'stat:' to identify it as on OStatus template).
 248          * If this isn't an email style address just return $webbie.
 249          * Return an empty string if email-style addresses but webfinger fails,
 250          * or if the resultant personal XRD doesn't contain a supported
 251          * subscription/friend-request attribute.
 252          *
 253          * amended 7/9/2011 to return an hcard which could save potentially loading
 254          * a lengthy content page to scrape dfrn attributes
 255          *
 256          * @param string $webbie    Address that should be probed
 257          * @param string $hcard_url Link to the hcard - is returned by reference
 258          *
 259          * @return string profile link
 260          * @throws HTTPException\InternalServerErrorException
 261          */
 262         public static function webfingerDfrn(string $webbie, string &$hcard_url)
 263         {
 264                 $profile_link = '';
 265
 266                 $links = self::lrdd($webbie);
 267                 Logger::debug('Result', ['url' => $webbie, 'links' => $links]);
 268                 if (!empty($links) && is_array($links)) {
 269                         foreach ($links as $link) {
 270                                 if ($link['@attributes']['rel'] === ActivityNamespace::DFRN) {
 271                                         $profile_link = $link['@attributes']['href'];
 272                                 }
 273                                 if (($link['@attributes']['rel'] === ActivityNamespace::OSTATUSSUB) && ($profile_link == "")) {
 274                                         $profile_link = 'stat:'.$link['@attributes']['template'];
 275                                 }
 276                                 if ($link['@attributes']['rel'] === 'http://microformats.org/profile/hcard') {
 277                                         $hcard_url = $link['@attributes']['href'];
 278                                 }
 279                         }
 280                 }
 281                 return $profile_link;
 282         }
 283
 284         /**
 285          * Check an URI for LRDD data
 286          *
 287          * @param string $uri     Address that should be probed
 288          *
 289          * @return array uri data
 290          * @throws HTTPException\InternalServerErrorException
 291          */
 292         public static function lrdd(string $uri)
 293         {
 294                 $data = self::getWebfingerArray($uri);
 295                 if (empty($data)) {
 296                         return [];
 297                 }
 298                 $webfinger = $data['webfinger'];
 299
 300                 if (empty($webfinger["links"])) {
 301                         Logger::info('No webfinger links found', ['uri' => $uri]);
 302                         return [];
 303                 }
 304
 305                 $data = [];
 306
 307                 foreach ($webfinger["links"] as $link) {
 308                         $data[] = ["@attributes" => $link];
 309                 }
 310
 311                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 312                         foreach ($webfinger["aliases"] as $alias) {
 313                                 $data[] = ["@attributes" =>
 314                                                         ["rel" => "alias",
 315                                                                 "href" => $alias]];
 316                         }
 317                 }
 318
 319                 return $data;
 320         }
 321
 322         /**
 323          * Fetch information (protocol endpoints and user information) about a given uri
 324          *
 325          * @param string  $uri     Address that should be probed
 326          * @param string  $network Test for this specific network
 327          * @param integer $uid     User ID for the probe (only used for mails)
 328          * @param boolean $cache   Use cached values?
 329          *
 330          * @return array uri data
 331          * @throws HTTPException\InternalServerErrorException
 332          * @throws \ImagickException
 333          */
 334         public static function uri($uri, $network = '', $uid = -1)
 335         {
 336                 // Local profiles aren't probed via network
 337                 if (empty($network) && Contact::isLocal($uri)) {
 338                         $data = self::localProbe($uri);
 339                         if (!empty($data)) {
 340                                 return $data;
 341                         }
 342                 }
 343
 344                 if ($uid == -1) {
 345                         $uid = local_user();
 346                 }
 347
 348                 if (empty($network) || ($network == Protocol::ACTIVITYPUB)) {
 349                         $ap_profile = ActivityPub::probeProfile($uri);
 350                 } else {
 351                         $ap_profile = [];
 352                 }
 353
 354                 self::$istimeout = false;
 355
 356                 if ($network != Protocol::ACTIVITYPUB) {
 357                         $data = self::detect($uri, $network, $uid, $ap_profile);
 358                         if (!is_array($data)) {
 359                                 $data = [];
 360                         }
 361                         if (empty($data) || (!empty($ap_profile) && empty($network) && (($data['network'] ?? '') != Protocol::DFRN))) {
 362                                 $data = $ap_profile;
 363                         } elseif (!empty($ap_profile)) {
 364                                 $ap_profile['batch'] = '';
 365                                 $data = array_merge($ap_profile, $data);
 366                         }
 367                 } else {
 368                         $data = $ap_profile;
 369                 }
 370
 371                 if (!isset($data['url'])) {
 372                         $data['url'] = $uri;
 373                 }
 374
 375                 if (empty($data['photo'])) {
 376                         $data['photo'] = DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO;
 377                 }
 378
 379                 if (empty($data['name'])) {
 380                         if (!empty($data['nick'])) {
 381                                 $data['name'] = $data['nick'];
 382                         }
 383
 384                         if (empty($data['name'])) {
 385                                 $data['name'] = $data['url'];
 386                         }
 387                 }
 388
 389                 if (empty($data['nick'])) {
 390                         $data['nick'] = strtolower($data['name']);
 391
 392                         if (strpos($data['nick'], ' ')) {
 393                                 $data['nick'] = trim(substr($data['nick'], 0, strpos($data['nick'], ' ')));
 394                         }
 395                 }
 396
 397                 if (!empty($data['baseurl']) && empty($data['gsid'])) {
 398                         $data['gsid'] = GServer::getID($data['baseurl']);
 399                 }
 400
 401                 if (empty($data['network'])) {
 402                         $data['network'] = Protocol::PHANTOM;
 403                 }
 404
 405                 // Ensure that local connections always are DFRN
 406                 if (($network == '') && ($data['network'] != Protocol::PHANTOM) && (self::ownHost($data['baseurl'] ?? '') || self::ownHost($data['url']))) {
 407                         $data['network'] = Protocol::DFRN;
 408                 }
 409
 410                 if (!isset($data['hide']) && in_array($data['network'], Protocol::FEDERATED)) {
 411                         $data['hide'] = self::getHideStatus($data['url']);
 412                 }
 413
 414                 return self::rearrangeData($data);
 415         }
 416
 417
 418         /**
 419          * Fetches the "hide" status from the profile
 420          *
 421          * @param string $url URL of the profile
 422          *
 423          * @return boolean "hide" status
 424          */
 425         private static function getHideStatus($url)
 426         {
 427                 $curlResult = DI::httpRequest()->get($url);
 428                 if (!$curlResult->isSuccess()) {
 429                         return false;
 430                 }
 431
 432                 // If the file is too large then exit
 433                 if (($curlResult->getInfo()['download_content_length'] ?? 0) > 1000000) {
 434                         return false;
 435                 }
 436
 437                 // If it isn't a HTML file then exit
 438                 if (($curlResult->getContentType() != '') && !strstr(strtolower($curlResult->getContentType()), 'html')) {
 439                         return false;
 440                 }
 441
 442                 $body = $curlResult->getBody();
 443                 if (empty($body)) {
 444                         return false;
 445                 }
 446
 447                 $doc = new DOMDocument();
 448                 @$doc->loadHTML($body);
 449
 450                 $xpath = new DOMXPath($doc);
 451
 452                 $list = $xpath->query('//meta[@name]');
 453                 foreach ($list as $node) {
 454                         $meta_tag = [];
 455                         if ($node->attributes->length) {
 456                                 foreach ($node->attributes as $attribute) {
 457                                         $meta_tag[$attribute->name] = $attribute->value;
 458                                 }
 459                         }
 460
 461                         if (empty($meta_tag['content'])) {
 462                                 continue;
 463                         }
 464
 465                         $content = strtolower(trim($meta_tag['content']));
 466
 467                         switch (strtolower(trim($meta_tag['name']))) {
 468                                 case 'dfrn-global-visibility':
 469                                         if ($content == 'false') {
 470                                                 return true;
 471                                         }
 472                                         break;
 473                                 case 'robots':
 474                                         if (strpos($content, 'noindex') !== false) {
 475                                                 return true;
 476                                         }
 477                                         break;
 478                         }
 479                 }
 480
 481                 return false;
 482         }
 483
 484         /**
 485          * Fetch the "subscribe" and add it to the result
 486          *
 487          * @param array $result
 488          * @param array $webfinger
 489          * @return array result
 490          */
 491         private static function getSubscribeLink(array $result, array $webfinger)
 492         {
 493                 if (empty($webfinger['links'])) {
 494                         return $result;
 495                 }
 496
 497                 foreach ($webfinger['links'] as $link) {
 498                         if (!empty($link['template']) && ($link['rel'] === ActivityNamespace::OSTATUSSUB)) {
 499                                 $result['subscribe'] = $link['template'];
 500                         }
 501                 }
 502
 503                 return $result;
 504         }
 505
 506         /**
 507          * Get webfinger data from a given URI
 508          *
 509          * @param string $uri
 510          * @return array Webfinger array
 511          */
 512         private static function getWebfingerArray(string $uri)
 513         {
 514                 $parts = parse_url($uri);
 515
 516                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 517                         $host = $parts['host'];
 518                         if (!empty($parts['port'])) {
 519                                 $host .= ':'.$parts['port'];
 520                         }
 521
 522                         $baseurl = $parts['scheme'] . '://' . $host;
 523
 524                         $nick = '';
 525                         $addr = '';
 526
 527                         $path_parts = explode("/", trim($parts['path'] ?? '', "/"));
 528                         if (!empty($path_parts)) {
 529                                 $nick = ltrim(end($path_parts), '@');
 530                                 // When the last part of the URI is numeric then it is most likely an ID and not a nick name
 531                                 if (!is_numeric($nick)) {
 532                                         $addr = $nick."@".$host;
 533                                 } else {
 534                                         $nick = '';
 535                                 }
 536                         }
 537
 538                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 539                         if (empty($webfinger)) {
 540                                 $lrdd = self::hostMeta($host);
 541                         }
 542
 543                         if (empty($webfinger) && empty($lrdd)) {
 544                                 while (empty($lrdd) && empty($webfinger) && (sizeof($path_parts) > 1)) {
 545                                         $host .= "/".array_shift($path_parts);
 546                                         $baseurl = $parts['scheme'] . '://' . $host;
 547
 548                                         if (!empty($nick)) {
 549                                                 $addr = $nick."@".$host;
 550                                         }
 551
 552                                         $webfinger = self::getWebfinger($parts['scheme'] . '://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 553                                         if (empty($webfinger)) {
 554                                                 $lrdd = self::hostMeta($host);
 555                                         }
 556                                 }
 557
 558                                 if (empty($lrdd) && empty($webfinger)) {
 559                                         return [];
 560                                 }
 561                         }
 562                 } elseif (strstr($uri, '@')) {
 563                         // Remove "acct:" from the URI
 564                         $uri = str_replace('acct:', '', $uri);
 565
 566                         $host = substr($uri, strpos($uri, '@') + 1);
 567                         $nick = substr($uri, 0, strpos($uri, '@'));
 568                         $addr = $uri;
 569
 570                         $webfinger = self::getWebfinger('https://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 571                         if (self::$istimeout) {
 572                                 return [];
 573                         }
 574
 575                         if (empty($webfinger)) {
 576                                 $webfinger = self::getWebfinger('http://' . $host . self::WEBFINGER, 'application/jrd+json', $uri, $addr);
 577                                 if (self::$istimeout) {
 578                                         return [];
 579                                 }
 580                         } else {
 581                                 $baseurl = 'https://' . $host;
 582                         }
 583
 584                         if (empty($webfinger)) {
 585                                 $lrdd = self::hostMeta($host);
 586                                 if (self::$istimeout) {
 587                                         return [];
 588                                 }
 589                                 $baseurl = self::$baseurl;
 590                         } else {
 591                                 $baseurl = 'http://' . $host;
 592                         }
 593                 } else {
 594                         Logger::info('URI was not detectable', ['uri' => $uri]);
 595                         return [];
 596                 }
 597
 598                 if (empty($webfinger)) {
 599                         foreach ($lrdd as $type => $template) {
 600                                 if ($webfinger) {
 601                                         continue;
 602                                 }
 603
 604                                 $webfinger = self::getWebfinger($template, $type, $uri, $addr);
 605                         }
 606                 }
 607
 608                 if (empty($webfinger)) {
 609                         return [];
 610                 }
 611
 612                 if ($webfinger['detected'] == $addr) {
 613                         $webfinger['nick'] = $nick;
 614                         $webfinger['addr'] = $addr;
 615                 }
 616
 617                 $webfinger['baseurl'] = $baseurl;
 618
 619                 return $webfinger;
 620         }
 621
 622         /**
 623          * Perform network request for webfinger data
 624          *
 625          * @param string $template
 626          * @param string $type
 627          * @param string $uri
 628          * @param string $addr
 629          * @return array webfinger results
 630          */
 631         private static function getWebfinger(string $template, string $type, string $uri, string $addr)
 632         {
 633                 // First try the address because this is the primary purpose of webfinger
 634                 if (!empty($addr)) {
 635                         $detected = $addr;
 636                         $path = str_replace('{uri}', urlencode("acct:" . $addr), $template);
 637                         $webfinger = self::webfinger($path, $type);
 638                         if (self::$istimeout) {
 639                                 return [];
 640                         }
 641                 }
 642
 643                 // Then try the URI
 644                 if (empty($webfinger) && $uri != $addr) {
 645                         $detected = $uri;
 646                         $path = str_replace('{uri}', urlencode($uri), $template);
 647                         $webfinger = self::webfinger($path, $type);
 648                         if (self::$istimeout) {
 649                                 return [];
 650                         }
 651                 }
 652
 653                 if (empty($webfinger)) {
 654                         return [];
 655                 }
 656
 657                 return ['webfinger' => $webfinger, 'detected' => $detected];
 658         }
 659
 660         /**
 661          * Fetch information (protocol endpoints and user information) about a given uri
 662          *
 663          * This function is only called by the "uri" function that adds caching and rearranging of data.
 664          *
 665          * @param string  $uri        Address that should be probed
 666          * @param string  $network    Test for this specific network
 667          * @param integer $uid        User ID for the probe (only used for mails)
 668          * @param array   $ap_profile Previously probed AP profile
 669          *
 670          * @return array uri data
 671          * @throws HTTPException\InternalServerErrorException
 672          */
 673         private static function detect(string $uri, string $network, int $uid, array $ap_profile)
 674         {
 675                 $hookData = [
 676                         'uri'     => $uri,
 677                         'network' => $network,
 678                         'uid'     => $uid,
 679                         'result'  => [],
 680                 ];
 681
 682                 Hook::callAll('probe_detect', $hookData);
 683
 684                 if ($hookData['result']) {
 685                         if (!is_array($hookData['result'])) {
 686                                 return [];
 687                         } else {
 688                                 return $hookData['result'];
 689                         }
 690                 }
 691
 692                 $parts = parse_url($uri);
 693
 694                 if (!empty($parts['scheme']) && !empty($parts['host'])) {
 695                         if (in_array($parts['host'], ['twitter.com', 'mobile.twitter.com'])) {
 696                                 return self::twitter($uri);
 697                         }
 698                 } elseif (strstr($uri, '@')) {
 699                         // If the URI starts with "mailto:" then jump directly to the mail detection
 700                         if (strpos($uri, 'mailto:') !== false) {
 701                                 $uri = str_replace('mailto:', '', $uri);
 702                                 return self::mail($uri, $uid);
 703                         }
 704
 705                         if ($network == Protocol::MAIL) {
 706                                 return self::mail($uri, $uid);
 707                         }
 708
 709                         if (Strings::endsWith($uri, '@twitter.com')
 710                                 || Strings::endsWith($uri, '@mobile.twitter.com')
 711                         ) {
 712                                 return self::twitter($uri);
 713                         }
 714                 } else {
 715                         Logger::info('URI was not detectable', ['uri' => $uri]);
 716                         return [];
 717                 }
 718
 719                 Logger::info('Probing start', ['uri' => $uri]);
 720
 721                 if (!empty($ap_profile['addr']) && ($ap_profile['addr'] != $uri)) {
 722                         $data = self::getWebfingerArray($ap_profile['addr']);
 723                 }
 724
 725                 if (empty($data)) {
 726                         $data = self::getWebfingerArray($uri);
 727                 }
 728
 729                 if (empty($data)) {
 730                         if (!empty($parts['scheme'])) {
 731                                 return self::feed($uri);
 732                         } elseif (!empty($uid)) {
 733                                 return self::mail($uri, $uid);
 734                         } else {
 735                                 return [];
 736                         }
 737                 }
 738
 739                 $webfinger = $data['webfinger'];
 740                 $nick = $data['nick'] ?? '';
 741                 $addr = $data['addr'] ?? '';
 742                 $baseurl = $data['baseurl'] ?? '';
 743
 744                 $result = [];
 745
 746                 if (in_array($network, ["", Protocol::DFRN])) {
 747                         $result = self::dfrn($webfinger);
 748                 }
 749                 if ((!$result && ($network == "")) || ($network == Protocol::DIASPORA)) {
 750                         $result = self::diaspora($webfinger);
 751                 }
 752                 if ((!$result && ($network == "")) || ($network == Protocol::OSTATUS)) {
 753                         $result = self::ostatus($webfinger);
 754                 }
 755                 if (in_array($network, ['', Protocol::ZOT])) {
 756                         $result = self::zot($webfinger, $result, $baseurl);
 757                 }
 758                 if ((!$result && ($network == "")) || ($network == Protocol::PUMPIO)) {
 759                         $result = self::pumpio($webfinger, $addr);
 760                 }
 761                 if (empty($result['network']) && empty($ap_profile['network']) || ($network == Protocol::FEED)) {
 762                         $result = self::feed($uri);
 763                 } else {
 764                         // We overwrite the detected nick with our try if the previois routines hadn't detected it.
 765                         // Additionally it is overwritten when the nickname doesn't make sense (contains spaces).
 766                         if ((empty($result["nick"]) || (strstr($result["nick"], " "))) && ($nick != "")) {
 767                                 $result["nick"] = $nick;
 768                         }
 769
 770                         if (empty($result["addr"]) && ($addr != "")) {
 771                                 $result["addr"] = $addr;
 772                         }
 773                 }
 774
 775                 $result = self::getSubscribeLink($result, $webfinger);
 776
 777                 if (empty($result["network"])) {
 778                         $result["network"] = Protocol::PHANTOM;
 779                 }
 780
 781                 if (empty($result['baseurl']) && !empty($baseurl)) {
 782                         $result['baseurl'] = $baseurl;
 783                 }
 784
 785                 if (empty($result["url"])) {
 786                         $result["url"] = $uri;
 787                 }
 788
 789                 Logger::info('Probing done', ['uri' => $uri, 'network' => $result["network"]]);
 790
 791                 return $result;
 792         }
 793
 794         /**
 795          * Check for Zot contact
 796          *
 797          * @param array $webfinger Webfinger data
 798          * @param array $data      previously probed data
 799          *
 800          * @return array Zot data
 801          * @throws HTTPException\InternalServerErrorException
 802          */
 803         private static function zot($webfinger, $data, $baseurl)
 804         {
 805                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
 806                         foreach ($webfinger["aliases"] as $alias) {
 807                                 if (substr($alias, 0, 5) == 'acct:') {
 808                                         $data["addr"] = substr($alias, 5);
 809                                 }
 810                         }
 811                 }
 812
 813                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
 814                         $data["addr"] = substr($webfinger["subject"], 5);
 815                 }
 816
 817                 $zot_url = '';
 818                 foreach ($webfinger['links'] as $link) {
 819                         if (($link['rel'] == 'http://purl.org/zot/protocol') && !empty($link['href'])) {
 820                                 $zot_url = $link['href'];
 821                         }
 822                 }
 823
 824                 if (empty($zot_url) && !empty($data['addr']) && !empty($baseurl)) {
 825                         $condition = ['nurl' => Strings::normaliseLink($baseurl), 'platform' => ['hubzilla']];
 826                         if (!DBA::exists('gserver', $condition)) {
 827                                 return $data;
 828                         }
 829                         $zot_url = $baseurl . '/.well-known/zot-info?address=' . $data['addr'];
 830                 }
 831
 832                 if (empty($zot_url)) {
 833                         return $data;
 834                 }
 835
 836                 $data = self::pollZot($zot_url, $data);
 837
 838                 if (!empty($data['url']) && !empty($webfinger['aliases']) && is_array($webfinger['aliases'])) {
 839                         foreach ($webfinger['aliases'] as $alias) {
 840                                 if (!strstr($alias, '@') && Strings::normaliseLink($alias) != Strings::normaliseLink($data['url'])) {
 841                                         $data['alias'] = $alias;
 842                                 }
 843                         }
 844                 }
 845
 846                 return $data;
 847         }
 848
 849         public static function pollZot($url, $data)
 850         {
 851                 $curlResult = DI::httpRequest()->get($url);
 852                 if ($curlResult->isTimeout()) {
 853                         return $data;
 854                 }
 855                 $content = $curlResult->getBody();
 856                 if (!$content) {
 857                         return $data;
 858                 }
 859
 860                 $json = json_decode($content, true);
 861                 if (!is_array($json)) {
 862                         return $data;
 863                 }
 864
 865                 if (empty($data['network'])) {
 866                         if (!empty($json['protocols']) && in_array('zot', $json['protocols'])) {
 867                                 $data['network'] = Protocol::ZOT;
 868                         } elseif (!isset($json['protocols'])) {
 869                                 $data['network'] = Protocol::ZOT;
 870                         }
 871                 }
 872
 873                 if (!empty($json['guid']) && empty($data['guid'])) {
 874                         $data['guid'] = $json['guid'];
 875                 }
 876                 if (!empty($json['key']) && empty($data['pubkey'])) {
 877                         $data['pubkey'] = $json['key'];
 878                 }
 879                 if (!empty($json['name'])) {
 880                         $data['name'] = $json['name'];
 881                 }
 882                 if (!empty($json['photo'])) {
 883                         $data['photo'] = $json['photo'];
 884                         if (!empty($json['photo_updated'])) {
 885                                 $data['photo'] .= '?rev=' . urlencode($json['photo_updated']);
 886                         }
 887                 }
 888                 if (!empty($json['address'])) {
 889                         $data['addr'] = $json['address'];
 890                 }
 891                 if (!empty($json['url'])) {
 892                         $data['url'] = $json['url'];
 893                 }
 894                 if (!empty($json['connections_url'])) {
 895                         $data['poco'] = $json['connections_url'];
 896                 }
 897                 if (isset($json['searchable'])) {
 898                         $data['hide'] = !$json['searchable'];
 899                 }
 900                 if (!empty($json['public_forum'])) {
 901                         $data['community'] = $json['public_forum'];
 902                         $data['account-type'] = User::PAGE_FLAGS_COMMUNITY;
 903                 }
 904
 905                 if (!empty($json['profile'])) {
 906                         $profile = $json['profile'];
 907                         if (!empty($profile['description'])) {
 908                                 $data['about'] = $profile['description'];
 909                         }
 910                         if (!empty($profile['keywords'])) {
 911                                 $keywords = implode(', ', $profile['keywords']);
 912                                 if (!empty($keywords)) {
 913                                         $data['keywords'] = $keywords;
 914                                 }
 915                         }
 916
 917                         $loc = [];
 918                         if (!empty($profile['region'])) {
 919                                 $loc['region'] = $profile['region'];
 920                         }
 921                         if (!empty($profile['country'])) {
 922                                 $loc['country-name'] = $profile['country'];
 923                         }
 924                         $location = Profile::formatLocation($loc);
 925                         if (!empty($location)) {
 926                                 $data['location'] = $location;
 927                         }
 928                 }
 929
 930                 return $data;
 931         }
 932
 933         /**
 934          * Perform a webfinger request.
 935          *
 936          * For details see RFC 7033: <https://tools.ietf.org/html/rfc7033>
 937          *
 938          * @param string $url  Address that should be probed
 939          * @param string $type type
 940          *
 941          * @return array webfinger data
 942          * @throws HTTPException\InternalServerErrorException
 943          */
 944         public static function webfinger($url, $type)
 945         {
 946                 $xrd_timeout = DI::config()->get('system', 'xrd_timeout', 20);
 947
 948                 $curlResult = DI::httpRequest()->get($url, ['timeout' => $xrd_timeout, 'accept_content' => $type]);
 949                 if ($curlResult->isTimeout()) {
 950                         self::$istimeout = true;
 951                         return [];
 952                 }
 953                 $data = $curlResult->getBody();
 954
 955                 $webfinger = json_decode($data, true);
 956                 if (!empty($webfinger)) {
 957                         if (!isset($webfinger["links"])) {
 958                                 Logger::info('No json webfinger links', ['url' => $url]);
 959                                 return [];
 960                         }
 961                         return $webfinger;
 962                 }
 963
 964                 // If it is not JSON, maybe it is XML
 965                 $xrd = XML::parseString($data, true);
 966                 if (!is_object($xrd)) {
 967                         Logger::info('No webfinger data retrievable', ['url' => $url]);
 968                         return [];
 969                 }
 970
 971                 $xrd_arr = XML::elementToArray($xrd);
 972                 if (!isset($xrd_arr["xrd"]["link"])) {
 973                         Logger::info('No XML webfinger links', ['url' => $url]);
 974                         return [];
 975                 }
 976
 977                 $webfinger = [];
 978
 979                 if (!empty($xrd_arr["xrd"]["subject"])) {
 980                         $webfinger["subject"] = $xrd_arr["xrd"]["subject"];
 981                 }
 982
 983                 if (!empty($xrd_arr["xrd"]["alias"])) {
 984                         $webfinger["aliases"] = $xrd_arr["xrd"]["alias"];
 985                 }
 986
 987                 $webfinger["links"] = [];
 988
 989                 foreach ($xrd_arr["xrd"]["link"] as $value => $data) {
 990                         if (!empty($data["@attributes"])) {
 991                                 $attributes = $data["@attributes"];
 992                         } elseif ($value == "@attributes") {
 993                                 $attributes = $data;
 994                         } else {
 995                                 continue;
 996                         }
 997
 998                         $webfinger["links"][] = $attributes;
 999                 }
1000                 return $webfinger;
1001         }
1002
1003         /**
1004          * Poll the Friendica specific noscrape page.
1005          *
1006          * "noscrape" is a faster alternative to fetch the data from the hcard.
1007          * This functionality was originally created for the directory.
1008          *
1009          * @param string $noscrape_url Link to the noscrape page
1010          * @param array  $data         The already fetched data
1011          *
1012          * @return array noscrape data
1013          * @throws HTTPException\InternalServerErrorException
1014          */
1015         private static function pollNoscrape($noscrape_url, $data)
1016         {
1017                 $curlResult = DI::httpRequest()->get($noscrape_url);
1018                 if ($curlResult->isTimeout()) {
1019                         self::$istimeout = true;
1020                         return $data;
1021                 }
1022                 $content = $curlResult->getBody();
1023                 if (!$content) {
1024                         Logger::info('Empty body', ['url' => $noscrape_url]);
1025                         return $data;
1026                 }
1027
1028                 $json = json_decode($content, true);
1029                 if (!is_array($json)) {
1030                         Logger::info('No json data', ['url' => $noscrape_url]);
1031                         return $data;
1032                 }
1033
1034                 if (!empty($json["fn"])) {
1035                         $data["name"] = $json["fn"];
1036                 }
1037
1038                 if (!empty($json["addr"])) {
1039                         $data["addr"] = $json["addr"];
1040                 }
1041
1042                 if (!empty($json["nick"])) {
1043                         $data["nick"] = $json["nick"];
1044                 }
1045
1046                 if (!empty($json["guid"])) {
1047                         $data["guid"] = $json["guid"];
1048                 }
1049
1050                 if (!empty($json["comm"])) {
1051                         $data["community"] = $json["comm"];
1052                 }
1053
1054                 if (!empty($json["tags"])) {
1055                         $keywords = implode(", ", $json["tags"]);
1056                         if ($keywords != "") {
1057                                 $data["keywords"] = $keywords;
1058                         }
1059                 }
1060
1061                 $location = Profile::formatLocation($json);
1062                 if ($location) {
1063                         $data["location"] = $location;
1064                 }
1065
1066                 if (!empty($json["about"])) {
1067                         $data["about"] = $json["about"];
1068                 }
1069
1070                 if (!empty($json["key"])) {
1071                         $data["pubkey"] = $json["key"];
1072                 }
1073
1074                 if (!empty($json["photo"])) {
1075                         $data["photo"] = $json["photo"];
1076                 }
1077
1078                 if (!empty($json["dfrn-request"])) {
1079                         $data["request"] = $json["dfrn-request"];
1080                 }
1081
1082                 if (!empty($json["dfrn-confirm"])) {
1083                         $data["confirm"] = $json["dfrn-confirm"];
1084                 }
1085
1086                 if (!empty($json["dfrn-notify"])) {
1087                         $data["notify"] = $json["dfrn-notify"];
1088                 }
1089
1090                 if (!empty($json["dfrn-poll"])) {
1091                         $data["poll"] = $json["dfrn-poll"];
1092                 }
1093
1094                 if (isset($json["hide"])) {
1095                         $data["hide"] = (bool)$json["hide"];
1096                 } else {
1097                         $data["hide"] = false;
1098                 }
1099
1100                 return $data;
1101         }
1102
1103         /**
1104          * Check for valid DFRN data
1105          *
1106          * @param array $data DFRN data
1107          *
1108          * @return int Number of errors
1109          */
1110         public static function validDfrn($data)
1111         {
1112                 $errors = 0;
1113                 if (!isset($data['key'])) {
1114                         $errors ++;
1115                 }
1116                 if (!isset($data['dfrn-request'])) {
1117                         $errors ++;
1118                 }
1119                 if (!isset($data['dfrn-confirm'])) {
1120                         $errors ++;
1121                 }
1122                 if (!isset($data['dfrn-notify'])) {
1123                         $errors ++;
1124                 }
1125                 if (!isset($data['dfrn-poll'])) {
1126                         $errors ++;
1127                 }
1128                 return $errors;
1129         }
1130
1131         /**
1132          * Fetch data from a DFRN profile page and via "noscrape"
1133          *
1134          * @param string $profile_link Link to the profile page
1135          *
1136          * @return array profile data
1137          * @throws HTTPException\InternalServerErrorException
1138          * @throws \ImagickException
1139          */
1140         public static function profile($profile_link)
1141         {
1142                 $data = [];
1143
1144                 Logger::info('Check profile', ['link' => $profile_link]);
1145
1146                 // Fetch data via noscrape - this is faster
1147                 $noscrape_url = str_replace(["/hcard/", "/profile/"], "/noscrape/", $profile_link);
1148                 $data = self::pollNoscrape($noscrape_url, $data);
1149
1150                 if (!isset($data["notify"])
1151                         || !isset($data["confirm"])
1152                         || !isset($data["request"])
1153                         || !isset($data["poll"])
1154                         || !isset($data["name"])
1155                         || !isset($data["photo"])
1156                 ) {
1157                         $data = self::pollHcard($profile_link, $data, true);
1158                 }
1159
1160                 $prof_data = [];
1161
1162                 if (empty($data["addr"]) || empty($data["nick"])) {
1163                         $probe_data = self::uri($profile_link);
1164                         $data["addr"] = ($data["addr"] ?? '') ?: $probe_data["addr"];
1165                         $data["nick"] = ($data["nick"] ?? '') ?: $probe_data["nick"];
1166                 }
1167
1168                 $prof_data["addr"]         = $data["addr"];
1169                 $prof_data["nick"]         = $data["nick"];
1170                 $prof_data["dfrn-request"] = $data['request'] ?? null;
1171                 $prof_data["dfrn-confirm"] = $data['confirm'] ?? null;
1172                 $prof_data["dfrn-notify"]  = $data['notify']  ?? null;
1173                 $prof_data["dfrn-poll"]    = $data['poll']    ?? null;
1174                 $prof_data["photo"]        = $data['photo']   ?? null;
1175                 $prof_data["fn"]           = $data['name']    ?? null;
1176                 $prof_data["key"]          = $data['pubkey']  ?? null;
1177
1178                 Logger::debug('Result', ['link' => $profile_link, 'data' => $prof_data]);
1179
1180                 return $prof_data;
1181         }
1182
1183         /**
1184          * Check for DFRN contact
1185          *
1186          * @param array $webfinger Webfinger data
1187          *
1188          * @return array DFRN data
1189          * @throws HTTPException\InternalServerErrorException
1190          */
1191         private static function dfrn($webfinger)
1192         {
1193                 $hcard_url = "";
1194                 $data = [];
1195                 // The array is reversed to take into account the order of preference for same-rel links
1196                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1197                 foreach (array_reverse($webfinger["links"]) as $link) {
1198                         if (($link["rel"] == ActivityNamespace::DFRN) && !empty($link["href"])) {
1199                                 $data["network"] = Protocol::DFRN;
1200                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1201                                 $data["poll"] = $link["href"];
1202                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1203                                 $data["url"] = $link["href"];
1204                         } elseif (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1205                                 $hcard_url = $link["href"];
1206                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1207                                 $data["poco"] = $link["href"];
1208                         } elseif (($link["rel"] == "http://webfinger.net/rel/avatar") && !empty($link["href"])) {
1209                                 $data["photo"] = $link["href"];
1210                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1211                                 $data["baseurl"] = trim($link["href"], '/');
1212                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1213                                 $data["guid"] = $link["href"];
1214                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1215                                 $data["pubkey"] = base64_decode($link["href"]);
1216
1217                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1218                                 if (strstr($data["pubkey"], 'RSA ')) {
1219                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1220                                 }
1221                         }
1222                 }
1223
1224                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1225                         foreach ($webfinger["aliases"] as $alias) {
1226                                 if (empty($data["url"]) && !strstr($alias, "@")) {
1227                                         $data["url"] = $alias;
1228                                 } elseif (!strstr($alias, "@") && Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"])) {
1229                                         $data["alias"] = $alias;
1230                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1231                                         $data["addr"] = substr($alias, 5);
1232                                 }
1233                         }
1234                 }
1235
1236                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == "acct:")) {
1237                         $data["addr"] = substr($webfinger["subject"], 5);
1238                 }
1239
1240                 if (!isset($data["network"]) || ($hcard_url == "")) {
1241                         return [];
1242                 }
1243
1244                 // Fetch data via noscrape - this is faster
1245                 $noscrape_url = str_replace("/hcard/", "/noscrape/", $hcard_url);
1246                 $data = self::pollNoscrape($noscrape_url, $data);
1247
1248                 if (isset($data["notify"])
1249                         && isset($data["confirm"])
1250                         && isset($data["request"])
1251                         && isset($data["poll"])
1252                         && isset($data["name"])
1253                         && isset($data["photo"])
1254                 ) {
1255                         return $data;
1256                 }
1257
1258                 $data = self::pollHcard($hcard_url, $data, true);
1259
1260                 return $data;
1261         }
1262
1263         /**
1264          * Poll the hcard page (Diaspora and Friendica specific)
1265          *
1266          * @param string  $hcard_url Link to the hcard page
1267          * @param array   $data      The already fetched data
1268          * @param boolean $dfrn      Poll DFRN specific data
1269          *
1270          * @return array hcard data
1271          * @throws HTTPException\InternalServerErrorException
1272          */
1273         private static function pollHcard($hcard_url, $data, $dfrn = false)
1274         {
1275                 $curlResult = DI::httpRequest()->get($hcard_url);
1276                 if ($curlResult->isTimeout()) {
1277                         self::$istimeout = true;
1278                         return [];
1279                 }
1280                 $content = $curlResult->getBody();
1281                 if (empty($content)) {
1282                         return [];
1283                 }
1284
1285                 $doc = new DOMDocument();
1286                 if (!@$doc->loadHTML($content)) {
1287                         return [];
1288                 }
1289
1290                 $xpath = new DomXPath($doc);
1291
1292                 $vcards = $xpath->query("//div[contains(concat(' ', @class, ' '), ' vcard ')]");
1293                 if (!is_object($vcards)) {
1294                         return [];
1295                 }
1296
1297                 if (!isset($data["baseurl"])) {
1298                         $data["baseurl"] = "";
1299                 }
1300
1301                 if ($vcards->length > 0) {
1302                         $vcard = $vcards->item(0);
1303
1304                         // We have to discard the guid from the hcard in favour of the guid from lrdd
1305                         // Reason: Hubzilla doesn't use the value "uid" in the hcard like Diaspora does.
1306                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' uid ')]", $vcard); // */
1307                         if (($search->length > 0) && empty($data["guid"])) {
1308                                 $data["guid"] = $search->item(0)->nodeValue;
1309                         }
1310
1311                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' nickname ')]", $vcard); // */
1312                         if ($search->length > 0) {
1313                                 $data["nick"] = $search->item(0)->nodeValue;
1314                         }
1315
1316                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' fn ')]", $vcard); // */
1317                         if ($search->length > 0) {
1318                                 $data["name"] = $search->item(0)->nodeValue;
1319                         }
1320
1321                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' searchable ')]", $vcard); // */
1322                         if ($search->length > 0) {
1323                                 $data["searchable"] = $search->item(0)->nodeValue;
1324                         }
1325
1326                         $search = $xpath->query("//*[contains(concat(' ', @class, ' '), ' key ')]", $vcard); // */
1327                         if ($search->length > 0) {
1328                                 $data["pubkey"] = $search->item(0)->nodeValue;
1329                                 if (strstr($data["pubkey"], 'RSA ')) {
1330                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1331                                 }
1332                         }
1333
1334                         $search = $xpath->query("//*[@id='pod_location']", $vcard); // */
1335                         if ($search->length > 0) {
1336                                 $data["baseurl"] = trim($search->item(0)->nodeValue, "/");
1337                         }
1338                 }
1339
1340                 $avatar = [];
1341                 if (!empty($vcard)) {
1342                         $photos = $xpath->query("//*[contains(concat(' ', @class, ' '), ' photo ') or contains(concat(' ', @class, ' '), ' avatar ')]", $vcard); // */
1343                         foreach ($photos as $photo) {
1344                                 $attr = [];
1345                                 foreach ($photo->attributes as $attribute) {
1346                                         $attr[$attribute->name] = trim($attribute->value);
1347                                 }
1348
1349                                 if (isset($attr["src"]) && isset($attr["width"])) {
1350                                         $avatar[$attr["width"]] = $attr["src"];
1351                                 }
1352
1353                                 // We don't have a width. So we just take everything that we got.
1354                                 // This is a Hubzilla workaround which doesn't send a width.
1355                                 if ((sizeof($avatar) == 0) && !empty($attr["src"])) {
1356                                         $avatar[] = $attr["src"];
1357                                 }
1358                         }
1359                 }
1360
1361                 if (sizeof($avatar)) {
1362                         ksort($avatar);
1363                         $data["photo"] = self::fixAvatar(array_pop($avatar), $data["baseurl"]);
1364                 }
1365
1366                 if ($dfrn) {
1367                         // Poll DFRN specific data
1368                         $search = $xpath->query("//link[contains(concat(' ', @rel), ' dfrn-')]");
1369                         if ($search->length > 0) {
1370                                 foreach ($search as $link) {
1371                                         //$data["request"] = $search->item(0)->nodeValue;
1372                                         $attr = [];
1373                                         foreach ($link->attributes as $attribute) {
1374                                                 $attr[$attribute->name] = trim($attribute->value);
1375                                         }
1376
1377                                         $data[substr($attr["rel"], 5)] = $attr["href"];
1378                                 }
1379                         }
1380
1381                         // Older Friendica versions had used the "uid" field differently than newer versions
1382                         if (!empty($data["nick"]) && !empty($data["guid"]) && ($data["nick"] == $data["guid"])) {
1383                                 unset($data["guid"]);
1384                         }
1385                 }
1386
1387
1388                 return $data;
1389         }
1390
1391         /**
1392          * Check for Diaspora contact
1393          *
1394          * @param array $webfinger Webfinger data
1395          *
1396          * @return array Diaspora data
1397          * @throws HTTPException\InternalServerErrorException
1398          */
1399         private static function diaspora($webfinger)
1400         {
1401                 $hcard_url = "";
1402                 $data = [];
1403
1404                 // The array is reversed to take into account the order of preference for same-rel links
1405                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1406                 foreach (array_reverse($webfinger["links"]) as $link) {
1407                         if (($link["rel"] == "http://microformats.org/profile/hcard") && !empty($link["href"])) {
1408                                 $hcard_url = $link["href"];
1409                         } elseif (($link["rel"] == "http://joindiaspora.com/seed_location") && !empty($link["href"])) {
1410                                 $data["baseurl"] = trim($link["href"], '/');
1411                         } elseif (($link["rel"] == "http://joindiaspora.com/guid") && !empty($link["href"])) {
1412                                 $data["guid"] = $link["href"];
1413                         } elseif (($link["rel"] == "http://webfinger.net/rel/profile-page") && (($link["type"] ?? "") == "text/html") && !empty($link["href"])) {
1414                                 $data["url"] = $link["href"];
1415                         } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1416                                 $data["poll"] = $link["href"];
1417                         } elseif (($link["rel"] == ActivityNamespace::POCO) && !empty($link["href"])) {
1418                                 $data["poco"] = $link["href"];
1419                         } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1420                                 $data["notify"] = $link["href"];
1421                         } elseif (($link["rel"] == "diaspora-public-key") && !empty($link["href"])) {
1422                                 $data["pubkey"] = base64_decode($link["href"]);
1423
1424                                 //if (strstr($data["pubkey"], 'RSA ') || ($link["type"] == "RSA"))
1425                                 if (strstr($data["pubkey"], 'RSA ')) {
1426                                         $data["pubkey"] = Crypto::rsaToPem($data["pubkey"]);
1427                                 }
1428                         }
1429                 }
1430
1431                 if (empty($data["url"]) || empty($hcard_url)) {
1432                         return [];
1433                 }
1434
1435                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1436                         foreach ($webfinger["aliases"] as $alias) {
1437                                 if (Strings::normaliseLink($alias) != Strings::normaliseLink($data["url"]) && ! strstr($alias, "@")) {
1438                                         $data["alias"] = $alias;
1439                                 } elseif (substr($alias, 0, 5) == 'acct:') {
1440                                         $data["addr"] = substr($alias, 5);
1441                                 }
1442                         }
1443                 }
1444
1445                 if (!empty($webfinger["subject"]) && (substr($webfinger["subject"], 0, 5) == 'acct:')) {
1446                         $data["addr"] = substr($webfinger["subject"], 5);
1447                 }
1448
1449                 // Fetch further information from the hcard
1450                 $data = self::pollHcard($hcard_url, $data);
1451
1452                 if (!$data) {
1453                         return [];
1454                 }
1455
1456                 if (!empty($data["url"])
1457                         && !empty($data["guid"])
1458                         && !empty($data["baseurl"])
1459                         && !empty($data["pubkey"])
1460                         && !empty($hcard_url)
1461                 ) {
1462                         $data["network"] = Protocol::DIASPORA;
1463                         $data["manually-approve"] = false;
1464
1465                         // The Diaspora handle must always be lowercase
1466                         if (!empty($data["addr"])) {
1467                                 $data["addr"] = strtolower($data["addr"]);
1468                         }
1469
1470                         // We have to overwrite the detected value for "notify" since Hubzilla doesn't send it
1471                         $data["notify"] = $data["baseurl"] . "/receive/users/" . $data["guid"];
1472                         $data["batch"]  = $data["baseurl"] . "/receive/public";
1473                 } else {
1474                         return [];
1475                 }
1476
1477                 return $data;
1478         }
1479
1480         /**
1481          * Check for OStatus contact
1482          *
1483          * @param array $webfinger Webfinger data
1484          * @param bool  $short     Short detection mode
1485          *
1486          * @return array|bool OStatus data or "false" on error or "true" on short mode
1487          * @throws HTTPException\InternalServerErrorException
1488          */
1489         private static function ostatus($webfinger, $short = false)
1490         {
1491                 $data = [];
1492
1493                 if (!empty($webfinger["aliases"]) && is_array($webfinger["aliases"])) {
1494                         foreach ($webfinger["aliases"] as $alias) {
1495                                 if (strstr($alias, "@") && !strstr(Strings::normaliseLink($alias), "http://")) {
1496                                         $data["addr"] = str_replace('acct:', '', $alias);
1497                                 }
1498                         }
1499                 }
1500
1501                 if (!empty($webfinger["subject"]) && strstr($webfinger["subject"], "@")
1502                         && !strstr(Strings::normaliseLink($webfinger["subject"]), "http://")
1503                 ) {
1504                         $data["addr"] = str_replace('acct:', '', $webfinger["subject"]);
1505                 }
1506
1507                 if (!empty($webfinger["links"])) {
1508                         // The array is reversed to take into account the order of preference for same-rel links
1509                         // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1510                         foreach (array_reverse($webfinger["links"]) as $link) {
1511                                 if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1512                                         && (($link["type"] ?? "") == "text/html")
1513                                         && ($link["href"] != "")
1514                                 ) {
1515                                         $data["url"] = $data["alias"] = $link["href"];
1516                                 } elseif (($link["rel"] == "salmon") && !empty($link["href"])) {
1517                                         $data["notify"] = $link["href"];
1518                                 } elseif (($link["rel"] == ActivityNamespace::FEED) && !empty($link["href"])) {
1519                                         $data["poll"] = $link["href"];
1520                                 } elseif (($link["rel"] == "magic-public-key") && !empty($link["href"])) {
1521                                         $pubkey = $link["href"];
1522
1523                                         if (substr($pubkey, 0, 5) === 'data:') {
1524                                                 if (strstr($pubkey, ',')) {
1525                                                         $pubkey = substr($pubkey, strpos($pubkey, ',') + 1);
1526                                                 } else {
1527                                                         $pubkey = substr($pubkey, 5);
1528                                                 }
1529                                         } elseif (Strings::normaliseLink($pubkey) == 'http://') {
1530                                                 $curlResult = DI::httpRequest()->get($pubkey);
1531                                                 if ($curlResult->isTimeout()) {
1532                                                         self::$istimeout = true;
1533                                                         return $short ? false : [];
1534                                                 }
1535                                                 $pubkey = $curlResult->getBody();
1536                                         }
1537
1538                                         $key = explode(".", $pubkey);
1539
1540                                         if (sizeof($key) >= 3) {
1541                                                 $m = Strings::base64UrlDecode($key[1]);
1542                                                 $e = Strings::base64UrlDecode($key[2]);
1543                                                 $data["pubkey"] = Crypto::meToPem($m, $e);
1544                                         }
1545                                 }
1546                         }
1547                 }
1548
1549                 if (isset($data["notify"]) && isset($data["pubkey"])
1550                         && isset($data["poll"])
1551                         && isset($data["url"])
1552                 ) {
1553                         $data["network"] = Protocol::OSTATUS;
1554                         $data["manually-approve"] = false;
1555                 } else {
1556                         return $short ? false : [];
1557                 }
1558
1559                 if ($short) {
1560                         return true;
1561                 }
1562
1563                 // Fetch all additional data from the feed
1564                 $curlResult = DI::httpRequest()->get($data["poll"]);
1565                 if ($curlResult->isTimeout()) {
1566                         self::$istimeout = true;
1567                         return [];
1568                 }
1569                 $feed = $curlResult->getBody();
1570                 $feed_data = Feed::import($feed);
1571                 if (!$feed_data) {
1572                         return [];
1573                 }
1574
1575                 if (!empty($feed_data["header"]["author-name"])) {
1576                         $data["name"] = $feed_data["header"]["author-name"];
1577                 }
1578                 if (!empty($feed_data["header"]["author-nick"])) {
1579                         $data["nick"] = $feed_data["header"]["author-nick"];
1580                 }
1581                 if (!empty($feed_data["header"]["author-avatar"])) {
1582                         $data["photo"] = self::fixAvatar($feed_data["header"]["author-avatar"], $data["url"]);
1583                 }
1584                 if (!empty($feed_data["header"]["author-id"])) {
1585                         $data["alias"] = $feed_data["header"]["author-id"];
1586                 }
1587                 if (!empty($feed_data["header"]["author-location"])) {
1588                         $data["location"] = $feed_data["header"]["author-location"];
1589                 }
1590                 if (!empty($feed_data["header"]["author-about"])) {
1591                         $data["about"] = $feed_data["header"]["author-about"];
1592                 }
1593                 // OStatus has serious issues when the the url doesn't fit (ssl vs. non ssl)
1594                 // So we take the value that we just fetched, although the other one worked as well
1595                 if (!empty($feed_data["header"]["author-link"])) {
1596                         $data["url"] = $feed_data["header"]["author-link"];
1597                 }
1598
1599                 if ($data["url"] == $data["alias"]) {
1600                         $data["alias"] = '';
1601                 }
1602
1603                 /// @todo Fetch location and "about" from the feed as well
1604                 return $data;
1605         }
1606
1607         /**
1608          * Fetch data from a pump.io profile page
1609          *
1610          * @param string $profile_link Link to the profile page
1611          *
1612          * @return array profile data
1613          */
1614         private static function pumpioProfileData($profile_link)
1615         {
1616                 $curlResult = DI::httpRequest()->get($profile_link);
1617                 if (!$curlResult->isSuccess() || empty($curlResult->getBody())) {
1618                         return [];
1619                 }
1620
1621                 $doc = new DOMDocument();
1622                 if (!@$doc->loadHTML($curlResult->getBody())) {
1623                         return [];
1624                 }
1625
1626                 $xpath = new DomXPath($doc);
1627
1628                 $data = [];
1629
1630                 $data["name"] = $xpath->query("//span[contains(@class, 'p-name')]")->item(0)->nodeValue;
1631
1632                 if ($data["name"] == '') {
1633                         // This is ugly - but pump.io doesn't seem to know a better way for it
1634                         $data["name"] = trim($xpath->query("//h1[@class='media-header']")->item(0)->nodeValue);
1635                         $pos = strpos($data["name"], chr(10));
1636                         if ($pos) {
1637                                 $data["name"] = trim(substr($data["name"], 0, $pos));
1638                         }
1639                 }
1640
1641                 $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-locality')]");
1642
1643                 if ($data["location"] == '') {
1644                         $data["location"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'location')]");
1645                 }
1646
1647                 $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'p-note')]");
1648
1649                 if ($data["about"] == '') {
1650                         $data["about"] = XML::getFirstNodeValue($xpath, "//p[contains(@class, 'summary')]");
1651                 }
1652
1653                 $avatar = $xpath->query("//img[contains(@class, 'u-photo')]")->item(0);
1654                 if (!$avatar) {
1655                         $avatar = $xpath->query("//img[@class='img-rounded media-object']")->item(0);
1656                 }
1657                 if ($avatar) {
1658                         foreach ($avatar->attributes as $attribute) {
1659                                 if ($attribute->name == "src") {
1660                                         $data["photo"] = trim($attribute->value);
1661                                 }
1662                         }
1663                 }
1664
1665                 return $data;
1666         }
1667
1668         /**
1669          * Check for pump.io contact
1670          *
1671          * @param array  $webfinger Webfinger data
1672          * @param string $addr
1673          * @return array pump.io data
1674          */
1675         private static function pumpio($webfinger, $addr)
1676         {
1677                 $data = [];
1678                 // The array is reversed to take into account the order of preference for same-rel links
1679                 // See: https://tools.ietf.org/html/rfc7033#section-4.4.4
1680                 foreach (array_reverse($webfinger["links"]) as $link) {
1681                         if (($link["rel"] == "http://webfinger.net/rel/profile-page")
1682                                 && (($link["type"] ?? "") == "text/html")
1683                                 && ($link["href"] != "")
1684                         ) {
1685                                 $data["url"] = $link["href"];
1686                         } elseif (($link["rel"] == "activity-inbox") && ($link["href"] != "")) {
1687                                 $data["notify"] = $link["href"];
1688                         } elseif (($link["rel"] == "activity-outbox") && ($link["href"] != "")) {
1689                                 $data["poll"] = $link["href"];
1690                         } elseif (($link["rel"] == "dialback") && ($link["href"] != "")) {
1691                                 $data["dialback"] = $link["href"];
1692                         }
1693                 }
1694                 if (isset($data["poll"]) && isset($data["notify"])
1695                         && isset($data["dialback"])
1696                         && isset($data["url"])
1697                 ) {
1698                         // by now we use these fields only for the network type detection
1699                         // So we unset all data that isn't used at the moment
1700                         unset($data["dialback"]);
1701
1702                         $data["network"] = Protocol::PUMPIO;
1703                 } else {
1704                         return [];
1705                 }
1706
1707                 $profile_data = self::pumpioProfileData($data["url"]);
1708
1709                 if (!$profile_data) {
1710                         return [];
1711                 }
1712
1713                 $data = array_merge($data, $profile_data);
1714
1715                 if (($addr != '') && ($data['name'] != '')) {
1716                         $name = trim(str_replace($addr, '', $data['name']));
1717                         if ($name != '') {
1718                                 $data['name'] = $name;
1719                         }
1720                 }
1721
1722                 return $data;
1723         }
1724
1725         /**
1726          * Check for twitter contact
1727          *
1728          * @param string $uri
1729          *
1730          * @return array twitter data
1731          */
1732         private static function twitter($uri)
1733         {
1734                 if (preg_match('=([^@]+)@(?:mobile\.)?twitter\.com$=i', $uri, $matches)) {
1735                         $nick = $matches[1];
1736                 } elseif (preg_match('=^https?://(?:mobile\.)?twitter\.com/(.+)=i', $uri, $matches)) {
1737                         $nick = $matches[1];
1738                 } else {
1739                         return [];
1740                 }
1741
1742                 $data = [];
1743                 $data['url'] = 'https://twitter.com/' . $nick;
1744                 $data['addr'] = $nick . '@twitter.com';
1745                 $data['nick'] = $data['name'] = $nick;
1746                 $data['network'] = Protocol::TWITTER;
1747                 $data['baseurl'] = 'https://twitter.com';
1748
1749                 return $data;
1750         }
1751
1752         /**
1753          * Checks HTML page for RSS feed link
1754          *
1755          * @param string $url  Page link
1756          * @param string $body Page body string
1757          * @return string|false Feed link or false if body was invalid HTML document
1758          */
1759         public static function getFeedLink(string $url, string $body)
1760         {
1761                 if (empty($body)) {
1762                         return '';
1763                 }
1764
1765                 $doc = new DOMDocument();
1766                 if (!@$doc->loadHTML($body)) {
1767                         return false;
1768                 }
1769
1770                 $xpath = new DOMXPath($doc);
1771
1772                 $feedUrl = $xpath->evaluate('string(/html/head/link[@type="application/rss+xml" and @rel="alternate"]/@href)');
1773
1774                 $feedUrl = $feedUrl ? self::ensureAbsoluteLinkFromHTMLDoc($feedUrl, $url, $xpath) : '';
1775
1776                 return $feedUrl;
1777         }
1778
1779         /**
1780          * Return an absolute URL in the context of a HTML document retrieved from the provided URL.
1781          *
1782          * Loosely based on RFC 1808
1783          *
1784          * @see https://tools.ietf.org/html/rfc1808
1785          *
1786          * @param string   $href  The potential relative href found in the HTML document
1787          * @param string   $base  The HTML document URL
1788          * @param DOMXPath $xpath The HTML document XPath
1789          * @return string
1790          */
1791         private static function ensureAbsoluteLinkFromHTMLDoc(string $href, string $base, DOMXPath $xpath)
1792         {
1793                 if (filter_var($href, FILTER_VALIDATE_URL)) {
1794                         return $href;
1795                 }
1796
1797                 $base = $xpath->evaluate('string(/html/head/base/@href)') ?: $base;
1798
1799                 $baseParts = parse_url($base);
1800                 if (empty($baseParts['host'])) {
1801                         return $href;
1802                 }
1803
1804                 // Naked domain case (scheme://basehost)
1805                 $path = $baseParts['path'] ?? '/';
1806
1807                 // Remove the filename part of the path if it exists (/base/path/file)
1808                 $path = implode('/', array_slice(explode('/', $path), 0, -1));
1809
1810                 $hrefParts = parse_url($href);
1811
1812                 if (!empty($hrefParts['path'])) {
1813                         // Root path case (/path) including relative scheme case (//host/path)
1814                         if ($hrefParts['path'] && $hrefParts['path'][0] == '/') {
1815                                 $path = $hrefParts['path'];
1816                         } else {
1817                                 $path = $path . '/' . $hrefParts['path'];
1818
1819                                 // Resolve arbitrary relative path
1820                                 // Lifted from https://www.php.net/manual/en/function.realpath.php#84012
1821                                 $parts = array_filter(explode('/', $path), 'strlen');
1822                                 $absolutes = array();
1823                                 foreach ($parts as $part) {
1824                                         if ('.' == $part) continue;
1825                                         if ('..' == $part) {
1826                                                 array_pop($absolutes);
1827                                         } else {
1828                                                 $absolutes[] = $part;
1829                                         }
1830                                 }
1831
1832                                 $path = '/' . implode('/', $absolutes);
1833                         }
1834                 }
1835
1836                 // Relative scheme case (//host/path)
1837                 $baseParts['host'] = $hrefParts['host'] ?? $baseParts['host'];
1838                 $baseParts['path'] = $path;
1839                 unset($baseParts['query']);
1840                 unset($baseParts['fragment']);
1841
1842                 return Network::unparseURL($baseParts);
1843         }
1844
1845         /**
1846          * Check for feed contact
1847          *
1848          * @param string  $url   Profile link
1849          * @param boolean $probe Do a probe if the page contains a feed link
1850          *
1851          * @return array feed data
1852          * @throws HTTPException\InternalServerErrorException
1853          */
1854         private static function feed($url, $probe = true)
1855         {
1856                 $curlResult = DI::httpRequest()->get($url);
1857                 if ($curlResult->isTimeout()) {
1858                         self::$istimeout = true;
1859                         return [];
1860                 }
1861                 $feed = $curlResult->getBody();
1862                 $feed_data = Feed::import($feed);
1863
1864                 if (!$feed_data) {
1865                         if (!$probe) {
1866                                 return [];
1867                         }
1868
1869                         $feed_url = self::getFeedLink($url, $feed);
1870
1871                         if (!$feed_url) {
1872                                 return [];
1873                         }
1874
1875                         return self::feed($feed_url, false);
1876                 }
1877
1878                 if (!empty($feed_data["header"]["author-name"])) {
1879                         $data["name"] = $feed_data["header"]["author-name"];
1880                 }
1881
1882                 if (!empty($feed_data["header"]["author-nick"])) {
1883                         $data["nick"] = $feed_data["header"]["author-nick"];
1884                 }
1885
1886                 if (!empty($feed_data["header"]["author-avatar"])) {
1887                         $data["photo"] = $feed_data["header"]["author-avatar"];
1888                 }
1889
1890                 if (!empty($feed_data["header"]["author-id"])) {
1891                         $data["alias"] = $feed_data["header"]["author-id"];
1892                 }
1893
1894                 $data["url"] = $url;
1895                 $data["poll"] = $url;
1896
1897                 $data["network"] = Protocol::FEED;
1898
1899                 return $data;
1900         }
1901
1902         /**
1903          * Check for mail contact
1904          *
1905          * @param string  $uri Profile link
1906          * @param integer $uid User ID
1907          *
1908          * @return array mail data
1909          * @throws \Exception
1910          */
1911         private static function mail($uri, $uid)
1912         {
1913                 if (!Network::isEmailDomainValid($uri)) {
1914                         return [];
1915                 }
1916
1917                 if ($uid == 0) {
1918                         return [];
1919                 }
1920
1921                 $user = DBA::selectFirst('user', ['prvkey'], ['uid' => $uid]);
1922
1923                 $condition = ["`uid` = ? AND `server` != ''", $uid];
1924                 $fields = ['pass', 'user', 'server', 'port', 'ssltype', 'mailbox'];
1925                 $mailacct = DBA::selectFirst('mailacct', $fields, $condition);
1926
1927                 if (!DBA::isResult($user) || !DBA::isResult($mailacct)) {
1928                         return [];
1929                 }
1930
1931                 $mailbox = Email::constructMailboxName($mailacct);
1932                 $password = '';
1933                 openssl_private_decrypt(hex2bin($mailacct['pass']), $password, $user['prvkey']);
1934                 $mbox = Email::connect($mailbox, $mailacct['user'], $password);
1935                 if (!$mbox) {
1936                         return [];
1937                 }
1938
1939                 $msgs = Email::poll($mbox, $uri);
1940                 Logger::info('Messages found', ['uri' => $uri, 'count' => count($msgs)]);
1941
1942                 if (!count($msgs)) {
1943                         return [];
1944                 }
1945
1946                 $phost = substr($uri, strpos($uri, '@') + 1);
1947
1948                 $data = [];
1949                 $data["addr"]    = $uri;
1950                 $data["network"] = Protocol::MAIL;
1951                 $data["name"]    = substr($uri, 0, strpos($uri, '@'));
1952                 $data["nick"]    = $data["name"];
1953                 $data["photo"]   = Network::lookupAvatarByEmail($uri);
1954                 $data["url"]     = 'mailto:'.$uri;
1955                 $data["notify"]  = 'smtp ' . Strings::getRandomHex();
1956                 $data["poll"]    = 'email ' . Strings::getRandomHex();
1957
1958                 $x = Email::messageMeta($mbox, $msgs[0]);
1959                 if (stristr($x[0]->from, $uri)) {
1960                         $adr = imap_rfc822_parse_adrlist($x[0]->from, '');
1961                 } elseif (stristr($x[0]->to, $uri)) {
1962                         $adr = imap_rfc822_parse_adrlist($x[0]->to, '');
1963                 }
1964                 if (isset($adr)) {
1965                         foreach ($adr as $feadr) {
1966                                 if ((strcasecmp($feadr->mailbox, $data["name"]) == 0)
1967                                         &&(strcasecmp($feadr->host, $phost) == 0)
1968                                         && (strlen($feadr->personal))
1969                                 ) {
1970                                         $personal = imap_mime_header_decode($feadr->personal);
1971                                         $data["name"] = "";
1972                                         foreach ($personal as $perspart) {
1973                                                 if ($perspart->charset != "default") {
1974                                                         $data["name"] .= iconv($perspart->charset, 'UTF-8//IGNORE', $perspart->text);
1975                                                 } else {
1976                                                         $data["name"] .= $perspart->text;
1977                                                 }
1978                                         }
1979
1980                                         $data["name"] = Strings::escapeTags($data["name"]);
1981                                 }
1982                         }
1983                 }
1984                 if (!empty($mbox)) {
1985                         imap_close($mbox);
1986                 }
1987                 return $data;
1988         }
1989
1990         /**
1991          * Mix two paths together to possibly fix missing parts
1992          *
1993          * @param string $avatar Path to the avatar
1994          * @param string $base   Another path that is hopefully complete
1995          *
1996          * @return string fixed avatar path
1997          * @throws \Exception
1998          */
1999         public static function fixAvatar($avatar, $base)
2000         {
2001                 $base_parts = parse_url($base);
2002
2003                 // Remove all parts that could create a problem
2004                 unset($base_parts['path']);
2005                 unset($base_parts['query']);
2006                 unset($base_parts['fragment']);
2007
2008                 $avatar_parts = parse_url($avatar);
2009
2010                 // Now we mix them
2011                 $parts = array_merge($base_parts, $avatar_parts);
2012
2013                 // And put them together again
2014                 $scheme   = isset($parts['scheme'])   ? $parts['scheme'] . '://' : '';
2015                 $host     = isset($parts['host'])     ? $parts['host']           : '';
2016                 $port     = isset($parts['port'])     ? ':' . $parts['port']     : '';
2017                 $path     = isset($parts['path'])     ? $parts['path']           : '';
2018                 $query    = isset($parts['query'])    ? '?' . $parts['query']    : '';
2019                 $fragment = isset($parts['fragment']) ? '#' . $parts['fragment'] : '';
2020
2021                 $fixed = $scheme.$host.$port.$path.$query.$fragment;
2022
2023                 Logger::debug('Avatar fixed', ['base' => $base, 'avatar' => $avatar, 'fixed' => $fixed]);
2024
2025                 return $fixed;
2026         }
2027
2028         /**
2029          * Fetch the last date that the contact had posted something (publically)
2030          *
2031          * @param string $data  probing result
2032          * @return string last activity
2033          */
2034         public static function getLastUpdate(array $data)
2035         {
2036                 $uid = User::getIdForURL($data['url']);
2037                 if (!empty($uid)) {
2038                         $contact = Contact::selectFirst(['url', 'last-item'], ['self' => true, 'uid' => $uid]);
2039                         if (!empty($contact['last-item'])) {
2040                                 return $contact['last-item'];
2041                         }
2042                 }
2043
2044                 if ($lastUpdate = self::updateFromNoScrape($data)) {
2045                         return $lastUpdate;
2046                 }
2047
2048                 if (!empty($data['outbox'])) {
2049                         return self::updateFromOutbox($data['outbox'], $data);
2050                 } elseif (!empty($data['poll']) && ($data['network'] == Protocol::ACTIVITYPUB)) {
2051                         return self::updateFromOutbox($data['poll'], $data);
2052                 } elseif (!empty($data['poll'])) {
2053                         return self::updateFromFeed($data);
2054                 }
2055
2056                 return '';
2057         }
2058
2059         /**
2060          * Fetch the last activity date from the "noscrape" endpoint
2061          *
2062          * @param array $data Probing result
2063          * @return string last activity
2064          *
2065          * @return bool 'true' if update was successful or the server was unreachable
2066          */
2067         private static function updateFromNoScrape(array $data)
2068         {
2069                 if (empty($data['baseurl'])) {
2070                         return '';
2071                 }
2072
2073                 // Check the 'noscrape' endpoint when it is a Friendica server
2074                 $gserver = DBA::selectFirst('gserver', ['noscrape'], ["`nurl` = ? AND `noscrape` != ''",
2075                         Strings::normaliseLink($data['baseurl'])]);
2076                 if (!DBA::isResult($gserver)) {
2077                         return '';
2078                 }
2079
2080                 $curlResult = DI::httpRequest()->get($gserver['noscrape'] . '/' . $data['nick']);
2081
2082                 if ($curlResult->isSuccess() && !empty($curlResult->getBody())) {
2083                         $noscrape = json_decode($curlResult->getBody(), true);
2084                         if (!empty($noscrape) && !empty($noscrape['updated'])) {
2085                                 return DateTimeFormat::utc($noscrape['updated'], DateTimeFormat::MYSQL);
2086                         }
2087                 }
2088
2089                 return '';
2090         }
2091
2092         /**
2093          * Fetch the last activity date from an ActivityPub Outbox
2094          *
2095          * @param string $feed
2096          * @param array  $data Probing result
2097          * @return string last activity
2098          * @throws \Friendica\Network\HTTPException\InternalServerErrorException
2099          */
2100         private static function updateFromOutbox(string $feed, array $data)
2101         {
2102                 $outbox = ActivityPub::fetchContent($feed);
2103                 if (empty($outbox)) {
2104                         return '';
2105                 }
2106
2107                 if (!empty($outbox['orderedItems'])) {
2108                         $items = $outbox['orderedItems'];
2109                 } elseif (!empty($outbox['first']['orderedItems'])) {
2110                         $items = $outbox['first']['orderedItems'];
2111                 } elseif (!empty($outbox['first']['href']) && ($outbox['first']['href'] != $feed)) {
2112                         return self::updateFromOutbox($outbox['first']['href'], $data);
2113                 } elseif (!empty($outbox['first'])) {
2114                         if (is_string($outbox['first']) && ($outbox['first'] != $feed)) {
2115                                 return self::updateFromOutbox($outbox['first'], $data);
2116                         } else {
2117                                 Logger::warning('Unexpected data', ['outbox' => $outbox]);
2118                         }
2119                         return '';
2120                 } else {
2121                         $items = [];
2122                 }
2123
2124                 $last_updated = '';
2125                 foreach ($items as $activity) {
2126                         if (!empty($activity['published'])) {
2127                                 $published =  DateTimeFormat::utc($activity['published']);
2128                         } elseif (!empty($activity['object']['published'])) {
2129                                 $published =  DateTimeFormat::utc($activity['object']['published']);
2130                         } else {
2131                                 continue;
2132                         }
2133
2134                         if ($last_updated < $published) {
2135                                 $last_updated = $published;
2136                         }
2137                 }
2138
2139                 if (!empty($last_updated)) {
2140                         return $last_updated;
2141                 }
2142
2143                 return '';
2144         }
2145
2146         /**
2147          * Fetch the last activity date from an XML feed
2148          *
2149          * @param array $data Probing result
2150          * @return string last activity
2151          */
2152         private static function updateFromFeed(array $data)
2153         {
2154                 // Search for the newest entry in the feed
2155                 $curlResult = DI::httpRequest()->get($data['poll']);
2156                 if (!$curlResult->isSuccess()) {
2157                         return '';
2158                 }
2159
2160                 $doc = new DOMDocument();
2161                 @$doc->loadXML($curlResult->getBody());
2162
2163                 $xpath = new DOMXPath($doc);
2164                 $xpath->registerNamespace('atom', 'http://www.w3.org/2005/Atom');
2165
2166                 $entries = $xpath->query('/atom:feed/atom:entry');
2167
2168                 $last_updated = '';
2169
2170                 foreach ($entries as $entry) {
2171                         $published_item = $xpath->query('atom:published/text()', $entry)->item(0);
2172                         $updated_item   = $xpath->query('atom:updated/text()'  , $entry)->item(0);
2173                         $published      = !empty($published_item->nodeValue) ? DateTimeFormat::utc($published_item->nodeValue) : null;
2174                         $updated        = !empty($updated_item->nodeValue) ? DateTimeFormat::utc($updated_item->nodeValue) : null;
2175
2176                         if (empty($published) || empty($updated)) {
2177                                 Logger::notice('Invalid entry for XPath.', ['entry' => $entry, 'url' => $data['url']]);
2178                                 continue;
2179                         }
2180
2181                         if ($last_updated < $published) {
2182                                 $last_updated = $published;
2183                         }
2184
2185                         if ($last_updated < $updated) {
2186                                 $last_updated = $updated;
2187                         }
2188                 }
2189
2190                 if (!empty($last_updated)) {
2191                         return $last_updated;
2192                 }
2193
2194                 return '';
2195         }
2196
2197         /**
2198          * Probe data from local profiles without network traffic
2199          *
2200          * @param string $url
2201          * @return array probed data
2202          * @throws HTTPException\InternalServerErrorException
2203          * @throws HTTPException\NotFoundException
2204          */
2205         private static function localProbe(string $url): array
2206         {
2207                 try {
2208                         $uid = User::getIdForURL($url);
2209                         if (!$uid) {
2210                                 throw new HTTPException\NotFoundException('User not found.');
2211                         }
2212
2213                         $profile   = User::getOwnerDataById($uid);
2214                         $approfile = ActivityPub\Transmitter::getProfile($uid);
2215
2216                         if (empty($profile['gsid'])) {
2217                                 $profile['gsid'] = GServer::getID($approfile['generator']['url']);
2218                         }
2219
2220                         $data = [
2221                                 'name' => $profile['name'], 'nick' => $profile['nick'], 'guid' => $approfile['diaspora:guid'] ?? '',
2222                                 'url' => $profile['url'], 'addr' => $profile['addr'], 'alias' => $profile['alias'],
2223                                 'photo' => Contact::getAvatarUrlForId($profile['id'], $profile['updated']),
2224                                 'header' => $profile['header'] ? Contact::getHeaderUrlForId($profile['id'], $profile['updated']) : '',
2225                                 'account-type' => $profile['contact-type'], 'community' => ($profile['contact-type'] == User::ACCOUNT_TYPE_COMMUNITY),
2226                                 'keywords' => $profile['keywords'], 'location' => $profile['location'], 'about' => $profile['about'],
2227                                 'hide' => !$profile['net-publish'], 'batch' => '', 'notify' => $profile['notify'],
2228                                 'poll' => $profile['poll'], 'request' => $profile['request'], 'confirm' => $profile['confirm'],
2229                                 'subscribe' => $approfile['generator']['url'] . '/follow?url={uri}', 'poco' => $profile['poco'],
2230                                 'following' => $approfile['following'], 'followers' => $approfile['followers'],
2231                                 'inbox' => $approfile['inbox'], 'outbox' => $approfile['outbox'],
2232                                 'sharedinbox' => $approfile['endpoints']['sharedInbox'], 'network' => Protocol::DFRN,
2233                                 'pubkey' => $profile['upubkey'], 'baseurl' => $approfile['generator']['url'], 'gsid' => $profile['gsid'],
2234                                 'manually-approve' => in_array($profile['page-flags'], [User::PAGE_FLAGS_NORMAL, User::PAGE_FLAGS_PRVGROUP])
2235                         ];
2236                 } catch (Exception $e) {
2237                         // Default values for non existing targets
2238                         $data = [
2239                                 'name' => $url, 'nick' => $url, 'url' => $url, 'network' => Protocol::PHANTOM,
2240                                 'photo' => DI::baseUrl() . Contact::DEFAULT_AVATAR_PHOTO
2241                         ];
2242                 }
2243
2244                 return self::rearrangeData($data);
2245         }
2246 }